Small Lesions Evaluation Based on Unsupervised Cluster Analysis of Signal-Intensity Time Courses in Dynamic Breast MRI

Figure 2

Diagram of the CAD system employed for the small mammographic lesion evaluation: the six MRI scans for each subject undergo a signal preprocessing, the ROI is determined and all voxels within it are subject to cluster analysis. The obtained time-signal intensity (SI) curves are compared to the four Kuhl classes from Figure 1 and automatically assigned to a class by the supervised neural network.