Technologies
This technology is an algorithm-assisted pipeline for detecting cerebral microbleeds in magnetic resonance imaging (MRI) scans, offering improved accuracy and efficiency compared to relying solely on human analysis.
Cerebral microbleeds (CMBs) are a type of chronic hemorrhage occurring in small brain vessels. While the clinical impact of these events is not well understood, there is a correlation between the presence of CMBs and increased risk for stroke or neurodegenerative diseases. While CMBs can be detected by MRI scans through the shape they produce and the presence of degraded blood products, current identification relies solely on manual assessment, which increases the time and expense of CMB detection and introduces an element of subjectivity. Therefore, there is a need for an algorithm-based approach that can quickly and accurately identify potential sites of CMBs to enhance efficacy and eliminate subjectivity from this process.
This technology describes an algorithm that analyzes MRI scans to detect sites containing likely cerebral microbleeds (CMBs). The algorithm employs a multi-step visual analysis approach to analyze MRI images slice by slice and highlights circular regions with irregular edges, characteristic of CMBs. The highlighted regions are then compared in size to filter out blood vessels and other potential artifacts, and output in a visual format convenient for human assessment and confirmation. This technology enhances the speed and efficacy of MRI-based CMB detection while reducing the subjectivity inherent in human analysis alone. The algorithm can be expanded for use in a wide range of clinical and research applications.
This technology has been validated pre-clinically in 64 subjects, where it performed CMB detection with 86.7% sensitivity and 2.6% precision.
Patent Pending (US 20230298170)
IR CU20112
Licensing Contact: Cynthia Lang