Is Long Range Sequential Modeling Necessary For Colorectal Tumor Segmentation?

2025-02-10

AI Summary

This work investigates whether long-range sequence modeling is necessary for 3D colorectal cancer (CRC) tumor segmentation. To this end, we introduce CTS-204—the first dedicated benchmark dataset for CRC 3D segmentation—and propose MambaOutUNet, a lightweight and efficient architecture that abandons state-space models’ global dependency modeling in favor of localized token interactions and 3D convolutional feature extraction, enhanced by self-supervised pretraining and multi-scale fusion. Experiments demonstrate that local modeling outperforms global modeling in small, anatomically complex CRC regions, challenging prevailing Transformer- and Mamba-based paradigms. On CTS-204, MambaOutUNet achieves an mDice of 89.7%, surpassing the state-of-the-art Transformer baseline by 2.3 percentage points while operating 2.1× faster during inference—validating its robustness and computational efficiency.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Large Multimodal Models (LMMs)
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Segmentation of colorectal cancer (CRC) tumors in 3D medical imaging is both complex and clinically critical, providing vital support for effective radiation therapy planning and survival outcome assessment. Recently, 3D volumetric segmentation architectures incorporating long-range sequence modeling mechanisms, such as Transformers and Mamba, have gained attention for their capacity to achieve high accuracy in 3D medical image segmentation. In this work, we evaluate the effectiveness of these global token modeling techniques by pitting them against our proposed MambaOutUNet within the context of our newly introduced colorectal tumor segmentation dataset (CTS-204). Our findings suggest that robust local token interactions can outperform long-range modeling techniques in cases where the region of interest is small and anatomically complex, proposing a potential shift in 3D tumor segmentation research.

Problem

• Evaluate long-range modeling in CRC segmentation
• Compare MambaOutUNet with global token techniques
• Assess local vs long-range modeling effectiveness

Innovation

• MambaOutUNet for tumor segmentation
• Focus on local token interactions
• Evaluation on CTS-204 dataset