MAU-Net: Mixed attention U-Net for MRI brain tumor segmentation

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Neuroscience

Background:

• Accurate brain tumor segmentation from MRI is crucial for effective clinical management.
• U-Net architecture has shown promise in automatic brain tumor segmentation.
• Attention mechanisms can improve the performance of deep learning models.

Purpose of the Study:

• To propose a novel Mixed Attention U-Net (MAU-Net) model for enhanced brain tumor segmentation in MRI.
• To integrate spatial-channel attention and self-attention mechanisms within the U-Net framework.
• To improve the accuracy and robustness of automatic brain tumor segmentation.

Main Methods:

• Developed MAU-Net by incorporating Shuffle Attention (spatial-channel) in the encoder and an enhanced Transformer module (self-attention) at the bottleneck.
• Employed U-Net architecture as the foundational model.
• Utilized magnetic resonance imaging (MRI) datasets for brain tumor segmentation.

Main Results:

• MAU-Net achieved improved Dice scores for enhancing tumor, whole tumor, and tumor core segmentation on BraTS 2019/2020 validation datasets.
• The model demonstrated an average performance improvement over the baseline U-Net.
• MAU-Net showed competitive results compared to other state-of-the-art methods.

Conclusions:

• The proposed MAU-Net effectively enhances brain tumor segmentation accuracy in MRI.
• The integration of mixed attention mechanisms significantly contributes to improved segmentation performance.
• MAU-Net offers a promising approach for clinical applications in neuro-oncology.