Knowledge distillation-based lightweight Convolutional Neural Networks (CNN) model for efficient breast cancer detection

Breast cancer (BC) remains a leading cause of cancer-related mortality among women globally, with early detection playing a crucial role in improving patient outcomes. While deep learning models—particularly Convolutional Neural Networks (CNNs)—have demonstrated exceptional performance in breast cancer detection, their high computational demands limit deployment in low-resource environments such as mobile devices and rural clinics. To bridge this gap, we propose a lightweight CNN model for breast cancer detection using Knowledge Distillation (KD), a technique that transfers knowledge from a complex, high-capacity "teacher" model to a compact and efficient "student" model. In this study, we develop and evaluate both teacher and student models using the Wisconsin Diagnostic Breast Cancer (WDBC), Breast Cancer Diagnosis (BCD), and Primary Breast Cancer vs Normal Breast Tissue (PBCT) datasets. The student model is designed to operate with only 0.06% of the parameters used by the teacher model, significantly reducing memory and computational overhead. Despite its lightweight architecture, the student model achieves up to 100% classification accuracy and demonstrates robust generalization across multiple datasets. Our approach enables high-accuracy breast cancer detection while ensuring fast inference and low resource consumption, making it well-suited for deployment in real-world, resource-constrained environments such as mobile health platforms and embedded medical devices. The findings highlight the transformative potential of knowledge distillation in democratizing access to advanced AI-driven diagnostics.