Related Experiment Video
Updated: Aug 13, 2025
04:25
Author Spotlight: Bridging Gaps in Anatomy and Establishing a Foundation for Algorithmic Studies
Published on: December 15, 2023
2.6K
Axial Attention Convolutional Neural Network for Brain Tumor Segmentation with Multi-Modality MRI Scans
Weiwei Tian1, Dengwang Li1, Mengyu Lv2
1Shandong Key Laboratory of Medical Physics and Image Processing, Shandong Institute of Industrial Technology for Health Sciences and Precision Medicine, School of Physics and Electronics, Shandong Normal University, Jinan 250358, China.
Brain Sciences
|January 21, 2023
Summary
This study introduces an axial attention brain tumor segmentation network (AABTS-Net) for automated tumor identification in MRI scans. The AABTS-Net improves diagnostic accuracy and efficiency compared to manual methods.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Neuro-oncology
Background:
- Manual segmentation of brain tumors from MRI scans is critical but time-consuming and prone to inter-observer variability.
- Accurate tumor segmentation is essential for effective clinical diagnostics and treatment planning.
Purpose of the Study:
- To develop an automated brain tumor segmentation network using multi-modality MRIs.
- To enhance the accuracy and efficiency of tumor subregion identification.
Main Methods:
- Developed an axial attention brain tumor segmentation network (AABTS-Net).
- Incorporated an axial attention mechanism for capturing local-global contextual information with simplified computational complexity.
- Utilized deep supervision and a hybrid loss function to improve feature representation and address class imbalance.
Main Results:
- The AABTS-Net demonstrated superior robustness and accuracy in segmenting brain tumors.
- Comprehensive experiments were conducted on the BraTS 2019 and 2020 datasets.
- The model effectively handles class imbalance and avoids vanishing gradients.
Conclusions:
- The proposed AABTS-Net offers a robust and accurate solution for automated brain tumor segmentation from multi-modality MRIs.
- This model shows potential for clinical application and advances medical image segmentation systems.
