Benign-Malignant Lesion Differentiation Using Functional ADC-Thresholding Allowing Expert Radiologist Interpretation Versus Conventional Thresholding Based on ADC Cut-Off Values

Wouter B. Veldhuis¹, Chunlei Liu¹, Yung Do¹, Thomas J. Brosnan¹, Michael E. Moseley¹, Bruce L. Daniel¹

Diffusion-weighted imaging (DWI) may aid in the discrimination of benign from malignant (breast) lesions. Approaches to benefit from the information contained in the DWI dataset have mostly been based on trying to define a cut-off value for the lesion ADC. This abstract compares the application of a new functionally-thesholded apparent diffusion coefficient (ftADC)-mapping algorithm to conventional ADC-thresholding by analyzing the diagnostic accuracy of both methods using a data set of 65 pathology-proven breast-MRI lesions. Using receiver-operating-characteristics (ROC)-curves it is shown that lower ADC correlates with a higher chance of malignancy and that ftADC-mapping improves diagnostic accuracy over conventional ADC thresholding.