Automated MR Imaging Aids New Stroke Therapy Development

Vidcast: https://www.instagram.com/p/DcWsUl8J3v5/ A new automated MRI system developed by researchers at the USC-Keck School of Medicine makes experimental stroke treatments easier to evaluate. It potentially helps prevent promising therapies from being discarded due to inconsistent laboratory data. The open-source system was tested on 2,442 mice and rats across six research centers as part of the NIH-sponsored Stroke Preclinical Assessment Network. It automatically checks image quality, adjusts for differences among MRI scanners, identifies the brain, and measures stroke injury, swelling, displacement, and longer-term tissue loss. When researchers compare its measurements with those made by imaging experts, the automated results are extremely close. The system is also able to reduce differences caused by scanners and imaging environments allowing researchers at multiple institutions to apply the same analytical standards to thousands of scans. More than 2,200 animals were scanned two days after experimental stroke, while about 1,750 were scanned again roughly a month later. The technology combines deep learning to identify the brain with transparent, rule-based methods to assess the resulting injury. That is important because researchers can see how the system reaches its conclusions rather than relying entirely on a “black box” AI model. Remember, though, that this technology does not treat stroke and is not designed to analyze human stroke scans. Its potential value is earlier in the process: by making animal studies more consistent and reproducible, it helps researchers identify which experimental stroke treatments deserve development for human clinical trials. References on my website. #StrokeResearch #MRI #BrainHealth #MedicalResearch #AIinMedicine