Artificial intelligence is one of the most ubiquitous emergent technologies today. It’s found itself in several industries and is now making inroads into the medical profession. Fast Company reports that Facebook and NYU partnered to develop a system that uses AI to speed up a patient’s MRI scan. If you’ve ever been in an MRI chamber, you probably know that scan times can be extended. This extensive scan time is necessary for the machine to get an accurate picture of your body. The longer it spends collecting scan data, the more precise the scan that comes out is. Details are crucial for technicians and doctors that need to use those results in diagnosis.
Why do MRI Scans Take so Long Now?
The journal BMJ informs us that MRI utilizes the body’s natural magnetic properties to form detailed images of specific anatomical structures. It’s a non-invasive technology but requires highly specialized and expensive equipment. For an MRI machine to get a picture of the anatomical structures it’s interested in, it requires a potent magnet. This magnet induces a field around the patient and then maps the resulting images in great detail. Each pass takes time, and the more passes you have, the more precise the image. However, because it takes multiple passes to get pictures clearly, scan times can be prolonged when trying to image large parts of the body. Small areas of the body such as the knee or wrist can be done in as little as ten minutes, ideally.
Dealing With the Problem of Time
MRI pictures are simply a method of compiling multiple passes into a single data set and outputting it into a visual medium. When we consider it like this, it shouldn’t be too difficult to get an artificial intelligence engine to collect this data and compile it faster than it takes to perform a full pass. The results can be astounding when it comes to how long a single MRI scan takes. The Facebook/NYU team cut down the average time for a knee scan from ten minutes to five.
MRIs utilize complex mathematical formulae to compile their pass data into images, but AI can take incomplete data and fill in the blanks. The scientists took incomplete data and used the AI engine to fill in the missing pieces, then compared it to a complete scan. When presented with both images side by side, several professionals noted that the AI image was clearer, crisper, and more visually distinct than the complete MRI scan. Of the six professional radiologists presented with pictures, only one of them could tell which were the MRI generated images and which were compiled by the AI engine.
Faster Scan Times Means Better Responses
MRIs can save lives, but as with most things, it’s time-sensitive. Getting accurate results, yet not taking a lot of time to generate them, can save more lives. Radiologists can depend on AI-generated images to help them get a picture of their patients’ illnesses faster than ever before. It’s an advance that humans should be proud of achieving since it helps our species detect and fix problems more quickly than ever before.