Characteristics of the femoral tunnel of anatomical and isometric single bundle anterior cruciate ligament reconstruction: a modeling analysis based on quadrant method and anatomical landmarks Authors: Li Yin, Dongfa Liao, Qingyun Xie, Jinbiao Liu, Bing Deng Published: 2024 ABSTRACT AbstractPurposeTo investigate the anatomical features of the femoral tunnel in anatomical and isometric single-bundle ACL reconstruction.MethodThirty-two 3-dimensional knee models were reconstructed based on CT scan (average age: 26.5 ± 6.7 years, 18 males and 14 females, 17 left and 15 right). Multiple anatomical landmarks were identified. Virtual femoral tunnels were created at the deep and high portion of ACL footprint, close to the lateral intercondylar ridge to achieve best anatomy and isometry, simulating an anteromedial portal reconstruction. Anatomical features of the femoral tunnels were analyzed. The position of the femoral tunnel was quantified by the distance to anatomical landmarks and using quadrant methods. The spatial angles, length and outer opening of the femoral tunnels were also evaluated.ResultsAcceptable tunnels were created in all models. The center of femoral tunnel was slightly higher than the apex of deep cartilage, near the deep one-third point across the shallow-deep dimension of the lateral femoral condyle. Using the quadrant method, the tunnel was located at 28.4% ± 2.2% and 22.2% ± 3.6%, parallel and perpendicular to the Blumensaat line, respectively. The spatial angles of the tunnel were 40°, 33.5° ± 4.1° and 38.2° ± 4.4° on the sagittal, transverse, and coronal planes, respectively. The average tunnel length was 34.8 mm ± 3.8 mm. The outer opening of the tunnels was located at the posterior one-third of the femoral metaphysis.ConclusionThe anatomical and isometric positioning of the femoral tunnel can be achieved through anteromedial portal with satisfied tunnel characteristics. The apex of deep cartilage may be used as an anatomical reference for tunnel positioning. When drilled at appropriate orientation, favorable tunnel length, integrity and position of the outer opening can be obtained.Keywords: Anatomical isometric ACL reconstruction, Femoral tunnel, Quadrant method, Anatomical landmarks, Model simulation Introduction The femoral tunnel is critical in anterior cruciate ligament (ACL) reconstruction. A mispositioned tunnel may come with multiple adverse changes in the behavior of the ACL graft, including high-tensioned or slacked graft fiber, decreased range of knee motion, instability, graft impingement, short tunnel with bad tendon-tunnel healing, broken posterior cortex with poor fixation strength and so on. All these matters may finally lead to graft failure and poor clinical outcomes [1–3]. Ideas of ACL femoral tunnel positioning went through progressive evolution over the past few decades. The “transtibial” approach had been a popular technique since the late 1980s, but later criticized for producing a vertically aligned ACL graft that lacks rotational stability [4]. The concept of anatomical reconstruction was popularized in the 21st century, aiming to restore the native ACL footprint [5]. However, the ACL footprint spans a large part of intercondylar notch, surpassing the area covered by the graft [6, 7], and debates regarding a more exact position within the anatomical footprint have raised ever since. Various positions have been subsequentially advocated, including the geometric center of the entire footprint [8–12], anteromedial (AM) bundle insertion [13–16] and posterolateral (PL) bundle insertion [17, 18]. In recent years, the integration of traditional biomechanical principles and advancement in ACL microanatomy has brought new perspective into femoral tunnel positioning. Specifically, it is proposed to locate the femoral tunnel within the native ACL footprint, attached to the direct fiber insertion, and eccentrically placed to grant best graft isometry during knee movements [19, 20]. Although it offers general principle on where to position a relatively small tunnel within the larger femoral footprint, the exact position is still vaguely described. Additionally, multiple tunnel characteristics associated with this proposed position, such as the posterior wall integrity, tunnel angle and length, position of the tunnel outer opening are also unclearly addressed. The aim of this study is to investigate the anatomical features of the femoral tunnel in anatomical and isometric single-bundle ACL reconstruction. Utilizing model simulation, virtual femoral tunnels that adhere to the principle of anatomy, isometry and the direct fiber proximity were created, replicating the actual anteromedial portal ACL reconstruction. The position and quality of the femoral tunnels were subsequentially analyzed and evaluated. The outcome of the current study may provide technical references for the contemporary idea of ACL reconstruction. Methods Subjects and imaging Computer tomography (CT) data of 32 knees were selected from the bone and joint imaging database in our department. The CT scans were performed using a 16-slice spiral scanner (SOMATOM, Siemens, Munich, Germany), with a slice interval of 0.6 mm and a matrix of 512*512. The average age of the subjects was 26.5 ± 6.7 years (range: 19 to 41 years), 18 males and 14 females, 17 left and 15 right knees. Knees were excluded if they exhibited lower limb deformity, degeneration, or a history of injury or surgery. Informed consents have been previously obtained from all individuals in the imaging database, who agreed to use their data for scientific research and publication. 