FACT: foundation model for assessing cancer tissue margins with mass spectrometry.

2025-04-04

International Journal of Computer Assisted Radiology and Surgery

AI Summary

To address insufficient model generalization in intraoperative cancer tissue margin assessment—caused by scarcity of labeled data—this work introduces the first foundation model specifically designed for Rapid Evaporative Ionization Mass Spectrometry (REIMS) data. Methodologically, we propose a novel supervised contrastive pretraining strategy based on triplet loss and, for the first time, adapt a cross-modal foundation model architecture—originally developed for text-audio tasks—to mass spectrometry signal modeling. The resulting model substantially enhances discriminative performance under few-shot settings, achieving 82.4% ± 0.8 AUROC on real surgical REIMS data. This outperforms existing self-supervised and semi-supervised baselines, establishing new state-of-the-art performance for this task. The framework enables real-time, accurate intraoperative assessment of residual tumor tissue and provides a deployable paradigm for clinical translation.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Cognitive Modeling & Cognitive Systems: Other Foundations of Cognitive Modeling & Systems
• Computer Vision: Multi-modal Vision

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

PURPOSE Accurately classifying tissue margins during cancer surgeries is crucial for ensuring complete tumor removal. Rapid Evaporative Ionization Mass Spectrometry (REIMS), a tool for real-time intraoperative margin assessment, generates spectra that require machine learning models to support clinical decision-making. However, the scarcity of labeled data in surgical contexts presents a significant challenge. This study is the first to develop a foundation model tailored specifically for REIMS data, addressing this limitation and advancing real-time surgical margin assessment. METHODS We propose FACT, a Foundation model for Assessing Cancer Tissue margins. FACT is an adaptation of a foundation model originally designed for text-audio association, pretrained using our proposed supervised contrastive approach based on triplet loss. An ablation study is performed to compare our proposed model against other models and pretraining methods. RESULTS Our proposed model significantly improves the classification performance, achieving state-of-the-art performance with an AUROC of 82.4 % ± 0.8 . The results demonstrate the advantage of our proposed pretraining method and selected backbone over the self-supervised and semi-supervised baselines and alternative models. CONCLUSION Our findings demonstrate that foundation models, adapted and pretrained using our novel approach, can effectively classify REIMS data even with limited labeled examples. This highlights the viability of foundation models for enhancing real-time surgical margin assessment, particularly in data-scarce clinical environments.

Problem

• Accurately classifying cancer tissue margins during surgery
• Overcoming labeled data scarcity in REIMS-based margin assessment
• Developing a foundation model for real-time surgical decisions

Innovation

• Adapts foundation model for REIMS data analysis
• Uses supervised contrastive pretraining with triplet loss
• Achieves state-of-the-art classification with 82.4% AUROC