Automated Vertebrae Diagnosis in Spinal X-ray Images using Artificial Intelligence

Spinal diseases affect millions of people worldwide, and early diagnosis is crucial for better treatment outcomes. However, the traditional methods of diagnosis can be time-consuming and inaccurate. Therefore, it is crucial to have a fast and accurate diagnosis of the spine to prevent potential problems that can arise if the condition is not treated on time. This study focuses on using a deep neural network that has been trained on an X-ray dataset to create an efficient and cost-effective framework for classifying spinal vertebrae X-ray scans. The proposed framework extracts feature from multiple layers using a lightweight deep neural network to diagnose the spine health, followed by three fully connected layers of classification for the retrieved information. The proposed framework achieved an accuracy of 91.33% on the validation set, and 92.22% on the test set, indicating the robustness of the proposed architecture. A comparison of the proposed model with other deep neural networks is proved its efficiency based on the results of the experiments. This will be beneficial in developing a diagnostic framework that can be easily adapted to low-cost computers, resulting in faster processing of image frames containing various spine vertebrae x-ray images, with low-time complexity. Further research is needed to validate our approach and integrate it into clinical practice.