Enhancing Breast Cancer Diagnosis with Integrated Dimensionality Reduction and Machine Learning Techniques

Abstract

Breast cancer remains a significant cause of cancer-related mortality worldwide, highlighting the critical need for advancements in diagnostic techniques. Recent diagnostic methods, while effective, often face limitations in accuracy and efficiency. This paper aims to differentiate between tumorous (malignant) and non-tumorous (benign) cases of breast cancer using three publicly available datasets: Wisconsin Breast Cancer (WBC), Wisconsin Diagnostic Breast Cancer (WDBC), and Wisconsin Prognostic Breast Cancer (WPBC) datasets. We applied popular supervised machine learning classifiers, including Multi-layer Perceptron (MLP), Support Vector Machine (SVM), Random Forest (RF), Decision Tree (DT), Naïve Bayes (NB), and K-Nearest Neighbor (KNN), in combination with dimensionality reduction techniques such as Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), and Factor Analysis (FA). The classifiers were evaluated based on accuracy, precision, recall and F1 score. The results show that, due to FA's emphasis on feature selection and noise reduction, the SVM with FA achieved the highest accuracy of 98.64% on the WBC dataset. MLP without any dimensionality reduction performed best with an accuracy of 98.26% on WDBC. Conversely, MLP and SVM with LDA yield an accuracy of 89.80% on the more complex and noisy WPBC dataset.