Deep learning-based detection and segmentation of diffusion abnormalities in acute ischemic stroke

Abstract

ABSTRACTBackgroundAccessible tools to efficiently detect and segment diffusion abnormalities in acute strokes are highly anticipated by the clinical and research communities.MethodsWe developed a tool with deep learning networks trained and tested on a large dataset of 2,348 clinical diffusion weighted MRIs of patients with acute and sub-acute ischemic strokes, and further tested for generalization on 280 MRIs of an external dataset (STIR).ResultsOur proposed model outperforms generic networks and DeepMedic, particularly in small lesions, with lower false positive rate, balanced precision and sensitivity, and robustness to data perturbs (e.g., artefacts, low resolution, technical heterogeneity). The agreement with human delineation rivals the inter-evaluator agreement; the automated lesion quantification of volume and contrast has virtually total agreement with human quantification.ConclusionOur tool is fast, public, accessible to non-experts, with minimal computational requirements, to detect and segment lesions via a single command line. Therefore, it fulfills the conditions to perform large scale, reliable and reproducible clinical and translational research.Plain Language SummaryQuantifying the volume and location of lesions caused by acute ischemic strokes is crucial for therapeutics and prognostic assessment. Research wise, it can be a window to observe relationships between anatomy and function. The definition of such lesions, mostly made by humans, is time- and work-consuming and uses subjective criteria. We used artificial intelligence in a large database of MRIs of patients with ischemic stroke to create an automated tool to detect and segment lesions derived from acute stroke. It outputs the lesion volumes and 3D segmentation maps. Our tool is fast, public, accessible to non-experts, run in local computers, with minimal computational requirements, fulfilling the conditions to perform large scale, reliable and reproducible clinical and translational research.