Related Experiment Video
Updated: Jul 26, 2025
06:08
A Cognitive Fusion-guided Prostate Biopsy Using Multiparametric Magnetic Resonance Imaging and Transrectal Ultrasound
Published on: March 21, 2025
293
Uncertainty quantification and attention-aware fusion guided multi-modal MR brain tumor segmentation
1School of Information Science and Technology, Hangzhou Normal University, Hangzhou 311121, China.
Computers in Biology and Medicine
|June 18, 2023
Summary
This study introduces a novel deep learning approach for brain tumor segmentation, incorporating uncertainty quantification to improve accuracy and safety in clinical applications. The method enhances segmentation refinement by utilizing uncertainty maps derived from multi-modal MRI data.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computational Biology
Background:
- Accurate brain tumor segmentation is crucial for diagnosis and treatment planning.
- Deep learning models excel at segmentation but lack uncertainty estimation.
- Clinical safety necessitates quantifying segmentation uncertainty.
Purpose of the Study:
- To develop a deep learning model for multi-modal brain tumor segmentation with uncertainty quantification.
- To improve the accuracy and reliability of brain tumor segmentation.
- To provide uncertainty maps for refining segmentation results.
Main Methods:
- A multi-encoder 3D U-Net architecture was used for initial segmentation.
- An estimated Bayesian model quantified segmentation uncertainty.
- Attention-aware multi-modal fusion learned complementary features from MRI modalities.
- Uncertainty maps were integrated as constraints for segmentation refinement.
Main Results:
- The proposed method demonstrated superior performance on BraTS 2018 and 2019 datasets.
- Outperformed state-of-the-art methods in Dice score, Hausdorff distance, and Sensitivity.
- Successfully generated uncertainty maps alongside segmentation results.
Conclusions:
- The integration of uncertainty quantification significantly enhances brain tumor segmentation.
- The attention-aware fusion method effectively leverages multi-modal MRI data.
- The proposed components are adaptable to other architectures and computer vision tasks.
