Human cerebral arterial morphogenesis explored by 3D microscopic imaging of cleared embryos: Bridging the gap between embryonic and adult anatomy Authors: Eimad Shotar, Raphaël Blain, Aymeric Nadjem, Nermine Saidi, Megumi Inoue, Daniel Smith‐Paredes, Miccaella Vergara, Armelle Lesaunier, Quentin Rappeneau, Joséphine Blevinal, Gérard Couly, Julien Burel, Julien Allard, Kevin Premat, Stéphanie Lenck, Pierre‐Marie Chiaroni, Erwah Kalsoum, Maryne Toupin, Séverine Mazaud‐Guittot, Diane Rebourcet, Paolo Giacobini, Olivier Naggara, Hugo Gortais, Grégoire Boulouis, Alexis Guédon, Philippe Gailloud, Frédéric Glicenstein, Anaïs Favre, Frédéric Clarençon, Yorick Gitton, Alain Chédotal Published: 2026 ABSTRACT AbstractThe anatomy of cerebral arteries impacts cerebrovascular diseases, their management, and their prognosis. Postnatal cerebrovascular anatomy can be understood only through detailed knowledge of the developmental events leading to mature cerebral vessels. Current knowledge about human embryology relies on data collected in the first half of the 20th century using traditional histological methods. In this study, we provide a narrative description of human embryonic cerebral arterial development based on whole‐mount immunostaining, tissue clearing, three‐dimensional (3D) light sheet microscopy imaging, and anatomically informed virtual segmentation of human embryos. The findings are compared with adult radiological anatomy explored by magnetic resonance imaging and 3D digital subtraction angiography. We provide a detailed account of the development of cerebral arteries including the circle of Willis and its major branches, the vertebrobasilar circulation, the ophthalmic artery, and related vessels. Explanations for several anatomical variants are provided. Pediatric and adult anatomy in the context of brain arteriovenous and vein of Galen malformations is compared with embryological data. Finally, we provide a direct comparison of embryonic and adult anatomical nomenclatures and clarify the identity of homonymous arterial segments. These data shed new light on both embryonic development of cerebral arteries and postnatal anatomy. INTRODUCTION Human evolution has been marked by a major increase in our ancestors' cognitive and behavioral abilities, underlined by brain morphological expansion (González‐Forero & Gardner, 2018; Neubauer et al., 2018; Püschel et al., 2024). As a highly metabolic organ, the brain needs an abundant supply of oxygen and, at rest, most of its ATP comes from oxidative metabolism of glucose (Mintun et al., 2001). As such, maintenance of consciousness and proper neurological functions is highly dependent on constant arterial blood flow. This blood supply is provided by cerebral arteries, whose dysfunction in the form of stroke is a major cause of mortality and neurological impairment worldwide (Feigin et al., 2024). Human cerebral arterial anatomy is constantly revisited in light of emerging surgical and endovascular procedures such as mechanical thrombectomy for the treatment of large vessel occlusion stroke (Berkhemer et al., 2015; Shapiro et al., 2020). The circle of Willis, for instance, having been recognized more than three centuries ago as an anastomotic ring at the base of the brain (Molnár, 2004, 2021), is still the object of active anatomical investigation (Ayre et al., 2022; Hendrikse et al., 2010; Krabbe‐Hartkamp et al., 1998; Lazorthes et al., 1979; Riggs & Rupp, 1963). Accurate characterization of variation in the circle of Willis underlies improved understanding of many cerebrovascular diseases, including development of intracranial aneurysms, subarachnoid hemorrhage or outcome following therapeutic or pathological carotid occlusion (Jacquens et al., 2020; Kaspera et al., 2014; Schomer et al., 1994; Sriwastwa et al., 2025). Improved descriptions of cerebrovascular development may provide further insights into poorly understood diseases like brain arteriovenous malformations (AVMs; a shortlist of anatomically and embryologically relevant abbreviations used in the manuscript is provided in Table 1; a full list of abbreviations used throughout the manuscript and figures is provided as Supplemental Material). In parallel, anatomical knowledge is also believed to improve therapeutic decision‐making (Picard, 2009). Understanding adult anatomical variability is rooted in improved comprehension of embryology, both via descriptive and experimental means (Li et al., 2015; Padget, 1948). This interplay between embryology, adult anatomy, and pathology, extensively reported for the circle of Willis, is expected to apply along the entire spectrum of cerebral arteries. TABLE 1Shortlist of abbreviations used in the manuscript.ShortformExpansionAnatomical abbreviationsA1Pre‐communicating segment of the anterior cerebral arteryACAAnterior cerebral arteryAChoAAnterior choroidal arteryAComAAnterior communicating arteryAICAAnterior inferior cerebellar arteryASAAnterior spinal arteryBABasilar arteryCCACommon carotid arteryCNSCentral nervous systemCoACollicular arteryDiADiencephalic arteryECAExternal carotid arteryHAHyaloid arteryICAInternal carotid arteryLPCALateral posterior choroidal arteryMCAMiddle cerebral arteryMMAMiddle meningeal arteryMMDMaxillo‐mandibular divisionMPCAMedial posterior choroidal arteryMSAMedial striate artery (recurrent artery of Heubner)OAOphthalmic arteryP1Pre‐communicating segment of the posterior cerebral arteryP2Proximal post‐communicating segment of the posterior cerebral arteryPAAPharyngeal arch arterypACAprimitive Anterior cerebral arterypAChoAprimitive Anterior choroidal arteryPCAPosterior cerebral arteryPComAPosterior communicating arteryPDOAPrimitive dorsal ophthalmic arteryPICAPosterior inferior cerebellar arteryPLBVAPrimitive lateral basilo‐vertebral anastomosisPOAPrimitive olfactory arterypPChoAprimitive Posterior choroidal arteryPVOAPrimitive ventral ophthalmic arterySAStapedial arterySCASuperior cerebellar arteryVAVertebral arteryVPAVentral pharyngeal arteryOther abbreviations3DThree dimensionalAVMArteriovenous malformationCD31Cluster of differentiation 31Col2Collagen 2DBEDibenzyl etherDCMDichloromethaneDICOMDigital imaging and communications in medicineDSADigital subtraction angiographyEDTAEthylenediaminetetraacetic acidEpCAMEpithelial cell adhesion moleculeMHCMyosin heavy chainMIPMaximum intensity projectionMRIMagnetic resonance imagingPACSPicture archiving and communication systemPCWPost conception weekPhox2BPaired‐like Homeobox 2BSMASmooth muscle actinSox9Sex‐determining region of chromosome Y (SRY)‐box containing gene 9VGAMVein of Galen aneurysmal malformation Note: A full list of abbreviations used throughout the figures is provided as supplemental material. Each figure legend provides the list of abbreviations used in the figure along with their expansions. Shortlist of abbreviations used in the manuscript. Note: A full list of abbreviations used throughout the figures is provided as supplemental material. Each figure legend provides the list of abbreviations used in the figure along with their expansions. Our understanding of the embryological origin of human cerebral arteries made a great leap with the descriptions of Hager Padget (Kretzer et al., 2004; Padget, 1948). However, since Padget's seminal 1948 publication, cerebral arterial embryology reports have been either limited in scope (Macchi et al., 2005; Takakuwa et al., 2016) or have consisted mainly of commentaries on earlier historical descriptions (Bertulli & Robert, 2021; Kathuria et al., 2011; Klostranec & Krings, 2022; Menshawi et al., 2015; Okahara et al., 2002; Ota, 2022; Raybaud, 2010). Most importantly, the bulk of recent work continues to rely on three‐dimensional (3D) reconstruction of serial histological sections for the description of cerebral arterial development (Macchi et al., 2005; Takakuwa et al., 2016). Such reconstructions provide limited resolution and are prone to errors when reconstructing vascular trees. Only recently have alternative strategies been explored, such as diffusible iodine‐based contrast‐enhanced computed tomography embryonic imaging (Jacobs et al., 2024, 2025). Much of what is known about human development is derived from early to mid‐20th‐century studies that employed conventional histological techniques (Padget, 1948; Streeter, 1905). In the past decade, tissue‐clearing methods, immunostaining and whole‐mount light sheet fluorescence microscopy (LSFM in figures) have permitted non‐destructive, tissue‐specific, high‐fidelity analysis of the cellular organization of intact human organs (Belle et al., 2014; Mai et al., 2022; Renier et al., 2014; Ueda et al., 2020). These methods provide a unique opportunity to revisit and improve our understanding of human embryology (Belle et al., 2017; Blain et al., 2023; Casoni et al., 2016). Virtual reality‐based computerized segmentation of 3D datasets, akin to historic cadaveric dissections and anatomical preparations (Narang et al., 2021), has proven to be particularly fruitful in depicting embryonic anatomy (Blain et al., 2023). Here, we applied these strategies to describe the development of human cerebral arteries during the first trimester of gestation and to bridge the gap with adult anatomy. METHODS Collection of human embryonic and fetal samples and ethical considerations As previously reported (Belle et al., 2017; Blain et al., 2023), the human samples included in this study were legally obtained from voluntary abortions. Written informed consent was obtained from all donors prior to tissue collection. The samples analyzed were provided by the Inserm HuDeCa Biobank (https://hudeca.com/, https://transparent‐human‐embryo.com). The French agency for biomedical research (Agence de la Biomédecine, Saint‐Denis La Plaine, France; N_PFS19‐012) and the Inserm Ethics Committee (IRB00003888) both granted authorizations to use human tissues. Samples were used in compliance with French regulations. Macroscopic morphological criteria were used to select samples and exclude those with damaged heads and obvious malformations. Sample staging The Chronological ages of the embryos and fetuses in the collection were provided by the obstetrical departments where the samples were collected. Restaging was considered whenever discrepancies were suspected between the declared chronological age, sample size, surface anatomy classified using Carnegie staging (O'Rahilly & Müller, 2010) and arterial developmental maturity. Embryonic restaging was based on concordance among observations of arterial maturity, surface anatomy, and internal development, including lung expansion and bronchial segmentation (Fujii et al., 2020). Fetuses were staged based on morphometric correction as previously described (Evtouchenko et al., 1996). Embryonic and fetal sample treatment workflow All steps of sample manipulation have been previously reported (Blain et al., 2023; Renier et al., 2016). For a detailed account of all the methods used, please refer to the open‐access manuscript of the 3D atlas of the developing human head (Blain et al., 2023). Briefly, this included: Sample fixation: 4% Paraformaldehyde in phosphate buffer (0.12 M, pH 7.4), or 10% formalin, at 4°C for 1–5 days depending on their size.Decalcification for samples older than postconception week 9 (PCW9): 1 week incubation in ethylenediaminetetraacetic acid (EDTA) 0.5 M under agitation at room temperature.Bleaching: samples were dehydrated in methanol and then placed overnight under white light (11 W 3000 K) and rolling agitation (IKA, 004011000) with a 6% hydrogen peroxide solution in 100% methanol.Whole‐mount immunolabeling: samples were incubated at 37°C with agitation at 20 rotations per minute for 7 days (≤PCW8) to 14 days (≥PCW10) with primary antibodies (the list of antibody targets is provided in Section 2.7 “Embryonic samples used in subsequent figures”). Next, secondary antibodies were diluted, and samples were incubated at 37°C in the secondary antibody solution for 3 (≤PCW8) to 7 days (≥PCW10). When conjugated antibodies were used in addition to non‐conjugated ones, they were added at the same time as the secondary antibodies, if the host species were different from the primary antibodies. Otherwise, conjugated antibodies were added after the round of secondary antibodies and incubated for 7 days.Agarose embedding: tissue samples were embedded prior to clearing in 1.5% agarose to facilitate manipulation.Clearing was performed in accordance with the previously published iDISCO+ protocol (Blain et al., 2023; Renier et al., 2016). Samples were dehydrated in methanol and subsequently incubated overnight in a solution of 67% dichloromethane (DCM) and 33% methanol followed by 100% DCM for 30 min at room temperature on a rotator. Finally, samples were cleared, imaged and stored in 100% dibenzyl ether (DBE).3D sample imaging was performed using a Blaze light sheet microscope (Miltenyi Biotec, Bergisch Gladbach, Germany). A 4‐μm‐thick light sheet was generated by lasers at four different wavelengths (488 nm, 561 nm, 639 nm, 785 nm). A combination of three objectives (1X, 4X or 12X) and three magnification lenses (0.6X, 1X, and 1.66X) was used for image acquisition at different resolutions. Sample fixation: 4% Paraformaldehyde in phosphate buffer (0.12 M, pH 7.4), or 10% formalin, at 4°C for 1–5 days depending on their size. Sample fixation: 4% Paraformaldehyde in phosphate buffer (0.12 M, pH 7.4), or 10% formalin, at 4°C for 1–5 days depending on their size. Decalcification for samples older than postconception week 9 (PCW9): 1 week incubation in ethylenediaminetetraacetic acid (EDTA) 0.5 M under agitation at room temperature. Decalcification for samples older than postconception week 9 (PCW9): 1 week incubation in ethylenediaminetetraacetic acid (EDTA) 0.5 M under agitation at room temperature. Bleaching: samples were dehydrated in methanol and then placed overnight under white light (11 W 3000 K) and rolling agitation (IKA, 004011000) with a 6% hydrogen peroxide solution in 100% methanol. Bleaching: samples were dehydrated in methanol and then placed overnight under white light (11 W 3000 K) and rolling agitation (IKA, 004011000) with a 6% hydrogen peroxide solution in 100% methanol. Whole‐mount immunolabeling: samples were incubated at 37°C with agitation at 20 rotations per minute for 7 days (≤PCW8) to 14 days (≥PCW10) with primary antibodies (the list of antibody targets is provided in Section 2.7 “Embryonic samples used in subsequent figures”). Next, secondary antibodies were diluted, and samples were incubated at 37°C in the secondary antibody solution for 3 (≤PCW8) to 7 days (≥PCW10). When conjugated antibodies were used in addition to non‐conjugated ones, they were added at the same time as the secondary antibodies, if the host species were different from the primary antibodies. Otherwise, conjugated antibodies were added after the round of secondary antibodies and incubated for 7 days. Whole‐mount immunolabeling: samples were incubated at 37°C with agitation at 20 rotations per minute for 7 days (≤PCW8) to 14 days (≥PCW10) with primary antibodies (the list of antibody targets is provided in Section 2.7 “Embryonic samples used in subsequent figures”). Next, secondary antibodies were diluted, and samples were incubated at 37°C in the secondary antibody solution for 3 (≤PCW8) to 7 days (≥PCW10). When conjugated antibodies were used in addition to non‐conjugated ones, they were added at the same time as the secondary antibodies, if the host species were different from the primary antibodies. Otherwise, conjugated antibodies were added after the round of secondary antibodies and incubated for 7 days. Agarose embedding: tissue samples were embedded prior to clearing in 1.5% agarose to facilitate manipulation. Agarose embedding: tissue samples were embedded prior to clearing in 1.5% agarose to facilitate manipulation. Clearing was performed in accordance with the previously published iDISCO+ protocol (Blain et al., 2023; Renier et al., 2016). Samples were dehydrated in methanol and subsequently incubated overnight in a solution of 67% dichloromethane (DCM) and 33% methanol followed by 100% DCM for 30 min at room temperature on a rotator. Finally, samples were cleared, imaged and stored in 100% dibenzyl ether (DBE). Clearing was performed in accordance with the previously published iDISCO+ protocol (Blain et al., 2023; Renier et al., 2016). Samples were dehydrated in methanol and subsequently incubated overnight in a solution of 67% dichloromethane (DCM) and 33% methanol followed by 100% DCM for 30 min at room temperature on a rotator. Finally, samples were cleared, imaged and stored in 100% dibenzyl ether (DBE). 3D sample imaging was performed using a Blaze light sheet microscope (Miltenyi Biotec, Bergisch Gladbach, Germany). A 4‐μm‐thick light sheet was generated by lasers at four different wavelengths (488 nm, 561 nm, 639 nm, 785 nm). A combination of three objectives (1X, 4X or 12X) and three magnification lenses (0.6X, 1X, and 1.66X) was used for image acquisition at different resolutions. 3D sample imaging was performed using a Blaze light sheet microscope (Miltenyi Biotec, Bergisch Gladbach, Germany). A 4‐μm‐thick light sheet was generated by lasers at four different wavelengths (488 nm, 561 nm, 639 nm, 785 nm). A combination of three objectives (1X, 4X or 12X) and three magnification lenses (0.6X, 1X, and 1.66X) was used for image acquisition at different resolutions. Adult and pediatric patients and ethical considerations The anatomy and brain AVM databases of the Pitié‐Salpêtrière Interventional Neuroradiology Department (Sorbonne Université, Paris, France) were screened for illustrative cases. The institution is a tertiary care teaching hospital providing care for neurocritically ill patients. The Ethical Committee for Research in Medical Imaging approved this study (CRM‐2506‐480). The requirement for patient informed consent was waived given the retrospective design of the study. Pediatric cases were provided by the Radiology department of the Necker pediatric university hospital (Université de Paris, France). The institution is a tertiary care teaching hospital providing care for neurocritically ill pediatric patients. Given the non‐interventional retrospective nature of the study, based on routinely collected data, the requirement for written informed consent was waived. The study is registered in the National Health Data Institute repository (MR04 #3618210420). A declaration of compliance (CNIL #2217698) was submitted to the French National Commission for Information Technology and Civil Liberties. Patients were informed of their right to object to the use of their health‐related data for research purposes. This study adhered to the World Medical Association Declaration of Helsinki. Digital subtraction angiography protocols In adult patients, digital subtraction angiography (DSA) was performed under local anesthesia, conscious sedation, or general anesthesia, depending on clinical setting. DSA was performed in one of two identical biplane angiography suites (Axiom Artis, Siemens, Erlangen, Germany). For diagnostic DSA, a catheter was advanced via a femoral or radial 4F or 5F sheath (5F–8F for therapeutic procedures) to an artery of interest at the cervical level (the internal carotid or vertebral artery [VA]) under fluoroscopic control. For superselective DSA, a microcatheter was navigated over a 0.007‐ to 0.014‐inch microwire under roadmap guidance into an intracranial artery. For DSA, intra‐arterial iohexol (300 mg I/mL) was injected into the parent vessel 2 s after image acquisition (2 images/s) at a rate of 4 mL/s (total volume of 8 mL). Injections were usually performed manually for superselective DSA. 3D DSA acquisitions were performed with a field of view of 22, 32 or 42 cm. Duration of rotation varied between 5, 6, and 10 s. Contrast media injection was initiated 2 s before acquisition at a rate of 3 mL/s and continued for the entire duration of the acquisition. 