A microwave super-resolution imaging approach towards breast cancer margin mapping

Harry Penketh, Sonal Saxena, Michal Mrnka, Cameron P. Gallagher, Caitlin Lloyd, Diksha Garg, Christopher R. Lawrence, Nicholas E. Grant, John D. Murphy, David B. Phillips, Ian R. Hooper, Nick Stone, Euan Hendry · 2026

Read A microwave super-resolution imaging approach towards breast cancer margin mapping by Harry Penketh, Sonal Saxena, Michal Mrnka, Cameron P. Gallagher, Caitlin Lloyd, Diksha Garg, Christopher R. Lawrence, Nicholas E. Grant, John D. Murphy, David B. Phillips, Ian R. Hooper, Nick Stone, Euan Hendry (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-21636.txt.

Introduction

Harry Penketh and 12 colleagues present microwave single-pixel imaging for margin assessment in excised breast tumours. A photo-induced transparency change in a silicon modulator under the sample maps reflectivity of tissue phantoms with sub-wavelength resolution. It maps hydration across 10 by 10 cm at about 1 mm resolution. Gelatin tumour phantoms with water shifts mimicking malignant-margin contrast test discrimination. The method finds, locates, and sizes inadequate margins up to 2 mm. Numerical modelling shows resilience across patients, making it promising for real-time use in surgery.

Catalog record

CREATORHarry Penketh, Sonal Saxena, Michal Mrnka, Cameron P. Gallagher, Caitlin Lloyd, Diksha Garg, Christopher R. Lawrence, Nicholas E. Grant, John D. Murphy, David B. Phillips, Ian R. Hooper, Nick Stone, Euan Hendry
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE42807
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.21636