3D modeling The femur models were reconstructed using an interactive medical image segmentation software (Mimics, Version 14.1, Materialise, Leuven, Belgium). Models were aligned to an initial position: the long axis of the femur shaft was horizontally placed, and the lateral and medial condyles were aligned to achieve maximal superposition. Subsequently, the femur model was bisected into lateral and medial parts along the Whiteside’s line, allowing for a clear display of the medial wall of the lateral condyle for later simulation and measurements. To describe locations within the intercondylar region, terminology such as “shallow (front) / deep (back)”, “high / low” was used. “Shallow (front) / deep (back)” corresponded to distal / proximal while “high / low” corresponded to anterior / posterior in standard anatomical terminology. This height and depth description of distal femur intuitively correlates arthroscopic view of the femoral notch and has been widely adopted in previous studies [11, 21, 22] (Fig. 1). The medial wall of lateral femoral condyle oriented in accordance with arthroscopy. The red and green dots represent the apex of deep cartilage (ADC) and the intercondylar roof, respectively. The black, red and green dotted lines represent the lateral intercondylar ridge (LIR), cartilage margin and the deep condylar edge (DCE), respectively. The red and blue areas represent the ACL femoral footprint of anteromedial (AM) bundle and posterolateral (PL) bundle, respectively Anatomical landmarks Several anatomical landmarks were carefully identified and marked in the intercondylar notch based on their morphology. The intercondylar roof was defined as the highest point of the intercondylar notch. The lateral intercondylar ridge (LIR) was identified along the medial wall of lateral condyle, extending from the intercondylar roof to cartilage margin [23, 24]. The apex of deep cartilage (ADC) was identified, which corresponds to the most deep and high rim of the cartilage of lateral femoral condyle at intercondylar notch. The ADC has recently been proposed as a reliable reference for femoral tunnel placement [12, 25, 26]. The deep condylar edge (DCE) was defined as the deepest edge of the lateral femoral condyle. These anatomical landmarks as well as the entire cartilage margin of the lateral femoral condyle were later used for positional evaluation of the femoral tunnel (Fig. 1). The footprint and centers of the ACL and its two bundles were also estimated and marked on the lateral femoral condyle based on published data [10, 27] (Fig. 1). Generally, the ACL borders the LIR in the front and top and extends backward and inferiorly to the cartilage margin, with an average distance of 3.0–3.5 mm. Using the quadrant method, the centers of the ACL, AM bundle and PL bundle are located at approximately 28.5%, 21.5%, and 32% from back to front along the Blumensaat line; and at 35.2%, 23.1%, and 48.8% from high to low perpendicular to the Blumensaat line, respectively. Femoral tunnel simulation Virtual femoral tunnels were created on the femur models, complying with the following rules: 1.Anatomy: the inner opening of the tunnel should be contained within the femoral footprint defined above [10, 27]. 2. Direct fiber: the attachment of direct fiber is located at the high portion of ACL footprint, close to the LIR, thus the tunnel should be placed eccentrically into that area, in close relationship to the LIR [7, 28]. 3.Isometry: the most isometric portion of the ACL footprint is its high and deep region, or high region of the AM bundle. This region is closest to the over-the-top point but does not superimpose it [15, 29]. The tunnel should mainly cover this area. 4.Tunnel size: the diameter was set to 8 mm, accommodating most patients using hamstring grafts. Moreover, a graft size less than 8 mm may be associated with an increased revision rate [30]. 