3D DSA generated three datasets: the baseline non‐contrasted dataset (rotation 1, noninjected), a contrast‐enhanced dataset (rotation 2, injected) and a subtracted dataset (arteries only). For a description of DSA institutional standard of care in the pediatric patients, please refer to (Hak et al., 2022). Image post processing Embryonic light sheet fluorescence microscopy images (OME.TIFF Z‐stacks) were initially stitched using either MACS iQ (Miltenyi Biotec) or the ImarisStitcher software (Oxford Instruments, Abingdon‐on‐Thames, Oxfordshire, United Kingdom). The stitched TIFF stack mosaics were converted to .ims format using the ImarisFileConverter software (Oxford Instruments) to be analyzed in the Imaris software v10.2.0 (Oxford Instruments). Embryonic light sheet fluorescence microscopy images in .ims format were subsequently imported into the syGlass software (IstoVisio, Morgantown, West Virginia, United States of America) for virtual reality‐based manipulation (see below). Pediatric and adult computed tomography (CT in figures; refer to Supplemental Materials for a full list of abbreviations), magnetic resonance imaging (MRI), and DSA datasets were exported from the hospital picture archiving and communication system (PACS, Carestream Health, Rochester, New York, USA) in digital imaging and communications in medicine (DICOM) format. DICOM data were subsequently imported into syGlass. Segmentations on prenatal (light sheet fluorescence microscopy images) and postnatal pediatric and adult datasets (3D DSA images) were both performed in the syGlass virtual reality environment (software v2.2.1+) using an Oculus Quest2 or 3 (Meta, Menlo Park, California, USA) headset. Briefly, a region of interest (ROI) was created, small enough to allow high‐resolution display. A unique channel was used to segment each individual anatomical structure, which can be exported by syGlass in a binarized TIFF stack (mask). Each individual mask, representing a binarized segmentation channel volume (individual structure), is identified in the exported TIFF images with a single grayscale level (all masks are exported in a single TIFF stack). To extract the raw signal of each anatomical structure (isolated in syGlass and represented by a binary mask in the export TIFF), custom Python code was used. Each segmented structure signal was extracted by the Python code in a new individual TIFF stack and imported using ImarisFileConverter (Oxford Instruments) as an independent channel in a new Imaris file. As they share the same dimensionality as the original datasets, all channels (each representing the signal of an individually segmented structure) could be imported simultaneously in a single Imaris file. This allowed a separate label and pseudo‐color to be assigned to each channel (anatomical structure) as well as an individually optimized signal threshold to be applied to each channel individually. Anatomical structures identifiable only on background noise (or the brain from MRI datasets) were further segmented directly in Imaris (Oxford Instruments) by manual contouring using the Surface tool to first create a surface Mesh. Signal within the Mesh boundaries was subsequently extracted using the Imaris Mask All tool to create a new channel. For adult samples, non‐subtracted 3D datasets were used, allowing 3D DSA–3D MRI fusion based on bony landmarks in Imaris (Oxford Instruments). Figure 1 summarizes the segmentation steps for both embryonic samples (Figure 1a–d) and medical imaging data (Figure 1e–g). Video S1 showcases a segmented adult dataset (related to Figure 1). Image post processing: Anatomically informed segmentation and annotation. Panels (a–d) are light sheet fluorescence microscopy (LSFM) images of a solvent‐cleared immunostained postconception week (PCW) 5.7 human embryo. Panels (e–g) are medical imaging data of an adult patient. All images are pseudo‐colored segmented or unsegmented signal except for the images noted Mesh is panels (b and c). All images are maximum intensity projections with the exception of the insets in panels (a and b), images noted Mesh and panel (g). Panel (a) images are left lateral views of a PCW5.7 embryo immunostained with anti‐SMA and anti‐EpCAM antibodies. The inset is a coronal oblique slice reconstruction showing the rhombencephalon, diencephalon, and trigeminal (V) nerves (nonspecific background noise in the 561 nm channel) in relationship with the internal carotid (ICA) and basilar (BA) arteries (SMA staining) and Rathke's pouch (EpCAM staining). Panel (b) illustrates segmentation by manual contouring. The inset shows central nervous system (CNS) structures identified in background noise and manually contoured (yellow lines). The subsequent images show CNS contouring, 3D surface Mesh reconstruction, extraction of raw signal within the 3D Mesh and pseudo‐coloring after further sub‐segmentation of the CNS. Panel (c) illustrates the process of anatomically informed segmentation and annotation in the syGlass virtual reality (VR) environment. The first inset on the left shows the ICAs, BA, and vertebral arteries (VAs) in raw signal (SMA staining). The second inset shows BA segmentation. Subsequent images are the segmented pseudo‐colored BA raw signal and BA surface Mesh reconstruction from the raw signal segmentation. Also shown is the entire segmented arterial vasculature with and without annotation by pseudo‐coloring. Panel (d) is a lateral view of the segmented CNS and arterial vasculature. As shown in panels (e–g), VR‐based and contouring segmentation can also be applied to adult anatomical datasets obtained by 3D digital subtraction angiography (DSA) and magnetic resonance imaging (MRI). Panel (f) is a lateral view of the segmented brain and arterial vasculature. Panel (g) is an axial reconstruction of the segmented brain and cerebral arteries. ACA, anterior cerebral artery; BA, basilar artery; CNS, central nervous system; Di, diencephalon; DSA, digital subtraction angiography; EpCAM, epithelial cell adhesion molecule; FLAIR, fluid‐attenuated inversion recovery; ICA, internal carotid artery; MCA, middle cerebral artery; Mes., mesencephalon; MRI, magnetic resonance imaging; PCA, posterior cerebral artery; PCW, postconception week; Rhomb., rhombencephalon; RP, Rathke's pouch; Seg., segmentation; SMA, smooth muscle actin; Telenc., telencephalon; V, trigeminal nerve; VA, vertebral artery; VR, virtual reality. Images for the figures were taken using the Snapshot tool in Imaris 10.2.0 (Oxford Instruments) in maximum intensity projection (MIP) or blend mode, the latter allowing customized opacity. Blend mode is unavailable in Imaris (Oxford Instruments) when manually fusing two independent datasets (3D DSA and MRI). Setting MIP signal intensity and thresholding parameters independently for each channel (single anatomical structure) allowed homogeneous visibility of all structures of interest in the final displayed anatomical scenes. In contrast, local signal heterogeneity in the raw datasets often hindered simultaneous visualization of the structures of interest. More specifically, if two structures have been marked with low (structure A) and high intensity (structure B), respectively, the former may only be visible in thresholding conditions of the raw dataset in which signal saturation hinders visibility of the latter. Conversely, modifying thresholding to optimize visibility of structure B (high signal intensity) may limit visibility of structure A (low signal intensity). Imaris downsamples image resolution when rendering multiple channels simultaneously in MIP mode, for any given scene. Therefore, each channel was displayed in isolation at highest resolution and a snapshot taken. Multiple channels representing a collection of anatomical structures for a given anatomical scene were subsequently collapsed into a single image using the superposition mode in Photoshop 2025 (San José, California, USA). In a minority of cases, pseudo‐color was added directly in Photoshop 2025 (San José, California, USA). MIP is advantageous as it allows visualizing structures based on microscopy signal rather than an artificial surface rendering. It also allows depicting several overlapping structures without objects in the foreground masking background objects. A drawback is that depth perception is lacking in 2D projections, a problem not encountered when manipulating the datasets in a 3D virtual reality environment. Therefore, in a limited number of occasions, when distinguishing background from foreground structures was deemed important for image comprehension in the figures, for visualization only, the background structure signal was erased manually at the crossing point with foreground objects to facilitate image comprehension. Embryonic samples used in subsequent figures The laboratory embryo/fetal collection (Belle et al., 2017; Blain et al., 2023) (www.hudeca.com, https://transparent‐human‐embryo.com) was screened for samples appropriate for segmentation, figure preparation, and narrative description. As part of the 3D atlas of the developing human head project conducted by one of the study authors (R.B.), an ontogenic series of human samples with arterial/vascular immunostaining was already available for retrospective analysis (Blain et al., 2023). A total of 41 samples with arterial/vascular specific immunostaining were screened for the current work (PCW5 to PCW13; 30 head or whole embryo/fetus and 11 isolated eyes). Samples were immunostained with either (1) an antibody against smooth muscle actin (SMA) and/or (2) antibodies against plasmalemma vesicle‐associated protein (PLVAP) or cluster of differentiation 31 (CD31). SMA immunostaining discriminates arterial vessel walls with high specificity (Belle et al., 2017; Blain et al., 2023). PLVAP and CD31 are both endothelial cell markers. Samples were immunostained with various other antibodies including different combinations of myosin heavy chain (MHC; muscles), Transient axonal glycoprotein 1 (Tag1; sensory nerves, spinal motor neurons, myelinated nerve fibers), synaptophysin (all nerves), epithelial cell adhesion molecule (EpCAM; epithelia), paired‐like Homeobox 2B (Phox2b; autonomic nervous system), and markers of skeletal cells, sex‐determining region of chromosome Y (SRY)‐box containing gene 9 (Sox9; skeleton/chondrocytes), and Collagen 2 (Col2). Samples with preserved cephalic structures, high‐quality arterial immunostaining and imaging were selected for segmentation and anatomical labeling and served as a basis for the narrative description provided in this manuscript. The selection of samples was made by the study lead author (E.S.) to maximize temporal coverage of developmental stages with samples with sufficiently high‐quality data for unambiguous anatomical segmentation and labeling. For illustrative purposes, nine embryonic head/whole samples were used in the different figures provided in this manuscript along with two isolated eye samples. Figure 2a shows the pseudo‐colored raw signal of the nine embryonic head/whole samples. Figure 2b–d illustrate anatomically informed segmentation of arteries along with various reference structures of interest. Table S1 provides specimen identifiers for the corresponding nine embryonic samples and two isolated eye samples as indexed in accordance with the OpenSpecimen database. Table S2 provides the list of antibodies used in the study. Collection of human embryonic samples used for the illustrations in subsequent figures. All panels are light sheet fluorescence microscopy (LSFM) images of solvent‐cleared immunostained postconception week (PCW) 5–8.5 human embryos. All images are maximum intensity pseudo‐colored segmented or unsegmented signal. Panel (a) images are left lateral views of all nine embryos with immunostaining antibodies indicated alongside each embryo. The dotted boxes around the PCW5.4 entire sample and PCW7 and PCW8.5 heads indicate the regions of interest in which structures were segmented and rendered in panels (b–d), respectively. Panel (b) is a left lateral view of different structures segmented in a PCW5.4 embryo based on anti‐EpCAM, PLVAP immunostainings and background noise (561 nm channel). Panel (c) is a left lateral view of different structures segmented in a PCW7 embryo head based on anti‐Sox9, PLVAP, and SMA immunostainings. Panel (d) is a left lateral view of different structures segmented in a PCW8.5 embryo head based on anti‐synaptophysin and SMA immunostainings. In each of the panels (b–d), the rightmost image represents the results of all segmented channels in the sample region of interest. Not all segmented structures are named in panels (b‐d). Ao, aorta; ASA, anterior spinal artery; BA, basilar artery; C1SN ISA, C1 spinal nerve intersegmental artery (proatlantal artery); CCA, common carotid artery; CD31, cluster of differentiation 31; CN I–XII, cranial nerves I–XII; CNS, central nervous system; CoeA, coeliac artery; EpCAM, epithelial cell adhesion molecule; ICA, internal carotid artery; InfMA, inferior mesenteric artery; ISA, intersegmental artery; MCA, middle cerebral artery; MesA, mesencephalic artery; MHC, myosin heavy chain; pAChoA, primitive anterior choroidal artery; PCA, posterior cerebral artery; PComA, posterior communicating artery; PCW, postconception week; Phox2b, paired‐like homeobox 2b; PICA/PLBVA, posterior inferior cerebellar artery/primitive lateral basilo‐vertebral anastomosis; PLVAP, plasmalemma vesicle‐associated protein; POA, primitive olfactory artery; Post‐com. ACA, post‐communicating anterior cerebral artery; Pulm. A., pulmonary artery; SCA, superior cerebellar artery; SMA, smooth muscle actin, Sox9, sex‐determining region of chromosome Y (SRY)‐box containing gene 9; SupMA, superior mesenteric artery; Tag1, transient axonal glycoprotein 1 (contactin2); VPA, ventral pharyngeal artery. Anatomical representation of the circle of Willis In most representations, the circle of Willis is shown from below (Figure 3a) and the anterior communicating artery (AComA) representing the ventral component of the circle is at the top. This convention derives from the way anatomists, including Willis himself, observed the anastomotic circle from a basal view of extracted brains. It also conforms to medical imaging convention (inferior view of axial reconstructions in supine position with the ventral side facing upwards in the image). The circle of Willis representation convention in adult anatomy and embryological constraints. Panel (a) is an illustration drawn by Christopher Wren in the 1664 Thomas Willis book Cerebri Anatome cui Accessit Nervorum Descriptio et Usus. The image shows an inferior view of the brain with the anastomotic circle of Willis. Panels (b–h) are light sheet fluorescence microscopy (LSFM) images of solvent‐cleared immunostained postconception week (PCW) 5–8.5 human embryos. Panels (b–f) and (h) are images taken in “Blend” mode in Imaris (opaque or partially opaque structures). Images in panel (g) are taken in maximum intensity projection. Panel (b) is an inferior view of the brain with cerebral arteries. Arteries are segmented according to adult anatomical convention and adult color coding is applied (related to subsequent figures). The primitive olfactory artery (POA) and ophthalmic arteries (OA) are not represented. Panel (c) images are lateral views of a PCW7 embryo with different degrees of opacity demonstrating the relative position of the circle of Willis beneath the brain. Panel (d) is an inferior view of the brain and circle of Willis. Panel (e) shows frontal views of a PCW7 embryo with different degrees of opacity demonstrating the relative position of the circle of Willis. Panel (f) is a frontal (superior) view of the circle of Willis. Notice that the “anterior” part of the circle (anterior communicating artery, AComA), belonging to the “cranial division” of the internal carotid artery (ICA; related to Figure 4) is in fact facing caudally if the embryo is positioned upright. Also, the “posterior” part of the circle (posterior communicating artery, PComA), belonging to the “caudal” division of the ICA, paradoxically faces cranially in an upright PCW7 embryo. Panel (g) is a growth series of the circle of Willis in frontal (superior) view. Panel (h) is a frontal (superior) view of the circle of Willis with its cerebral branches. Arteries are segmented according to adult anatomical convention and adult color coding is applied (related to subsequent figures). The POAs, OAs, and basilar artery (BA) branches are not represented. A1, anterior cerebral artery pre‐communicating segment; ACA, anterior cerebral artery; AChoA, anterior choroidal artery; AComA, anterior communicating artery; AICA, anterior inferior cerebellar artery; ASA, anterior spinal artery; BA, basilar artery; CPA, circumferential pontine artery; Di., diencephalon; DiA, diencephalic artery; ICA, internal carotid artery; LPCA, lateral posterior choroidal artery; MCA, middle cerebral artery; Mes., mesencephalon; MesA, mesencephalic artery; MSA, medial striate artery (recurrent artery of Heubner); P1, pre‐communicating segment of the posterior cerebral artery; PCA, posterior cerebral artery; PComA, posterior communicating artery; PCW, postconception week; PICA, posterior inferior cerebellar artery; PLBVA, primitive lateral basilo‐vertebral anastomosis; Rhomb., rhombencephalon; SC, spinal cord; Telenc., telencephalon; VA, vertebral artery. In her seminal 1948 paper, Padget provided exquisitely clear lateral graphical reconstructions of human embryos (Padget, 1948). In contrast, the ventral graphical reconstructions are often more difficult to understand (see figs. 4b, 5b and 7b in Padget, 1948 for instance). This is especially true regarding the full anatomy of the circle of Willis and its posterior branches, which are difficult to discern on the ventral graphical reconstructions. The difficulty in understanding the circle of Willis anatomy on these views probably stems from the fact that the posterior part of the anastomotic circle in human embryos is deeply embedded in the groove between diencephalon, mesencephalon, and rhombencephalon (Figure 3b–d). Padget probably chose this inferior view for the ontogenic series to conclude with the conventional inferior view of the circle of Willis (fig. 11 in Padget, 1948). Tissue clearing eliminates the constraints of anatomical dissection that impose looking at the circle of Willis from below. The circle can be observed through the transparent brain, in a view that is frontal with regard to the embryo but superior with regard to the brain (Figure 3e–g). In this view, the anterior communicating artery (AComA) is in the inferior part of the image (Figure 3f). Notice the paradox that, owing to the embryonic brain flexures and ventral flexion of the head (Figure 3c), the so‐called cranial division of the internal carotid artery (ICA) is oriented “downwards” or caudally while its caudal division is facing “upwards,” in effect cranially/rostrally in an embryo presenting upright (Figure 3e,f). Likewise, in the embryo, the anterior part of the circle of Willis (i.e., the AComA) is oriented caudally (Figure 3e,f). Also, the posterior segment of the circle (posterior communicating arteries) is cranial rather than posterior in the embryo (Figure 3e,f). The “superior” view of the circle of Willis with the AComA facing downwards (in fact an embryonic frontal view) allows for a clearer depiction of the branches of the caudal division of the ICA (posterior part of the circle, Figure 3h). This is especially relevant given that the developmental fate of this posterior portion of the circle of Willis and its branches lacks satisfactory descriptions. In fact, Padget herself provided a frontal (superior) graphical reconstruction of the circle of Willis through a transparent brain, confirming better visualization of the aforementioned branches in this projection (fig. 9 in Padget, 1948). The frontal (superior) view of the circle of Willis has therefore been used throughout the current manuscript figures, including for adult anatomy to facilitate comparison. For simplicity, this view is hereafter referred to as the superior view. RESULTS Closure of the circle of Willis The starting point of this developmental time series is a PCW5 embryo (head and cervical region) immunostained with anti‐PLVAP antibodies labeling immature vessels (Figure 4a). The central nervous system (CNS) in this sample is covered by a dense