Delayed disease diagnosis leads to postponed treatment, negatively impacting patient outcomes and broader public health. The need for a rapid diagnostic framework is evident, as analyzing biological signals and medical images is essential for assessing health conditions and the severity of the disease. Advanced imaging techniques, including X-rays, ultrasounds, and MRIs, play a crucial role in diagnosis, especially when enhanced by artificial intelligence and deep learning. This study focuses on X-ray imaging, the most accessible and cost-effective medical diagnostic tool. It introduces a hybrid metaheuristic optimization technique for hyperparameter tuning of convolutional neural networks (CNNs), advancing the field of image classification. The Xception model was employed for feature extraction, with data augmentation techniques enhancing its robustness. This research provides valuable insights for improving deep learning models in complex visual tasks, particularly in X-ray image classification, where limited datasets and performance optimization remain key challenges. While numerous techniques have been explored for disease detection, only a few have achieved exceptionally high accuracy. Comparative analysis with previous studies demonstrates that the proposed model attained outstanding accuracy coefficients of 0.9747, 0.9966, and 1.00 for four-class, three-class, and two-class classifications, respectively.
