Distinguishing Molecular Subtypes of Breast Cancer Based on Computer-aided Diagnosis of DCE-MRI

Introduction: Previous studies [1,2] have suggested that genetic subtypes of breast cancer are associated with distinct imaging phenotypes on DCEMRI. However, previous attempts to distinguish molecular subtypes of breast cancer are based on qualitative, visual examination of the tumors. In our previous work, we developed a computer aided diagnosis (CAD) system that exploited textural changes within the lesion as a function of contrast uptake (textural kinetics) to distinguish benign from malignant breast lesions [3]. In this study, we developed and evaluated a CAD system that uses dynamic texture for distinguishing triple negative (TN) tumors, which are estrogen receptor (ER) negative, from ER positive (ER+) tumors. We also evaluate the discriminability of textural kinetics with respect to static texture, dynamic contrast enhancement, pharmacokinetic, and morphological features for distinguishing different molecular breast cancer subtypes on DCE-MRI.