Deep Learning-assisted Fracture Diagnosis: Real-time Femur Fracture Diagnosis and Categorization

Abstract

Fractures necessitate accurate and timely diagnosis to enable proper medical treatment, and are a frequent occurrence in the human body. X-ray image-based fracture detection through manual means is prone to human error and is a time-intensive process. This study proposes a new method for fracture diagnosis using AI-assisted techniques, particularly deep learning. The goal is to allay worries regarding fracture detection. A DNN model was created in order to quickly identify and classify femur fractures. For improving model accuracy and preventing overfitting when working with tiny datasets, data augmentation techniques are essential. Three trials were used in the study to assess the model’s effectiveness utilizing the Adam optimizer and softmax activation. Fivefold cross-validation was used to assess the proposed model’s capability to distinguish between healthy and fractured femurs. A 92.44% accuracy percentage was attained. On numerous subsets, the data accuracy was examined, and it was discovered to be greater than 95% on 10% of the data and 93% on 20% of the data. The model is more effective than currently used methods. The study examines how quickly and precisely artificial intelligence (AI) and deep learning can diagnose femur fractures. The study suggests an AI system that might revolutionize the identification of femur fractures. To improve speed and detection accuracy, the technique incorporates real-time analysis and classification of femur fractures.

Ähnliche Arbeiten