CRC-SAM: SAM-Based Multi-Modal Segmentation and Quantification of Colorectal Cancer in CT, Colonoscopy, and Histology Images

TL;DR

CRC-SAM introduces a unified, modality-agnostic framework for colorectal cancer segmentation across colonoscopy, CT, and histology images, leveraging lightweight LoRA adaptation for efficient domain transfer.

Contribution

It presents a novel multi-modal segmentation framework built on MedSAM, enabling consistent analysis across different imaging modalities with minimal additional training.

Findings

Outperforms state-of-the-art baselines across multiple datasets.

Demonstrates effective domain transfer with lightweight LoRA layers.

Provides consistent segmentation results across diverse clinical imaging modalities.

Abstract

We present CRC-SAM, a unified framework for colorectal cancer segmentation across colonoscopy, CT, and histopathology images. Unlike prior single-modality methods, CRC-SAM provides consistent, modality-agnostic segmentation throughout the clinical workflow. Built on MedSAM, it incorporates low-rank adaptation (LoRA) layers into a frozen encoder, enabling efficient domain transfer to underrepresented modalities with minimal trainable parameters. Experiments on MSD-Colon, CVC-ClinicDB, and EBHI-Seg demonstrate superior performance across modalities, outperforming state-of-the-art baselines and highlighting the effectiveness of lightweight LoRA adaptation for foundation-model-based colorectal cancer analysis.