J-CaPA : Joint Channel and Pyramid Attention Improves Medical Image Segmentation

TL;DR

This paper introduces J-CaPA, a transformer-based model with joint Channel and Pyramid Attention mechanisms, significantly enhancing medical image segmentation accuracy and robustness over existing methods.

Contribution

It proposes a novel joint attention architecture combining Channel and Pyramid Attention for improved multi-scale feature extraction in medical imaging.

Findings

Achieved 6.9% higher Mean Dice score on Synapse dataset.

Reduced Hausdorff Distance (HD95) by 39.9%.

Outperformed existing state-of-the-art segmentation methods.

Abstract

Medical image segmentation is crucial for diagnosis and treatment planning. Traditional CNN-based models, like U-Net, have shown promising results but struggle to capture long-range dependencies and global context. To address these limitations, we propose a transformer-based architecture that jointly applies Channel Attention and Pyramid Attention mechanisms to improve multi-scale feature extraction and enhance segmentation performance for medical images. Increasing model complexity requires more training data, and we further improve model generalization with CutMix data augmentation. Our approach is evaluated on the Synapse multi-organ segmentation dataset, achieving a 6.9% improvement in Mean Dice score and a 39.9% improvement in Hausdorff Distance (HD95) over an implementation without our enhancements. Our proposed model demonstrates improved segmentation accuracy for complex anatomical structures, outperforming existing state-of-the-art methods.