Classification of Breast Masses in Mammograms Using 2D Homomorphic Transform Features and Supervised Classifiers

Barkana B. D., SARIÇİÇEK İ.

JOURNAL OF MEDICAL IMAGING AND HEALTH INFORMATICS, cilt.7, sa.7, ss.1566-1571, 2017 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 7 Sayı: 7
  • Basım Tarihi: 2017
  • Doi Numarası: 10.1166/jmihi.2017.2167
  • Dergi Adı: JOURNAL OF MEDICAL IMAGING AND HEALTH INFORMATICS
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Sayfa Sayıları: ss.1566-1571
  • Anahtar Kelimeler: Computer-Aided Diagnosis, Mammogram, Breast Cancer, Descriptive Statistics, Homomorphic Transform, Cepstral Coefficients, CANCER DETECTION, MRI
  • Eskişehir Osmangazi Üniversitesi Adresli: Evet

Özet

There are ongoing research efforts to develop accurate computer-aided diagnosis systems for detecting benign and malignant breast masses. Studies are mainly focusing on imaging modalities, new feature sets, and classification methods. In this paper, a new feature set is developed by using the two-dimensional homomorphic transform. The compatibility of the proposed features with the commonly used fractal dimension, perimeter, area, and the compactness features is investigated. The proposed set is based on the spectral information of the breast masses and achieved promising classification accuracies with well-balanced true positive and true negative rates. It is observed that the proposed set works as complementary to the other feature sets. Our experimental results verify that the 2 dimensional homomorphic transform (2D-HT) feature set improves the performance of the systems in classifying benign and malignant breast masses on mammograms. Digital Database for Screening Mammography database is used for testing. Artificial neural network and support vector machines classifiers are designed to evaluate the performance the features. Two-decision rule scenarios are defined by two threshold values. The proposed design allows radiologists to adjust the system's sensitivity and specificity by changing the threshold values.