perineural vascular plexus (PNVP in figures; refer to supplemental materials for a full list of abbreviations). Noticeably, at this stage, immature anatomical features are still conspicuous. This includes individualized intersegmental arteries coursing along the cervical spinal nerves branching from the paired dorsal aortas (rather than mature VAs) and unfused longitudinal neural arteries (LNAs in figures). These longitudinal neural arteries appear as vertical vascular condensations at the boundary between the perineural vascular plexus and a ventral midline avascular band. The right ICA is still connected to the right longitudinal neural artery by primitive carotid‐basilar anastomoses (pCBA in figures) including the trigeminal artery (the most proximal of the 2, projecting at the expected level of the trigeminal nerve). The most distal of the two anastomoses could not be identified by name. The ICAs give rise to the primitive maxillary arteries (PMAs in figures) before terminating as cranial and caudal divisions. At this stage, the cranial divisions are only represented by, and therefore synonymous to, the primitive olfactory arteries (POAs). The caudal divisions have already anastomosed with the termination of the ipsilateral longitudinal neural arteries, an event reported in prior studies in both human embryos and animal models (Padget, 1948; Ulrich et al., 2011). The circle of Willis is still open, as seen in the two rightmost images of Figure 4a (and the leftmost image in Figure 3g). The next major event is fusion of the longitudinal neural arteries over the midline, giving rise to the basilar artery (BA), as seen in a PCW5.4 embryo (Figure 4b). An intermediate step of this fusion process has previously been reported by Padget (fig. 2 in Padget, 1948). This event interconnects the two caudal divisions and therefore closes the posterior portion of the nascent circle of Willis (Figures 3g and 4b). The cranial divisions (synonymous to the POAs at this stage), on the other hand, still course toward the nasal pits (NPs in Figure 4c) but have yet to anastomose medially (Figure 4b). Closure of the circle of Willis. All panels are light sheet fluorescence microscopy (LSFM) images of solvent‐cleared immunostained postconception week (PCW) 5–7 human embryos rendered in maximum intensity projection. Panel (a) represents raw signal and vascular segmentation of a PLVAP‐stained PCW5 embryo head and neck. Superior and oblique views show the absence of a closed arterial circle of Willis. Panel (b) is a superior view of the posteriorly closed circle of Willis in a PCW5.4 embryo. Panel (c) is a superior view of the circle of Willis in a PCW5.7 embryo (left) and a lateral view of the same embryo brain and cerebral arteries (right). Panel (d) is a superior view of the circle of Willis in the process of anterior closure in a PCW6.1 embryo. The left image represents segmentation based on embryological arterial nomenclature. The image on the right shows how the proximal segment of the POA is relabeled as the internal carotid artery (ICA) termination (white arrowheads) with the emergence of the middle cerebral artery (MCA) while the middle segment of the POA will be relabeled as the proximal part of the A1 pre‐communicating segment of the anterior cerebral artery (ACA; white arrows). Panel (e) illustrates complete circle of Willis closure in a PCW7 embryo with consolidation of the anterior communicating artery (AComA). The superior views of the circle of Willis illustrate how the middle segment of the POA (proximal A1) is integrated into the ACA A1 segment in anatomical nomenclature as the ACA overshadows the POA. A1, pre‐communicating segment of the anterior cerebral artery; ACA, anterior cerebral artery; AComA, anterior communicating artery; AICA, anterior inferior cerebellar artery; BA, basilar artery; CN I, cranial nerve I (olfactory nerve); CPA, circumferential pontine artery; Di., diencephalon; DiA, diencephalic artery; DiA‐pPChoA CT, diencephalic artery—primitive posterior choroidal artery common trunk; ICA, internal carotid artery; ICA CaD, internal carotid artery caudal division; IE, inner ear; ISA, intersegmental artery (ISAs of the C1 to C8 spinal nerves); LNA, longitudinal neural artery; MCA, middle cerebral artery; Mes., mesencephalon; MesA, mesencephalic artery; MSA, medial striate artery (recurrent artery of Heubner); NP, nasal pit; pACA, primitive anterior cerebral artery; pAChoA, primitive anterior choroidal artery; pCBA, primitive carotid‐vertebrobasilar anastomosis; PCW, postconception week; PLBVA, primitive lateral basilo‐vertebral anastomosis; PLVAP, plasmalemma vesicle‐associated protein; PMA, primitive maxillary artery; PNVP, perineural vascular plexus; POA, primitive olfactory artery; PP, pineal primordium; pPChoA, primitive posterior choroidal artery; Rhomb., rhombencephalon; SCA, superior cerebellar artery; SMA, smooth muscle actin; Telenc., telencephalon; VA, vertebral artery. At PCW5.4 (Figure 4b), the nascent circle of Willis gives rise cranially to three major cerebral embryonic arteries: the primitive anterior choroidal artery (pAChoA), the mesencephalic artery (MesA in figures) and a primitive choroido‐diencephalic artery. At this stage, the only cranial branch of the cranial division (synonymous to the POA at this stage) is the pAChoA. This artery is labeled as such to adhere to existing nomenclature (Padget, 1948) even though no choroidal plexus is visible at this stage. The caudal division gives rise to two major branches: the mesencephalic artery and primitive choroido‐diencephalic artery. The primitive ophthalmic arteries, also seen at this stage, are discussed in a later chapter (Section 3.10 “The ophthalmic artery”). At PCW5.7 (Figure 4c), the cranial division of the ICA gives rise to the pAChoA (first of the three branches to be individualized as early as PCW5.4, Figure 4b), the middle cerebral artery (MCA) and a vessel designated herein as the primitive anterior cerebral artery (pACA). On the left side of this sample, two distinct vessels arising from the cranial division correspond to the MCA territory (Figure 4c). The caudal division gives rise to the primitive posterior choroidal artery (pPChoA), the diencephalic artery (DiA) and the mesencephalic artery. Next, as seen in a PCW6.1 embryo (Figure 4d), the pACAs give rise to an anastomotic midline vascular plexus, foreshadowing the constitution of the AComA and ensuing anterior closure of the circle of Willis. As the MCA grows with the telencephalic vesicles, it overshadows the POA as the major branch of the cranial division. As a result, the proximal segment of the POA is relabeled as the ICA termination (white arrowheads, Figure 4d). With this event, the anterior choroidal artery (AChoA), previously the first branch of the POA/ICA cranial division in the form of a pAChoA, becomes the last collateral branch of the ICA before the ICA termination as the MCA–anterior cerebral artery (ACA) bifurcation (Figure 4d). Also, with the pACA growth, the POA is further overshadowed, and the middle segment of the POA/ICA cranial division is relabeled as the proximal portion of the A1 (i.e., pre‐communicating) segment of the definitive ACA (white arrows, Figure 4d). Relabeling of arterial segment boundaries with the growth of collateral branches that overshadow the primitive vessels from which they arise is a process that will be seen over and over during vascular development. This, in large part, explains discrepancies between embryonic and adult anatomical descriptions of homonymous structures. Complete closure of the circle of Willis is achieved when the AComA consolidates within the anterior communicating plexus, as seen in a PCW7 embryo (Figure 4e). At this stage, the ACA has overshadowed the POA and the mature A1 segment (white arrows, Figure 4e) can therefore be interpreted as a combination of the middle portion of the POA/ICA cranial division and the proximal pre‐communicating segment of the pACA. The ACAs further grow beyond the AComA and give rise to the medial striate arteries (MSAs, or recurrent arteries of Heubner) and cortical interhemispheric branches (Figure 4e). Maturation of the anterior and middle cerebral arteries The MCA and ACA undergo major expansion accompanying the growth of the telencephalon as seen in a PCW8.5 embryo (Figure 5a). A number of observations made in the embryonic samples shed light on anatomical observations in adult patients. As mentioned earlier, the AComA consolidates from a plexiform anterior communicating plexus at PCW6.1 (Figure 4d) into a single vessel at PCW7 (Figure 4e). This plexiform organization can persist later into embryonic life (PCW8.5, Figure 5a,b upper inset) and likely explains AComA fenestrations observed in adults (Figure 5b). A midline branch, labeled the median artery of the corpus callosum (MACC in figures), originates from the AComA at PCW8.5 (Figure 5a). We maintain the name of the median artery of the corpus callosum in this sample even though the corpus callosum is yet to develop at this stage, in accordance with the precedent of the choroidal arteries. This median artery of the corpus callosum likely gives rise to the adult subcallosal artery, also originating from the AComA (Figure 5c). In adult cases when the territory of the subcallosal artery extends to include cortical branches more often originating from the ACA, it is designated as a median artery of the corpus callosum (Figure 5c). Anterior and middle cerebral artery growth. All panels are either light sheet fluorescence microscopy (LSFM) images of a solvent‐cleared immunostained postconception week (PCW) 8.5 human embryos (panels (a), (b) inset, (d) inset and (e)), adult 3D digital subtraction angiography (DSA) images (panels b–d) or 2D DSA images of an adult patient (panel (e) insets). All images except panel (e) insets are segmented pseudo‐colored maximum intensity projection images. Panel (a) is a frontal view of the anterior cerebral circulation in a PCW8.5 embryo showing extension of the anterior and middle cerebral arteries (ACA and MCA) territories. Panel (b) inset is a zoom on the plexiform anterior communicating artery (AComA) complex in the same PCW8.5 embryo. The upper and lower panel (b) images show a fenestrated AComA in an adult patient. Panel (c) shows the course of the subcallosal artery (upper left image, left lateral view of the anterior cerebral arteries) originating from AComA (upper right image frontal view) in an adult patient. The lower image in panel (c) (left antero‐inferior oblique view of the ACAs) shows the extended territory of the median artery of the corpus callosum variant. Panel (d) inset shows the left medial striate arteries (MSAs) originating from the A1–A2 junction of the ACA and anastomosing with the left MCA (white arrowhead) in a PCW8.5 embryo. Panel (d) middle and lower images are frontal‐superior and frontal‐inferior views, respectively, of 3D DSA images from a patient with the so‐called accessory MCA anatomical variant (left side, white arrow). On the right side, the lenticulostriate arteries (LSAs) are seen to originate either from the MCA basal segment (M1) (blue‐colored LSAs) or from the MSA (double white arrows). Panel (e) left image shows the leptomeningeal anastomosis (only two are marked by white stars) between the ACAs and MCAs at the surface of the telencephalon vertex. The insets are postero‐anterior projections of the right internal carotid artery (ICA) 2D DSA in a patient undergoing a mechanical thrombectomy (MT) procedure for a right M1 MCA occlusion (white arrowhead). The white arrows mark the direction of circulation of cortical MCA branches, initially retrograde via ACA–MCA leptomeningeal anastomosis at baseline vs. antegrade following M1 recanalization by MT. A1, pre‐communicating segment of the anterior cerebral artery; A2, post‐communicating segment of the anterior cerebral artery; ACA, anterior cerebral artery; AComA, anterior communicating artery; DSA, digital subtraction angiography; ICA, internal carotid artery; LSA, lenticulostriate artery; M1, basal segment of the middle cerebral artery; MACC, median artery of the corpus callosum; MCA, middle cerebral artery; MSA, medial striate artery (recurrent artery of Heubner); MT, mechanical thrombectomy; PCW, postconception week. The MSAs (recurrent arteries of Heubner) originate from the peri‐communicating segment of the ACAs as seen at PCW7 (Figure 4e) and subsequently expand laterally to anastomose with the MCA (Figure 5d inset). This anastomosis is the most likely explanation of the so‐called accessory MCA variant in which an MCA cortical branch originates from the ACA, usually from the peri‐communicating segment. In the example provided (Figure 5d), the left accessory MCA (white arrow) gives rise to lenticulostriate arteries (LSA in figures). On the right side, in the same patient, the medial striate artery (recurrent artery of Heubner) territory extends to include a subset of the lenticulostriate arteries (Figure 5d, double white arrows) in what can be considered a minimal form of an accessory MCA variant. These observations make the case for interpreting the adults' lenticulostriate arteries as originating from the embryonic medial striate artery (recurrent artery of Heubner) territory. With MCA and ACA expansion, the distalmost cortical branches of these arteries meet at the telencephalic vertex and often anastomose (Figure 5e). These anastomoses most likely explain adult leptomeningeal anastomoses that account for distal retrograde filling of cortical arteries from adjacent vessels in case of large vessel occlusion stroke (Figure 5e insets). The primitive olfactory artery As mentioned earlier, the cranial division of the ICA is primitively only represented by the primitive olfactory artery (POA, Figure 4a). With the emergence and growth of the MCA and pACA (Figure 4c–e), the POA is overshadowed and subsequently appears as a collateral of the A1 pre‐communicating segment of the ACA. The POA persists as a vessel coursing toward the ipsilateral olfactory bulb and nerves (Figure 6a,b), leaving the braincase into the nasal capsule (Figure 6c). Interestingly, the pACAs, in parallel to midline fusion (Figure 4d) also send branches that secondarily anastomose with the POA distally (Figure 6d white arrowheads). This could potentially explain reported cases of A1 segment fenestrations (Dimmick & Faulder, 2009). The distal POA‐ACA anastomosis translates at later stages into the observation of multiple POAs originating from A1 (primitive position), the peri‐communicating segment of the ACA, the AComA and even from the medial striate artery (recurrent artery of Heubner) as seen in a PCW8.5 embryo (Figure 6e). In adult anatomy, the ACA does not have macroscopic nasal branches. A putative derivative of the POA in adult anatomy may be the orbitofrontal branch of the ACA, originating from the post‐communicating portion of the artery and coursing above the olfactory grooves of the cribriform plate (Figure 6f upper image) but this is speculative. An interesting adult anatomical variant in this context is the so‐called persistent POA (example provided in Figure 6f). Usually, the persistent POA does not have any ethmoidal or nasal territory and therefore does not strictly represent the embryonic POA. However, as would be expected if the so‐called adult persistent POA relates to the embryonic POA, the persistent POA originates from the A1 segment of the ACA (Figure 6f). Finally, the embryonic POAs secondarily anastomose with the ophthalmic arteries (OAs) (Figure 6g). This leads us to hypothesize that POA nasal branches could be secondarily transferred to the OAs and may represent the ethmoidal branches of the adult OA, a hypothesis warranting further investigation. The primitive olfactory artery. All panels are either light sheet fluorescence microscopy (LSFM) images of solvent‐cleared immunostained postconception week (PCW) 6.6 or 8.5 human embryos (panels (a–d) and (g)) or adult 3D digital subtraction angiography (DSA) images (panel f). All images are segmented pseudo‐colored maximum intensity projections. Panel (a) upper images are left lateral views of a PCW6.6 embryo brain with segmented cranial nerves (left) and cerebral arteries (right). The lower image in panel (a) is a superior view of the primitive olfactory artery (POA) reaching the olfactory bulbs and nerves (cranial nerve I, CN I). Panel (b) is a left lateral view of the same sample demonstrating the relationship of the POA and CN I. Panel (c) shows the POA leaving the braincase and coursing into the nasal capsule (presumably along CN I) in a PCW7 embryo. Panel (d) is a superior view of the circle of Willis in a PCW6.6 embryo showing the primitive anterior cerebral arteries (pACA) giving rise to branches that secondarily anastomose with the POA distally (white arrowheads). Panel (e) images are different views of a PCW8.5 embryo showing the various origins of the POA at this stage from the A1 segments of the ACAs, the A1–A2 junction and the medial striate arteries (MSAs). Panel (f) is an adult case of the so‐called persistent POA, an artery originating from mid‐A1, coursing along the base of the anterior cranial fossa before making an acute angle and giving rise to ACA cortical branches. Panel (g) is a frontal (superior) view of anastomosis between the POAs and ophthalmic arteries (OA) in a PCW8.5 embryo. A1, pre‐communicating segment of the anterior cerebral artery; A2, post‐communicating segment of the anterior cerebral artery; ACA, anterior cerebral artery; AComA, anterior communicating artery; AICA, anterior inferior cerebellar artery; BA, basilar artery; CN I–XII, cranial nerves I–XII; CPA, circumferential pontine artery; Di., diencephalon; DiA, diencephalic artery; DSA, digital subtraction angiography; ICA, internal carotid artery; MACC, median artery of the corpus callosum; MCA, middle cerebral artery; Mes., mesencephalon; MesA, mesencephalic artery; MSA, medial striate artery (recurrent artery of Heubner); OA, ophthalmic artery; pACA, primitive anterior cerebral artery; pAChoA, primitive anterior choroidal artery; PCW, postconception week; PLBVA/PICA, primitive lateral basilo‐vertebral anastomosis/posterior inferior cerebellar artery; POA, primitive olfactory artery; pPChoA, primitive posterior choroidal artery; proxA1, proximal A1; Rhomb., rhombencephalon; SCA, superior cerebellar artery; Telenc., telencephalon; VA, vertebral artery. The posterior cerebral artery As a consequence of the events described in Figure 4, the three branches of the cranial division of the ICA (primitive anterior choroidal artery, MCA, and pACA) translate into adult anatomy in a relatively straightforward manner. The adult ICA gives rise to the AChoA before terminating in an MCA–ACA bifurcation. It was however unclear from historical descriptions of human neurovascular embryology how the branches of the caudal division of the ICA (pPChoA, diencephalic and mesencephalic arteries) translate into adult anatomy and conversely, how the posterior cerebral artery (PCA) emerges from embryonic anatomy in humans (Padget, 1948). As stated earlier, the ICA caudal division gives rise to a mesencephalic artery, a DiA and a pPChoA, the last two arising from a common trunk in most instances (Figure 4d,e for instance). At PCW7, the pAChoA artery and pPChoA have reached the choroid plexus of the lateral ventricle (LV ChoP in figures; Figure 7a). Interestingly, these primitive choroidal arteries (unambiguously seen in Figure 4c,d in PCW5.7 and PCW6.1 embryos) are identifiable before the choroid plexus of the lateral ventricles develop. We have nevertheless chosen to uphold their classical designation to avoid confusion. Figure 7a upper left inset (same PCW5.7 embryo as in Figure 4c) demonstrates the absence of macroscopic choroid plexus in the lateral ventricles at this stage at which the pAChoA artery and pPChoA are well developed. Subsequently, these vessels are primitively not choroidal arteries. As mentioned earlier, the pAChoA directly translates into the AChoA in adult anatomy. On the other hand, it is unclear before PCW7 where the PCA is and how the pPChoA relates to the adult medial and lateral posterior choroidal