We describe a new methodology for discriminating benign from malignant breast lesions on DWI for women with dense breasts and show that the new methodology results in statistically significant improvement over standard ADC methods. The methodology uses computer models that can be constructed independently from the three directional DWI signals or from the trace signal. Preliminary results show improved discrimination is obtained using the multi-directional models when compared to using the trace. The methodology can be adapted for computer-aided-detection by tiling the image, independently marking each tile with areas of suspicion, and constructing a mosaic from the marked tiles.
This abstract and the presentation materials are available to members only; a login is required.