Klasifikasi Kanker Payudara pada Citra Mammogram dengan Metode Ekstraksi Fitur Compound Local Binary Pattern

Abstract

Breast cancer is a serious disease that can be life-threatening if not diagnosed and treated in a timely manner. Therefore, early detection through mammography screening is crucial for detecting abnormalities in the breasts before the onset of noticeable symptoms. However, mammogram images have inherent limitations in quality, which can pose challenges for doctors or radiologists in analyzing them. To address this issue, computer vision techniques are employed using the local binary pattern (LBP) and compound local binary pattern (CLBP) feature extraction methods, along with support vector machine (SVM) classification. The CLBP method is an extension of the LBP method, where instead of using P-bits to encode the difference between the grayscale value of a central pixel and its neighbors, CLBP utilizes 2P-bits. This additional bit is expected to enhance the robustness of the feature descriptor by incorporating additional local information that is discarded by the LBP operator. The Mammographic Image Analysis Society (MIAS) dataset consisting of 322 images, with 207 normal breast images and 115 abnormal breast images, is utilized for this study. Testing is performed with various data training and testing splits, and the performance of the model is evaluated using a confusion matrix. The results show that the CLBP-SVM model with a 90:10 data split ratio achieves the best performance, with an accuracy of 96.97%, precision of 98.00%, and recall of 94.44%.