Breast cancer is a more common neoplasm among women (not considering non-melanoma skin cancer). The estimate for the coming years is still growing and poses a threat to human health. Currently, the methods used in the diagnosis of breast cancer are performed through analysis of mammography images. Allowed, an analysis made by two specialists, which are subject to errors due to factors such as fatigue and lack of capacity. Not only the factor of human errors in diagnoses, certainly the long periods of time until the final diagnosis is another factor to be taken into account, because cancer is a progressive disease over time. In this sense, the present work applied a solution through the automatic classification of mammography images, in order to determine as normal or cancer. In addition, for simulations, two machine learning techniques were added independently, as they can eventually serve as a support in the diagnosis of breast cancer, that is, a CAD system, which means “computer-aided diagnosis”. As machine learning techniques applied for classification referenced as convolutional neural networks and support vector machines. Subsequently, the construction of the classification algorithms, they were subjected to the testing phase, which was found to be more than 85% accurate in the classification of mammography images.
Analysis of Machine Learning Techniques Applied to the Classification of Masses and Microcalcification Clusters in Breast Cancer Computer-Aided Detection
