Enhancing Breast Cancer Detection via Optimized Machine Learning

Abstract

Breast cancer stands as the most prevalent malignancy among women in India. Typically, when a patient presents with breast symptoms or exhibits abnormalities on imaging tests like mammography suggestive of breast cancer, a recommendation for breast biopsy ensues. This study zeroes in on one prevalent biopsy method, namely Fine Needle Aspiration (FNA). FNA entails extracting a small sample of breast tissue or fluid from a suspicious area using a fine, hollow needle, subsequently scrutinizing it for cancerous cells. The objective of this research centers on leveraging the Breast Cancer Wisconsin Dataset, which encompasses numerical attributes characterizing cell nuclei features post FNA biopsy. The aim is to employ various machine learning classifiers on this dataset to discern between benign and malignant types of breast cancer. Experimental findings indicate that the accuracy of classification improves through hyperparameter optimization and refinement of classifiers. Such enhancements lead to more precise predictions, potentially resulting in the preservation of more lives through timely and accurate identification of breast cancer types.