5.Tunnel integrity: “posterior wall blowout” should be strictly avoided during tunnel creation, thus the thickness of the remaining cortex deep to the tunnel was set to no smaller than 2 mm. 6.Tunnel orientation: In the sagittal plane, the tunnel was set to 40° in relation to the longitudinal axis of the femoral shaft, corresponding to roughly 120° of knee hyper flexion intraoperatively, achievable by most patients [31]. In the transverse plane, the tunnel was placed so that the virtual drill was oriented as close to the medial condyle as possible, to yield more room for the posterior wall of the tunnel. The position and orientation of the virtual tunnel were repeatedly adjusted until all requirements could be best satisfied. Evaluation of the position of the femoral tunnel. After the femoral tunnel was created, its position was evaluated in relation to the anatomical landmarks mentioned above. Specifically, in the high-low direction, the distances of the tunnel center to the ADC, intercondylar roof, and low cartilage margin were measured. In the shallow-deep direction, the distances to the DCE and the front cartilage margin were measured. Additionally, the position of the tunnel center was analyzed by the quadrant method (Fig. 2). Quantification of the position of the femoral tunnel. (a) Distance to anatomical landmarks. (b) Quadrant method. The red, green and gray dots represent the apex of deep cartilage (ADC), the intercondylar roof and the center of femoral tunnel, respectively. The black, red and green dotted lines represent the lateral intercondylar ridge (LIR), the cartilage margin and the deep condylar edge (DCE), respectively. The red area and the gray area represent the femoral footprint of ACL and the inner opening of the femoral tunnel, respectively. L1 and L2 represent the distances to the DCE and the front cartilage margin, respectively. h1, h2 and h3 represent the distances to the ADC, intercondylar ridge and low cartilage margin, respectively Evaluation of the quality of femoral tunnel A few parameters were used to evaluate the quality of femoral tunnel. 1. Tunnel length. 2.Tunnel angles. 3. The position of tunnel outer opening on the lateral femoral condyle, which was quantified by: (a) distance to the lateral femoral epicondyle along the proximodistal direction; (b) distance to the posterior and anterior femoral cortex along the anteroposterior direction (Fig. 3). Spatial parameters of the femoral tunnel. (a) The angle with the femoral shaft on the sagittal plane (θ1). (b) The angle with the condylar tangent on the transverse plane (θ2). (c) The angle with the condylar tangent on the coronal plane (θ3). (d) The position of the outer opening of the femoral tunnel. L1, L2 and L3 represent the distance to lateral femoral epicondyle along proximodistal direction, distance to posterior and anterior femoral cortex along the anteroposterior direction, respectively Results Position of the femoral tunnel The desired femoral tunnel was in the high and deep region of the ACL femoral footprint, or in other words, in the high region of the AM bundle. In the high-low direction, the center of the tunnel was positioned midway between the low cartilage margin and the intercondylar roof, but closer to the latter. The ADC was slightly lower than the tunnel center, being the closest anatomical landmark in the high-low direction. In the shallow-deep direction, the center was located at nearly deep one-third across the shallow-deep dimension of the lateral femoral condyle. When using the quadrant method, the tunnel center was close to the AM bundle center, but slightly higher and shallower (Table 1). Table 1The location of the center of the femoral tunnel in relation to anatomical landmarks and according to the quadrant method. Aberrations: ADC: apex of deep cartilage, DCE: deep condylar edge. The tunnel center is lower than the intercondylar roof, higher than the low cartilage margin and ADC, deeper than the front cartilage margin and shallower than the DCE. For the quadrant method, the “parallel” and “perpendicular” were referred to the Blumensaat line. The origin (0% and 0%) was defined as the high and deep point of the quadrantHigh-low distance (mm)Deep-shallow distance (mm)Quadrant (%)Intercondylar roofLow cartilage marginADCDCEFront cartilage marginParallelPerpendicularMean ± STD.5.0 ± 0.87.6 ± 1.42.6 ± 1.27.2 ± 1.315.5 ± 1.728.4 ± 2.222.2 ± 3.6Range3.6–6.85.3–10.50.2–5.45.0–9.710.9–18.223.5–32.913.3–29.4 The location of the center of the femoral tunnel in relation to anatomical landmarks and according to the quadrant method. Aberrations: ADC: apex of deep cartilage, DCE: deep condylar edge. The tunnel center is lower than the intercondylar roof, higher than the low cartilage margin and ADC, deeper than the front cartilage margin and shallower than the DCE. For the quadrant method, the “parallel” and “perpendicular” were referred to the Blumensaat line. The origin (0% and 0%) was defined as the high and deep point of the quadrant Quality of the femoral tunnel All simulated tunnels yielded a posterior wall with a thickness no smaller than 2 mm. The tunnels exited at the metaphysis, anterior and proximal to the lateral femoral epicondyle, and close to the posterior cortex. The characteristics of the tunnels were summarized in Table 2. Table 2The characteristics of simulated femoral tunnels. The tunnel angles were measured with respect to the femoral shaft on the sagittal plane and the tangent lines of femoral condyles on the transverse and coronal planes. For the outer opening of tunnels, the distance to the lateral femoral epicondyle was measured along the proximodistal direction. The distances to the