CA-Diff: Collaborative Anatomy Diffusion for Brain Tissue Segmentation

TL;DR

CA-Diff is a novel framework that integrates anatomical spatial features into diffusion models to improve brain tissue segmentation accuracy from MRI, outperforming existing methods.

Contribution

The paper introduces a collaborative diffusion framework with anatomical features, a consistency loss, and a time-adapted attention module for enhanced brain MRI segmentation.

Findings

CA-Diff outperforms state-of-the-art segmentation methods.

Incorporating anatomical features improves diffusion model accuracy.

The proposed modules enhance feature fusion and spatial context understanding.

Abstract

Segmentation of brain structures from MRI is crucial for evaluating brain morphology, yet existing CNN and transformer-based methods struggle to delineate complex structures accurately. While current diffusion models have shown promise in image segmentation, they are inadequate when applied directly to brain MRI due to neglecting anatomical information. To address this, we propose Collaborative Anatomy Diffusion (CA-Diff), a framework integrating spatial anatomical features to enhance segmentation accuracy of the diffusion model. Specifically, we introduce distance field as an auxiliary anatomical condition to provide global spatial context, alongside a collaborative diffusion process to model its joint distribution with anatomical structures, enabling effective utilization of anatomical features for segmentation. Furthermore, we introduce a consistency loss to refine relationships between the distance field and anatomical structures and design a time adapted channel attention module to enhance the U-Net feature fusion procedure. Extensive experiments show that CA-Diff outperforms state-of-the-art (SOTA) methods.