Adaptive classification model based on artificial immune system for breast cancer detection

Early stage asymmetric signs in breast that can be captured by the screening-digital mammography can be used for a precocious diagnosis of breast cancer. Conventional mammography screening fails to detect subtle anomalies, so computer-aided methods are studied in order to improve the accuracy of image analysis. To classify the images into asymmetric and normal cases, in this paper we investigated the performance of an Adaptive Artificial Immune System (A
²
INET) classifier. To test the efficiency of the algorithm, two public datasets have been considered: 32 pairs of mammographic images including MLO projection retrieved from Digital Database for Screening mammographic (DDSM) and 30 ones from Mammographic Image Analysis Society (mini-MIAS) databases. Results show that A
²
INET yields best results with respect to the other more conventional classifiers.