arteries (MPCA and LPCA). The adult MPCA is usually a proximal branch of the P2 (post‐communicating) segment of the PCA (Figure 7b). The LPCA is a branch or series of branches originating distally from the PCA. Importantly, only the LPCA is closely related to and anastomotic with the AChoA in the temporal horn and atrial portion of the choroid plexus of the lateral ventricles in adults. Figure 7c–g illustrate in a single PCW7 embryo how the PCA can be interpreted to replace the primitive posterior choroidal artery (pPChoA) in adult anatomy and how this leads to cascading consequences in terms of relabeling of arterial segments. Figure 7c represents left lateral and right superolateral views of cerebral arteries in the same PCW7 embryo. In Figure 7c, the ICA caudal division branches are labeled according to embryonic nomenclature. At PCW7, and contrary to what is observed earlier (PCW5.7, 6.1, and 6.6 embryos depicted in Figures 4c,d and 6a), the pPChoA gives rise to collateral branches coursing along the medial‐dorsal surface of the telencephalic vesicles (Figure 7d, double white arrowheads), that is, the territory of the adult PCA. This leads to the reinterpretation of the proximal stem of the pPChoA as being the PCA (Figure 7e, double white arrows). The distalmost choroidal termination of the pPChoA can subsequently be interpreted as the LPCA (Figure 7f, white arrowhead), respecting both the distal origin of the LPCA from the PCA and close relationship of the LPCA with the adult AChoA (pAChoA in embryological nomenclature). The adult PCA is therefore a composite vessel combining (1) the most dorsal segment of the caudal division of the ICA, (2) the common trunk of the DiA and primitive posterior choroidal artery (pPChoA), (3) the stem of the pPChoA and 4/ the non‐choroidal branches of the pPChoA coursing along the medial and dorsal surface of the telencephalon (Figure 7g). The posterior communicating artery (PComA) is the residual segment of the caudal division of the ICA. The PComA is the segment of the caudal division between the ICA and the origin of the common trunk of the DiA and pPChoA (Figure 7c–g). Embryological origin of the posterior cerebral artery. All panels are either light sheet fluorescence microscopy (LSFM) images of solvent‐cleared immunostained postconception week (PCW) 5.7 and 7 human embryos or segmentations of MRI–3D digital subtraction angiography (DSA) data in adult patients (panel (b) and images noted “Adult” in panel (h)). All images except the insets in panel (a) are segmented pseudo‐colored maximum intensity projections. Panel (a) is a lateral view of the brain and cerebral arteries in a PCW7 embryo. The insets are coronal oblique views showing emergence of the telencephalic choroid plexus primordia (white arrowheads, lower left inset) and later expansion (white arrows, lower right inset). Panel (b) shows the position of the lateral ventricles choroid plexus within the brain and the three major choroidal arteries reaching the plexus at adult age. Panels (c–g) illustrate the reinterpretation of embryonic anatomy at PCW7 to account for the emergence of the posterior cerebral artery (PCA). In panel (c), the internal carotid artery caudal division (ICA CaD) and its three major branches are segmented according to embryonic nomenclature. In panel (d), the primitive posterior choroidal artery (pPChoA) at PCW7 is seen to have previously absent non‐choroidal branches (double white arrowheads), reaching the medial‐dorsal surface of the telencephalon, the territory of the adult PCA. In panel (e), the proximal stem of the pPChoA is reinterpreted as the PCA (double white arrows). In panel (f), the terminal choroidal branch of the pPChoA is relabeled as a lateral posterior choroidal artery (LPCA, white arrowhead). In panel (g), the PCA is shown to be constituted by the most dorsal segment of the ICA CaD, the common trunk of the pPChoA with the diencephalic artery (DiA), the stem of the pPChoA and finally the non‐choroidal branches of the pPChoA. Panel (h) represents superior views of the circle of Willis cerebral arteries in a PCW7 embryo and adult each interpreted both in terms of embryonic and adult anatomical nomenclature. The basilar artery (BA) and internal carotid arteries (ICA), whose labels are placed inside the circle, are the arteries feeding the circle of Willis. The labels placed outside the circle are those of the individual segments of the circle of Willis. A1, pre‐communicating segment of the anterior cerebral artery; ACA, anterior cerebral artery; AChoA, anterior choroidal artery; AComA, anterior communicating artery; AICA, anterior inferior cerebellar artery; BA, basilar artery; CPA, circumferential pontine artery; Di., diencephalon; DiA, diencephalic artery; DiA‐pPChoA CT, diencephalic artery—primitive posterior choroidal artery common trunk; EpCAM, epithelial cell adhesion molecule; ICA, internal carotid artery; ICA CaD, internal carotid artery caudal division; LPCA, lateral posterior choroidal artery; LV ChoP, lateral ventricle choroid plexus; MCA, middle cerebral artery; Mes., mesencephalon; MesA, mesencephalic artery; MPCA, medial posterior choroidal artery; P1, pre‐communicating segment of the posterior cerebral artery; pACA, primitive anterior cerebral artery; pAChoA, primitive anterior choroidal artery; PCA, posterior cerebral artery; PComA, posterior communicating artery; PCW, postconception week; PLBVA, primitive lateral basilo‐vertebral anastomosis; PLVAP, plasmalemma vesicle‐associated protein; POA, primitive olfactory artery; PP, pineal primordium; pPChoA, primitive posterior choroidal artery; Rhomb., rhombencephalon; RP, Rathke's pouch; SCA, Superior cerebellar artery; SMA, smooth muscle actin; Telenc., telencephalon; V4 ChoP, IVth ventricle choroid plexus; VA, vertebral artery. In adult anatomy, the circle of Willis is fed by both ICAs and the BA and is per se composed of the AComA, A1 segments of the ACAs, the ICA terminal segments (giving rise to the AChoAs), the PComAs, and the P1 pre‐communicating segments of the PCAs. Figure 7h illustrates how the boundaries and identities of segments of the circle of Willis vary when interpreted according to embryonic and adult anatomical nomenclatures. Adult derivative of the embryonic mesencephalic artery As a consequence of the aforementioned events (Section 3.4 “The posterior cerebral artery”), if the embryonic mesencephalic artery is to give rise to an adult derivative, this artery must (1) be destined to a mesencephalic territory, (2) arise from the P1 segment of the PCA. Indeed, the mesencephalic artery arises as the most dorsal branch of the ICA caudal division, a segment dorsal to the DiA‐pPChoA common trunk and therefore destined to become the pre‐communicating P1 segment of the PCA (Figure 7g,h). The most likely adult candidate meeting both aforementioned specifications is the collicular artery (CoA), a small vessel described as arising from P1 and coursing around the mesencephalon to reach the tectum mesencephali. Because this is a small vessel, it is best depicted in pathological situations where increased flow leads to increased arterial caliber. Figure 8 illustrates the anatomy of enlarged CoAs in a patient with a ruptured tectum mesencephali micro‐arteriovenous malformation (AVM). The patient presented with acute onset headache and altered consciousness and was diagnosed with an intracerebral hemorrhage (ICH in figures) of the tectum mesencephali with secondary intraventricular hemorrhage and obstructive hydrocephalus (Figure 8a, upper images). The patient underwent DSA as part of the routine workup, revealing the microAVM (Figure 8a, lower images). MRI–3D DSA fusion and segmentation demonstrated the microAVM to be in the tectal plate, adjacent to the intracerebral hemorrhage, as expected (Figure 8b). This tectal microAVM was fed bilaterally by enlarged CoAs, originating from the P1 segments of the right and left PCAs and coursing around the mesencephalon (Figure 8b,c). The origin and course of the CoAs, as seen in this patient and described in the literature (Cortese et al., 2022), conform to the interpretation of the embryonic mesencephalic artery as being represented by the CoA in adult anatomy. Collicular arteries supplying a tectal arteriovenous malformation. All panels are from the same adult patient presenting with a ruptured tectum mesencephali arteriovenous malformation (AVM) supplied bilaterally by collicular arteries emerging from the P1 segments of the posterior cerebral arteries (PCAs). Panel (a) is a summary of the acute phase imaging workup. Axial and sagittal non‐contrast computed tomography (NCCT) reconstructions show a tectal intracerebral hematoma (ICH; white arrow) and secondary 3rd and 4th ventricles intraventricular hemorrhage (white arrowheads) with obstructive hydrocephalus. The magnetic resonance imaging (MRI) susceptibility weighed imaging (SWI) axial reconstruction (including the upper right inset) shows the position of the ICH within the mesencephalon. The lower part of the image shows the result of vertebrobasilar 2D and 3D digital subtraction angiography (DSA) demonstrating an arteriovenous shunt, as evidenced by early venous opacification (double white arrows), confirming the diagnosis of a ruptured microAVM. The rightmost image is a lateral view of segmented and pseudo‐colored arteries and AVM of the vertebrobasilar circulation. Panel (b) shows the relative position of the AVM nidus in the tectum with deep veinous drainage and arterial supply by collicular arteries. Vascular segmentations in panel (b) were performed on 3D DSA data. Panel (c) represents the spatial relationship of the collicular arteries, mesencephalon and brain AVM nidus. The collicular artery is a vessel arising from P1 and coursing around the mesencephalon and most likely represents the remnant of the embryonic mesencephalic artery at adult age. The middle image is an inferior view of the collicular arteries (rather than a superior view) to facilitate comparison with NCCT and MRI images in panels (a) and (b). AVM, arteriovenous malformation; BA, basilar artery; DSA, digital subtraction angiography; Mes., mesencephalon; MRI, magnetic resonance imaging; NCCT, non‐contrast computed tomography; PCA, posterior cerebral artery; SWI, susceptibility weighted imaging; VA, vertebral artery; VB, vertebrobasilar. Adult derivative of the embryonic diencephalic artery Identification of an adult derivative of the embryonic DiA is less straightforward because of the remodeled anatomy of the diencephalon in the adult human brain, hidden deeply between the cerebral hemispheres. According to the sequence of events described earlier (Section 3.4 “The posterior cerebral artery”), if the embryonic DiA is to give rise to an adult macroscopic derivative, this artery is expected to: (1) be destined to a diencephalic territory, (2) arise from the P2 segment of the PCA. Indeed, the DiA most often arises from a common trunk with the pPChoA (Figure 4d,e). This places the DiA on the PCA beyond the PComA‐P1 divide of the caudal division of the ICA (Figure 7g,h). A major derivative of the diencephalon in adult anatomy is the thalamus (Figure 9a). A way forward is therefore to speculate that one of the major adult thalamic arteries may represent the adult derivative of the embryonic DiA. The thalamic blood supply is provided by the vertebrobasilar circulation and PComAs (Figure 9b). Each thalamus receives arterial supply from four major arteries (Figure 9c): the paramedian artery (PmedA in figures), the tuberothalamic artery (TTA in figures) and the previously mentioned lateral and medial posterior choroidal arteries (LPCA and MPCA). For a list of all alternative synonymous names of these vessels, please refer to (Rauch et al., 2024). The thalamus is sometimes also described to be supplied by inferolateral or thalamogeniculate arteries. This latter group of vessels is not readily identifiable in our 3D DSA datasets and likely represents a smaller contribution to thalamic blood supply. Schmahmann states that the inferolateral arteries are a set of –10 arteries arising from P2 (Schmahmann, 2003), likely explaining why these small vessels are not discernable macroscopically on 3D DSA. The paramedian artery arises from the P1 segment of the PCA (Figure 9d). The tuberothalamic artery arises from the PComA (Figure 9e). Neither of these two vessels conforms to the prerequisite stated above (a thalamic artery originating proximally from P2). The LPCA has already been interpreted in our paradigm as representing, in adult anatomy, the distalmost choroidal branch of the pPChoA (Figure 9f). The MPCA is an artery feeding both the lateral ventricle choroid plexus and thalamus and originating proximally from the P2 segment of the PCA (Figure 9g). Adult thalamic arteries. All panels are from the same adult patient and serve to illustrate the anatomy of the major thalamic arteries. All images except panel (a) are segmented pseudo‐colored maximum intensity projections. Panel (a) is an axial reconstruction of a 3D T1‐weighted magnetic resonance imaging (MRI) scan with the thalamus segmented by contouring. Panel (b) shows different views of segmented pseudo‐colored vertebrobasilar circulation arteries, the segmented thalami and choroid plexus of the lateral ventricles. Panel (c) images are different views of the thalami with the thalamic arteries arising from the posterior communicating arteries (PComAs), P1 pre‐communicating, proximal P2 post‐communicating and distal segments of the posterior cerebral arteries (PCAs). Panel (d) shows the paramedian arteries arising from the P1 segment of the PCA. Panel (e) shows the tuberothalamic arteries (TTAs) arising from the PComAs. The right TTA is hypoplastic. Panel (f) shows the lateral posterior choroidal arteries (LPCAs) arising distally from the PCAs. Panel (g) shows the medial posterior choroidal artery (MPCA) arising proximally from the P2 post‐communicating segment of the PCA. BA, basilar artery; DSA, digital subtraction angiography; LPCA, lateral posterior choroidal artery; MPCA, medial posterior choroidal artery; MRI, magnetic resonance imaging; P1, pre‐communicating segment of the posterior cerebral artery; PCA, posterior cerebral artery; PComA, posterior communicating artery; PmedA, paramedian artery; TTA, tuberothalamic artery. We speculate that the MPCA represents the embryonic DiA in adult anatomy (Figure 10a). Indeed, both the thalamic territory of the MPCA and its proximal P2 origin conform to this interpretation. Interestingly, the MPCA has been described as supplying the pineal gland (Kahilogullari et al., 2013) and courses along this structure in our datasets (Figure 10b). This also conforms to the interpretation of MPCA as deriving from the embryonic DiA since the pineal gland originates along the diencephalon (see the position of the pineal primordium noted PP in Figures 4c and 7a), albeit at least partially from non‐neural progenitors (Staudt et al., 2019). The medial posterior choroidal artery as a putative representative of the embryonic diencephalic artery in adult anatomy. All panels are either segmentations of MRI—3D digital subtraction angiography (DSA) data in adult patients or light sheet fluorescence microscopy (LSFM) images of solvent‐cleared immunostained postconception Week (PCW) 7 (panel a) and PCW5.7 (panel c) human embryos. All images except panel (b) MRI are segmented pseudo‐colored maximum intensity projections. Panel (a) serves to illustrate the possible homologous origin of the embryonic diencephalic artery (DiA) and adult medial posterior choroidal artery (MPCA). The images represent right superolateral (PCW7 embryo) or superior (Adult) views of dorsal part or the circle of Willis and downstream arterial branches in a PCW7 embryo and adult, each interpreted both in terms of embryonic and adult anatomical nomenclature. Panel (b) demonstrates the spatial relationship between the MPCA and the pineal gland. Panel (c) serves to illustrate how an anatomical variant of the origin of the MPCA in adult anatomy (proximal origin from P1) can be understood as a consequence of an anatomical variant of the primitive posterior choroidal artery (pPChoA) during embryonic development (separate origin from the diencephalic artery). The images represent superior (Adult) and left superolateral (PCW7 embryo) views of the dorsal part or the circle of Willis and arterial downstream branches in an adult and PCW5.7 embryo, each interpreted both in terms of adult and embryonic anatomical nomenclature. Panel (d) shows two superior views of a posterior cerebral artery (PCA) in an adult, interpreted in terms of adult and embryonic anatomical nomenclature. The left image shows a variant of the MPCA arising from a common trunk with the collicular artery (CoA) from P1 while the right image predicts the embryonic anatomy that would have resulted in such a variant (common trunk of the mesencephalic and diencephalic arteries, separate from the pPChoA). Panel (e) shows superior views of three PCAs with distal origin of the MPCA, a variant lacking an embryonic explanation. A1, pre‐communicating segment of the anterior cerebral artery; AChoA, anterior choroidal artery; AComA, anterior communicating artery; BA, basilar artery; CoA, collicular artery; DiA, diencephalic artery; DiA‐pPChoA CT, diencephalic artery—primitive posterior choroidal artery common trunk; ICA, internal carotid artery; ICA CaD, internal carotid artery caudal division; LPCA, lateral posterior choroidal artery; LV ChoP, lateral ventricle choroid plexus; MCA, middle cerebral artery; MesA, mesencephalic artery; MPCA, medial posterior choroidal artery; P1, pre‐communicating segment of the posterior cerebral artery; pACA, primitive anterior cerebral artery; pAChoA, primitive anterior choroidal artery; PCA, posterior cerebral artery; PComA, posterior communicating artery; PCW, postconception week; POA, primitive olfactory artery; pPChoA, primitive posterior choroidal artery. To further argue in favor of the above‐mentioned interpretations, we sought to compare embryonic and adult anatomical variants. A classically described anatomical variation of the MPCA is a more proximal origin from the P1 PCA segment (Figure 10c). We have identified at least one PCW5.7 embryo in which the three branches of the ICA caudal division arise independently on the right side, namely, from proximal to distal relative to the ICA: the pPChoA, the DiA and the mesencephalic artery (Figure 10c). In this setting, the absence of a DiA‐pPChoA common trunk implies that the DiA and its putative MPCA derivative would originate from the P1 rather than the P2 segment of PCA. Another potential variant is the CoA and MPCA common trunk (left image in Figure 10d). We have not encountered an embryo explaining this conformation. Nonetheless, we can predict the embryonic anatomical variant that is expected to provide the basis of the above‐mentioned adult variant (Figure 10d): a common trunk of the embryonic mesencephalic and diencephalic arteries. A more problematic, and more common, variant of the MPCA is a distal origin, often beyond the origin of the LPCA branches (Figure 10e). Our interpretation of embryonic‐adult anatomy correspondence does not provide an explanation for this variant. This either contradicts the hypothesis that the DiA is represented in adult anatomy by the MPCA or points to a previously unrecognized vascular remodeling event involving the PCA and MPCA later in embryonic or fetal life. Illustrative case related to the mesencephalic and diencephalic arteries Figure 11 illustrates the above‐mentioned anatomical observations in an adult patient with a ruptured thalamo‐mesencephalic microAVM. The patient presented with acute onset vertigo and right brachiofacial sensory deficit and was diagnosed with a left thalamic intracerebral hemorrhage (Figure 11a). The patient underwent DSA as part of the routine workup, revealing the microAVM (Figure 11b). MRI–3D DSA fusion and segmentation demonstrated the microAVM to be at the boundary of the diencephalon and mesencephalon, at the medial aspect of the intracerebral hemorrhage sequela (Figure 11c,d). Given the topography of the microAVM, it is expected to be fed