DeepSpine: Multi-Class Spine X-Ray Conditions Classification Using Deep Learning
Sheshang Degadwala, Vinay Nagarad Dasavandi Krishnamurthy, Dhairya Vyas
Abstract
Addressing the complex challenges inherent in the automated analysis of spine X-rays, our research introduces DeepSpine, a deep learning model designed for multi-class classification of diverse spine conditions. Employing convolutional neural networks (CNNs), DeepSpine exhibits remarkable proficiency in identifying various abnormalities, including Scoliosis, Osteochondrosis, Osteoporosis, Spondylolisthesis, Vertebral Compression Fractures (VCFs), Disability, Other and Healthy. Trained on a Kaggle dataset, the model achieves high accuracy and robustness. Incorporating transfer learning techniques enhances DeepSpine's generalization across different datasets, presenting a promising avenue for automated diagnosis and decision support in musculoskeletal radiology. This research contributes to the ongoing intersection of artificial intelligence and medical imaging, showcasing the potential of deep learning to revolutionize spine X-ray analysis and improve clinical outcomes.