anterior and posterior cortex were measured along the anteroposterior direction. The posterior distance percentage was defined as the ratio of the distance to the posterior cortex to the anteroposterior dimension of femoral metaphysis at the level of tunnel outer openingTunnel angles (°)Tunnel length (mm)Position of the tunnel outer openingSagittalTransverseCoronalDistance to lateral femoral condyle (mm)Distance to posterior cortex (mm)Distance to anterior cortex (mm)Posterior distance percentage (%)Mean ± STD.40 ± 033.5 ± 4.138.2 ± 4.434.8 ± 3.819.9 ± 3.911.2 ± 2.329.4 ± 3.927.8 ± 5.6Range4026.8–40.331–45.328–4514.3–29.87.0–16.122.6–37.319.4–39.1 The characteristics of simulated femoral tunnels. The tunnel angles were measured with respect to the femoral shaft on the sagittal plane and the tangent lines of femoral condyles on the transverse and coronal planes. For the outer opening of tunnels, the distance to the lateral femoral epicondyle was measured along the proximodistal direction. The distances to the anterior and posterior cortex were measured along the anteroposterior direction. The posterior distance percentage was defined as the ratio of the distance to the posterior cortex to the anteroposterior dimension of femoral metaphysis at the level of tunnel outer opening Discussion In this study, we utilized 3D models based on CT scans to simulate the creation of femoral tunnels for anteromedial portal ACL reconstruction. The use of 3D-CT model simulation had several advantages, including clearer visualization of bony landmarks, more precise planning and control of tunnel angle and position, easy and accurate measurements [32]. To best correlate the model simulation with the actual arthroscopy procedure, the femur model was positioned and studied in a view that intuitively mimicked the arthroscopic view of the knee at 90° of flexion [9]. Anatomical landmarks and their quantified relationship with the tunnel may be useful as references during arthroscopy. Previous biomechanical and anatomical studies of the ACL have provided insights into femoral tunnel placement. Isometry is a key aspect of ACL reconstruction. Despite numerous studies conducted in the past decades, the exact isometric point associated with the femoral attachment of the ACL remains ambiguously documented. Some studies have indicated that the most isometric area is at the “over-the-top” region [29, 33, 34], while others suggest it to be slightly shallower (anatomically distal) [35, 36]. However, consensus has been reached regarding the pattern of isometry associated with the ACL femoral footprint. Generally, the high (anatomically anterior) portion of the footprint, closer to the LIR, is more isometric than the low (anatomically posterior) portion [15]. These fibers experience minimal length change during knee flexion-extension. When moved astray from the native footprint to a high and shallow (anatomically distal and anterior) position, the distance to tibial insertion tends to significantly increase with knee flexion; conversely, moving to a low (anatomically posterior) point, a decreased distance is seen [36, 37]. Therefore, placing the femoral tunnel eccentrically within the high portion of the AM bundle footprint neighboring the LIR aligns with the principle of isometry. Anatomy of the femoral footprint of the ACL has also been extensively studied previously but inconsistently documented on the exact size and position. Generally, the footprint is oval-shaped at the deep and low (anatomically proximal and posterior) portion of the lateral wall of the intercondylar notch, with the long axis aligned with the extended line of the posterior femoral cortex [28]. The footprint is surrounded by the LIR at the high and shallow (anatomically distal and anterior) border, by the cartilage margin at the low and deep (anatomically posterior and proximal) border, and by the over-the-top region at the high and deep (anatomically anterior and proximal) border [10, 28, 38, 39]. Recent histological studies suggested that the insertion within the femoral footprint was not identical. The so-called direct insertion distributes in the high portion of the footprint, approaching the LIR, while the smaller indirect insertion encompasses the residual low area, which extends to the low and deep condylar edge [7, 28]. These two types of insertion corresponded to the so-called mid-substance and fan-like fibers in other studies [40]. Direct insertion has been proven to not only play a more essential mechanical role but also is associated with the isometric region in anatomy mentioned above [28]. In the current study, the femoral tunnel was placed within the high portion of the ACL footprint, centered slightly higher and shallower than the center of the AM bundle. This position corresponded with the isometric region and the direct insertion of ACL. Locating the femoral tunnel under arthroscopy can be significantly more challenging than in-vitro or model simulations due to the limited field of view, interference of soft tissue and lens parallax. Although