Pixel-Based Machine Learning in Medical Imaging

Figure 12

Illustrations of various types of nontraining nodules and nonnodules and corresponding output images of the trained MTANN. Nodules are represented by bright pixels, whereas nonnodules are almost dark around the centers of ROIs.

(a)

(b)