Abstract
BACKGROUND: Open fractures are critical injuries that require prompt and accurate diagnosis to optimize treatment outcomes. Traditional methods often rely on manual interpretation of radiological images, which can be prone to human error. With advancements in deep learning, there is a significant opportunity to enhance the precision of fracture classification through automated systems. METHODS: Based on this, we developed a deep learning image processing model for the binary classification of tibiofibula open and closed fractures. By incorporating hybrid convolution with multi-scale convolutional kernels, we enhanced the model's feature extraction capabilities. To further optimize the feature selection process for orthopedic images, we integrated a channel attention mechanism, improving feature extraction without significantly increasing computational costs. RESULTS: The results of the comparative experiments with other models show that our model's accuracy (ACC) surpasses that of the state-of-the-art classification models by 4.71%. Additionally, the F1 score of our model demonstrates a 4.77% improvement in precision. Moreover, the results of the ablation experiments indicate that each component of the constructed model effectively enhances performance. CONCLUSION: The findings of this study suggest that the proposed deep learning model significantly enhances the detection accuracy of open fractures. By integrating advanced feature extraction techniques and leveraging pre-trained models, our approach offers a promising tool for clinical application in orthopedic imaging, potentially leading to improved patient outcomes through more accurate diagnoses. Future work will focus on expanding the model's capabilities to include other fracture types and real-world clinical scenarios.