Making MRI faster

Mevan Ekanayake, Making MRI faster

Winner – Monash University VYT local competition (2022)  

Magnetic Resonance Imaging (MRI) provides the most accurate medical  images inside the human body and is often utilized by radiologists and  medical practitioners to search for abnormalities, make diagnoses, and  recommend treatment options. However, the scan time in MRI can easily  exceed 30 minutes which leads to low patient throughput, problems with  patient comfort and compliance, artefacts from patient motion, and high  examination costs. In our research, we employ Artificial Intelligence  (AI) to reduce the scan time of MRI. This would be beneficial not only  to patients in critical conditions requiring quick scans but also to  vulnerable groups such as pregnant women, elderly persons, small  children, and persons suffering from special conditions such as  Parkinson's disease, Claustrophobia, etc. Achieving faster scan times  will also increase throughput and reduce complications in resource  allocation in regional areas where accessibility to MRI scanners is  scarce. Our methodology involves combining MRI physics with AI to build  deep learning models which could produce high-quality MR images  utilising undersampled raw MRI measurements as input. The outputs of our  research will fundamentally reduce risks and costs in MRI.