Cancer Mammography Detection Using Four Features Extractions on Gray Level Co-occurrence Matrix with SVM Kernel Analysis

Cancer is one of the most dangerous diseases that often threaten human life. This study aims to detect breast cancer using the best kernel by performing a kernel analysis to obtain high accuracy breast cancer detection based on image analysis. The kind of SVM Kernel in this research uses Polynomial, Gaussian, and Radial Basis Function (RBF). The proposed method can help the medical personnel to make it easier to detect breast cancer by sending emails to doctors to immediately notice results so that patients caught with cancer can directly get special treatment. Suppose the image is detected as cancer, then the systems sending the result by e-mail to the Doctor who treats them so that patients who are detected with cancer can immediately get special treatment. This research uses a combination of Gray Level Co-occurrence Matrix (GLCM) with a distance equal to 1 and angle direction (0°, 45°, 90°, 135°). The feature extractions of the GLCM matrix are obtained from Energy, Homogeneity, Contrast, and Entropy. Furthermore, Support Vector Machine (SVM) is the classification method to classify non-cancer and cancer. Analysis, an accuracy result in different types of SVM Kernel were conducted in this experimental research. The detailed accuracy result is 93%; The sensitivity is 91%; The precision is 96%; The specificity is 95%, and The F1-score is 93%. The best accuracy of SVM Kernel is RBF. In the future, this study can be used in hospitals to make it easier to check for breast cancer.