The University of Michigan has displayed that it is possible to analyze brain tumor tissue inside the operating room accurately and also assess its nature with the help of artificial intelligence.
FREMONT, CA: Tumors being a common disease these days have experienced a lot of advancements in its treatment. The tumor tissues traditionally look healthy similar to the area around them. On the removal of the tumor, the parts that are near the margins are collected and sent to the pathology lab for further tests. After observing and staining the area using a microscope, the pathologist can inform the surgical team whether they have successfully removed the entire tumor or not. This process consumes a very long time, and generally, a follow-up surgery is needed if the margins are not entirely excised.
The new innovative technology comes in the form of the NIO Imaging System. This technology uses stimulated Raman histology that is developed in the University of Michigan, to instantly capture tissues at the microscopic scale without any staining, entirely bypassing the pathology lab. The technology is so quick that the surgeons can take follow up actions, which can prevent the tumor from relapsing without having to schedule another costly procedure.
Realizing the fact that surgeons are not pathologists and even pathologists make mistakes, the imaging system was boosted by artificial intelligence software, which was instructed to learn how different types of brain tumor look. This included feeding a convolutional neural network empowering the software with more than 2.5 million tissue sample images from 415 patients. It is very precise and offers an instant prediction of the type of tissue it is looking at. After evaluating 278 patients undergoing brain surgery, this new method has had a diagnostic accuracy a little better than the conventional histology (94.6 percent versus 93.9 percent respectively).
This is marked as the first prospective trial evaluating the usage of artificial intelligence in the operating room, stated Todd Hollon, M.D., and the lead author of the research appearing in Nature Medicine. They have successfully performed the clinical translation of an AI-based workflow, which so fast that it enables the researchers to image numerous specimens from right by the patient’s bedside and better evaluate the success rate of tumor removal.