Abstract #2480

Principal Component Analysis of Breast DCE-MRI: Evaluation of Clinical Protocols at Two Temporal Resolutions

Daria Badikhi¹, Myra Shapiro-Feinberg², Erez Eyal³, Edna Furman-Haran⁴, Dov Grobgeld¹, Hadassa Degani¹

¹Biological Regulation, Weizmann Institute of Science, Rehovot, Israel; ²Radiology, Meir Medical Center, Kfar Sabah, Israel; ³Biological Regulation, Weizmann Institute of Science, Israel; ⁴Biological Services, Weizmann Institute of Science, Rehovot

Principal component analysis (PCA) of clinical breast DCE MRI datasets, recorded at two different temporal resolutions (80 s and 120 s), was tested and evaluated for its diagnostic ability. We found that PCA can differentiate with high accuracy between benign and malignant lesions at both temporal resolutions, however, discriminative ability between invasive ductal and lobular carcinoma can be reached only at the higher temporal resolution. Overall, PCA was found to be a useful, standardized, fast, and objective tool for computer aided diagnosis of breast lesions