Mammographic mass discrimination using K-Nearest Neighbor and BIRADS attribute

Abstract

The mammography is the most effective method for the detection of breast cancer, however, the predictive value is low, and it can lead to unnecessary biopsies with benign results. This research aims to develop a predictive model for discrimination of mammographic masses using KNN and BIRADS attributes with an acceptable level of Accuracy, Precision, Recall and F1-Score. For this, we carried out the following phases: Data cleaning, KNN algorithm training and selection. The result obtained was a mammographic mass discrimination model with an accuracy=85% and acceptable levels of precision, sensitivity and F1-score. We concluded that it is possible to use this model as an element of judgment for the diagnosis of breast cancer; also that through the error rate it is possible to find optimal KNN models.