Peering Inside the Black Box: Explainable AI to Interpret Advanced Computer Vision Fungal Pathogen Prediction

Antimicrobial resistance is a growing concern, with pathogenic fungi making a substantial contribution to untreatable life-threatening infections across the globe. Artificial intelligence is increasingly used in microbiology and antimicrobial resistance research, with promise to improve clinical diagnostics and infectious disease treatments. However, how AI models make predictions remains largely unknown, which is a major hurdle for human trust and regulatory approval. We trained convolutional neural networks (DenseNet121 and InceptionV3) and vision transformers (Swin Transformer-Tiny and Vision Transformer-Base 16) to quickly and accurately identify human fungal pathogens from microscopy images. Using explainable AI (Occlusion Sensitivity and Grad-CAM), we identified biologically relevant features (organelle, cell interior, cell wall, budding patterns/scars, and optical patterns) and irrelevant image features (background artifacts) high-performance computer vision models used to make predictions. These findings advance our understanding of how computer vision models make predictions on microbial pathogens and are anticipated to have profound implications for AI-based diagnostics.