Mammographic Image Classification using Texture Features

In this work, the statistical Haralick features from the texture description methods such as Surrounding Region Dependency Matrix (SRDM), Spatial Grey Level Dependency Matrix (SGLDM), Grey Level Difference Matrix (GLDM) and Run-length features from the texture description method Grey Level Run-Length Matrix (GLRLM) are widely extracted features in mammogram images. The classification process is done using Support vector machine. The performance is evaluated using sensitivity, specificity, and accuracy .The performance of GLRLM features ,GLDM features ,SGLDM features are considerably poor where as SRDM features outperforms. Based performance we can analyze that the SRDM distinguish between malignant and benign on mammogram images with high accuracy.