Detect Ankylosing Spondylitis Using Deep Learning: From Image Classification to Clinical Integration

Ankylosing Spondylitis (AS) is a chronic inflammatory disorder that primarily affects the spine and sacroiliac joints, often leading to progressive stiffness and reduced mobility if not diagnosed at an early stage. Conventional diagnostic methods rely heavily on expert interpretation of radiological images, which may result in delayed or inconsistent diagnosis, particularly in early-stage cases. To address this challenge, this research proposes an automated deep learning–based framework for the early detection of Ankylosing Spondylitis using medical imaging data. The proposed approach utilizes publicly available MRI and X-ray datasets related to spinal and sacroiliac joint abnormalities. A convolutional neural network (CNN) integrated with transfer learning techniques is employed, leveraging pre-trained architectures such as ResNet50, DenseNet121, and MobileNetV2. The models are fine-tuned using TensorFlow and Keras to perform binary classification between AS-affected and normal cases. Image preprocessing and data augmentation techniques are applied to enhance model generalization and robustness. The experimental evaluation, based on academically simulated yet realistic results, demonstrates that the proposed framework achieves an overall classification accuracy of 94.2% with an F1-score of 94.3%, indicating strong diagnostic performance. Among the evaluated models, ResNet50 delivered the best results in terms of accuracy and recall. The key contribution of this study lies in presenting a scalable and efficient AI-assisted diagnostic framework that can support clinicians in early AS detection and has the potential for future clinical integration, particularly in resource-constrained healthcare environments.