Predicting Ischemic Stroke Prognosis based on Lesion Structural and Functional Disconnections

Motivation: Assessing stroke patients' prognosis is challenging due to complex neurophysiological mechanisms involved, with only lesion location accessible from DWI sequence. Goal(s): This study aims to use patients' lesion information, alongside its structural and functional disconnections, to predict their recovery. Approach: We designed a retrospective study using lesion information at admission along with its strctural and funcitonal disconnetion, combined with machine learning to predict the prognosis of 148 stroke patients six months post-stroke. Results: Our results suggested that the structural and functional disruptions of the lesion could explain and predict National Institutes of Health Stroke Scale score and prognosis of stroke. Impact: The results not only help us understand the neurophysiological mechanisms underpinning stroke prognosis from the perspective of brain structural and functional connections, but also reveal potential targets for intervention treatments aimed at stroke recovery.