arXiv 2601.08732
ISLA: A U-Net for MRI-based acute ischemic stroke lesion segmentation with deep supervision, attention, domain adaptation, and ensemble learning
By Vincent Roca, Martin Bretzner, et al.
Published 2026-01-13
Mindmap
Browse the paper's core ideas, clusters, and relationships in a structured outline.
Accurate delineation of acute ischemic stroke lesions in MRI is a key component of stroke diagnosis and management. In recent years, deep learning models have been successfully applied to the automatic segmentation of such lesions. While most proposed architectures are based on the U-Net framework, they primarily differ in their choice of loss functions and in the use of deep supervision, residual connections, and a…