Wildan Iswahyudi, Mochamad Farhan Ali Irfani, Yogi Dwi Mahandi, Ilham Ari Elbaith Zaeni, Siti Sendari, Triyanna Widiyaningtyas
This study evaluates the performance of three deep learning models-MobileNetV2, DenseNet, and Xception-in the detection of lung diseases using chest X-ray images. The datasets used in this study consist of four classes: Normal, Pneumonia, Pneumothorax, and Cardiomegaly. Among the models tested, MobileNetV2 achieved the highest accuracy at 99.60%, surpassing DenseNet, which recorded 99.20%, and Xception, which achieved 98.96%. These results highlight MobileNetV2's superior ability to accurately diagnose pulmonary conditions from medical images, making it a highly effective tool in the context of automated medical diagnosis. The study's findings underscore the critical role that model selection plays in the field of medical image analysis, where diagnostic accuracy is paramount. By demonstrating higher performance, MobileNetV2 presents itself as a promising candidate for integration into clinical workflows, offering the potential to enhance the reliability and efficiency of lung disease detection. This research contributes to the growing body of evidence supporting the application of deep learning models in healthcare, emphasizing the need for continued innovation and optimization in medical imaging technologies to improve patient outcomes. © 2024 IEEE.
State University of Malang, Department of Electrical and Informatics Engineering, Malang, Indonesia
