Ingrid Reiser, PhD, and colleagues are developing a CAD tool for finding masses in breast CT volumes. Their study included 71 breast CT volumes that contained 80 masses (56 malignant, 24 benign). Each breast imaged was segmented into fatty and glandular tissues; the glandular breast regions were sampled at a resolution of 2 mm to develop features and feature images that would allow the CAD software to distinguish normal breast glandular tissue from masses.
How did CAD for breast CT perform? The technology's overall sensitivity rate per case was 87%, with 3.2 false positives per volume. Its sensitivity rate for malignant masses only was 83% per case, with 3.5 false positives, leading Reiser's team to conclude that adding CAD technology to breast CT imaging could help radiologists in the