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

A microwave super-resolution imaging approach towards breast cancer margin mapping Harry Penketh, Sonal Saxena, Michal Mrnka, Cameron P. Gallagher, Caitlin Lloyd, Diksha Garg, Christopher R. Lawrence, Nicholas E. Grant, John D. Murphy, David B. Phillips, Ian R. Hooper, Nick Stone, Euan Hendry Abstract Accurate characterisation of margins in excised breast cancer tumours is critical to the success of surgical interventions, yet margin status is typically confirmed post‑operatively using histopathology. Here we present a new approach to intraoperative margin assessment based on microwave single pixel imaging, demonstrating tissue phantom hydration mapping across large areas ( ∼ \sim 10 cm × 10 ~\times~10 cm) at ∼ \sim 1 mm resolution. By leveraging the photo-induced change in microwave transparency of a silicon modulator placed under the sample, we map the microwave reflectivity and identify positive margins with deeply sub-wavelength resolution. We test the discriminatory capabilities of our approach using gelatine-based tumour phantoms with variations in water density representative of the margin and cancerous tissues of a resected tumour. We demonstrate the capability to identify, locate and quantify inadequate margins up to the typically targeted minimum thickness of 2 mm. Furthermore, using numerical modelling, we show that our approach is expected to be resilient to patient-specific tissue differences. Our technique has potential for future deployment as a real-time intraoperative tissue margin analysis tool. I Introduction Breast cancer is the most commonly diagnosed cancer in women worldwide, with around 2.3 million new cases and 670,000 related deaths each year 5 . Breast cancer is mostly treated using breast conserving surgery (BCS). The objectives of BCS are twofold; the complete removal of the tumour including a margin of healthy tissue, whilst minimising the volume of healthy tissue removed. The presence of a cancer-free margin surrounding an excised tumour has been shown to reduce the chance of local disease recurrence significantly 20 . Whilst margin guidelines vary internationally and with disease stage, the UK guidelines for breast cancer generally classify margins as positive (0 mm), close ( \textless 2 mm) and negative ( \geq 2 mm), as depicted in Fig. 1 inset. Histopathology remains the ‘gold standard’ for margin assessment in an excised tumour, but is typically carried out days after BCS. A 2017 study revealed that in some UK units up to 40\% of patients undergoing BCS required reoperation to remove additional breast tissue, placing a significant burden on patients and healthcare providers 45 . Intraoperative margin assessment (IMA) aims to identify inadequate margins during the initial surgery, thereby avoiding additional operations. However, there remains no clear leading approach which offers both high diagnostic accuracy and clinical feasibility 40 ; 18 . Typically, techniques that yield high resolution images such as optical coherence tomography 14 ; 1 ; 2 ; 8 , microscopy 4 ; 6 ; 7 ; 46 ; 47 ; 51 or Raman spectroscopy 48 ; 43 ; 16 suffer from limited penetration depth through margins and/or long acquisition times, whilst approaches with excellent depth penetration such as X-ray computed tomography 44 ; 29 ; 37 and magnetic resonance imaging 12 ; 32 have limited contrast and therefore low specificity. Figure 1: Method for margin depth mapping. (a) Experimental setup: a vector network analyser (VNA) records the photoinduced change in microwave reflection spectrum from a tumour phantom atop a silicon modulator. By projecting a sequence of binary Hadamard patterns (of N~\times~N pixels - shown in right hand inset) and correlating each with a recorded spectrum, a hyperspectral image is formed. (Inset top left) lateral margins are classified as negative ( > 2 mm healthy tissue), close ( < 2 mm healthy tissue) and positive (no healthy tissue) as shown. Modelled changes in the normalised complex reflection \hat{\delta\textbf{r}} when photoexcited for different margin depths are shown in (b) and (c) (phase and amplitude respectively). The UK target margin thickness of 2 mm is indicated by a dotted white line. Radiofrequency (RF) and terahertz electromagnetic radiation can discriminate cancerous from healthy tissues due to their differing hydration levels 9 ; 28 ; 22 ; 21 . However, high absorption by water impairs the use of terahertz radiation for evaluating deep breast cancer margins of millimetric thickness 50 ; 39 . RF margin assessment probes, which operate at longer wavelengths, are commercially available, have been clinically tested, and have proven successful in reducing secondary surgery rates 38 ; 10 . However, these RF probes have major limitations due to their wide sensing fields ( \sim 7 mm \times~ 7 mm 10 ), within which they are unable to localise small regions of positive margin and are limited to a binary discrimination between positive and negative margins. In addition, they require laborious manual point measurements with limited sampling of the complete surface of the excised tumour. An RF imaging approach to margin assessment could transcend these constraints. However, the long wavelengths of RF radiation severely restrict resolution in standard imaging. In this work, we overcome the spatial resolution problem of RF approaches to margin analysis using a single-pixel camera that exploits near-field photo-generated masks to enable deeply sub-wavelength resolution. Our single-pixel camera operates in the microwave portion of the RF band, specifically the frequency range 11 - 20 GHz which exhibits the maximal sensitivity to tissue depth. We experimentally demonstrate quantification of positive and close margins using tissue phantoms, up to the 2 mm limit appropriate for breast cancer margins 3 ; 36 . In this proof of concept demonstration, a 100 mm \times 100 mm area of excised tumour phantom is imaged hyperspectrally, and margin depth is algorithmically determined with \sim 1 mm lateral resolution. Our approach offers a safe, low cost, sensitive method for intraoperative margin analysis, with significant potential to reduce the reoperation rate for breast cancer patients. II Results and discussion II.1 System design and simulations Our microwave single-pixel imaging system is shown schematically in Fig. 1 (a). It relies on the interaction of a series of optical projections with a microwave beam probing the sample under test. This interaction is facilitated by a silicon photomodulator, which converts incident visible photons into free charge carriers within the silicon, thereby locally blocking the microwave beam. In single-pixel imaging, discussed in detail in Method section III.4 , an image is formed by correlating the sequence of measured signals (here the complex reflected electric field) with the known sequence of spatial masks, usually forming a Hadamard basis set 31 ; 11 . In our approach, as the microwave image resolution is governed by the dimensions of the pixels in the photoexcited masks, deeply subwavelength resolution is possible, with resolution of 10 s of microns demonstrated previously in this frequency band 34 . The microwave probe beam is provided by a portable two port vector network analyser (VNA) coupled to two lens horn antennas in a reflection geometry, as shown in Fig. 1 (a). This gives collimated illumination of the photomodulator at 35 ∘ with transverse-electric (TE) polarisation, allowing convenient normal incidence photomodulation from a digital micromirror device (DMD). The compact VNA serves as both source and detector and monitors the frequency-dependent complex reflection coefficient \mathbf{r}(f) , capturing the change in both the amplitude and phase of reflected microwave radiation, over the range 11-20 GHz. Since the

Read the complete text