A research team led by Harvard University School of Medicine (HMS) in the US has created an AI tool called the Pathological Characterization Tool, featuring rapid assessments (photos) that recognize uncertainty to distinguish between types of brain cancer.
This tool can distinguish glioblastoma and primary central nervous system lymphoma (PCNSL). In particular, two brain cancers are often confused by the similar appearance under a microscope.
Discover the B2B marketing to run
Combining business intelligence with editorial excellence, it reaches enthusiastic experts from 36 leading media platforms.
Please see more
Glioblastoma is known as a wide range of aggressive brain tumors derived from brain cells, but the less common type of PCNSL is said to arise from immune cells.
HMS states that misdiagnosis between these two may lead to inappropriate treatment strategies, which could have serious consequences for patients.
The National Institutes of Health is partially funding the work.
HMS' Picture Tool also features an uncertainty component that warns you when encountering unfamiliar tumor types, encouraging reviews by healthcare professionals.
The laboratory said during brain surgery, tumor tissues are often evaluated quickly by freezing them with liquid nitrogen and examining them under a microscope.
This initial diagnosis may change the following day after a more thorough examination by a pathologist. The photo model aims to reduce the risk of uncertainty and error at this critical stage.
Photos tested at five hospitals are said to have “better than both human pathologists and other AI models.”
Kun-Hsing Yu, Associate Professor of Biomedical Informatics and Senior Research Authority at HMS Blavatnik Institute, said:
HMS added that the model, co-developed by Yu and co-first authors Shih-Yen Lin and Junhan Zhao, was trained and evaluated on 2,141 brain pathology slides around the world, including both rare cases and Holmann fixed samples.
The current focus is on glioblastoma and PCNSL, but there are plans to expand the capabilities of the tool to other cancer types and integrate genetic and molecular data for more comprehensive analysis.
The team said that the AI model is publicly available, so other scientists can use it further and enhance it.
