Advancing Bone Fracture Risk Assessment and prediction through Spiking Neural Network

Bone Fracture (BF) is one of the most prevalent health issues affecting people, caused by accidents or medical conditions like bone cancer. This paper proposes a Spiking Neural Network (SNN), an innovative approach for detecting bone fractures. To identify fracture and non fracture X ray images, the input image is taken from bone fracture detection dataset that incorporates advanced techniques. Initially, image resizing and the Adaptive Gabor Filter (AGF) is applied to the dataset to resize the images and remove unwanted noise. Next, the processed images are segmented using Kernel K means clustering, which helps locate clusters that are not linearly separable with bone fractures. The segmented images are then extracted by using Principal Component Analysis (PCA) addressing the over fitting problem and aiding in data reduction for further analysis. Finally, a SNN framework is proposed to enhance the prediction of bone fractures from X ray images, overcoming challenges in risk analysis. The SNN classifies X ray images as either fractured or non fractured. The proposed approach is implemented using Python software and demonstrates that the SNN classifier achieves an improved accuracy of 96% compared to other techniques.