NYU and Facebook collaborate for ‘fastMRI'

NYU is entering into a partnership with Facebook to accelerate the process of MRI scanning. The aim of the collaboration is to cut down the scan time which generally spans anything between 15 minutes to an hour. They intend to do this by using AI to reconstruct MRI in innovative ways.

The partnership has been named “fastMRI” that will bring together the Facebook AI Research (FAIR) group and the Department of Radiology in New York School of Medicine. According to the NYU imaging specialist Dan Sodickson, the research could cut down a 50-minute evaluation to 5 minutes.

Traditionally what an MRI scan does is that it takes 2D images from multiple angles and stacks them up to create a 3D view of a body part being scanned. This new research is going to try and capture lesser data and make the process faster using AI. It is expected to preserve or even enhance the quality of the content of the MRI.

“The key is to train artificial neural nets to recognize the underlying structure of the images, and what types of things tend to be clustered together so that the nets can then fill in views that are emitted from the accelerated scan,” said Sodickson. If and when successful, this research will give a major boost to healthcare by reducing the overall time taken in delivering diagnosis significantly.