by both mesencephalic and diencephalic arteries. This thalamo‐mesencephalic microAVM is fed by the left CoA and left MPCA as demonstrated by superselective DSA (early venous filling proving arteriovenous shunting) and comparison with 3D segmentation data (Figure 11e). As described earlier (Section 3.5 “Adult derivative of the embryonic mesencephalic artery”), both CoAs arose from the P1 segments of the PCAs with a subsequent perimesencephalic course (Figure 11f). Also, the MPCAs originated proximally from the P2 segments of the PCAs (Figure 11f; as described in Section 3.6. “Adult derivative of the embryonic diencephalic artery”). Comparison of adult and embryonic anatomical nomenclature is provided for the three embryonic branches of the ICA caudal division for this patient (Figure 11f). Illustrative case: Thalamo‐mesencephalic micro‐arteriovenous malformation. All panels are from the same adult patient presenting with a ruptured left thalamo‐mesencephalic arteriovenous malformation (AVM) supplied by the left collicular artery (CoA; originating from P1) and the left medial posterior choroidal artery (MPCA; originating from the post‐communicating segment of the posterior cerebral artery). Panel (a) is a summary of the acute phase non‐contrast computed tomography (NCCT). Axial and coronal reconstructions show a left thalamic intracerebral hematoma (ICH; white arrow). Panel (b) is the follow‐up diagnostic workup with magnetic resonance imaging (MRI) axial T2‐weighted fluid‐attenuated inversion recovery (FLAIR) imaging showing a thalamic sequela (white arrowhead). Postero‐anterior projection vertebrobasilar (VB) digital subtraction angiography (DSA) further demonstrates an arteriovenous shunt, as evidenced by early venous opacification (double white arrows), confirming the diagnosis of a previously ruptured microAVM. Panel (c) represents left lateral and frontal views of the segmented and pseudo‐colored 3D VB DSA. Panel (d) images are axial and coronal reconstructions of the fused 3D T2‐FLAIR MRI and segmented 3D DSA. Only the brain AVM nidus is represented from segmentation data (in red). The nidus is shown to be at the medial aspect of the ICH sequela (white arrowhead). The images demonstrate the relative position of the AVM nidus at the boundary between the diencephalon (Di.) and mesencephalon (Mes.). Panel (e) shows postero‐anterior (PA) and lateral projections of superselective DSAs of the two main AVM feeders (insets). These are identified on 3D DSA as being the left CoA and left MPCA. Panel (f) shows the emergence of the CoAs from the P1 segments of the posterior cerebral arteries (PCAs) and the circummesencephalic course of these arteries. Also illustrated is the course of the left MPCA relative to the left thalamus (lateral view). Next (right half of the panel), the basilar artery (BA) termination into PCAs and proximal branches of interest are represented with both adult and embryonic segmentation and color coding. AICA, anterior inferior cerebellar artery; ASA, anterior spinal artery; AVM, arteriovenous malformation; BA, basilar artery; CoA, collicular artery; CPA, circumferential pontine artery; Di., diencephalon; DiA, diencephalic artery; DiA‐pPChoA CT, diencephalic artery—primitive posterior choroidal artery common trunk; DSA, digital subtraction angiography; FLAIR, fluid‐attenuated inversion recovery; ICA CaD, internal carotid artery caudal division; L, left; LPCA, lateral posterior choroidal artery; LV ChoP, lateral ventricle choroid plexus; Mes., mesencephalon; MesA, mesencephalic artery; MPCA, medial posterior choroidal artery; MRI, magnetic resonance imaging; NCCT, non‐contrast computed tomography; P1, pre‐communicating segment of the posterior cerebral artery; PA, postero‐anterior (projection); PCA, posterior cerebral artery; PComA, posterior communicating artery; PICA, posterior inferior cerebellar artery; PLSA, postero‐lateral spinal artery; PmedA, paramedian artery; pPChoA, primitive posterior choroidal artery; R, right; SCA, superior cerebellar artery; VA, vertebral artery; VB, vertebrobasilar. Vein of Galen aneurysmal malformations and the choroidal branch of the anterior cerebral arteries A vein of Galen aneurysmal malformation (VGAM) is a rare congenital/pediatric disease of the velum interpositum and the ambient cistern (Raybaud, 2010). VGAMs are characterized by multiple arteriovenous shunts draining into a dilated median cerebral venous collector thought to be a persistent median prosencephalic vein of Markowski (Raybaud, 2010). Although their pathophysiology remains poorly understood, VGAMs are sometimes described as choroidal malformations (Klostranec & Krings, 2022) with arterial supply mostly derived from choroidal arteries (Gailloud et al., 2005). The posterior choroidal arteries are the most common source of supply to VGAM, but individual vessels (MPCA and LPCA) are often difficult to distinguish in this setting (Raybaud et al., 1989). The anterior cerebral artery (ACA) is also a frequent primary arterial feeder of VGAMs (Raybaud et al., 1989). This is interpreted by stating that embryologically, the distal pericallosal branch of the ACA is a choroidal artery (Raybaud et al., 1989). The existence of this putative choroidal ACA branch is poorly documented. Padget indeed noted a choroidal branch of the ACA in an injected embryo of 24 mm but the vessel was not labeled in the figures provided (Padget, 1948). She then stated that at a succeeding stage in a 39‐mm embryo (no. 6203 in her dataset), this choroidal branch was less direct. We have screened our datasets in light of this and unambiguously identified a choroidal branch of the ACA in a PCW8.5 embryo, at which stage the ACAs have significantly expanded in the interhemispheric fissure (Figure 12a). In PCW7 embryos and beyond, cerebral veins and dural sinuses appear to be labeled by SMA immunostaining, albeit with fainter signal than arteries. Figure 12b shows the spatial relationship of the ACA, ACA choroidal branches, choroidal veins, and general cerebral venous anatomy. Choroidal branches of the anterior cerebral arteries. Panels (a) and (b) are light sheet fluorescence microscopy (LSFM) images of a solvent‐cleared immunostained postconception week (PCW) 8.5 human embryos. Panels (c) to (g) are medical images of an infant with a vein of Galen aneurysmal malformation (VGAM) between 4 and 12 months of postnatal age. Panels (a), (b), and (d) are segmented pseudo‐colored maximum intensity projection images. Panel (a) shows different views of the choroidal branches of the anterior cerebral arteries (ACA ChoB). Panel (b) shows different views illustrating the spatial relationship of the ACA ChoB, the choroid plexus of the lateral ventricles and their draining veins. The olfactory bulb and nerves (CN I) and cochleovestibular nerve (CN VIII) are represented as anatomical landmarks. Panel (c) shows a non‐contrast computed tomography (NCCT) scan (upper image) and magnetic resonance imaging (MRI) scan (lower image) of a 4 and then 5‐month‐old infant with a VGAM (white arrowhead). Panel (d) represents various views of the VGAM. The left image is a fusion of a right internal carotid (R ICA) and left vertebrobasilar (L VB) 3D digital subtraction angiography (DSA) acquisitions. The middle image is a frontal view of the Left VB 3D DSA showing the posterior cerebral arteries (PCAs) feeders of the VGAM. The right image shows the right ACA feeder of the VGAM. Panel (e) is a lateral projection of the right ICA 2D DSA following partial embolization of the VGAM (white arrowhead) at 9 months (9 mo) via the ACA feeder (double white arrows) and right PCA feeder (double white arrowheads). Persistent shunting is demonstrated with the VGAM opacified at the arterial phase. Panel (f) shows the occurrence at 10 months (10 mo) of obstructive triventricular hydrocephalus (black stars) with transependymal resorption (white stars) secondary to aqueductal stenosis (red arrow). The white arrowhead points to the possible thrombosis of the VGAM. Panel (g) is a follow‐up MRI at 12 months (12 mo) showing resolution of the hydrocephalus and a patent ventriculocisternostomy stoma with adequate flow (red arrowhead). The VGAM is seen to have retracted (white arrowhead) following arteriovenous shunting obliteration, subsequently demonstrated on follow‐up DSA (not shown). A1, pre‐communicating segment of the anterior cerebral artery; ACA, anterior cerebral artery; ACA ChoB, anterior cerebral artery choroidal branch; AChoA, anterior choroidal artery; AComA, anterior communicating artery; BA, basilar artery; CN I, cranial nerve I (olfactory bulb and nerves); CN VIII, cochleovestibular nerve; DiA, diencephalic artery; DSA, digital subtraction angiography; HA, hyaloid artery; ICA, internal carotid artery; MACC, median artery of the corpus callosum; MCA, middle cerebral artery; MesA, mesencephalic artery; MRI, magnetic resonance imaging; MSA, medial striate artery (recurrent artery of Heubner); NCCT, non‐contrast computed tomography; NCilA, nasociliary artery; OA, ophthalmic artery; PCA, posterior cerebral artery; PComA, posterior communicating artery; PCW, postconception week; PICA, posterior inferior cerebellar artery; POA, primitive olfactory artery; Post‐com ACA, post‐communicating segment of the anterior cerebral artery; SCA, superior cerebellar artery; TCilA, temporociliary artery; VA, vertebral artery; VB, vertebrobasilar; VGAM, vein of Galen aneurysmal malformation. For illustrative purposes, we provide a case report of a VGAM (Figure 12c, white arrowhead) diagnosed in a 4‐month‐old infant presenting with macrocrania. At 9 months, the infant underwent VGAM embolization (Figure 12d,e). The 3D DSA (Figure 12d) clearly demonstrates that the feeders of the VGAM are the distal pericallosal branches of the right ACA (presumably the embryonic ACA choroidal branch) and multiple branches of the PCAs (presumably enlarged MPCAs and LPCAs). The right ACA (double white arrows) and PCA (double white arrowheads) were partially embolized (Figure 12e). The immediate postoperative course was uneventful. At 10 months, the infant presented with obstructive hydrocephalus secondary to aqueductal stenosis in the setting of possible VGAM thrombosis (Figure 12f). The patient was treated by ventriculocisternostomy with favorable outcome (Figure 12g). VGAM obliteration was confirmed by DSA (not shown). In the VGAM case report described in Figure 12, the AChoAs are not arterial feeders, a finding not unusual in this disease and contradicting the view that VGAMs are purely choroidal lesions per se (Klostranec & Krings, 2022) given that the AChoA is the most prominent choroidal artery, both in utero and post‐natally. Elucidating the pathophysiology of VGAM is beyond the scope of this manuscript. We nevertheless sought to compare our data with commonly held views in this setting. It is usually stated that the vascular pattern of VGAMs reflects the anatomy of a so‐called embryonic‐fetal choroid stage (Raybaud, 2010). This developmental stage has been described as a period between Weeks 6 and 11 during which the brain is fed by the prominent choroid plexus and before the intrinsic vascularization is established (Raybaud et al., 1989). According to a prevailing model, because of the developing choroid plexus metabolic activity, the choroid plexus induce a clear selection of their arterial feeders: the ACA, the AChoA, and the posterior choroidal arteries. Our data contradict these statements. At PCW5, the central nervous system (CNS) is covered by a dense perineural vascular plexus, but no intrinsic CNS vessels are seen and the telencephalic vesicles have yet to clearly individualize from the prosencephalon (Figure 13a). At PCW5.4 and 5.7 (Figure 13b,c), primitive choroidal arteries are clearly individualized as branches of the cranial (POA) and caudal divisions of the ICAs. As mentioned earlier (Section 3.4 “The posterior cerebral artery”), these arteries are identifiable before the choroid plexus of the lateral ventricles develop. Interestingly, in both these samples and at PCW6.1 (Figure 13c,d), intrinsic vessels are conspicuous in the entire CNS with the notable exception of the telencephalon. At PCW6.6 (Figure 13e), the roof of the boundary between the diencephalon and telencephalic vesicles (foramen of Monro) invaginates into the ventricles, foreshadowing the development of the choroid plexus (white arrowheads). The primitive choroidal arteries course toward these invaginations (white arrows). At PCW7, the choroid plexus of the lateral ventricles has developed and concomitantly, vascular invasion of the telencephalic parenchyma is visible (Figure 13f). The data provided indicate that vascular invasion of the CNS, with the exception of the telencephalon, precedes development of the choroid plexus. Notably, we were unable to ascertain whether the CD31 and PLVAP‐positive vessels invading the CNS are arteries, veins, or both. Intrinsic central nervous system vasculature. All panels are light sheet fluorescence microscopy (LSFM) images of solvent‐cleared immunostained postconception week (PCW) 5 to 7 human embryos. Panel (a) left image and panel (b) upper left and upper middle images represent pseudo‐colored and segmented structures taken in “Blend” mode in Imaris (opaque structures). Panel (c) left image and panel (f) middle image represent pseudo‐colored and segmented structures in maximum intensity projections. All other images are 100 μm thick non‐segmented raw signal slices in various orientations. In panel (e), the arrowheads indicate the origin of the future choroid plexus and arrows indicate choroidal arteries. BA, basilar artery; CD31, cluster of differentiation 31; CNS, central nervous system; Di., diencephalon; ICA, internal carotid artery; ICA CaD, internal carotid artery caudal division; LPCA, lateral posterior choroidal artery; LV ChoP, lateral ventricle choroidal plexus; Mes., mesencephalon; MesA, mesencephalic artery; OC, otic capsule; pAChoA, primitive anterior choroidal artery; PCA, posterior cerebral artery; PCW, postconception week; PLVAP, plasmalemma vesicle‐associated protein; POA, primitive olfactory artery; pPChoA, primitive posterior choroidal artery; Pros., prosencephalon; Rhomb., rhombencephalon; SMA, smooth muscle actin; Sox9, sex‐determining region of chromosome Y (SRY)‐box containing gene 9; Tel., telencephalon. The basilar artery branches The arteries of the mesencephalon, diencephalon, and telencephalon primarily originate from the circle of Willis. In contrast, the major rhombencephalic arteries originate directly from the vertebral and basilar arteries (Figure 14a). The BA can be considered a differentiated extension of the anterior spinal artery (ASA), enlarged to accommodate the increased metabolic demand of the rhombencephalon compared with the spinal cord, a notion defended by Charpy more than a century ago (Charpy, 1899). Like the ASA, the BA forms by midline fusion of primitive bilateral paramedian channels called the longitudinal neural arteries in the case of BA precursors (Figure 4a,b). Midline fusion of the BA is, however, heterochronous as it precedes ASA fusion which is observed to occur between PCW6.1 and PCW6.6 in our series (Figure 14b). The origin of the BA and ASA as two paramedian channels likely explains BA and ASA fenestrations, observed both during development (Figure 14c) and in adults (Figure 14d). The rhombencephalic arteries. All panels except panel (d) are light sheet fluorescence microscopy (LSFM) images of solvent‐cleared immunostained postconception week (PCW) 6.1–8.5 human embryos. Panel (d) images are adult 3D digital subtraction angiography (DSA). All images represent pseudo‐colored and segmented structures. The left image in panel (a) was taken in “Blend” mode in Imaris (opaque structures). All other images are maximum intensity projections. Panel (a) shows frontal and oblique views of the basilar artery (BA) branches in relation to the rhombencephalon (rhomb.) and spinal cord. Panel (b) illustrates the continuity of the BA with the anterior spinal artery (ASA). The ASA is seen to be unfused on the midline at PCW6.1 (left image) and partially fused at PCW6.6 (right image). Panels (c) and (d) show a BA fenestration in a PCW8.5 and adult subject, respectively. Panel (e) left images show the primitive lateral basilo‐vertebral anastomosis (PLBVA) before (PCW6.1) and after (PCW6.6) vascular pruning. The two images on the right illustrate the spatial relationship of the posterior inferior cerebellar artery (PICA, primitively the PLBVA) with cranial nerves IX to XII, particularly the hypoglossal nerve (CN XII). Panel (f) shows the choroidal branch of the PICA. Panel (g) shows the anterior inferior cerebellar artery (AICA) and labyrinthine artery coursing along the cochleovestibular nerve (CN VIII). Panel (h) upper image shows the superior cerebellar arteries (SCA) to be the penultimate set of branches of the BA. The lower image illustrates the oculomotor (CN III) and trochlear (CN IV) nerves coursing between the SCA and posterior cerebral artery (PCA). AICA, anterior inferior cerebellar artery; ASA, anterior spinal artery; BA, basilar artery; C2, 2nd (cervical) spinal nerve dorsal root ganglion; C8, 8th (cervical) spinal nerve dorsal root ganglion; CN, cranial nerves; CPA, circumferential pontine artery; DRG, dorsal root ganglion; DSA, digital subtraction angiography; P1, pre‐communicating segment of the posterior cerebral artery; PCA, posterior cerebral artery; PComA, posterior communicating artery; PCW, postconception week; PLBVA/PICA, primitive lateral basilo‐vertebral anastomosis/posterior inferior cerebellar artery; PLSA, postero‐lateral spinal artery; Rhomb., rhombencephalon; SCA, superior cerebellar artery; V4, 4th ventricle; VA, vertebral artery. The major cerebellar arteries are the posterior inferior cerebellar artery (PICA), anterior inferior cerebellar artery (AICA) and superior cerebellar artery (SCA). During development, the terminal VAs and BA give rise to relatively symmetrical pairs of circumferential branches that course around the rhombencephalon (Figure 14a). The PICA, AICA, and SCA can be identified and distinguished from the other circumferential pontine arteries (CPAs in figures). The PICA likely originates from the primitive lateral basilo‐vertebral anastomosis (PLBVA). This structure appears to be a cranial extension of the postero‐lateral spinal artery (PLSA in figures), akin to the basilar artery—ASA relation. The PLBVA primitively presents with multiple connections to the vertebrobasilar junction at PCW6.1 (Figure 14e). The definitive stem of the PICA is only apparent after vascular pruning leads to the loss of most PLBVA connections to the vertebral and basilar artery. As expected from the relation of the adult PICA with the hypoglossal nerve (cranial nerve XII), the PLBVA multiple origins course between the hypoglossal nerve rootlets (Figure 14e). The prominent branch of the PICA in our embryonic series is a choroid artery reaching the choroid plexus of the fourth ventricle (V4; Figure 14f). The PICA also seems anastomotic with the AICA (white arrow, Figure 14f). The AICA is identified as a prominent branch of the BA giving rise to a labyrinthine branch supplying the cochleovestibular nerve (Figure 14g). The SCAs are the penultimate set of BA branches before it terminates as the P1 segments of the PCAs (Figure 14h). The SCAs course below the oculomotor (cranial nerve III) and trochlear (cranial nerve IV) nerves. The emergence of the PICA stem from a process of vascular pruning of the PLBVA likely explains the important variability in the position of the origin of this artery in adult anatomy (Figure 15a). Incomplete pruning can also explain the so‐called PICA duplications (Figure 15b). The PICA can exceptionally originate extracranially from the extradural atlantoaxial segment of the VA (Figure 15c). To explain this anatomical variant, the PICA origin in this case can be interpreted as a posterior radiculomedullary artery coursing along the C2 nerve in the atlantoaxial (C1C2) intervertebral foramen followed by the cranial segment of the postero‐lateral spinal artery (white arrows; Figure 15c inset). Cerebellar arteries anatomical