various methods have been utilized to localize femoral tunnels, including fluoroscopy, ruler measurements, offset aimer, and sole anatomical estimation [22], anatomical landmarks always serve as basic reference. ACL remnants can be an intuitive reference, but it may be absent in chronic injuries, and identifying a precise border can also be challenging. The LIR and bifurcate ridges are strongly associated with ACL footprint but are also hard to identify under arthroscopy due to the coverage of ACL remnants and synovial [23]. Cartilage margins are also suggested, with the distances to ACL footprints quantified. The results of the current study were close to the position of AM bundle centers quantified by Davis et al., which are 9.6 ± 1.1 mm, 16.5 ± 2.1 mm and 8.2 ± 1.5 to the low, shallow and deep cartilage border, respectively [11], although the tunnel center in the current study was moved slightly higher and shallower than the AM bundle center. Nevertheless, identifying the cartilage borders and performing precise measurements under arthroscopy could be difficult in practice. Some recent studies suggest the use of ADC as a new reference for femoral tunnel positioning [12, 25, 26]. The main advantage of the ADC is that it presents in all patients with intact condyle cartilage and is easily identifiable under arthroscopy. The ADC is found at the same or a bit lower level than the center of ACL footprint, and much lower than the AM bundle center [12, 25, 26]. These data are comparable to the results in the current study, suggesting it may be a valuable high-low reference in the placement of femoral tunnel. Other important aspects of femoral tunnels include length and posterior wall integrity. Compared with the transtibial technique, the anatomical reconstruction drilled through the AM portal tends to create oblique, horizontal and shorter femoral tunnels, increasing the risk of posterior wall blowout [31]. Short tunnel length has been proven to decrease the pull-out strength of the tendon-tunnel complex [41]. Moreover, for a suspensory fixation, a short tunnel length results in a shorter graft length inside the tunnel, which has been observed to positively correlate with tunnel enlargement [42]. The mean tunnel length in the current study was 34.8 mm, comparable to previous studies of anatomical single-bundle reconstruction [43]. Posterior wall blowout can be a devastating complication as it may lead to loss of graft fixation and early graft failure [44]. Tunnel length and violation of the posterior wall are not only influenced by the inner opening of the tunnel, but also by its orientation. Knee flexion during tunnel drilling has been suggested to have a significant impact on the spatial angles of the tunnel, as well as its length and distance to the posterior cortex [31, 42]. In this study, the femoral tunnel was established with a sagittal angle of 40° in relation to the femoral shaft. This angle corresponds to roughly 120° of knee flexion intraoperatively [31], which could be achieved by most patients. The thickness of the posterior wall was kept at no smaller than 2 mm. The length, orientation and integrity of the tunnels simulated in the current study were feasible for regular suspensory fixation of the ACL graft. Limited attention had been paid to the outer opening of the femoral tunnel on the lateral femoral condyle previously. It has been suggested that the position of the outer opening does not essentially affect the outcome of ACL reconstruction [45]. However, a distally placed endobutton close to the lateral femoral epicondyle was reported to be associated with iliotibial band irritation [46]. Posteriorly placed openings increase the risk of blowout of the posterior cortex, which may further lead to failure of suspensory fixation. On the other hand, a too anteriorly placed opening may place the endobutton within the knee capsule, inducing the risk of trochlear injury. The average position of the tunnel outer opening lay at 2 cm above the lateral epicondyle, and at the posterior third across the anteroposterior dimension of the metaphysis, ensuring a safe distance from the structures mentioned above. Our results differ from previous studies where the femoral tunnels were targeted at the center of the entire ACL footprint. Xu et al. reported that the theoretical center of the femoral tunnel was located at 28.4% ± 5.1% and 35.7% ± 6.9% parallel and perpendicular to the Blumensaat’s line using the quadrant method, compared to 28.4% ± 2.2% and 22.2% ± 3.6% in our study. This position approximates the midpoint of the shallow-deep dimension of the lateral femoral condyle, compared to the deep one-third point in our study [47]. Lee et al. also positioned femoral tunnels at the center of femoral insertion, with average positions of 26.7% ± 2.7% and 30.0% ± 2.9% using the quadrant method. In addition, their tunnels had larger sagittal and coronal angles of 57.7° ± 6.2° and 49.9° ± 5.6°, respectively [48]. Unlike their insertion-center strategy, the tunnels were intentionally placed off-center