Texture feature dimensionality reduction-based mammography classification using Random Forest

Abstract

Breast cancer is the most frequent cancer and the leading cause of death among females. Diagnosis mass from mammogram correctly can reduce the unnecessary biopsy to a large extent. In this paper, we present a novel mammogram classification method combining the Random Forest and the Locally Linear Embedding (LLE) dimensionality reduction algorithm for texture features. The proposed method consists of three stages. In the first stage, preprocessing is performed to enhance the contrast and suppress the noise of the ROI images. Then, the sixteen-dimensional texture features are extracted from Grey Level Co-occurrence Matrix (GLCM) as the input dataset of LLE and being mapped into a five-dimensional subspace. Finally, a Random Forest classifier is investigated for the mammogram classification and compared with the other four classifiers (SVM, KNN, Logistic Regression, MLPC). The experimental results show that the Random Forest classifier outperforms than the others, with an average accuracy of 92.87% and the AUC value of 0.99, that indicates that the combination of LLE algorithm and Random Forest classifier is a promising method for the mammogram classification.