variability. All panels are segmentations of 3D digital subtraction angiography (DSA) data in adult patients except panels (c, inset) and (e, images noted PCW) which represent light sheet fluorescence microscopy (LSFM) images of solvent‐cleared immunostained postconception week (PCW) 6.6–8.5 human embryos. All images represent pseudo‐colored and segmented structures in maximum intensity projections except the upper right image in panel (c) which was taken in “Blend” mode in Imaris (opaque structures). Panel (a) images are oblique views illustrating variability in the origin of the posterior inferior cerebellar artery (PICA) in adult patients (white arrowheads). Panel (b) illustrates PICA duplications, either in the form of dual origins that subsequently fuse (upper images, white arrowheads) or two distinct sub‐PICA arteries (lower images, white arrows for each of the two PICA arteries). Panel (c) images are different views of a PICA artery originating from the extradural segment of the left vertebral artery. The inset illustrates a likely embryonic explanation for this variant whereas the primitive lateral basilo‐vertebral anastomosis (PLBVA) would be fed by the C2 nerve posterior radiculomedullary artery (PRMA) and the cranial segment of the postero‐lateral spinal artery (PLSA). The proximal segment of the PICA (white star) is lost in this scenario. Panel (d) illustrates the anterior inferior cerebellar artery (AICA) origin and course in modal anatomy (upper images) and the so‐called AICA–PICA variant (lower images) in which part of the PICA lateral territory is fed by the AICA. Panel (e) illustrates various configurations of the superior cerebellar arteries (SCAs) being either unique or duplicated in embryonic samples and adult patients. Panel (f) is a vertebrobasilar 3D DSA segmentation from a patient with multiple anatomical variants including basilar artery (BA) and anterior spinal artery (ASA) fenestrations, a right complete AICA–PICA with the entire PICA territory (both medial and lateral) supplied by the right AICA and a left partial AICA–PICA. Part of the left PICA territory is also supplied by the SCA. AICA, anterior inferior cerebellar artery; ASA, anterior spinal artery; BA, basilar artery; C2–C3–C8, 2nd, 3rd, 8th (cervical) spinal nerves dorsal root ganglions; CPA, circumferential pontine artery; DRG, dorsal root ganglion; DSA, digital subtraction angiography; P1, pre‐communicating segment of the posterior cerebral artery; PCA, posterior cerebral artery; PComA, posterior communicating artery; PCW, postconception week; PICA, posterior inferior cerebellar artery; PLBVA/PICA, primitive lateral basilo‐vertebral anastomosis/posterior inferior cerebellar artery; PLSA, postero‐lateral spinal artery; PRMA, posterior radiculomedullary artery; SCA, superior cerebellar artery; VA, vertebral artery. The AICA has a relatively stable origin as compared to the posterior inferior cerebellar artery (PICA). The most frequent AICA variant is the so‐called AICA–PICA configuration in which the AICA partially or completely overtakes the PICA territory (Figure 15d). The PICA‐AICA embryological anastomosis (Figure 14f) is a possible explanation of the adult AICA–PICA variant. The SCA topographic origin is also stable with the most frequently observed variant being the so‐called SCA duplication. In our embryonic samples, the SCA appears as either a single or paired set of arteries with the dominant branch being either the proximal or distal artery depending on cases (Figure 15e). Figure 15f illustrates the frequent co‐occurrence of these different anatomical variants of the vertebrobasilar circulation. As mentioned earlier, the BA gives rise to several macroscopic circumferential pontine arteries. One of these arteries, originating distal to the AICA, either terminates at the root of the trigeminal nerve or courses below the nerve (Figure 16a,b). This artery will henceforth be designated as the subtrigeminal circumferential pontine artery. It gives rise to an artery coursing within the trigeminal ganglion before exiting cranially along the ophthalmic branch of CN V (Figure 16b,c). This trigeminal nerve artery (TNA in figures) was found to either terminate within or above the trigeminal ganglion or anastomose with the OA (Figure 16c,d) or middle meningeal artery (MMA; Figure 16d). These observations are the most likely explanation for a BA origin of the OA (Sade et al., 2004) or MMA (Figure 16e), both of which are exceptionally rare variants. The trigeminal nerve artery. All except panel (e) are light sheet fluorescence microscopy (LSFM) images of solvent‐cleared immunostained postconception week (PCW) 6.6–8.5 human embryos. In panels (a–d), the insets are raw data while the other images are pseudo‐colored and segmented structures rendered in maximum intensity projections. Panel (a) is a frontal view of the basilar artery (BA) and its branches. The highlighted circumferential pontine artery (CPA) is seen to course below the roots of the trigeminal nerves (cranial nerve V). Panel (b) shows the origin of the trigeminal nerve artery (TNA) from the subtrigeminal CPA and its relation to CN V. Panel (c) shows the right TNA to be anastomotic with the ophthalmic artery (OA). The insets demonstrate the course of the subtrigeminal CPA (white arrowhead) below the origin of CN V and the subsequent course of the TNA (white arrows) within the trigeminal ganglion (noted V). Panel (d) from a PCW8 embryo shows the left TNA to be anastomotic with the left OA while the right TNA is anastomotic with the middle meningeal artery (MMA). Panel (e) are digital subtraction angiography (DSA) and magnetic resonance imaging (MRI) data from a patient with a MMA (double white arrowheads) originating from the BA. MRI and DSA were performed as part of the diagnostic workup of an intracerebral hemorrhage unrelated to the anatomical variant depicted herein. The white arrowheads are presumably the proximal course of the subtrigeminal CPA while the white arrows indicate the origin of the TNA, demonstrated on the MRI to join the right trigeminal nerve. AICA, anterior inferior cerebellar artery; BA, basilar artery; CN V, cranial nerve V (trigeminal nerve); CN VIII, cranial nerve VIII (cochleovestibular nerve); CPA, circumferential pontine artery; DSA, digital subtraction angiography; HA, hyaloid artery; IMA, internal maxillary artery; MMA, middle meningeal artery; MRI, magnetic resonance imaging; NCilA, nasociliary artery; OA, ophthalmic artery; P1, pre‐communicating segment of the posterior cerebral artery; PA, Postero‐anterior (projection); PCA, Posterior cerebral artery; PCW, postconception week; PICA, posterior inferior cerebellar artery; POA, primitive olfactory artery; SCA, superior cerebellar artery; SMA, smooth muscle actin; TCilA, temporociliary artery; TNA, trigeminal nerve artery; VA, vertebral artery. To understand why an artery coursing along the ophthalmic branch (V1) of the CN V is related to both the OA and MMA, a description of the partially shared origin of these two arteries is warranted. The ophthalmic artery In the earliest sample in the ontogenic series (PCW5), both the primitive maxillary artery and the cranial division of the ICA (only represented by the POA at this stage) terminate in the perineural vascular plexus covering the CNS (Figure 17a). The plexus extends circumferentially around the stem of the optic vesicles (OV in figures). At this stage, the outermost surface of the optic vesicles appears to be avascular and no vascular invagination in the vesicle is observed. At PCW5.4, vascular invagination of the perineural vascular plexus is observed at the level of the optic (choroid) fissure of the optic cup (Figure 17b). The invaginated vessels appear to target the lens pit and constitute the first occurrence of a hyaloid vasculature. At this stage, two arteries are identified as targeting the optic cup. The most proximal of the two, arising from the ICA and observed bilaterally, is the primitive dorsal ophthalmic artery (PDOA). The second, observed only on the left, possibly because of limited spatial resolution, arises from the ICA cranial division opposite the ostium of the pAChoA. This second artery is identified as the primitive ventral ophthalmic artery (PVOA). As previously reported (Blain et al., 2023), the arterial vasculature of the developing eye can be clearly divided into choroid vessels and a hyaloid vasculature (Figure 17c). The hyaloid vasculature of the lens is provided by the hyaloid artery (HA) which penetrates into the optic nerve and reaches the lens where it undergoes major expansion. At PCW5.7, the relative positions of the primitive dorsal and ventral ophthalmic arteries appear to be unchanged (Figure 17d). The PDOA originates proximally to the ICA caudal division (the future PComA), while the PVOA originates from the ICA cranial division (POA) opposite and slightly proximal to the origin of the pAChoA, an arterial segment which will constitute the supraclinoid portion of the ICA. The distal portions of these vessels are at the limit of the spatial resolution in this sample but their territory is clearly depicted in a PCW6.1 sample (Figure 17e). The PDOA bifurcates into a temporociliary artery (TCilA in figures) and a HA while the PVOA terminates as the nasociliary artery (NCilA in figures). The temporociliary and nasociliary arteries are the arterial feeders of the eye choroid vasculature. At this stage, the PDOA is proximal to the future PComA while the PVOA is a branch of what is now the ICA terminal segment, opposite and slightly proximal to the pAChoA. In the PCW6.6 sample, the right nasociliary artery has anastomosed with the HA while the proximal stem of the PVOA is dwindling (Figure 17f), foreshadowing complete supply of the eye choroidal vasculature by the PDOA. This process has yet to occur on the left side in this sample but the PVOA coursing along the optic nerve appears in close proximity to both PDOA branches and it can be assumed that nasociliary artery transfer to the PDOA and subsequent PVOA involution is to follow. The primitive ophthalmic arteries. All panels are light sheet fluorescence microscopy (LSFM) images of solvent‐cleared immunostained postconception week (PCW) 5–6.6 human embryos or isolated human embryonic PCW6.1 and PCW9.7 eyes (panel c). The insets are raw data with reconstructed slice thickness of 50 and 5 μm in panels (a) and (b), respectively. In panel (c), the left inset is a 100 μm‐thick reconstructed slice while the three insets on the right are 5 μm thick images. All other images are pseudo‐colored and segmented structures rendered in maximum intensity projections. Panel (a) shows the stem of the optic vesicle (OV) to be covered by the perineural vascular plexus (PNVP) while the lateral surface of the OV appears to be avascular. In the insets, the diencephalon (Di.), otic capsule (OC), the first (mandibular) pharyngeal arch (PA I) and second (hyoid) pharyngeal arch (PA II) are labeled as reference anatomical landmarks. Panel (b) shows the first occurrence of hyaloid vascular invasion of the optic cup and visible primitive dorsal (bilaterally) and ventral (on the left) ophthalmic arteries (PDOA and PVOA). Panel (c) illustrated the distinct organization of the hyaloid and choroid eye vasculatures. The hyaloid artery (HA, white arrow, lower left image and insets) is seen to change course (white arrowhead) to dive into the optic nerve (ON). Panel (d) illustrates the position of the PDOA and PVOA relative to other internal carotid artery (ICA) branches. Panel (e) illustrates how the PVOA terminates as the nasociliary artery (NCilA) while the PDOA divides into a HA and temporociliary artery (TCilA). Panel (f) shows the right NCilA (PVOA) being transferred to the HA. ACA, anterior cerebral artery; ACom, anterior communicating; Ao, aorta; BA, basilar artery; CCA, common carotid artery; CNS, central nervous system; CPA, circumferential pontine artery; DC, ductus caroticus; Di, diencephalon; HA, hyaloid artery; ICA, internal carotid artery; ICA CaD, internal carotid artery caudal division; ICA CrD, internal carotid artery cranial division; LNA, longitudinal neural artery; MCA, middle cerebral artery; Mes., mesencephalon; MesA, mesencephalic artery; MMD, maxillo‐mandibular division; NCilA, nasociliary artery; NP, nasal pit; OA, Ophthalmic artery; OC, Otic capsule; ON, optic nerve; OV, optic vesicle; PA I, pharyngeal arch I (mandibular); PA II, pharyngeal arch II (hyoid); PAA, pharyngeal arch artery; pACA, primitive anterior cerebral artery; pAChoA, primitive anterior choroidal artery; pCBA, primitive carotid‐vertebrobasilar anastomosis; pCDA, primitive choroido‐diencephalic artery; PCW, postconception week; PDOA, primitive dorsal ophthalmic artery; PLVAP, plasmalemma vesicle‐associated protein; PMA, primitive maxillary artery; PNVP, perineural vascular plexus; POA, primitive olfactory artery; pPA, primitive pulmonary artery; PVOA, primitive ventral ophthalmic artery; Rhomb., rhombencephalon; SA, stapedial artery; SMA, smooth muscle actin; TCilA, temporociliary artery; Tel., telencephalon; VPA, ventral pharyngeal artery. Padget in her description of the OA development stated that the stem of the definitive adult OA originates independently from the primitive dorsal and ventral OAs. Furthermore, she speculated that the stem of the OA reaches its adult position by a caudal “migration” of the PDOA along the carotid, effected through a series of anastomotic loops (Padget, 1948). Our findings contradict these assertions as no third OA is identified in our series and the PDOA consistently originates from the ICA proximal to the PComA. Based on our findings, we interpret the PDOA as representing the stem of the definitive OA. However, at the stages described above, the OAs lack any orbital non‐ocular branches. A description of the process by which these branches are acquired is provided below. To give such an account, a concomitant brief description of the derivatives of pharyngeal arch arteries (PAAs) and the origin of the external carotid artery (ECA) is warranted. At PCW5, the mandibular (PAA I) and hyoid (PAA II) pharyngeal arch arteries are regressing and are represented by dwindling segments (Figure 18a). The carotid pharyngeal arch artery (PAA III) is connected both cranially to the ICA and caudally to the ductus caroticus (DC in figures) segment of the aorta. The penultimate (aortic) pharyngeal arch artery (PAA IV) is already well developed and the last (pulmonary) pharyngeal arch artery is emerging. At PCW5.4, the 1st (mandibular) PAA has disappeared as a macroscopic vessel, the dorsal remnant of the 2nd (hyoid) PAA is identified as the stapedial artery (SA), the ductus caroticus is starting to regress and the last (pulmonary) PAA has increased in size and gives rise to the primitive pulmonary arteries (pPA in figures). The ventral pharyngeal artery (VPA) is a vessel originating from the ventral aspect of the third (carotid) PAA. At PCW5.7, the SA gives rise to a maxillo‐mandibular division (MMD). Concomitantly, with the regression of the ductus caroticus, the 3rd (carotid) PAA is incorporated into the definition of the ICA that can now be considered to originate at the point of bifurcation with the VPA which is the future stem of the ECA. In the same manner, the left 4th PAA is incorporated into the definitive aorta (a broader description of the origin of the supra‐aortic arteries is beyond the scope of this work). At PCW6.1, the SA gives rise to a second ascending branch identified as the supraorbital division (SOD in figures). At PCW6.6, the ductus caroticus has completely disappeared, and the common carotid artery (CCA) continues its progressive elongation observed in earlier timepoints. Figure 18b illustrates the relationship of the CCA, ICA, VPA, SA, and SA branches. In particular, the ventral pharyngeal artery divides into two distinct territories, a lateral and a medial branch. The medial branch appears to be anastomotic with the contralateral counterpart. The lateral branch of the VPA courses in close relation to the termination of the facial nerve while the medial branch is identified as the lingual artery coursing along the hypoglossal nerve (Figure 18c upper left image). The MMD of the SA is closely coupled with the mandibular nerve (V3) while the supraorbital division courses lateral to the trigeminal ganglion toward and along the ophthalmic division (V1) of the trigeminal nerve (Figure 18c lower left image). On the left side of the PCW6.6 embryo, the supraorbital division reaches the same target as the termination of the trochlear nerve, foreshadowing the emergence of the orbital non‐ocular branches of the OA (Figure 18c middle and right images). In a PCW7 sample, concomitant cartilage immunostaining allows for informative colocalization of the SA and its branches with the developing chondrocranium (Figure 19a). The MMD has a maxillary branch below the nasal capsule and a mandibular branch coursing along Meckel's cartilage (Figure 19a lower right image). The SA courses through the stapes foramen (Figure 19a upper right image). Notice that the SA, a derivative of the second (hyoid) pharyngeal arch, courses along the only one of the three auditory ossicles derived from the second pharyngeal arch (O'Gorman, 2005). By contrast, the malleus and incus auditory ossicles detach from Meckel's cartilage (Figure 19a lower right image) and as such are first (mandibular) pharyngeal arch derivatives. At this stage, the supraorbital division, which is still attached to the SA, has anastomosed with the OA which gives rise to the hyaloid, temporociliary, and nasociliary arteries (Figure 19b). The position of the OA at PCW7 is unchanged from that of the PDOA in previous stages. As stated earlier, the most parsimonious interpretation of our data is that the definitive stem of the OA is the PDOA. In the same sample aged PCW7, the MMD is now anastomotic with the VPA (Figure 19c left image). In Figure 19c, we also provide an interpretation of the origin of the middle meningeal artery (MMA). A model of hypothetical blood flow prior to MMD–VPA anastomosis is proposed (lower left part of Figure 19c). The supraorbital division at PCW7 has a growing non‐orbital branch designated in this manuscript as the primitive MMA (pMMA in figures). With the subsequent SA regression, the MMA can be interpreted as comprising the proximal segment of the MMD, the proximal segment of the supraorbital division and the primitive MMA. With this in mind, the distal supraorbital division is now relabeled the primitive orbital artery providing the orbital non‐ocular branches of the OA (Figure 19b) while an intermediate segment can persist as a MMA‐OA anastomosis. Also, with the future regression of the SA, the distal MMD can now be relabeled as part of the internal maxillary artery (IMA in figures) and the VPA is relabeled as the ECA. SA regression implies blood flow inversion in the proximal segment of the MMD which is now the origin of the MMA (lower right part of Figure 19c). The distal segment of the SA is interpreted as the superior tympanic branch of the MMA while the proximal SA stem is relabeled the carotico‐tympanic branch of the ICA. At PCW8.5, the SA has completely regressed and the MMD distal branches are incorporated as branches of the newly labeled ECA which has undergone major territorial expansion (Figure 19d upper left image). Because the ECA is a relabeled VPA with an expanded maxillo‐mandibular territory, a relation to the facial nerve is expected based on observations previously provided (Figure 18c upper left image). Indeed, and as expected, the ECA provides the so‐called stylomastoid artery coursing along the facial nerve (Figure 19d upper left dotted box and lower left image). In this sample, the OA, via its non‐ocular branches (previously the primitive orbital artery branch of the supraorbital division) is unilaterally anastomotic with the MMA on the left but not on the right (Figure 19d superior view), consistent with inconstant MMA‐OA anastomosis in the adult. Reminiscent of the supraorbital division targeting the trochlear nerve termination earlier (Figure 18c), the right OA gives rise to a branch coursing along the