Answers are grounded in this paper and cite their sources.
Pneumonia remains a significant global health problem, predominantly impacting vulnerable groups such as young children and the elderly. Timely and accurate diagnosis is essential to ensure effective treatment and reduce the likelihood of complications. While chest X-rays (CXR) are a first-line imaging modality in the diagnosis of pneumonia, they can be challenging and subject to human error to interpret. This work proposes an advanced deep learning-based approach for the enhancement of pneumonia diagnosis from CXRs using convolutional neural networks (CNNs) and Explainable AI (XAI) techniques. Various pre-trained models, including Inception V3(94% accuracy), MobileNetV2 (95%), NASNet (94%), ResNet152 (86%), Xception (93%), and a hyperparameter-optimized version of MobileNetV2 with the highest accuracy of 98%, were employed. Saliency maps were used to provide visual interpretability to allow clear and reliable decision-making in revealing the most important areas in the diagnosis. The integration of XAI not only increases transparency but also allows clinical validation. Generally, the designed method has good medical performance and gives a reliable and explainable tool to assist doctors in alleviating diagnostic duties, especially in limited-resource settings.
No takes yet. Share an insight, caveat, or question.
Mazumder et al. (2025) studied this question.