I-Corps: Artificial Intelligence (AI) Brain Lesion Detection Diagnostic Software

University Of Arkansas Medical Sciences Campus, Little Rock AR

Investigators

Abstract

The broader impact/commercial potential of this I-Corps project is the development of a simple, workflow-driven solution for detecting focal cortical dysplasia (FCD) with high sensitivity and specificity. As many rural hospitals do not have epilepsy specialists, the proposed cloud-based solution can bring the diagnostic tool into their patients’ care plan. Commercialization of this technology may enable: clinical neurology teams - enabling them to better evaluate patients with suspected FCD for surgery; hospitals - allowing them to provide better patient care; patients - ensuring they live seizure-free lives; and scientists, researchers, and biomedical engineers - enabling them to use artificial intelligence to detect challenging medical conditions not limited to FCD. The project combines biology, medicine, artificial intelligence (AI) technology, and business. This I-Corps project is based on the development of a cloud-based software system to automatically identify focal cortical dysplasia (FCD) lesions with at least 90% accuracy by using adaptive deep machine learning (ML) to segment magnetic resonance imaging (MRI) images and indicate suspected lesions. The output will be an MRI sequence with lesions highlighted and a clinician-friendly report indicating the probability that a FCD lesion is present, and if so, the location. Automated detection of FCD lesions may improve the care of epileptic patients. FCD lesions result in a form of epilepsy characterized by medication-resistant seizures. These lesions are difficult to detect due to their characteristics and location. The current standard of care using visual MRI inspection misses up to 50% of cases, even when employed by experienced neuroradiologists. These highly specialized physicians are usually only available in select academic medical centers or urban environments. Because of the limited availability of specialists and the highly treatable nature of these lesions with surgery, there is a significant unmet need for efficient and accurate identification of FCD lesions in MRI scans. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.