Is Grad-CAM Explainable in Medical Images?
Subhashis Suara, Aayush Jha, Pratik Sinha, Arif Ahmed Sekh
TL;DR
Medical imaging DL models are often opaque, motivating explainable AI approaches. This work reviews explainable DL and Grad-CAM, and empirically evaluates Grad-CAM's ability to localize regions driving predictions in a metastasis-detection task on 96x96 histopathology patches from the PCam dataset, using a pre-trained DenseNet169 with a one-cycle training regime. Grad-CAM heatmaps are generated via fastai hooks and overlaid on grayscale images to assess interpretability, with a validation accuracy around 96.7% and qualitative heatmap examples highlighting tumor-region focus. The study suggests Grad-CAM can enhance interpretability in medical imaging, while acknowledging robustness limitations and the need for broader modality coverage and clinician-friendly interfaces.
Abstract
Explainable Deep Learning has gained significant attention in the field of artificial intelligence (AI), particularly in domains such as medical imaging, where accurate and interpretable machine learning models are crucial for effective diagnosis and treatment planning. Grad-CAM is a baseline that highlights the most critical regions of an image used in a deep learning model's decision-making process, increasing interpretability and trust in the results. It is applied in many computer vision (CV) tasks such as classification and explanation. This study explores the principles of Explainable Deep Learning and its relevance to medical imaging, discusses various explainability techniques and their limitations, and examines medical imaging applications of Grad-CAM. The findings highlight the potential of Explainable Deep Learning and Grad-CAM in improving the accuracy and interpretability of deep learning models in medical imaging. The code is available in (will be available).