An Improved Model for Breast Cancer Diagnosis by Combining PCA and Logistic Regression Techniques

Abstract

Breast cancer is weighed one of the most life-threatening illnesses confronting women. It happens when the multiplication of cells in breast tissue is uncontrollable. Several studies have been performed in the healthcare field for early breast cancer diagnosis. However, traditional methods can generate incomplete or misleading outcomes. To overcome these limitations, computer-aided diagnosis (CAD) systems are extensively exploited in the healthcare domain. It is designed to improve accuracy, decrease complexity, and reduce misclassification costs. The goal of this study is to present a breast cancer CAD system based on combining the Principal Component Analysis (PCA) method for feature reduction and Logistic Regression (LR) for BC tumors classification. The experiments have been conducted on Wisconsin Diagnosis Breast Cancer (WDBC) and Wisconsin Original Breast Cancer (WOBC) datasets from UCI repository using different training and testing subsets. Moreover, we carried out extensive comparisons of our approach with other existing approaches. Multiple metrics like precision, F1 score, recall, accuracy, and Area Under Curve (AUC) were used in this study. Experimental results indicate that the proposed approach records a remarkable performance rate with an accuracy of 1.00 and 0.98 for WDBC and WOBC respectively and outperforms the previous works by decreasing the number of features, improving the data quality, and reducing the response time.