Unmet Need
According to the World Health Organization annually, 15 million people worldwide suffer a stroke. Of these, 5 million die and another 5 million are left permanently disabled, placing a burden on family and community. Within the United States more than 795,000 people suffer from a stroke every year with more than 131,000 of these resulting in death (see CDC). Stroke diagnosis and treatment relies on a variety of imaging tests that allows clinicians to estimate the volume, location and characteristics of brain lesions caused by the stroke. Magnetic Resonance Imaging (MRI) is a widely used method to rapidly diagnose strokes. Evaluating lesions from MRI images requires the manual segmentation of the stroke regions which is a time consuming process that is subject to variation and can result in inaccurate representations of cerebral lesions. Therefore, there is a need for tools that can provide automated, unbiased, and accurate segmentation of stroke lesions in a timely matter.
Technology Overview
Johns Hopkins University researchers have created a comprehensive user friendly tool to accurately segment acute stroke lesions from brain MRIs, in real time. Using a large database of MRIs the researchers have developed and tested a tool for automated segmentation of acute strokes, that involves three main steps: 1) “pre-processing, 2) “classical” artificial intelligence, 3) supervised AI (deep learning, DL). This tool will allow quantification of brain damage in real time, and provide the 3D lesion segmentation required for lesion-based studies and more accurate quantitative information by accurately measuring stroke volume.
Stage of Development
The code for this tool is available at github.com/Chin-Fu-Liu/Acute-stroke_Detection_Segmentation
