Explainable AI and GAN-based Approaches for Lung Cancer Diagnosis: A Comprehensive Survey
Abstract
Globally, lung cancer continues to be the leading cause of cancer-related death, and treatment effectiveness significantly declines with an advanced diagnosis. The diagnostic accuracy of multimodal imaging methods, such as radiology, cytology, and histopathology, may be improved by recent developments in artificial intelligence (AI), particularly in deep learning and generative models. This paper provides a comprehensive analysis of recent studies using Explainable AI (XAI) and Generative Adversarial Networks (GANs) for the detection of lung cancer. We examine key architectures, datasets, and multimodal fusion methods, emphasizing methods for data augmentation and data processing optimization. Together with the survey, a fictitious dataset is used to demonstrate how Grad-CAM can be used to create and explain real-time diagnostic predictions using a basic CNN model. By demonstrating the model’s interpretation of various test samples, this case study demonstrates how the model might be applied in practical situations where privacy is a top priority or resources are scarce. There is also discussion of the difficulties, open questions, and potential directions for further clinical translation research.