Quantum Convolutional Neural Network for Bone Fracture Classification from X-Ray Images

Hiren Mewada, Ivan Miguel Pires, Mrugendrasinh Rahevar, Narendra Khatri

Accurate and efficient classification of bone fractures from X-ray images is crucial for timely diagnosis, effective treatment planning, and improved patient outcomes in orthopedic medicine. Convolutional neural networks (CNNs) have demonstrated their ability to automatically extract relevant features from X-ray images, making them a powerful tool for the classification of complex bone fracture patterns. However, as the complexity of the classification task increases, the computational requirements of classical CNNs also grow rapidly. This poses a challenge in high-stakes medical applications, where the accuracy and efficiency of the classification system are paramount. In this work, we present a quantum convolutional neural network (QCNN) for the classification of bone fractures from X-ray images. A quantum convolution layer is integrated as the first layer of the network, which leverages the unique properties of quantum mechanics to extract more informative features from the input X-ray images. A traditional CNN network follows this quantum-based feature extraction step to perform the final bone fracture classification. We tested the proposed QCNN model on the Bone Fracture Multi-Region X-ray Data, a benchmark dataset for evaluating bone fracture classification systems. The QCNN architecture achieved an impressive 96% accuracy and 95.81% F1-score, demonstrating its superior performance compared to state-of-the-art classical CNN models.