A Research of Pneumonia Detection Using EffcientNetV2L with Grad-Cam

The application of EfficientNetV2-L, an advanced deep learning model, for detecting pneumonia from chest X-ray images. Traditional diagnostic techniques, such as chest X-rays and lung biopsies, require skilled medical practitioners and can be time-consuming and error-prone, especially in resource-limited settings.By training EfficientNetV2-L on a dataset of chest X-ray images, the study classifies scans as either "Pneumonia" or "Normal." To enhance explainability, the model integrates Grad-CAM (Gradient-weighted Class Activation Mapping), which generates heatmaps highlighting critical areas influencing the model’s predictions. These visual explanations help clinicians understand the model's decision-making process.The proposed model achieved an impressive 94.02% accuracy, outperforming widely used architectures like CNN, ResNet50, and VGG16. Grad-CAM visualizations confirmed that the model effectively identifies key pneumonia indicators, such as consolidation and infiltration. The study concludes that combining EfficientNetV2-L with Grad-CAM offers a reliable and interpretable solution for pneumonia detection, improving diagnostic accuracy and aiding healthcare professionals in clinical decision-making.