trochlear nerve (Figure 19d superior view). Interestingly, an analogous branch on the left arises from the meningohypophyseal trunk (MHT in figures), an arterial trunk originating from the ICA proximal to the OA and giving rise to an inferior hypophyseal artery (IHyA in figures). A frontal view (Figure 19d) illustrates the origin of the inferior hypophyseal artery relative to the more distal superior hypophyseal arteries (SHyA in figures) which also give rise to chiasmastic/pituitary stalk anastomotic arteries (ChiaA in figures). At PCW8.5, all major branches are identified along the axis of the ICA (Figure 19d right image). These are identified from proximal to distal as the meningohypophyseal trunk (giving rise to inferior hypophyseal artery), the inferolateral trunk (ILT in figures), the OA and superior hypophyseal arteries, the PComA, the AChoA and ICA termination into an MCA and ACA. The stapedial artery. All panels are light sheet fluorescence microscopy (LSFM) images of solvent‐cleared immunostained postconception week (PCW) 5–6.6 human embryos. All images are pseudo‐colored and segmented structures rendered in maximum intensity projections. Panel (a) illustrates remodeling of pharyngeal arch arteries (PAA) I–IV on lateral views in PCW5 to PCW6.6 embryos. Panel (b) shows frontal and left antero‐lateral oblique views in PCW5.4 to 6.6 embryos illustrating the relationship of the common carotid artery (CCA), internal carotid artery (ICA), ventral pharyngeal artery (VPA), stapedial artery (SA), and SA branches. Panel (c) upper left image shows a frontal view showing the relation of the VPA branches to cranial nerves. The medial (single white arrow) branch of the VPA courses along the hypoglossal nerve (cranial nerve XII) toward the tongue. The glossopharyngeal nerve (cranial nerve IX) and chorda tympani branch of the facial nerve (cranial nerve VII) are also shown. The lateral branch courses (double white arrow) along the termination of the facial nerve (cranial nerve VII). The double white arrowheads point toward branches of the sympathetic chains (SCh) coursing along the ICAs. Panel (c) lower left image is a lateral view showing the relation of the SA branches with the trigeminal nerve (CN V) and in particular, the close proximity of the maxillo‐mandibular division (MMD) with the mandibular nerve (V3 branch of CN V). Panel (c) middle and right images show the supraorbital division (SOD) of the left SA to be directed toward the same target as the termination of the trochlear nerve (white arrowhead). AChoA, anterior choroidal artery; Ao, aorta; CCA, common carotid artery; CN, cranial nerve (IV: Trochlear, V: Trigeminal, VII: Facial, IX: Glossopharyngeal, XII: Hypoglossal); CTN, chorda tympani nerve; DC, ductus caroticus; HA, hyaloid artery; ICA, internal carotid artery; ICA CaD, internal carotid artery caudal division; LNA, longitudinal neural artery; MCA, middle cerebral artery; MMD, maxillo‐mandibular division; MMP, maxillo‐mandibular vascular plexus; NCilA, nasociliary artery; OC, otic capsule; ON, optic nerve; PAA, pharyngeal arch artery; pCBA, primitive carotid‐vertebrobasilar anastomosis; PCW, postconception week; PDOA, primitive dorsal ophthalmic artery; PMA, primitive maxillary artery; PNVP, perineural vascular plexus; POA, primitive olfactory artery; pPA, primitive pulmonary artery; PVOA, primitive ventral ophthalmic artery; SA, stapedial artery; SCh, sympathetic chain; SOD, supraorbital division; TCilA, temporociliary artery; VPA, ventral pharyngeal artery. Ophthalmic artery maturation. All panels are light sheet fluorescence microscopy (LSFM) images of two solvent‐cleared immunostained postconception week (PCW) 7 and PCW8.5 human embryos. All images are pseudo‐colored and segmented structures rendered in maximum intensity projections. Panel (a) is a left inferior oblique view of the stapedial artery (SA) and branches in relation to the chondrocranium and auditory ossicles. The SA is seen to course through the stapes foramen. Panel (b) is a superior view of the skull base in a PCW7 embryo showing the supraorbital division (SOD) of the SA to be anastomotic with the ophthalmic artery at this stage. Panel (c) is a lateral view of the common carotid artery (CCA) bifurcation and branches in a PCW7 embryo rendered with adult nomenclature and color coding and mature nomenclature. At this stage, the maxillo‐mandibular division (MMD) of the SA is anastomotic with the ventral pharyngeal artery (VPA). One of the non‐ophthalmic branches of the SOD is identified as a primitive middle meningeal artery (pMMA). Panel (d) are various views illustrating ophthalmic artery (OA) anatomy in a PCW8.5 embryo. Panel (d) upper left image is a lateral view showing the mature middle meningeal artery (MMA) as being a branch of the external carotid artery (ECA). The dotted box signals the stylomastoid artery shown to course along the facial nerve in the lower left image. Next is a superior view of the OA with major branches visible at this stage. The white arrowheads point toward the persistent relation of the trochlear nerve (CN IV) termination with medial branches of the OA. On the right side, the OA is seen to give rise to a branch coursing along CN IV (white arrows). On the left side, the artery coursing along CN IV (double white arrows) is a branch of the meningohypophyseal trunk (MHT), also giving rise to the inferior hypophyseal arteries (IHyA). The IHyA and superior hypophyseal arteries (SHyA) are demonstrated in the subsequent frontal view on the right. The oblique lateral view in the lower right side of panel (d) shows all identifiable branches of the internal carotid artery to be visible at this stage. ACA, anterior cerebral artery; AChoA, anterior choroidal artery; AComA, anterior communicating artery; CCA, common carotid artery; ChiaA, chiasmatic artery; CN VII, cranial nerve VII (facial nerve); ECA, external carotid artery; HA, hyaloid artery; Hyp., hypophysis; ICA, internal carotid artery; IHyA, inferior hypophyseal artery; ILT, inferolateral trunk; IMA, internal maxillary artery; MACC, median artery of the corpus callosum; MCA, middle cerebral artery; MHT, meningohypophyseal trunk; MMA, middle meningeal artery; MMD, maxillo‐mandibular division; MSA, medial striate artery (recurrent artery of Heubner); NCilA, nasociliary artery; OA, ophthalmic artery; OC, otic capsule; PComA, posterior communicating artery; PCW, postconception week; pMMA, primitive middle meningeal artery; POA, primitive olfactory artery; SA, stapedial artery; SHyA, superior hypophyseal artery; SOD, supraorbital division; STympA, superior tympanic artery; TCilA, temporociliary artery; VPA, ventral pharyngeal artery. Ophthalmic artery origin variants and persistent stapedial artery The account we provide contradicts previous interpretations of several anomalous origins of the OA. A classical variant of the OA is an anomalous origin from the cavernous ICA, entering the orbit via the superior orbital fissure (SOF in figures). This has previously been described as a persistent PDOA (Lasjaunias et al., 2011; Pretterklieber et al., 1994; Uchino, Saito, Takahashi, et al., 2013), a notion contested based on reinterpretation of data provided by Padget (Gregg et al., 2016; Padget, 1948). In this erroneous paradigm, the embryonic PDOA was described as a transient artery, represented in adult anatomy by the inferolateral trunk (Lasjaunias et al., 2011; Pretterklieber et al., 1994; Uchino, Saito, Takahashi, et al., 2013). Contrary to this claim, our data consistently locate the embryonic PDOA (Figure 20a) at a position compatible with that of the definitive OA stem in adult anatomy (Figure 20b). In fact, we describe an interesting case of an anomalous origin of the OA from the cavernous segment of the ICA in Figure 20c (left inset lateral projection DSA; white arrow for the anomalous proximal OA). On 3D DSA, both the origin of this anomalous OA from the cavernous ICA and its course in the superior orbital fissure are confirmed. At close inspection, a second OA is identified, smaller in size, more distally originating from the usual OA position along the clinoid segment of the ICA. This more distal OA is labeled with a white arrowhead in Figure 20c (both segmented maximum intensity data and the lower right inset). This OA courses along the optic nerve in the optic canal and gives rise to the temporociliary, nasiciliary, and central retinal arteries (CRA in figure), the latter being the adult derivative of the HA. In this regard, this more distally located OA can be interpreted as the PDOA (the parent vessel of the embryonic temporociliary and hyaloid arteries), contradicting the notion that the anomalous proximal OA is a persistent PDOA. Also, this small OA gives rise to the nasociliary artery, a derivative of the PVOA. As such, this OA can be interpreted as a PDOA that has annexed the PVOA nasociliary branch, an event depicted in Figure 17f, but that either failed to annex the primitive orbital artery branch of the supraorbital division or subsequently transferred it. To account for the anomalous proximal OA, a possible explanation is that the primitive orbital artery has been transferred to the inferolateral trunk rather than the PDOA (i.e., the true OA stem). Indeed, the non‐ocular branches of the OA could be anastomotic with the inferolateral trunk via the recurrent meningeal artery, a branch of the lacrimal artery coursing through the superior orbital fissure (Figure 20b) or the so‐called deep recurrent OA branch of the inferolateral trunk (Gregg et al., 2016) that is however often depicted as arising from the stem of the OA. The primitive ophthalmic arteries in adult anatomy. Images are either pseudo‐colored and segmented structures rendered in maximum intensity projection (MIP) or unsegmented raw two‐dimensional (2D) and three‐dimensional (3D) digital subtraction angiography (DSA) images (insets except the one in panel (d)). Panel (a) is a light sheet fluorescence microscopy (LSFM) image of a solvent‐cleared immunostained postconception week (PCW) 6.6 human embryo. Panels (b) to (d) represent adult anatomy. Panel (b) upper image is a lateral view of the left ophthalmic artery (OA) and OA branches. The OA is seen to arise from the internal carotid artery, distal to the meningohypophyseal trunk (MHT) and proximal to the position of the posterior communicating artery (PComA, absent in this patient). In the lower left inset, the OA (white arrowhead) is seen to course in the optic canal (double white arrowheads, shown bilaterally). The other insets illustrate the position and course of various OA branches, namely the nasociliary artery (NCilA), temporociliary artery (TCilA), central retinal artery (CRA), anterior ethmoidal artery (AEA) and lacrimal artery (LA). The AEA is seen to give rise to an anterior falcine artery (AFA). Also, the recurrent meningeal artery (RMA) branch of the LA courses back into the cranium through the superior orbital fissure (SOF). Panel (c) illustrates a case of proximal origin of the OA (white arrow) in a patient with a left carotid‐cave aneurysm (unrelated to the anatomical variant depicted herein). The right upper inset shows proximal OA (white arrow) reaching the orbit via the SOF. The lower right inset shows the “true” OA (white arrowhead) coursing in the optic canal. Panel (d) is a case of OA originating from the A1 segment of the anterior cerebral artery. The upper left image is a lateral view of the right internal carotid artery (ICA) 3D DSA. The OA is seen to arise distal to its usual origin on the ICA (compare to panel (b)). The lower left inset is an oblique view of a time of flight (TOF) magnetic resonance imaging (MRI) with 3D DSA segmented structures in MIP. The inset shows the OA to course along the optic nerve (ON) in the optic canal. The image on the right is a superior view showing the origin of the OA from the A1 segment of the ACA (white arrowhead). The origin of the OA from A1 is incompatible with it being a persistent primitive ventral ophthalmic artery (PVOA; see panel (a)). 2D, two‐dimensional; 3D, three‐dimensional; ACA, anterior cerebral artery; AChoA, anterior choroidal artery; AEA, anterior ethmoidal artery; AFA, anterior falcine artery; CRA, central retinal artery; DSA, digital subtraction angiography; HA, hyaloid artery; ICA, internal carotid artery; ICA CaD, internal carotid artery caudal division; LA, lacrimal artery; MCA, middle cerebral artery; MHT, meningohypophyseal trunk; MIP, maximum intensity projection; MRI, magnetic resonance imaging; NCilA, nasociliary artery; OA, ophthalmic artery; ON, optic nerve; P1, P1 pre‐communicating segment of the posterior cerebral artery; PCA, posterior cerebral artery; PComA, posterior communicating artery; PCW, postconception week; PDOA, primitive dorsal ophthalmic artery; POA, primitive olfactory artery; PVOA, primitive ventral ophthalmic artery; RMA, recurrent meningeal artery; SOF, superior orbital fissure; TCilA, temporociliary artery; TOF, time of flight. Exceptionally, the OA can arise from the ACA (Figure 20d). This has been interpreted as a persistence of the PVOA (Li et al., 2011), a notion also contested based on reinterpretation of data provided by Padget (Gregg et al., 2016; Padget, 1948). Indeed, the PVOA originates from a vascular segment of the POA/ICA cranial division destined to become the supra‐clinoid segment of the ICA (Figure 17). As such, the PVOA originates distal to the AChoA and proximal to the MCA–ACA bifurcation. Alternative speculative interpretations of the OA originating from the ACA could implicate the anastomosis observed between the OA branches and the POA (Figures 6g, 16d and 19d) or the anastomotic ring of the perineural vascular plexus between the POA and primitive maxillary artery around the stem of the optic vesicle. In contrast, a persistent PVOA would most likely result in an OA originating from the distal ICA as previously proposed (Gregg et al., 2016). A hypothesis emerging from our observations that we are however unable to ascertain is that the POA may be transferred to the OA (Figures 6g, 16d and 19d) and may constitute the origin of the OA ethmoidal branches (Figure 20b). To explain the OA or MMA originating from the BA, previous authors have cited Hogg and invoked a possible perineural vascular plexus along the trigeminal nerve and ganglion, described as receiving branches both from the developing BA and SA precursors (Hogg et al., 1972; Kumar & Mishra, 2012). This is reminiscent of the data provided herein (Section 3.9 “The basilar artery branches”). Unfortunately, this explanation appears to result from misquotation or misreading of the original report. Indeed, Hogg, describing rat embryo development, reported finding an anastomosis between the BA and SA along the facial nerve and geniculate ganglion rather than the trigeminal nerve (Hogg et al., 1972). As stated earlier, the trigeminal nerve artery anastomosing with the MMA or OA (Figure 16c,d) is the most likely explanation for a BA origin of either the MMA (Figure 16e) or the OA (Sade et al., 2004), respectively. Another exceptionally rare OA variant can also arise from the position of the meningohypophyseal trunk (Uchino, Saito, Kurita, & Ishihara, 2013). The existence of a trochlear artery arising from either the OA or the meningohypophyseal trunk (Figure 19d) provides a potential framework for interpreting meningohypophyseal trunk origin of the OA and vice versa. Anatomical variants related to the MMA and OA are readily interpretable given data provided herein (Figures 19 and 21a). The OA can arise from the MMA rather than the ICA (Figure 21b). The most likely embryological scenario explaining this variant is a transfer of the PDOA branches (after annexation of the nasociliary branch of the PVOA) to the primitive orbital artery that persists as a branch of the MMA. The proximal PDOA in this scenario is lost. More commonly, the MMA originates from the OA (Figure 21c). The most likely scenario explaining this variant is the persistence of the proximal stem of the primitive orbital artery (labeled as the OA‐MMA anastomosis in Figure 21c) and loss of the proximal MMD and supraorbital division segments usually constituting the proximal MMA in adult anatomy. Finally, the MMA can originate from the ICA in a configuration that could represent a partial persistence of the SA (Figure 21d). Two patients with an ICA origin of the MMA are shown in Figure 21d. Unfortunately, in neither of these patients was 3D imaging available to ascertain a middle ear course of the MMA through the stapes. Also, in patient B (lower insets), the MMA originated from a position proximal to the expected origin of the hypothetical carotico‐tympanic branch of the ICA. Patient B may alternatively represent a pharyngo‐tympano‐stapedial artery that has been previously speculated to implicate both the SA and an inferior tympanic branch of the ascending pharyngeal artery (Baltsavias, Kumar, & Valavanis, 2012; Sheriff et al., 2025). Variants of the middle meningeal and ophthalmic arteries. Images are either pseudo‐colored and segmented structures rendered in maximum intensity projections or unsegmented raw two‐dimensional (2D) digital subtraction angiography (DSA) and three‐dimensional (3D) non‐contrast computed tomography (NCCT) images. Panel (a) is a lateral view segmented light sheet fluorescence microscopy (LSFM) image of a solvent‐cleared immunostained postconception week (PCW) 7 human embryo. Panel (a) shows the anatomy of the stapedial artery (SA) with its maxillo‐mandibular division (MMD) and supraorbital division (SOD) branches anastomosing with the ventral pharyngeal artery (VPA) and ophthalmic artery (OA), respectively. The same image is reproduced in panels (b) to (d) to illustrate alternative remodeling scenarios leading to anatomical variants implicating the middle meningeal artery (MMA) and OA. Panel (b) is an illustration of the OA originating from the MMA. The upper right inset is a lateral projection 2D DSA of the internal carotid artery (ICA) not giving rise to the OA. The lower right inset shows the external carotid artery (ECA) giving rise to the MMA (white arrowheads) in turn giving rise to the OA (double white arrows). The choroidal blush (triple white arrows) signals the opacification of choroidal arteries (and presumably the central retinal artery). Panel (c) is an illustration of the MMA originating from the OA. The upper left inset is a lateral projection 2D DSA of the ECA whereas the MMA is not opacified. The upper right inset is an ICA DSA showing the MMA (white arrowheads) originating from the OA. The NCCT (from another patient) serves to illustrate the absence of the right foramen spinosum (black circle on the right) when the MMA does not arise from the internal maxillary artery. On the left, the white arrow points to a patent foramen spinosum. Panel (d) is an illustration of partial SA persistence giving rise to a MMA. 2D, two‐dimensional; 3D, three‐dimensional; anast., anastomosis; CCA, common carotid artery; DSA, digital subtraction angiography; ECA, external carotid artery; HA, hyaloid artery; ICA, Internal carotid artery; MMA, middle meningeal artery; MMD, maxillo‐mandibular division; NCCT, non‐contrast computed tomography; NCilA, nasociliary artery; OA, ophthalmic artery; PA, postero‐anterior (projection); PCW, postconception week; pMMA, primitive middle meningeal artery; SA, stapedial artery; SOD, supraorbital division; TCilA, temporociliary artery; VPA, ventral pharyngeal artery. Growth and boundary shifts of the carotid arteries As previously mentioned (Section 3.1 “Closure of the circle of Willis”), the definitions of arterial segments' boundaries shift during ontogeny. Regarding the ICA, both distal (Section 3.1 “Closure of the circle of Willis”, Figure 4) and proximal (Section 3.10 “The ophthalmic artery,” Figure 18) boundaries shift at various embryological stages. Herein, we aim to provide a comprehensive account of the stepwise extension of the ICA boundaries from the primitive ICA (prICA in figures) to the mature ICA configuration. As stated earlier, the youngest embryo in this ontogenic series is a PCW5 sample (Figure 2). At PCW5, the pharyngeal arch arteries (PAAs) and future