Meme Kanseri Tanısı İçin Derin Öznitelik Tabanlı Karar Destek Sistemi ; A Deep Feature Based Decision Support System for Breast Cancer Diagnosis

The breast cancer is the second leading cause of cancer deaths among women after lung cancer. Early diagnosis is quite significant with breast cancer treatment. Mammography is the most commonly used imaging technique for the early detection of breast cancer. Researches have been shown that the mortality rate can decrease by 30% for women who have mammogram regularly over 50 years of age (Jadoon et al. 2017). However, interpreting mammograms is often subjective. In this study, an integrated system for automated detection, classification, and content based retrieval of breast masses is presented. In this manner, physician’s decisions on mass were expressed by high-level deep features and low-level feature set. In proposed framework, to extract low-level features, a graph based visual saliency (GBVS) method is used for mass detection and the nonsubsampled contourlet transform (NSCT) and eig(Hess)-HOG are used for feature extraction. In addition, high-level convolutional neural network features have been used. Then, two extreme learning machine (ELM) classifiers rely on the features mentioned above is employed to predict category of test images. And outputs of classifiers based on each feature were examined together to define the kind of test image. The image retrieval and classification performances are evaluated and compared on IRMA mammographic dataset by using both the precision-recall (PR) and classification accuracies. Experimental results demonstrate the effectiveness of the proposed system and the viability of a real-time clinical application. ; Meme kanseri, akciğer kanserinden sonra kadınlarda kanser ölümlerinin ikinci önemli sebebidir. Erken tanı, meme kanseri tedavisinde oldukça önemlidir. Mamografi, meme kanserinin erken teşhisinde en çok kullanılan görüntüleme tekniğidir. Yapılan araştırmalar, 50 yaşın üstünde düzenli mamografi çektirmenin kadınlar için ölüm oranını oranında azaltabileceğini göstermektedir. Ancak, mamogramların yorumlanması genellikle özneldir. Bu çalışmada, göğüs kitlelerinin otomatik ...