Contrastive Learning-Driven Representation and Feature Selection for Spinal Fracture Detection on CT Images

The objective of this project is to develop an automated system for detecting spinal fractures from CT images using deep learning techniques. By extracting local and global anatomical features, the system aims to accurately classify images as fractured or normal. It leverages advanced convolutional and residual neural network architectures to learn meaningful representations, reduce overfitting, and improve prediction reliability. The goal is to assist medical professionals in faster and more accurate diagnosis, enhancing patient care while minimizing manual inspection effort.