The list of artificial intelligence-powered medical diagnostic applications keeps growing, particularly for cancer-related applications. We already wrote about work on AI-assisted diagnosis for prostate cancer at Radboud University Medical Center in the Netherlands, colonoscopy scans at Wuhan University, beast cancer screening at Google Health, lung cancer at the NYU School of Medicine, skin cancer at the Stevens Institute of Technology, and more. In 2018 we wrote about Penn Medicine’s Brain Tumor Center’s telemedicine second opinion program for brain cancer.
Researchers at UT Southwestern recently published a study in Neuro-Oncology about an extremely accurate deep-learning-based AI method to classify genetic factors in gliomas. Gliomas are brain cancers in the brain and spinal cord that occur in the supportive tissue around nerve cells. Prior to beginning treatment for gliomas, the conventional practice is to perform surgery to take glioma samples. The UT Southwestern technology uses the AI model to examine 3D MRI scans. According to the published study, the deep-learning model identified specific glioma genetic mutations with greater than 97% accuracy using a single MRI scan series and employing just one algorithm. Previous research used multiple imagery types and additional algorithms and models, according to the UT study authors.
Surgery by definition is an invasive procedure and to be avoided if there is an acceptable alternative. This AI-based system of screening brain imaging could reduce the need for brain biopsies when diagnosing gliomas. Chalk up another possible win for computer-assisted diagnosis tools that can help in the fight against cance.