in this study, shifted to a deeper and higher region within the footprint, which has been shown to be more favorable for graft isometry. Despite the feasibility of creating such satisfactory tunnel as demonstrated in our study, tunnel creation under arthroscopy is much more challenging due to the depth of the intercondylar notch, which allows for more limited room to achieve acceptable tunnel orientations. A thorough debridement of the intercondylar area and clear visualization of anatomical landmarks are necessary to avoid tunnel malposition and posterior wall blow-out. In summary, by utilizing 3D model simulation and analysis, this study confirmed the feasibility of achieving anatomical and isometric positioning of the femoral tunnel through AM portal, yielding satisfactory tunnel characteristics. The location of the femoral tunnel was quantified using established methods, including the quadrant method and anatomical references. These findings may contribute to the modern evidence base regarding the surgery and rehabilitation of ACL reconstruction [49–55]. There are a few limitations in the current study. Firstly, tunnel simulations were performed on reconstructed models from CT data instead of cadaver knees. Thus, non-bony anatomical structures such as ACL footprint could not be directly visualized. Also, the isometry with regards to anatomical regions could not be directly tested. Although numerous published research facilitates the identification of these featured areas according to bone geometry, errors may be exacerbated. Secondly, the femoral tunnel was only simulated in general settings, with a fixed shaft angle of 40° and fixed diameter of 8 mm. Although this may be suitable for most patients, variations in individual graft size and knee flexion angle which may affect tunnel parameters are predictable. Thirdly, the current study was based on the technique of single bundle anatomical ACL reconstruction with rigid drilling through the AM portal. Although it is currently the most widely accepted surgical technique, new instruments have enabled new techniques including flexible drilling and retro-drilling reconstruction, which may bring new methods in femoral tunnel positioning. Most importantly, although measurements in the current study were conducted in a way to achieve the best simulation-to-operation coherence, caution should be taken when applying the quantified positions of the femoral tunnel in relation to the anatomical landmarks as intraoperative references. The discrepancies between simulation and real surgeries were inherent, due to factors such as the absence of soft tissue, the parallax effect of arthroscopes, soft tissue deformation, changes in knee flexion angle and portal positions by individual surgeons. Therefore, it’s still important to rely on the surgeons’ expertise founded upon their personal operative habits. Conclusion The anatomical and isometric positioning of the femoral tunnel can be achieved through AM portal with satisfied tunnel characteristics. The center of the desired femoral tunnel was in the high and deep region of the ACL femoral footprint, slightly higher and shallower than the center of AM bundle. In the high-low direction, it’s slightly higher than the apex of the deep cartilage. In the shallow-deep direction, it’s close to the deep one-third point across the dimension from the deep condyle edge to the front cartilage margin. When drilled at appropriate orientation, favorable tunnel length, integrity and position of the outer opening can be obtained. Acknowledgements The authors appreciate Dr. Ran Xiong from the department of joint surgery, Southwest hospital, Chongqing, China, for his assistance in software support. Abbreviations Anterior cruciate ligament Anteromedial Posterolateral Computer tomography Lateral intercondylar ridge Apex of deep cartilage Deep condylar edge Author contributions L.Y. designed the study, conducted model analysis and was the main composer of the manuscript; D.L. managed subject selection, built models and participated in manuscript writing; Q.X. processed models and conducted tunnel simulation; J.L. conducted literature review and data preparation for the study; B.D. provided resources and managed the project. Funding No funding from any organization has been received for the current study. Data availability No datasets were generated or analysed during the current study. Declarations Ethics approval and consent to participate The current study was approved by the ethics committee of the authors’ hospital. Informed consent was obtained from all individuals in the imaging database from which the modeling data of the current study were drawn, granting permission to use their personal imaging data for scientific research and publication. Consent for publication Consent has been obtained from individual in the imaging database to use their personal imaging data for scientific publication anonymously. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement No datasets were generated or analysed during the current study.