Meeting Banner
Abstract #2654

Automated machine learning-based brain lesion segmentation on structural MRI acquired from traumatic brain injury patients

Ilkay Yildiz1, Rachael Garner1, Michael Douglas Morris2, Jesus Ruiz Tejeda2, Courtney Real2, Manuel Buitrago Blanco2, Paul Vespa2, and Dominique Duncan1
1Keck School of Medicine, University of Southern California, Los Angeles, CA, United States, 2David Geffen School of Medicine, University of California, Los Angeles, LOS ANGELES, CA, United States


Traumatic brain injury (TBI) can cause severe disorders, including post-traumatic epilepsy (PTE). Lesion segmentation is an MRI-based analysis to identify brain structures that correlate with PTE development post-TBI. Unfortunately, manual segmentation, considered the gold standard, is highly tedious and noisy. Thus, we propose the first automated machine-learning based lesion segmentation method for MRI of TBI patients enrolled in the Epilepsy Bioinformatics Study for Antiepileptogenic Therapy (EpiBioS4Rx). Experimental validation demonstrates considerable visual overlap of lesion predictions and ground-truths with 61% precision. Early and automated lesion segmentation via our approach can aid experts in MRI analysis and successful PTE identification following TBI.

This abstract and the presentation materials are available to members only; a login is required.

Join Here