Pixel-Based Machine Learning in Medical Imaging

Table 1

Classes of PMLs, their functions, and their applications.

PMLs	Functions	Applications
Neural filters (including neural edge enhancers)	Image processing	Edge-preserving noise reduction [38, 39]. Edge enhancement from noisy images [40]. Enhancement of subjective edges traced by a physician [41].
Convolution neural networks (including shift-invariant neural networks)	Classification	FP reduction in CAD for lung nodule detection in CXR [42–44]. FP reduction in CAD for detection of microcalcifications [45] and masses [46] in mammography. Face recognition [47]. Character recognition [48].
Massive-training artificial neural networks (MTANNs, including a mixture of expert MTANNs, a LAP-MTANN, an MTSVR)	Classification (image processing + scoring), pattern enhancement and suppression, object detection (pattern enhancement followed by thresholding or segmentation)	FP reduction in CAD for detection of lung nodules in CXR [57] and CT [17, 52, 63]. Distinction between benign and malignant lung nodules in CT [58]. FP reduction in CAD for polyp detection in CT colonography [53, 59–62]. Bone separation from soft tissue in CXR [54, 55]. Enhancement of lung nodules in CT [56].
Others	Image processing or classification	Segmenting posterior ribs in CXR [64]. Separation of ribs from soft tissue in CXR [65].