Performance analysis of various classifiers on deep learning network for breast cancer detection
D. Selvathi, A AarthyPoornila · 2017
Cancer is a deadly disease that plagues mankind. 12.5%of women acquire breast cancer in their lifetime. Though, the mortality rate in women due to breast cancer is the second highest behind lung cancer, periodic clinical checkups and self-tests help in early detection and thereby significantly increases the odds of survival. Mammography is considered as the “Gold Standard” of non-invasive breast cancer detection technique. Screening for cancer through manual analysis of the medical images is tedious, time-consuming and impractical for large data. Thus, an automated, accurate and efficient detection system is needed to diagnose breast cancer within a short period of time. Deep learning techniques can revolutionize modern medicine by automating the detection process without compromising its accuracy. In this work, an automated system is proposed for achieving error-free detection of breast cancer using a Sparse Autoencoder (SAE) which learns feature representations from the mammogram and a classifier which is cascaded with the SAE performs the classification based on these learned features. Different classifiers like decision trees, K-nearest Neighbor (KNN), Support Vector Machine (SVM) and Random Forest classifier are used and their performances are compared. From the analysis, it is observed that the Random Forest Classifier performs better compared to other methods.