Tumor Detection Using Deep Learning and Explainable Artificial Intelligence

Deep learning methods in artificial intelligence (AI) have transformed the automated detection of tumors in vital organs due to their exceptional image classification capabilities. However, their lack of transparency frequently impedes the establishment of clinical trust and the widespread adoption of these models. This study presents a deep learning methodology for automatically classifying and detecting brain and lung tumors using a medical imaging dataset. In order to tackle the inherent opacity of deep learning models and improve its interpretability, two XAI(explainable AI) techniques: Integrated Gradients and Saliency Maps are used in this study. The latest Convolutional Neural Network(CNN) architecture, ResNet50, is used in this study to provide a thorough method for identifying lung and brain tumors from medical imaging data. Additionally saliency maps and integrated gradients provide the visual depiction of the model's focus by highlighting the areas in images that have the highest impact on the model's classification decision. This approach not only attains high accuracy in tumor detection but also provides the interpretability and transparency of black-box AI models, resulting in increased trust in healthcare providers. Future work will investigate the adoption of other categories of XAI techniques and expand this framework to include other forms of medical imaging data, thereby emphasizing the signifi-cance of AI in healthcare.