Breast Cancer Classification in Mammograms Using Support Vector Machines (SVMs) and Histogram of Oriented Gradients (HOG)

This work focuses on breast cancer classification in mammograms using SVMs classifier and histogram of oriented gradients features. The envisioned system is designed in such a way to enhance the diagnostic precision through integration of the virtues of both SVMs and HOG in identifying discriminative characteristics of breast tissue. Special emphasis was given on comparing and testing the performance of the intended model with other models on a public mammogram dataset containing both malignant and benign cases; the optimized SVM model provided an accuracy of 92%. 5%, sensitivity of 91. 0%, specificity of 93. 74 (fig 5) and 8%, and an AUC-ROC of 0. 95. This study shows that the method is efficient in discriminating between the malignant and benign first axillary lymph node tissue. The developed CAD system was found to possess the prospect of becoming a valuable support system for radiologists in the early detection of breast cancer. Keywords: Breast cancer; Mammograms; Support Vector Machines; Histogram of Oriented Gradient Descent; Computer-aided diagnosis.