ABSTRACT
Studies on computer analysis of medical images began in the 1960s. However, extensive research under the concept of computer-aided diagnosis (CAD) on mammograms and other medical images started in the 1980s. This chapter introduces broad research topics on computer analysis of mammograms. One of the problems in breast cancer diagnosis is the difficulty in distinguishing between benign and malignant lesions on mammograms. The utility of the system was evaluated by the observer study, in which radiologists interpreted mammograms unaided, with regular CAD prompts and with interactive CAD. S. K. Kinoshita et al. proposed visual features to retrieve similar mammograms with comparable breast density levels. J. W. Byng et al. proposed a computerized method to analyze mammograms using the brightness histogram and fractal dimension. In assisting to reduce interpretation errors, some commercial CAD systems provide additional information, such as the suspiciousness of lesions, size of lesions, morphology of masses, and the number of microcalcifications.
