Optical coherence tomography for surgical demarcation of GI tumor margins for accelerated decision-making during the surgical resection of tumors

Gastric cancer, including stomach tumors, poses significant health challenges due to late-stage diagnosis and limited early detection methods. Accurate imaging and precise tumor margin demarcation are critical for effective treatment planning and improved patient outcomes. Conventional imaging techniques, such as endoscopy and histopathology, provide valuable diagnostic information but cannot offer real-time assessment. We aim to explore the application of optical coherence tomography (OCT), combined with statistical and machine learning methods, for rapid tumor margin demarcation in gastrointestinal (GI) cancer tissues in ex-vivo. GI tumor specimens from 17 patients were imaged intraoperatively with OCT. Quantitative features were extracted from the images, and statistical and machine learning analyses were applied to distinguish tissue types. Subsequent histopathological evaluation was used as the reference standard for validation. The combination of OCT imaging and data-driven analysis enables clear differentiation between healthy, marginal and tumor tissues in near real-time. The KNN model achieved the highest classification accuracy ([Formula: see text]), with SVM ([Formula: see text]) and Extra Trees ([Formula: see text]) also demonstrating robust performance in discriminating tissue margins. OCT findings demonstrate strong agreement with histopathology. Integrated with statistical and machine learning techniques, OCT enables rapid assessment of GI tumor margins. This approach provides quantitative objectivity and real-time feedback, closely approximating histopathological analysis and supporting improved surgical decision-making.