A multi-institutional team of researchers from the University of California, San Francisco, including the Department of Radiology and Biomedical Imaging and Center for Intelligent Imaging (ci2), investigated whether a probabilistic brain metastasis artificial intelligence (AI) risk model could enhance the therapeutic ratio of whole-brain radiotherapy by targeting high-risk areas while preserving cortical and subcortical brain regions of functional significance and low metastasis risk, potentially reducing the cognitive side effects of therapy.
The retrospective study, "Multi-institutional atlas of brain metastases informs spatial modeling for precision imaging and personalized therapy," authored by UCSF ci2's Javier Villanueva-Meyer, MD, Soonmee Cha, MD, Christopher Hess, MD, PhD, and Daniel Cooke, MD, was published in Nature Communications.
Brain metastases (BM) patients previously treated after initial diagnosis of BM with radiosurgery were investigated. The clinical data template, tumor characteristics, lesion distance to the MNI white-gray matter junction and lesion centroid presence in the region-of-interest atlases were consolidated into a SQL database for plotting and analysis.
"Our findings corroborate historical studies and clinical experience, which have demonstrated the non-uniform distribution of BM, favoring the gray-white matter junction and vascular watershed regions," the authors write. "The infrastructure, code base, and database comprising 13,067 BM across 3065 MRIs from as many unique patients from four institutions, are intended as a community resource in support of efforts to enhance auto-segmentation, detection, and quantitative analysis of BMs, and is easily extensible to incorporate data from additional institutions as well as longitudinal data on existing patients."
Visit the UCSF ci2 blog to read more about research and news.
