Attention Xception UNet (AXUNet): A Novel Combination of CNN and Self-Attention for Brain Tumor Segmentation

2025-03-26

AI Summary

To address the challenge of precise multi-sequence (T1CE/T2/FLAIR) MRI segmentation for gliomas, this paper proposes AXUNet—a novel UNet-based architecture that integrates an Xception backbone with a large-language-model-inspired dot-product self-attention mechanism into both encoder and decoder pathways, thereby enhancing joint spatial-contextual modeling. This design represents the first integration in medical image segmentation of a lightweight, high-efficiency backbone with a highly interpretable attention mechanism, enabling end-to-end co-optimization of multi-sequence features. Evaluated on the BraTS 2021 test set, AXUNet achieves a mean Dice score of 93.73—surpassing state-of-the-art methods including Inception-UNet (90.88) and Xception-UNet (93.24)—and demonstrates superior performance across all three tumor subregions: whole tumor (WT), tumor core (TC), and enhancing tumor (ET). The method thus provides more reliable imaging support for clinical diagnosis and treatment planning.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Multimodal Learning
• Humans and AI: Explainable AI (XAI) for Human Understanding

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Accurate segmentation of glioma brain tumors is crucial for diagnosis and treatment planning. Deep learning techniques offer promising solutions, but optimal model architectures remain under investigation. We used the BraTS 2021 dataset, selecting T1 with contrast enhancement (T1CE), T2, and Fluid-Attenuated Inversion Recovery (FLAIR) sequences for model development. The proposed Attention Xception UNet (AXUNet) architecture integrates an Xception backbone with dot-product self-attention modules, inspired by state-of-the-art (SOTA) large language models such as Google Bard and OpenAI ChatGPT, within a UNet-shaped model. We compared AXUNet with SOTA models. Comparative evaluation on the test set demonstrated improved results over baseline models. Inception-UNet and Xception-UNet achieved mean Dice scores of 90.88 and 93.24, respectively. Attention ResUNet (AResUNet) attained a mean Dice score of 92.80, with the highest score of 84.92 for enhancing tumor (ET) among all models. Attention Gate UNet (AGUNet) yielded a mean Dice score of 90.38. AXUNet outperformed all models with a mean Dice score of 93.73. It demonstrated superior Dice scores across whole tumor (WT) and tumor core (TC) regions, achieving 92.59 for WT, 86.81 for TC, and 84.89 for ET. The integration of the Xception backbone and dot-product self-attention mechanisms in AXUNet showcases enhanced performance in capturing spatial and contextual information. The findings underscore the potential utility of AXUNet in facilitating precise tumor delineation.

Problem

• Improving brain tumor segmentation accuracy using deep learning
• Integrating Xception and self-attention in UNet for better performance
• Evaluating AXUNet against SOTA models on BraTS 2021 dataset

Innovation

• Combines Xception backbone with self-attention modules
• Integrates CNN and self-attention in UNet architecture
• Uses dot-product self-attention for spatial context