The Analysis of Digital Mammograms Using HOG and GLCM Features
Krishna Chaitanya Tatikonda, Chandra Mohan Bhuma, Srinivas Kumar Samayamantula · 2018
An algorithm for early detection of breast cancer is proposed in this paper. Breast cancer is one disease if detected early, can be cured effectively. Failure of early detection is causing many deaths among woman worldwide. Early detection requires proper screening. Digital mammograms are useful for this purpose as they are non-invasive. But data available from the mammograms is often difficult to interpret and analyse. Many algorithms are available in the literature for analysis of digital mammograms. In this work, an algorithm using the concatenation of simple Histogram of Oriented Gradients (HOG) and Grey Level Co-Occurrence Matrix (GLCM) features are used for classification. The database used is a benchmark database MIAS. All the abnormal images of the database are chosen for the experimentation. The images in the database are of poor contrast and are having some noisy pixels. ROIs are extracted from the images. All the ROI (Region of Interest) images are pre-processed with median filtering and Contrast Limited Adaptive Histogram Equalization (CLAHE). Several classifiers are used to assess the performance of the proposed algorithm i.e., KNN, CT, SVM (RBF kernel), ANN, LDA, and Naive Bayes classifier. Performance metrics used in this work are sensitivity, specificity, precision and accuracy. Results are compared with the existing algorithms available in the literature and superiority of the proposed method is presented as the proposed method obtained an accuracy of 99.11%.