Medical Physics Seminar – Monday, March 29, 2021
Studies of classification repeatability and feature harmonization in computer-aided diagnosis using radiomic features from breast DCE-MRI
Heather Whitney, PhD
Associate Professor at Wheaton College and Visiting Scholar with the Department of Radiology, The University of Chicago
Computer-aided diagnosis presents opportunity for decision-making support in cancer diagnosis and prognosis. Human-engineered radiomic features extracted from breast DCE-MRI have shown promise in this area, and there are ongoing efforts to better understand the prospect and limitations of its implementation. Current studies are being conducted to characterize the repeatability of classification in multiple classification tasks (malignant vs. benign, ductal carcinoma in situ (DCIS) vs. mixed invasive ductal carcinoma/DCIS, and cancers of molecular subtype luminal A or luminal B vs. other molecular subtypes) and to harmonize features across populations. In this talk, I will share the progress made by two groups with which I work, undergraduate student researchers at Wheaton College and collaborators in the Giger Lab at the University of Chicago, in these two areas of investigation.