Fig. 16
From: Trustworthy deep learning for malaria diagnosis using explainable artificial intelligence
Grad-CAM Visual Explanations for Malaria Detection Using Xception-Based Model (Left) and Inception-ResNetV2 Model (Right). The heatmaps illustrate the discriminative regions used by both models to identify parasitized cells in a blood smear. Warmer colors (red/yellow) indicate areas with higher importance. The Xception model focuses on broader contextual features, while the Inception-ResNetV2 model provides more centered and precise localization of the infected regions.
