Enhanced Fibonaccinet: Attention-Based Deep Learning For Balanced Bone Fracture Classification In X-Ray Images

Abstract

This study introduces Enhanced FibonacciNet, a cutting-edge deep learning model designed to automatically identify bone fractures in X-ray pictures..It focuses on differentiating between comminuted fractures (many fragments) and simple fractures (single-line breaks), an essential step for precise diagnosis and prompt treatment. The model, which was trained on a dataset of 16,061 original and augmented images, includes features like Depthwise Separable Convolutions for increased efficiency, Avg2Max Pooling for combining texture and intensity cues, Area Attention for capturing detailed features, and Central Region Focus for highlighting possible fracture areas. With an accuracy of 91%, an F1-score of 0.91, and a ROC AUC of 0.98, the model demonstrated strong dependability under practical circumstances. It is appropriate for hospitals, diagnostic facilities, and remote healthcare settings because it provides radiologist-level performance with fewer false predictions and quick processing while being lightweight and scalable.

Ähnliche Arbeiten