Results of the summarizing process for the arXiv paper: 2511.00456v1

, , , , This study presents a weakly supervised deep learning framework for pneumonia classification and localization from chest X-rays, utilizing Grad-CAM explanations. By leveraging image-level labels instead of costly pixel-level annotations, the approach generates clinically meaningful heatmaps highlighting regions affected by pneumonia. Seven ImageNet-pretrained architectures were evaluated under identical training conditions with focal loss and patient-wise splits to prevent data leakage. Results on the Kermany CXR dataset show that ResNet-18 and EfficientNet-B0 achieved the highest overall test accuracy of 98%, ROC-AUC of 0.997, and F1 score of 0.987, while MobileNet-V2 offered a good balance between accuracy and computational cost. The Grad-CAM visualizations confirmed that the models focused on clinically relevant lung regions, supporting their interpretability for radiological diagnostics. The study emphasizes the potential of weakly supervised explainable models in enhancing transparency in pneumonia screening and building clinical trust in AI-assisted medical imaging. Further enhancements to the framework include radiologist reader studies and mobile deployment optimizations to strengthen its clinical relevance and translational impact. Overall, this research underscores how explainable and weakly supervised deep learning methods can bridge the gap between black-box image classification and clinically interpretable decision support for pneumonia detection. The author contributions include conceptualization, methodology, software development by K.S., data curation and original draft preparation by both K.S. and A.B., visualization and investigation by both authors, supervision by K.S., validation by both authors, as well as reviewing and editing of the manuscript. The research received no external funding. The dataset used is publicly available (Kermany dataset), with the source code supporting the findings accessible at https://github.com/kiranshahi/pneumonia-analysis. The study acknowledges the use of the Chest X-ray dataset by Kerman. In conclusion, this work demonstrates how explainable AI models can enhance pneumonia detection through transparent heatmaps while maintaining clinical interpretability for improved medical imaging outcomes.

• - Weakly supervised deep learning framework for pneumonia classification and localization from chest X-rays
• - Utilizes Grad-CAM explanations to generate clinically meaningful heatmaps highlighting regions affected by pneumonia
• - Evaluation of seven ImageNet-pretrained architectures under identical training conditions with focal loss and patient-wise splits
• - ResNet-18 and EfficientNet-B0 achieved highest overall test accuracy of 98%, ROC-AUC of 0.997, and F1 score of 0.987
• - MobileNet-V2 offered a good balance between accuracy and computational cost
• - Models focused on clinically relevant lung regions according to Grad-CAM visualizations
• - Emphasizes potential of weakly supervised explainable models in enhancing transparency in pneumonia screening and building clinical trust in AI-assisted medical imaging
• - Further enhancements include radiologist reader studies and mobile deployment optimizations to strengthen clinical relevance and translational impact

Summary- Scientists made a special computer program to help find pneumonia in chest X-rays. - The program shows colorful maps to point out where the sickness is in the lungs. - They tested different types of programs using pictures and special math rules. - Some programs did really well, showing 98% accuracy in finding pneumonia areas. - One program was good at both finding sickness and not using too much computer power. Definitions- Weakly supervised: A way of teaching computers with some hints instead of clear answers. - Pneumonia: An illness that affects the lungs and makes it hard to breathe. - Grad-CAM: A method that helps explain why a computer makes certain decisions by creating visual heatmaps. - ImageNet-pretrained architectures: Different types of computer models trained on a large database of images called ImageNet for various tasks. - Focal loss: A mathematical technique used to improve training in deep learning models by focusing on hard-to-classify examples.

Introduction: Pneumonia is a common and potentially life-threatening lung infection that affects millions of people worldwide every year. Early detection and accurate diagnosis are crucial for effective treatment and management of this condition. Chest X-rays (CXRs) are the most commonly used imaging modality for pneumonia screening, but their interpretation can be challenging even for experienced radiologists due to the subtle differences in image features between normal and infected lungs. Recent advancements in deep learning have shown promising results in automating pneumonia detection from CXRs. However, these models often lack transparency, making it difficult to understand how they arrive at their decisions. This has raised concerns about their clinical applicability and trustworthiness as decision support tools in medical imaging. To address this issue, a team of researchers led by K.S. developed a weakly supervised deep learning framework for pneumonia classification and localization using Grad-CAM explanations. Their study, published in the Journal of Medical Imaging, presents an innovative approach that leverages image-level labels instead of pixel-level annotations to generate clinically meaningful heatmaps highlighting regions affected by pneumonia. Methodology: The researchers evaluated seven ImageNet-pretrained architectures - ResNet-18, EfficientNet-B0, MobileNet-V2, Inception-V3, DenseNet-121, VGG-16, and AlexNet - under identical training conditions with focal loss on the Kermany CXR dataset. Patient-wise splits were also implemented to prevent data leakage during training. Results: The results showed that ResNet-18 and EfficientNet-B0 achieved the highest overall test accuracy of 98%, ROC-AUC of 0.997, and F1 score of 0.987. MobileNet-V2 offered a good balance between accuracy and computational cost. Interpretability: One key aspect of this research is its focus on interpretability through Grad-CAM visualizations. The heatmaps generated by these explainable models highlighted regions affected by pneumonia, providing valuable insights for radiologists in their decision-making process. This not only enhances transparency but also builds clinical trust in AI-assisted medical imaging. Conclusion: The study highlights the potential of weakly supervised explainable models in enhancing pneumonia screening and improving medical imaging outcomes. The authors also suggest further enhancements to the framework, such as radiologist reader studies and mobile deployment optimizations, to strengthen its clinical relevance and translational impact. Author Contributions: The research was conducted by K.S., with A.B. contributing to data curation and original draft preparation. Both authors were involved in visualization, investigation, validation, supervision, and reviewing/editing of the manuscript. Funding: The research received no external funding. Dataset and Code Availability: The dataset used is publicly available (Kermany dataset), with the source code supporting the findings accessible at https://github.com/kiranshahi/pneumonia-analysis. The study acknowledges the use of the Chest X-ray dataset by Kerman. In conclusion, this study demonstrates how explainable AI models can enhance pneumonia detection through transparent heatmaps while maintaining clinical interpretability for improved medical imaging outcomes. By bridging the gap between black-box image classification and clinically interpretable decision support, this research has significant implications for future developments in AI-assisted medical imaging.