Adaptive Hybrid Models for Enhanced Breast Cancer Classification

Abstract Breast cancer remains one of the leading causes of mortality among women worldwide, making early and accurate diagnosis critical. Artificial intelligence (AI)–based approaches have demonstrated significant potential for automated breast cancer classification; however, existing models often suffer from overfitting, loss of fine image details, and extensive hyperparameter tuning. To address these limitations, this paper proposes an Adaptive Hybrid Model Infused with Physics Insights (AHM-PI) for breast cancer classification. A transformed dynamic and adaptive filtering strategy is first introduced to balance noise suppression and edge preservation. Subsequently, a robotic physics-informed deep learning model incorporating diffusion-based convolution, batch normalization, activation, pooling, physics-driven optimization, and classification layers is developed. The proposed approach is evaluated on a breast ultrasound dataset and compared with state-of-the-art methods including AOADL-HBCC, DTLRO-HCBC, Inception-v3 variants, VGG-16, and ResNet. Experimental results demonstrate a superior classification accuracy of 70.56%, validating the robustness and effectiveness of the proposed framework for breast cancer diagnosis.