Hybrid Attention Mechanisms and Bio-Inspired Optimization for Enhanced Breast Cancer Diagnosis from Ultrasound Images

Breast cancer is one of the most prevalent cancers worldwide. Early detection through medical imaging screening and timely diagnosis are crucial for improving prognosis. How-ever, interpretation of complex ultrasound scans is an arduous task even for experienced radiologists. Computer-aided diagnosis (CADx) systems can aid clinicians by automatically analyzing images to identify malignant cases. This research presents a novel deep learning framework with hybrid attention mechanisms and bio-inspired optimization (HAM-BIO) to enhance ultrasound-based breast cancer diagnosis. Custom residual blocks equipped with multi-level attention modules help selectively focus on discriminative imaging attributes indicative of tumors. Grey Wolf Optimization adaptively evolves model hyperparameters for superior topology and convergence. Evaluations on two breast ultrasound datasets achieve state-of-the-art classification accuracy, 96.7% on dataset 1 and 88.9% on dataset 2, sur-passing existing CADx systems. The models optimization and interpretability enhancements improve automated breast cancer screening without additional data. These results highlight the techniques clinical utility and potential. By integrating compu-tational intelligence and explainable AI, the approach sets new standards for supporting radiologists with accurate diagnostic assistance.