Computer-aided diagnosis of digital mammograms using Gabor wavelets

Abstract

Digital mammogram X-ray is commonly used for breast cancer diagnosis, where computer aided diagnosis (CADx) algorithms are used to help the radiologists process the large volume of data with more accurate diagnosis. In this study, we developed a new CADx algorithm applied and tested on digital X-ray mammogram images from a standard test database from the Mammographic Image Analysis Society (MIAS). The algorithm starts by extracting features using Gabor wavelet transform of different level of orientation and wavelengths. After that, the dimension of the extracted features is reduced through Principal Component Analysis (PCA) followed by Support Vector Machine (SVM) classifier of Gaussian kernel. The model perfectly fitted the training data with 100% accuracy, sensitivity, and specificity in detecting malignant cases versus benign ones. Furthermore, the model performed well on testing set with 90% accuracy, 100% sensitivity, and 89% specificity.