MSACL 2025 Abstract

Real-time Intraoperative Mass Spectrometry Imaging to Guide Robotic-Assisted Cancer Surgery

Amelia Fraser-Dale (1), Jinshi Zhao (1), Lauren Ford (1), István Pap (2), Eftychios Manoli (1), James Higginson (3), Vinidh Paleri (4,5), Zoltan Takats (1,2), Burak Temelkuran (1)
(1) Imperial College London, London, United Kingdom, (2) University of Regensburg, Regensburg, Germany, (3) Brighton and Sussex Medical School, Brighton, United Kingdom, (4) The Royal Marsden NHS Foundation Trust, London, United Kingdom, (5) The Institute of Cancer Research, London, United Kingdom

Amelia Fraser-Dale, BSc (Presenter) Imperial College London

Presenter Bio: Amelia is a clinical PhD student at Imperial College London. Her research interests include AI in surgery and real-time cancer identification using Rapid Evaporative Ionisation Mass Spectrometry (REIMS).

Relevant Financial Disclosures (within past 24 months, reported on Mar 19, 2025) No relevant financial relationship(s) to disclose.

INTRODUCTION:
Successful cancer resection is predicated upon the complete removal of cancerous tissue with minimal disruption to the surrounding healthy tissue. Incomplete resection significantly worsens mortality, whilst excessive damage to healthy tissue reduces postoperative quality-of-life. There is an evident clinical need for solutions to improve the differentiation between cancer and healthy tissues during surgery. Rapid evaporative ionization mass spectrometry (REIMS) has demonstrated high-accuracy, real-time, objective, metabolomic tissue classification without requiring sample preparation, making the technique ideally suited for intraoperative use [1]. Mass spectrometry imaging (MSI) could provide a rich molecular map to guide cancer surgery. To achieve spatially resolved REIMS, we present a high-precision minimally-invasive fiber-based robotic platform with integrated REIMS imaging and tissue-ablation capabilities.

METHODS:
An electrothermally actuated robotic fiber, capable of movement with micrometer-level precision, was produced using a thermal-drawing manufacturing technique [1]. A CO2 laser fiber and REIMS system were integrated with the fiber robot to enable tissue ablation and metabolomic classification. For proof-of-concept validation, a cervical cancer phantom was 3D-printed with different materials representing cancer and healthy regions. Machine learning techniques were applied to single-point REIMS data obtained from each material, demonstrating accurate real-time material classification. The robotic fibre system was then mounted using a commercially available colposcope to provide a stable platform from which robotic end-effector position tracking was introduced to demonstrate MSI in the cervical phantom. The system was also mounted to a commercially available Da Vinci surgical robotic platform to perform MSI in a cadaver during simulated natural-orifice tonsil surgery.

RESULTS:
The robotic fibre, with a diameter of 2 mm and motion precision below 50 um, is well-suited for enhancing precision in minimally invasive surgery. Coupling the robotic fibre with a CO₂ surgical laser fibre enables superficial tissue ablation scanning with a spot size of approximately 200 um. The aerosol produced is transmitted to a mass spectrometer through a silicone tube, enabling real-time tissue classification. We have demonstrated robotic fibre-coupled mass spectrometry in the transoral region through a cadaver study, where surgical ablation scanning and aerosol collection were performed in the vocal cord, tonsil, and tongue regions. This study demonstrated the accessibility and precision of the system within complex anatomical structures. By combining fibre tip positioning with real-time diagnostic information, the system successfully generated a diagnostic tissue map that visually corresponded to the anatomical features of the tissues. Additionally, we developed a 3D-printed cervical model using PLA materials in two distinct colours to represent cancerous and healthy tissues, respectively. REIMS data collected from these two PLA materials revealed distinct spectral signatures, allowing differentiation between the simulated healthy and diseased tissues. The cervical model was then integrated into a gynecological phantom setup, where we evaluated the robotic system's performance. By correlating the robotic fibre's position with real-time diagnostic information, we successfully generated diagnostic tissue maps. The molecularly-aware robotic surgical platform precisely ablates targeted tissue layers to create diagnostic maps, enabling accurate, layer-by-layer removal of an entire tumour mass. Each ablated layer produces diagnostic information guiding the removal of subsequent layers. This iterative approach continues until the diagnostic analysis confirms the absence of residual tumour tissue. We will discuss the practical feasibility of this method and its benefits for improving the precision of tumour removal procedures.

CONCLUSION:
This work shows the potential for real-time objective tissue classification in cancer surgery. We demonstrate a proof-of-concept robotic surgical tool to enable metabolomic tissue imaging using REIMS during minimally invasive surgery.

REFERENCES:
[1] M. Tzafetas, A. Mitra et al., The intelligent knife (iKnife) and its intraoperative diagnostic advantage for the treatment of cervical disease, Proceedings of the National Academy of Sciences 117 (2020) 7338–7346. https://doi.org/10.1073/pnas.1916960117.
[2] M.E.M.K. Abdelaziz, J. Zhao et al., Fiberbots: Robotic fibers for high-precision minimally invasive surgery, Science Advances 10 (2024) eadj1984. https://doi.org/10.1126/sciadv.adj1984.