arXiv 2404.01946
Synthetic Data for Robust Stroke Segmentation
By Liam Chalcroft, Ioannis Pappas, et al.
Published 2024-04-02
Mindmap
Browse the paper's core ideas, clusters, and relationships in a structured outline.
Current deep learning-based approaches to lesion segmentation in neuroimaging often depend on high-resolution images and extensive annotated data, limiting clinical applicability. This paper introduces a novel synthetic data framework tailored for stroke lesion segmentation, expanding the SynthSeg methodology to incorporate lesion-specific augmentations that simulate diverse pathological features. Using a modified n…