Sharma, Pankaj

Background and Purpose— We evaluated deep learning algorithms’ segmentation of acute ischemic lesions on heterogeneous multi-center clinical diffusion-weighted magnetic resonance imaging (MRI) data sets and explored the potential role of this tool for phenotyping acute ischemic stroke. Methods— Ischemic stroke data sets from the MRI-GENIE (MRI-Genetics Interface Exploration) repository consisting of 12 international genetic research centers were retrospectively analyzed using an automated deep learning segmentation algorithm consisting of an ensemble of 3-dimensional convolutional neural networks. Three ensembles were trained using data from the following: (1) 267 patients from an independent single-center cohort, (2) 267 patients from MRI-GENIE, and (3) mixture of (1) and (2). The algorithms’ performances were compared against manual outlines from a separate 383 patient subset from MRI-GENIE. Univariable and multivariable logistic regression with respect to demographics, stroke subtypes, and vascular risk factors were performed to identify phenotypes associated with large acute diffusion-weighted MRI volumes and greater stroke severity in 2770 MRI-GENIE patients. Stroke topography was investigated. Results— The ensemble consisting of a mixture of MRI-GENIE and single-center convolutional neural networks performed best. Subset analysis comparing automated and manual lesion volumes in 383 patients found excellent correlation (ρ=0.92; P <0.0001). Median (interquartile range) diffusion-weighted MRI lesion volumes from 2770 patients were 3.7 cm 3 (0.9–16.6 cm 3 ). Patients with small artery occlusion stroke subtype had smaller lesion volumes ( P <0.0001) and different topography compared with other stroke subtypes. Conclusions— Automated accurate clinical diffusion-weighted MRI lesion segmentation using deep learning algorithms trained with multi-center and diverse data is feasible. Both lesion volume and topography can provide insight into stroke subtypes with sufficient sample size from big heterogeneous multi-center clinical imaging phenotype data sets.

Introduction: Genome-wide association (GWA) studies have contributed substantially to the understanding of complex vascular traits including ischemic stroke; however, genetic determinants of acute cerebral ischemia remain to be elucidated. Objective: To investigate the genetic architecture of cerebral infarct lesion burden in patients with acute ischemic stroke (AIS). Methods: Twelve international sites contributed 3,301 AIS subjects with acute MRI and genome-wide genotyping to the MRI-GENIE study. In a preliminary analysis of 657 subjects of European ancestry, acute cerebral infarct lesions were outlined on diffusion-weighted images (DWI) using a semi-automated volumetric method. Standard quality control measures were performed per single nucleotide polymorphism (SNP) and per subject. SNPs were imputed to the Haplotype Reference Consortium v1.1 panel. Natural log-transformed DWI volume was used for GWA analysis with allelic dosage per SNP, age, sex and principal components 1-5 of genetic ancestry as covariates. Results: GWA testing for DWI volume in 7.7 million SNPs yielded no signal crossing the Bonferroni-corrected genome-wide significance threshold of p <5*10 -8 . However, several loci passed the nominal significance threshold of p <1*10 -6 (Figure 1), including a locus on chromosome 2 in the SPATA3-AS1 (spermatogenesis associated 3 antisense RNA 1) gene (lead SNP: rs2368999, MAF=36%, p =8.4*10 -7 ) and an intronic locus on chromosome 10 in the ATRNL1 (Attractin like 1) gene (lead SNP: rs592284, MAF=10%, p =9.1*10 -7 ). Conclusion: In this first-to-date, preliminary GWAS of acute cerebral infarct volume in AIS patients, we identified several new, nominally significant loci including a locus in ATRNL1, a gene previously linked to carotid plaque burden. Further analyses of the MRI-GENIE cohort to replicate these findings, to include stroke subtype specific analyses, and to increase the overall statistical power of the study are ongoing.