Prateek Prasanna1, Nathaniel Braman1, Salendra Singh1, Donna Plecha2, Hannah Gilmore2, Lyndsay Harris2, Tao Wan3, Vinay Varadan1, and Anant Madabhushi1
1Case Western Reserve University, Cleveland, OH, United States, 2University Hospitals, Cleveland, OH, United States, 3Beihang University, Beijing, China, People's Republic of
We present the initial results of using a novel
radiogenomic descriptor, CoLlAGe, on breast DCE-MRI to identify associations with HER2+ breast
cancer subtypes. Current method involves
using a PAM50 assay to analyze primary tumor tissues. CoLlAGe is a quantitative measurement of the degree of
order/disorder of localized image gradient orientations. We extract CoLlAGe
entropy from the regions of interest. Unsupervised hierarchical clustering of
the entropy statistics show that we can segregate the cohort into three
distinct subtypes (enriched, basal and luminal), as identified by PAM50 assay. CoLlAGe resulted in higher clustering accuracy as compared to pharmacokinetic
parameters and signal intensities.