supra‐aortic arteries have already undergone early remodeling. Therefore, to give a complete account of ICA maturation, we provide schematic drawings of the PAAs at Carnegie stages (CS in figures) 11 and 12 (both earlier than PCW5; Figure 22a) redrawn from previous literature (de Bakker et al., 2016). At Carnegie stage 11, the aortic sac gives rise to bilaterally symmetrical mandibular pharyngeal arch arteries (PAA I) that extend into paired dorsal aortas. The primitive ICAs arise as branches from the cranial apex of the dorsal aortas at this stage. At Carnegie stage 12, the aortic sac and bilateral aortic horns give rise to the mandibular (PAA I), hyoid (PAA II) and carotid (PAA III) pharyngeal arch arteries. The primitive ICA cranially terminates as cranial and caudal divisions. At PCW5 (Figure 22b), the 4th and nascent last PAAs are conspicuous. On the other hand, the first (mandibular) and second (hyoid) PAAs have regressed and are only visible as dwindling remnants. An indirect consequence is that a segment of the paired dorsal aortas is now incorporated into the definition of the ICA (three white arrowheads in Figure 22a,b), that is, the segment between the 1st (mandibular) and 3rd (carotid) PAAs. At PCW5.4, the segment of the dorsal aortas between the dorsal aspects of 3rd (carotid) and 4th (aortic) PAAs is beginning to regress and is relabeled the ductus caroticus (double white arrows). At PCW5.7, regression of the ductus caroticus is near complete. With the separation of the 3rd (carotid) PAA from the dorsal aorta resulting from ductus caroticus regression, the 3rd (carotid) PAA is further incorporated into the definition of the ICA (double white arrowheads). The proximal boundary of the ICA is now the bifurcation with the VPA, future stem of the ECA. Proximal to the VPA‐ICA bifurcation is the CCA. Notice that because the VPA originates at the most ventral aspect of the 3rd (carotid) PAA at PCW5, the CCA, by definition proximal to this point, can be interpreted as an elongated aortic horn rather than a segment of the 3rd (carotid) PAA. An alternative less likely scenario regarding the origin of the CCA is VPA migration along the ventro‐dorsal axis of the 3rd (carotid) PAA. Another consequence of the regression of the ductus caroticus is that the 4th (aortic) PAA is now relabeled, on the left, as the aortic arch distal to the origin of the left CCA (white star). At PCW6.1, because of the growth of the MCA and ACA, the proximal segment of the POA is relabeled as the terminal segment of the ICA (three white arrows), a process previously described in Section 3.1 “Closure of the circle of Willis” (Figure 4). Embryonic to mature internal carotid artery. All embryonic images except panel (a) are light sheet fluorescence microscopy (LSFM) images of solvent‐cleared immunostained postconception week (PCW) 5–8.5 human embryos. Adult images are 3D digital subtraction angiography (DSA) images. With the exception of panel (a) images that are schematic drawings, all other panels represent either embryonic or adult imaging data. Images in panels (b–d) are pseudo‐colored and segmented structures rendered in maximum intensity projections (MIP). Panel (e) images are either pseudo‐colored segmented structures rendered in MIP (left image) and “Blend” mode in Imaris (opaque structures; lower right images) or a noninjected cone beam computed tomography (CBCT) in axial reconstruction (upper right inset). Panel (a) images are schematic drawings of pharyngeal arch arteries at Carnegie stages 11 and 12 (earlier than PCW5) redrawn from (de Bakker et al., 2016). Panel (b) images are lateral views of the internal carotid artery (ICA) shifting boundaries in PCW5 to PCW6.1 human embryos. Panel (c) images are lateral views of the carotid arteries (internal and external) in PCW5.7–PCW8.5 embryos and an adult patient illustrating differential growth of distinct vascular segments during development. Panel (d) images are two views of the carotid arteries in a PCW5.7 embryo showing an anastomosis between the proximal internal carotid artery (ICA) and the stapedial artery (SA). Panel (e) are medical images of a patient with a left aberrant ICA (courtesy of Professor Emmanuel Houdart, Lariboisière Hospital, Paris). ACA, anterior cerebral artery; AChoA, anterior choroidal artery; Ao, aorta; CBCT, cone beam computed tomography; CCA, common carotid artery; CS, carnegie stage; DC, ductus caroticus; DSA, digital subtraction angiography; ECA, external carotid artery; HA, hyaloid artery; ICA, internal carotid artery; ICA CaD, internal carotid artery caudal division; MCA, middle cerebral artery; MMA, middle meningeal artery; MMD, maxillo‐mandibular division; MMP, maxillo‐mandibular vascular plexus; OA, ophthalmic artery; PAA, pharyngeal arch artery; pACA, primitive anterior cerebral artery; pAChoA, primitive anterior choroidal artery; PComA, posterior communicating artery; PDOA, primitive dorsal ophthalmic artery; PMA, primitive maxillary artery; POA, primitive olfactory artery; pPA, primitive pulmonary artery; prICA, primitive internal carotid artery; PVOA, primitive ventral ophthalmic artery; SA, stapedial artery; SOD, supraorbital division; TCilA, temporociliary artery; VPA, ventral pharyngeal artery. Along with the relabeling of arterial segments depicted in Figure 22b, these various segments appear to undergo distinct growth regimes, a likely consequence of both elongation of the neck and progressive descent of the heart in the thorax. Fixed landmarks are chosen to facilitate appreciation of differential growth between PCW6.1 and adult age in Figure 22c: (1) the origin of the CCA, (2) the ICA‐VPA/ECA bifurcation, (3) the OA, and (4) the posterior communicating artery (remnant of the ICA caudal division). As already seen between PCW5 and PCW6.1 (Figure 22b), the CCA continues to grow disproportionately beyond PCW6.1 when compared with the OA‐PComA intracranial segment (Figure 22c). On the right side of Figure 22c, the carotid artery of a PCW5.7 embryo is shown with an additional landmark labeled: the origin of the SA. A corresponding point is provided in adult anatomy despite the fact that the SA is a transient embryological structure absent as a macroscopic structure in adult anatomy. This putative position of the SA in the adult anatomy is interpreted based on observations made in patients with a rare anatomical variant: the aberrant carotid artery (Hitier et al., 2013). In such a configuration, the proximal ICA is thought to be replaced by an enlarged inferior tympanic artery branch of the ascending pharyngeal artery, entering the skull through the tympanic canal and joining the horizontal intrapetrous part of the carotid artery via the proximal SA (Hitier et al., 2013). A discussion of the aberrant ICA embryology is provided below. Figure 22d shows frontal and left anterolateral views of a PCW5.7 embryo in which a vessel, originating from the most dorsal aspect of the 3rd PAA remnant (incorporated in the definition of the ICA), is seen to bilaterally anastomose with the SA. We are unable to definitively identify this vessel but speculate that it is related to the ascending pharyngeal artery embryology because the latter artery can arise from the ICA in adults (Hacein‐Bey et al., 2002). The existence of this anastomosis serves as a basis to interpret aberrant ICA anatomy. An illustrative case is provided in Figure 22e. In this adult patient, an aberrant ICA courses more laterally than would be expected in standard anatomy and is associated with an ICA origin of the MMA, interpreted as a form of persistent SA (Hitier et al., 2013). The aberrant ICA passes through the middle ear as seen in the upper right inset (Figure 22e). The lower right part of the panel is an interpretation of the aberrant ICA anatomy based on observations in the PCW5.7 embryo provided in Figure 22d. In the aberrant ICA configuration, the proximal ICA is absent (triple white arrows, Figure 22d,e). The extracranial aberrant ICA is interpreted to be the ICA–SA anastomosis (light green, putative ascending pharyngeal artery), followed by the proximal SA in the middle ear (dark pink). At the point of transition between the anastomosis and the SA component of the aberrant ICA, the SA continues as an MMA, proven to be the SA trunk as it passes through the stapes (lower right inset). For a discussion of the origin of the MMA as a branch of the supraorbital division of the SA, refer to Sections 3.10 (“The ophthalmic artery”) and 3.11 (“Ophthalmic artery origin variants and persistent stapedial artery”). These various observations (Figure 22d,e) serve as a basis to identify the hypothetical position of the embryonic SA, and consequently the 2nd (hyoid) pharyngeal arch artery (PAA), along the length of the definitive ICA anatomy. Regarding the 1st (mandibular) PAA, it has been proposed to be represented in adult anatomy in the form of the vidian artery (Takeuchi et al., 2005), a hypothesis we are unable to uphold or contradict based on our data. DISCUSSION In this work, we provide a detailed account of the development of human cerebral arteries during the first trimester of gestation using a combination of tissue clearing, whole‐mount immunostaining, 3D imaging and virtual reality‐based anatomically informed segmentation and labeling. We further expanded virtual dissections to pediatric and adult radiological anatomy (3D DSA datasets) to bridge the gap between embryonic and postnatal anatomy. In the process, we clarify key developmental events, provide new insights into previously open questions and report original observations that remain unexplained and may motivate future work. Classifying our environment into categories is a basic feature of human cognition (Minda et al., 2024). The world as perceived by the human senses is continuous. Yet, people often perceive it as a succession of discrete entities (Zacks et al., 2007). An event for instance is defined as “a segment of time at a given location that is perceived by an observer to have a beginning and an end” (Zacks et al., 2001). Visible light is a continuum of wavelengths but is distorted into nonlinearly perceived perceptual categories we call colors (Huette, 2016). Anatomy does not deviate from the rule, nowhere more so than in the description of the vasculature. The arterial tree is a continuum dividing into ever‐smaller branches. To make sense of the arterial tree, anatomists have long individualized and named segments bounded by defined anatomical landmarks, changes in direction, variations in caliber or specific bifurcations. The exercise of naming arterial segments is as old as anatomical science itself as illustrated by the designation of the trunk of the arterial tree as the aorta (Antoniou et al., 2011). Implicit in the use we make of many categorical entities is the notion that their boundaries are universal and fixed. This, however, does not stand up to scrutiny in many instances. Going back to colors, there is evidence cross‐linguistic variation in color categorization. Differences in color categorization between languages may be caused by differences in overall usefulness of color to different cultures (Gibson et al., 2017; Roberson et al., 2005). With regard to vascular anatomy, beyond understandable disagreements between anatomists, boundaries of arteries or arterial segments shift during both ontogeny and phylogeny. The embryonic dorsal aorta is an artery coursing along the entire length of the ventral aspect of the spine. Yet, the adult aorta is said to terminate at the level of the origin of the common iliac arteries, with the post‐iliac sacro‐coccygeal aorta in effect relabeled as the adult median sacral artery because of the major difference in size. Descriptive embryology is a great way of making sense of adult anatomy and anatomical variability. Beyond this obvious fact, we believe explicitly comparing embryological and adult anatomical descriptions and nomenclature sheds new light on both. The process of segmenting embryonic arterial datasets has forced us to explicitly define arterial segments. In the process, it has come to our attention that loose definitions are responsible for cascading interpretive consequences that may lead to inaccurate statements in the literature. Vessels are often designated in illustrations by pointing to their middle section, avoiding explicit definitions of boundaries with adjacent segments. Together with lack of firsthand access to original embryological material and intermingled use of adult and embryonic nomenclatures, this may lead to inaccurate descriptions and ultimately to conclusions that may not be supported by the underlying anatomy. For instance, the adult ACA has a fairly consensual definition. Few would argue against its description as being the medial branch of the ICA termination, with all downstream branches and bifurcations (with a few exceptions like the anterior communicating artery or recurrent artery of Heubner). In the embryological literature however, several authors alternatively use the designations POA, ICA cranial division and ACA interchangeably when describing cerebrovascular arteriogenesis (Klostranec & Krings, 2022; Raybaud, 2010). This leads to a number of problematic statements contradicted by our data including that the MCA grows as a lateral branch of the ACA (Raybaud, 2010) or that the PVOA is a branch of the ACA (Baltsavias, Türk, & Valavanis, 2012; Belotti et al., 2016). Likewise, the caudal division of the ICA is sometimes confused with an insufficiently defined mesencephalic artery (Raybaud, 2010), leading to the erroneous statement that dorsal branches of the mesencephalic arteries expand to form the cortical territory of the PCA (Raybaud, 2010). Padget's seminal paper published in 1948 (Padget, 1948) was truly disruptive in the sense that subsequent research was more likely to cite Padget than the earlier sources on which her work drew, in effect making previous research obsolete (Kozlov, 2023). The quality of the description provided remains remarkable even by current standards. However, the work, hindered by the techniques of its time, is not devoid of limitations. For example, Padget fell short of describing the PCA or the embryonic trigeminal nerve artery and provided a confusing description of the OA, invoking a definitive OA stem described as distinct from the PDOA. The current work clarifies these points. Our own work is not devoid of limitations itself. Indeed, the account provided is based on a limited number of human samples. A larger sample collection could have provided a better temporal resolution of embryonic remodeling events. Also, a larger collection would have provided a clearer distinction between developmental events leading to textbook anatomy as opposed to anatomical variations. Another limitation of this work is the lack of embryonic samples prior to PCW5, hindering description of early developmental events. We are unable to uphold or disprove the observation made by Moffat in rat embryos that the medial aspect of the posterior part of the telencephalic vesicles is primitively fed by the primitive anterior choroidal artery (pAChoA) before being taken over by the primitive posterior choroidal artery (pPChoA) (Moffat, 1961). Also, developmental events during the fetal period are beyond reach of the methodologies employed given sample size. We therefore lack direct evidence for some of the hypotheses generated by this work. For instance, the interpretation of the PCA as representing the pPChoA appears intrinsically coherent and external validation is provided by a comparable description during later rat development (Moffat, 1961) but direct fetal observations of the latter stages of this process are lacking. Further exploration of fetal anatomy could nonetheless reinforce this finding. Other hypothetical hypotheses merit further investigation including the suggestion that the CoA derives from the embryonic mesencephalic artery, the MPCA may derive from the embryonic DiA or that the POA branches may be annexed by the OA and translate into ethmoidal arteries. Finally, this work is descriptive in essence, providing a qualitative account of the developmental events observed in the sample collection. A quantitative analysis of the differential growth of arterial segments, geometrical properties and bifurcation patterns could provide new insights into the transition between embryonic and mature anatomical patterns of the cerebral arteries. CONCLUSIONS This work provides a detailed account of the embryonic development of human cerebral arteries using 3D imaging of optically cleared human embryos. In the process, key developmental events are clarified and new insights into previously open questions are provided. AUTHOR CONTRIBUTIONS Eimad Shotar: Concept/design, acquisition of data, data analysis/interpretation, drafting of the manuscript, critical revision of the manuscript, and approval of the article. Raphaël Blain: Acquisition of data, critical revision of the manuscript and approval of the article. Aymeric Nadjem: Data analysis/interpretation, critical revision of the manuscript, and approval of the article. Nermine Saidi: Acquisition of data, critical revision of the manuscript, and approval of the article. Megumi Inoue: Data analysis/interpretation, critical revision of the manuscript and approval of the article. Daniel Smith‐Paredes: Data analysis/interpretation, critical revision of the manuscript and approval of the article. Miccaella Vergara: Data analysis/interpretation, critical revision of the manuscript and approval of the article. Armelle Lesaunier: Acquisition of data, critical revision of the manuscript and approval of the article. Quentin Rappeneau: Acquisition of data, critical revision of the manuscript and approval of the article. Joséphine Blevinal: Acquisition of data, critical revision of the manuscript and approval of the article. Gérard Couly: Concept/design, critical revision of the manuscript and approval of the article. Julien Burel: Acquisition of data, critical revision of the manuscript and approval of the article. Julien Allard: Acquisition of data, critical revision of the manuscript and approval of the article. Kevin Premat: Acquisition of data, critical revision of the manuscript and approval of the article. Stéphanie Lenck: Acquisition of data, critical revision of the manuscript and approval of the article. Pierre‐Marie Chiaroni: Acquisition of data, critical revision of the manuscript and approval of the article. Erwah Kalsoum: Acquisition of data, critical revision of the manuscript and approval of the article. Maryne Toupin: Acquisition of data, critical revision of the manuscript and approval of the article. Séverine Mazaud‐Guittot: Acquisition of data, critical revision of the manuscript and approval of the article. Diane Rebourcet: Acquisition of data, critical revision of the manuscript and approval of the article. Paolo Giacobini: Acquisition of data, critical revision of the manuscript and approval of the article. Olivier Naggara: Acquisition of data, critical revision of the manuscript and approval of the article. Hugo Gortais: Acquisition of data, critical revision of the manuscript and approval of the article. Grégoire Boulouis: Acquisition of data, critical revision of the manuscript and approval of the article. Alexis Guédon: acquisition of data, critical revision of the manuscript and approval of the article. Philippe Gailloud: Critical revision of the manuscript and approval of the article. Frédéric Glicenstein: Acquisition of data, critical revision of the manuscript and approval of the article. Anaïs Favre: Acquisition of data, critical revision of the manuscript and approval of the article. Frédéric Clarençon: Concept/design, acquisition of data, critical revision of the manuscript and approval of the article. Yorick Gitton: Concept/design, acquisition of data, critical revision of the manuscript, and approval of the article. Alain Chédotal: Concept/design, funding, critical revision of the manuscript and approval of the article. Supporting information Table S1. List of samples. Table S2. List of antibodies. Video S1. Example of a segmented adult dataset. Data S1. List of Abbreviations.