Infervision, a Chinese company, has been integrating deep learning artificial intelligence to the practice of radiology in order to improve the detection rates of early stage tumors. The technology relies on analyzing tens of thousands of X-rays and CT scans that have been used to perform a diagnosis and to use that base of knowledge to evaluate future scans. The software is already being used at a number of Chinese treatment centers and the company claims that its X-ray capability has “achieved extremely high accuracy so that it is close to that of a deputy chief physician in the diagnosis of cardiothoracic diseases at one of the top Chinese hospitals.”
The software works with standard PACS (Picture Archiving and Communication Systems) and Infervision was able to obtain nearly 200,000 CT scans and X-rays from Chinese medical records in order to train it. It’s certainly not doing diagnostics on its own, but is rather used to double check and point out potential trouble spots. The company hopes that its software can help cut down the amount of time radiologists spend reviewing each scan, easing a clinical bottleneck that’s often leads to delayed care.
More from Infervision:
Infervision’s technology and a group of radiologists recently went head-to-head in a report reading experiment with different types and sizes of nodules. Infervision’s AI-CT predicted more accurately than radiologists in every category.
“In no way will this technology ever replace doctors. It is intended to eliminate much of the highly repetitive work. Infervision empowers doctors and helps them deliver more accurate reports and do it much faster,” continued Kuan.
The Infervision artificial intelligence continues to learn as more data becomes available and also as it analyzes past results.