D Kalaivani
Mammography is one of the best methods is early detection of breast cancer. Mammographic Institute Society Analysis [MIAS] dataset is used for experimentation. The use of feature selection can improve accuracy, efficiency, applicability and understandability of a learning process. For this reason, many methods feature selection have been developed. In this work, the rough set theory based unsupervised feature selection method using relative dependency measures is proposed. The method employs a backward elimination-type search to remove features from original features. The statistical Haralick feature from the texture description methods GLCM,GLDM and SRDM are widely used to extract feature in mammogram images for analysis and classification of microcalcification. Finally, we compared results of the proposed algorithm by using the mammogram image datasets for breast cancer diagnosis. The performance of USRR algorithm is compared with the USQR algorithm. From that the USRR produces high accuracy reat when compare with the USQR. The mean absolute error also reduced. An experimental result shows the ability and high performance of the algorithms.
-