Meeting Banner
Abstract #0417

Information-based assessment of the radiomic-histomic relationship in brain cancer patients

Samuel Bobholz1, Allison Lowman2, Alexander Barrington3, Michael Brehler2, Sean McGarry1, Jennifer Connelly4, Elizabeth Cochran5, Anjishnu Banerjee6, and Peter LaViolette2,3
1Biophysics, Medical College of Wisconsin, Wauwatosa, WI, United States, 2Radiology, Medical College of Wisconsin, Wauwatosa, WI, United States, 3Biomedical Engineering, Medical College of Wisconsin, Wauwatosa, WI, United States, 4Neurology, Medical College of Wisconsin, Wauwatosa, WI, United States, 5Pathology, Medical College of Wisconsin, Wauwatosa, WI, United States, 6Biostatistics, Medical College of Wisconsin, Wauwatosa, WI, United States

This study sought to provide a biological basis of radiomics-based analyses by assessing the relationship between MR features and analogous histomic features of the underlying tissue using coregistered histology samples taken at autopsy from brain cancer patients. Several radiomic features demonstrated substantial mutual information with their histomic analogs, with first order features showing the strongest associations. These histomic-preserving features were shown to be stable across potential confounds such as differences in scanner vender and acquisition field strength. These findings suggest that MR radiomic features reflect information about the texture of the underlying tissue.

This abstract and the presentation materials are available to members only; a login is required.

Join Here