Akademik Veri Yönetim Sistemi
Tezin Türü: Yüksek Lisans
Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Mühendislik Fakültesi, Elektrik ve Elektronik Mühendisliği Bölümü, Türkiye
Tezin Onay Tarihi: 2015
Öğrenci: CANSU AKBAY
Danışman: NEVZAT GÜNERİ GENÇER
Özet:Breast cancer is the second leading cause of cancer deaths for women. Mammography is the most effective technology presently available for breast cancer screening, despite the fact that there are still some limitations of the imaging technique, such as insufficient resolution, low local contrast and noise combined with the subtle nature of the usual radiographic findings. One of the most important radiographic findings associated to the existence of breast cancer is the clustered microcalcifications. Especially, it has been shown that some characteristics concerning the clustering parameters of microcalcifications are of great diagnostic value. However, the mentioned limitations of mammography make the detection and interpretation of microcalcifications a complicated task. The main purpose of this thesis is to develop Computer Aided Diagnosis (CAD) system in order to increase the efficiency of the mammographic screening process. The system may provide automated detection of microcalcification clusters leading a considerable decrease in misdiagnosis rates. To make microcalcifications more visible than their surrounding tissues, image enhancement on mammograms is required. There are many contrast enhancement algorithms that can be employed for the same purpose. However, by contrast enhancement it is expected to reduce overlap between tonal values that belongs to microcalcifications and their surrounding tissues instead of stretching the histogram of an image. The algorithms based on multiresolution analysis such as wavelet transform, contourlet transform and the detail enhancement on local frequency algorithms are considered to achieve this purpose. In this study, these algorithms have been implemented on clinical data and their performances are compared by using quantitative methods and evaluated under supervision of radiologists. To observe the efficiency of enhancement on classification of microcalcification clusters, selected regions from original images and images enhanced by using 2 enhancement methods chosen by a radiologist are used. To classify microcalcification clusters as benign and suspicious, features are extracted by using texture analysis. For classification the Support Vector Machine (SVM) is employed. As a result, best classification is obtained by features obtained from Gabor filter banks and enhanced images with a detail enhancement method with using mean and standart deviation by 77 % truth rate .(the area under the ROC curve is 0.81) .