This research demonstrates that ResNet152 outperforms VGG16 and MobileNet when classifying pneumonia types in chest X-rays, highlighting advanced AI's clinical utility.
Key Points
ResNet152 achieved the highest accuracy of 89% for pneumonia detection, showing superior performance for COVID-19 and viral pneumonia cases.
The comparative analysis included three models: VGG16, MobileNet, and ResNet152, with a dataset of 5,863 chest X-ray images.
Transfer learning and data augmentation techniques were employed to enhance the robustness of the models during training.
Utilizing larger, balanced datasets can significantly improve the diagnostic performance of AI systems in clinical settings.