Comparative Evaluation of AI models for Early Detection of Breast Cancer

Abstract

Most of the deaths caused by breast cancer are due to lack of early detection of tumors. Mammography, biopsies, and ultrasounds are commonly used for the detection of breast cancer, which have major limitations like high false positives, subjective interpretation, and the variability in radiologists’ skills. However, advancement in the Artificial Intelligence (AI) has significantly revolutionized the detection of cancers using image classification and analysis of historical mammography and MRI data. In this paper, six different models were evaluated on the Wisconsin Diagnostic Breast Cancer (WDBC) dataset. The performance of traditional machine learning models, particularly logistic regression (LR), random forest (RF), support vector machine (SVM), multi-layer perceptron (MLP), and deep learning approaches, particularly 1D convolutional neural networks (CNN) and deep neural networks, was compared using multiple evaluation metrics, including accuracy, precision, recall, and F1-score. The result highlights Logistic Regression as the most effective model for early detection of breast cancer, offering a reliable prediction on whether a tumor is malignant or benign. The research contributes invaluable insights into the performance of AI models and the implementation of such models for breast cancer detection.