The Prediction Of The Hemorrhagic Transformation Locations After Reperfusion Therapy In Acute Stroke Patients: A Perfusion Study Using Deep Learning

Yannan Yu,Danfeng Guo,Min Lou,David Liebeskind,Fabien Scalzo

DOI: https://doi.org/10.1212/wnl.90.15_supplement.p3.212

IF: 9.9

2018-01-01

Neurology

Abstract:Objective: We aimed to explore the feasibility of hemorrhagic transformation (HT) location prediction from baseline magnetic resonance perfusion (MRP) source images. Background: HT is a major complication of reperfusion therapy in acute stroke patients. Accurately predicting HT spatial location is challenging. Design/Methods: Consecutive acute ischemic stroke patients who had HT after reperfusion therapy (IV tPA and/or endovascular thrombectomy) were reviewed from two stroke centers. Patients with MRP and diffusion-weighted imaging (DWI) before and 24 hours after treatment and follow-up susceptibility weighted imaging (SWI) were included. The HT was depicted on follow-up SWI. The DWI lesions and HT were classified into 12 regions based on cerebral vascular territories. Then a total of 80,000 tissue-voxels were extracted from both HT region and non-HT region randomly with the ratio of 1:1. Each voxel included a set of pixel values of MRP source image and baseline DWI. Convolutional neural networks (CNN) were trained with 3-fold cross-validation. Prediction maps were generated from the CNN for each patient. The prevalence of HT after infarction and the sensitivity and specificity of HT prediction in each vascular territory were calculated. Results: Seventy-seven HT patients were analyzed (40 Male, Age 73 ±14 years, median baseline NIHSS score 14 [IQR 8–19], median Onset-to-Treatment time 242 min [IQR 152–355], median modified Rankin score 3 [IQR 1–4], 28 with Parenchymal hematoma). HT occurred most frequently in the territory of middle cerebral artery superficial branches (67.5%) and basal ganglia (37.6%), and least frequently in brain stem or thalamus (3.9%). The best model was a locally-connected CNN with the area under curve of 0.88. The average sensitivity and specificity for HT location prediction in anterior circulation was 89% and 60%. Conclusions: The CNN trained from DWI and MRP source images could predict the HT locations in acute ischemic stroke patients. Disclosure: Dr. Yu has nothing to disclose. Dr. Guo has nothing to disclose. Dr. Lou has nothing to disclose. Dr. Liebeskind has received personal compensation for consulting, serving on a scientific advisory board, speaking, or other activities with Stryker, Medtronic. Dr. Scalzo has nothing to disclose.