The Artificial Intelligence tool helps to advance the breast cancer imaging accuracy

Oct, 2021 - By SMI

The computer software that has been trained to identify trends in thousands of breast ultrasound pictures may help doctors properly identify breast cancer, according to new research.

While evaluated independently on 44,755 previously completed ultrasound tests, the AI tool increased radiologists' ability to accurately diagnose the disease by 37% and decreased the number of clinical specimens, or samples, required to confirm suspected tumors by 27%. The team's AI analysis, led by researchers from Laura and Isaac Perlmutter Cancer Center and NYU Langone Health's Department of Radiology, is supposed to have been the biggest of this kind, that included 288,767 independent ultrasound tests carried from 143,203 women being treated at NYU Langone hospitals from 2012 and 2018.

Ultrasound tests create real-time pictures of the breast or other tissues by using sound waves of high-frequency that travel through tissue. Although it is not commonly used as a testing technique for breast cancer, it has acted as an additional (to mammography) or follow-up diagnostics for several women. According to the researchers, ultrasound is less expensive, more readily accessible in local hospitals, and does not entail radiation exposure. Furthermore, ultrasound penetrates thick breast tissue improved than mammography and distinguishes packed yet healthy cells from compact tumor cells.

While the team's early results are encouraging, the researcher warns that his team only examined prior tests in their newest study, and clinical studies of the AI tool in patients and realistic scenarios are required before it is routinely used. He also intends to improve the AI program to integrate more patient data, such as a woman's increased risk of breast cancer due to genetic mutation or family history, which does not include their most recent research.

About half of all ultrasound breast tests were utilized to develop the computer software for the study. Following that, ten radiologists independently examined a different group of 663 breast examinations, with an accuracy rate of 92 %