Extraction and computational estimation of malignant microcalcification on mammography
Seiji Yabashi, Masayasu Hata, Kanji Kubo, Taro Ishikawa · Systems and Computers in Japan · 1989
Abstract Mammography is one of the tools in the diagnosis of breast cancer. By a new method of mammography based on the soft X‐ray and the ultrafine X‐ray film, the X‐ray exposure is reduced and the fine internal structure of the breast can be observed. With the increase of the subjects for breast cancer, it is required to improve the accuracy and efficiency of the diagnosis. A diagnosis‐supporting equipment is required which provides an image display to assist in recognition of pathological changes, and provides an objective feature evaluation. This paper considers the microcalcification which is important in the diagnosis of the early breast cancer. The microcalcification appears as opaque points on the mammograph, and when the points are concentrated locally, it is diagnosed as malignant. After the emphasis by Laplacian processing, cell division was applied. The cell with aggregated microcalcification was extracted, and the image display was made so that the feature difference between cancerous and noncancerous lesions can be recognized. To arrive at a numerical representation of the feature difference, the following parameters were considered: (1) effective distance between cells; and (2) rate of change of the number of cells with regard to the change of the threshold. As the result of processing 37 cases, including 14 benign cases and three normal cases, up to 14 of 15 malignant cases were able to be separated from other benign and normal cases, using those two parameters.