A number of fascinating projects have emerged recently around the successful analysis of medical imaging, with AI capable of spotting the onset of various conditions considerably faster than trained professionals.
Most of these algorithms are quite specialized in that they look out for specific conditions, but a recent paper documents some new software from the University of Missouri that is capable of analyzing any series of images faster and more accurately than ever before.
The software, which is free for academic use, is called Tracking Equilibrium and Nonequilibrium shifts in Data (TREND), and gives users the ability to rapidly analyze any images, including CT scans, MRI images, ultrasound and video.
The software is designed to examine a set of images to resolve and then track the changes throughout the series. For instance, one might feed it a video of a heart beating, and the software can then be used to spot any irregularities in that heartbeat, and the potential health problems this may cause.
“TREND allows accurate, rapid analysis of incredibly complex and nuanced images, which can potentially save doctors, patients and scientists countless hours and money,” the developers say. “TREND has allowed us to advance our own research into enzyme interactions considerably. Previously, it would take us weeks to analyze a single group of images. With TREND, that analysis now takes only a few minutes and is more accurate and consistent than if a human performed the work.”