Interpretation of mammograms with rotation forest and PCA

Abstract

Discrimination of benign and malignant mammographic masses based on supervised and unsupervised learning methods help physicians in their decision to perform a breast biopsy on a suspicious lesion seen in a mammogram. For predicting the outcomes of breast biopsies, we propose Rotation Forest with twelve decision trees algorithms as base classifiers and Principal Component Analysis (PCA) as filter used to project the data. Experimental results demonstrate the effectiveness of the proposed method compared to one single classification system: higher classification accuracy and smaller number of leaf nodes and size of tree.