Facebook’s Artificial Intelligence Research (FAIR) group is working with New York University’s medical school to make magnetic resonance imaging (MRI) scans 10 times faster with the help of artificial intelligence (AI). The new collaborative research project is named as fastMRI.
MRI scan usually takes anywhere from 15 minutes to over an hour. It is a traumatic experience for people who are claustrophobic or for whom lying down is painful. Moreover, in rural areas, there is a shortage of MRI facilities resulting in long scheduling backlogs for the scanners. These problems will be solved if AI trims down the scanning time. It will be good news to patients all around the world.
How Can AI Change MRI
An MRI machine gathers raw numerical data in a series of sequential views. It turns the data into cross-sectional images of internal body structures. The duration of the scan depends on the data to be collected. Researchers aim to make use of AI-based imaging tools to change the process.
AI can be used to capture less raw data during MRI scans while preserving the rich information content of magnetic resonance images. Thus, the time taken for scanning will be less. “The key is to train artificial neural networks to recognize the underlying structure of the images in order to fill in views omitted from the accelerated scan. This approach is similar to how humans process sensory information,” the research team said in a blog post on the Facebook website.
The researchers use approximately 3 million MRI images of the brain, liver, and knees from 10,000 clinical cases collected exclusively by the NYU School of Medicine. The research team has removed patients’ names and medical information from these images.
The impact of the research will not be limited to MRI scans. The findings can be used to improvise other medical imaging applications. The research team believes that AI-based tools will have the potential to revolutionize CT scans as well.