GASA-UNet: Global Axial Self-Attention U-Net for 3D Medical Image Segmentation

TL;DR

GASA-UNet introduces a novel 3D self-attention mechanism within a U-Net framework to improve segmentation accuracy of small and ambiguous structures in medical images.

Contribution

The paper proposes GASA-UNet, a new model with a Global Axial Self-Attention block that processes 3D medical images more effectively than previous methods.

Findings

Improved Dice scores on BTCV, AMOS, and KiTS23 datasets.

Enhanced segmentation of small anatomical structures.

Better delineation of organ boundaries.

Abstract

Accurate segmentation of multiple organs and the differentiation of pathological tissues in medical imaging are crucial but challenging, especially for nuanced classifications and ambiguous organ boundaries. To tackle these challenges, we introduce GASA-UNet, a refined U-Net-like model featuring a novel Global Axial Self-Attention (GASA) block. This block processes image data as a 3D entity, with each 2D plane representing a different anatomical cross-section. Voxel features are defined within this spatial context, and a Multi-Head Self-Attention (MHSA) mechanism is utilized on extracted 1D patches to facilitate connections across these planes. Positional embeddings (PE) are incorporated into our attention framework, enriching voxel features with spatial context and enhancing tissue classification and organ edge delineation. Our model has demonstrated promising improvements in segmentation performance, particularly for smaller anatomical structures, as evidenced by enhanced Dice scores and Normalized Surface Dice (NSD) on three benchmark datasets, i.e., BTCV, AMOS, and KiTS23.