Charles S. Springer1, Luminita A. Tudorica1, Xin Li1, Sunitha Thakur2, Elizabeth A. Morris2, Karen Y. Oh1, Mark D. Kettler1, Yiyi Chen1, Ian J. Tagge1, Stephanie L. Hemmingson1, Maayan Korenblit2, John W. Grinstead3, Gerhard Laub4, Jason A. Koutcher2, Wei Huang1
1Oregon Health & Science University, Portland, OR, United States; 2Memorial Sloan Kettering Cancer Center, New York, United States; 3Siemens Healthcare, Portland, OR, United States; 4Siemens Healthcare, San Francisco, CA, United States
The Ktrans DCE-MRI parametric biomarker was tested on a meta-population of 137 positively screened breast lesions (129 patients), before these were biopsied. Two of the three cohorts had positive mammographic screening and the other (a high risk group) negative mammography but positive clinical MRI. A combination of ROI averaged and pixel-by-pixel mapping/histographic Ktrans analyses discriminated all of the 32 malignant tumors from all but one of the 105 benign lesions. The >105 biopsy/pathology procedures on the latter could have been avoided. This discrimination was independent of screening modality, scanner vendor (platform/software), data acquisition, CR, and magnetic field strength.