New dual-branch network segments brain Magnetic Resonance Imaging lesions accurately in ISLES 2022 stroke data
DBENet: Dual-Branch Encoder Network for brain MRI lesion segmentation.
Qiang Zhao … Tinghua Cao
Frontiers in neurology · 2026
Background
Accurate brain lesion segmentation in Magnetic Resonance Imaging (Magnetic Resonance Imaging) remains challenging due to heterogeneous lesion appearance, variable scales, and ambiguous boundaries.
Methods
We propose DBENet, a Dual-Branch Encoder Network for brain Magnetic Resonance Imaging lesion segmentation.
Results
over 86% overlap between predicted and true lesion areas, near-perfect match
ISLES 2022
0.8621Dice s
BraTS 2018
0.8261Dice s
More results
Ablation studies confirm the complementary benefits of Spatial and Frequency Fusion and MAF.
DISCUSSION: Furthermore, DBENet achieves a favorable balance between segmentation accuracy and computational cost, supporting practical implementation under moderate hardware constraints.
Future work will extend DBENet to volumetric 3D segmentation.
“
Conclusion
We will also investigate multimodal learning and foundation-model adaptation to improve robustness across diverse imaging protocols.