Related Experiment Video
Updated: Jan 13, 2026
10:25
Brain Infarct Segmentation and Registration on MRI or CT for Lesion-symptom Mapping
Published on: September 25, 2019
49.3K
SGAFNet: Robust brain tumor segmentation via learnable sequence-guided adaptive fusion in available MRI acquisitions
Zhuoneng Zhang1, Luyi Han2, Dengqiang Jia1
1Faculty of Applied Sciences, Macao Polytechnic University, Rua de Luis Gonzaga Gomes, Macao, 999078, China.
Summary
This study introduces a new network for segmenting brain tumors in MRI scans, even with missing data. The method adaptively fuses MRI sequences, improving accuracy for clinical use.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Neuroscience
Background:
- Accurate brain tumor segmentation from MRI is crucial for clinical applications.
- Challenges arise with incomplete or missing MRI sequences, impacting automated methods.
- Existing methods often fail to account for the varying importance of different MRI sequences.
Purpose of the Study:
- To develop a robust brain tumor segmentation method for incomplete MRI sequences.
- To introduce a novel network architecture that adaptively fuses multi-sequence MRI data.
- To enhance the clinical applicability of automated brain tumor segmentation.
Main Methods:
- Proposed a Learnable Sequence-Guided Adaptive Fusion Network (SGAFNet).
- Utilized parallel encoder-decoders for sequence-specific feature extraction.
- Introduced a Learned Sequence-Guided Weighted Average (SGWA) module for adaptive feature fusion.
- Incorporated a Sequence-Specific Attention (SSA) module for cross-sequence dependency modeling.
Main Results:
- SGAFNet achieved state-of-the-art performance on BraTS2018 and BraTS2020 datasets for incomplete MRI sequences.
- Ablation studies confirmed the effectiveness of the SGWA and SSA modules.
- Demonstrated superior robustness compared to existing approaches in handling missing MRI data.
Conclusions:
- The proposed SGAFNet effectively segments brain tumors from incomplete MRI sequences.
- Adaptive fusion and attention mechanisms are critical for robust performance.
- The method enhances clinical applicability by improving consistency in diagnostic workflows.
