A team of researchers at The University of Texas Health Science Center at San Antonio has developed an artificial intelligence (AI) tool to accurately count brain lesions on MRI scans in seconds. This tool, once adapted for clinical use, has the potential to aid neuroradiologists in evaluating patients’ brain diseases at earlier stages.
Brain lesions, particularly enlarged perivascular spaces (ePVS), can be challenging to quantify without the assistance of AI. ePVS are spaces filled with cerebrospinal fluid surrounding arteries and veins, serving as a marker for cerebral small-vessel disease, which can lead to stroke and dementia. The researchers conducted a study involving 1,026 individuals who participated in the Multi-Ethnic Study of Atherosclerosis (MESA) to demonstrate the utility of the AI tool in identifying and counting ePVS.
The study, published in JAMA Network Open, showcased the innovative deep-learning tool’s ability to quantify every ePVS in the brain precisely, providing a map of the patient’s small-vessel disease. Traditionally, counting ePVS manually on MRI scans was time-consuming and labor-intensive, resulting in these lesions often needing to be addressed.
On average, a middle-aged person might have hundreds of these small spaces on an MRI, making it impractical for neuroradiologists to count them due to the significant time investment manually. The AI tool developed by the researchers addresses this challenge by employing automated deep-learning methods to detect ePVS. By training an algorithm with expert knowledge, the tool can recognize, locate, count, and provide detailed information about ePVS, surpassing the capabilities of human observers.
The researchers focused on studying enlarged perivascular spaces throughout the brain, as previous studies had yet to explore whole-brain ePVS quantification. The advanced deep-learning tools developed in this study allowed for comprehensive analysis, identifying two significant regions, the basal ganglia and the thalamus, which showed associations with stroke and small-vessel disease markers. The basal ganglia, a deep-brain region, plays a crucial role in neurodegenerative disorders and is involved in movement and decision-making. The thalamus, located near the basal ganglia, is associated with sensory functions like taste and touch.
The researchers anticipate that the AI tool for counting brain lesions will be further investigated at the Alzheimer’s Disease Research Centers (ADRCs), U.S. National Institute on Aging-Designated Centers of Excellence. The South Texas ADRC, a collaboration between UT Health Science Center San Antonio’s Biggs Institute and The University of Texas Rio Grande Valley, focuses on cerebrovascular disease and its contribution to dementia. The researchers believe their novel AI methodology can benefit large-scale studies across the nation’s ADRCs.
The AI tool developed by the team utilizes the computational power of UT Health Science Center San Antonio’s Genie supercomputer. By leveraging this computing resource, the researchers have made significant strides in automating the quantification of brain lesions, ultimately facilitating earlier detection and evaluation of brain diseases. This breakthrough could enhance the clinical workflow and improve patient care by reducing the time and effort required for lesion quantification, enabling clinicians to make more informed decisions.
