Sliding Window Based Support Vector Machine System for Classification of Breast Cancer Using Histopathological Microscopic Images

Abstract

Breast cancer is one of the most common cancers in women's community, which is responsible for millions of death cases. The early diagnosis of breast cancer in its early stages increases the chances of healing; the most common current diagnosis methods are based on digital mammogram images and histopathological images. Automatic diagnosis of breast cancer and classifying the type of cancer is recommended to decrease the error that can be caused by humans. Recently, many systems have been developed to diagnose breast cancer by extracting textural and non-textural features from digital mammograms or histopathological images. This paper proposes a new sliding window technique for feature extraction where the local binary pattern (LBP) features are used. In this technique, each image produces 25 sliding windows. Features extracted from each window are saved and used to build a Support Vector Machine (SVM) classifier. The SVM classifier is used to classify each image into benign and malignant based on its most common windows classes. The system can be used to localize the cancerous tissues from the whole histopathological image. The proposed method achieved an overall accuracy of 91.12%, sensitivity of 85.22%, and specificity of 94.01%. Which is considered high when compared with other systems in the literature. The system can extend to extract more features and a comparison between different machine learning algorithms can performed.