Synthetic Data for Robust Stroke Segmentation
This work addresses limitations in clinical neuroimaging workflows for stroke pathology by enabling robust segmentation with less annotated data and cross-sequence applicability, though it is incremental as it builds on existing SynthSeg methodology.
The paper tackles the problem of stroke lesion segmentation in neuroimaging by introducing a synthetic data framework that reduces reliance on high-resolution images and extensive annotations, achieving state-of-the-art performance in-domain and significantly outperforming existing methods on out-of-domain data.
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 nnUNet architecture, our approach trains models with label maps from healthy and stroke datasets, facilitating segmentation across both normal and pathological tissue without reliance on specific sequence-based training. Evaluation across in-domain and out-of-domain (OOD) datasets reveals that our method matches state-of-the-art performance within the training domain and significantly outperforms existing methods on OOD data. By minimizing dependence on large annotated datasets and allowing for cross-sequence applicability, our framework holds potential to improve clinical neuroimaging workflows, particularly in stroke pathology. PyTorch training code and weights are publicly available at https://github.com/liamchalcroft/SynthStroke, along with an SPM toolbox featuring a plug-and-play model at https://github.com/liamchalcroft/SynthStrokeSPM.