Empowering Skin Cancer Diagnosis: Integrating Advanced Deep Learning Models with Explainable AI for Lesion Classification

Skin cancer poses a significant global health challenge, demanding precise and timely diagnosis to enhance patient outcomes and tailor treatment effectively. This study addresses the complexity of skin cancer detection by employing cutting-edge deep learning algorithms alongside explainable artificial intelligence (XAI) methodologies. Utilizing recent pre-trained models such as XceptionNet, EfficientNetV2S, InceptionResNetV2, and EfficientNetV2M, we aim to classify skin lesions accurately. Additionally, we employ image augmentation strategies to improve model generalization and mitigate class imbalances. Through XAI, we elucidate the decision-making processes of our models, a critical step in establishing trust and facilitating the integration of AI-powered diagnostic tools into clinical workflows. Our findings highlight the superiority of the XceptionNet architecture, achieving an accuracy of 88.72%. By demonstrating the potential of deep learning and XAI in refining skin cancer diagnosis, this study lays the foundation for further advancements in medical image analysis. The adoption of these technologies holds promise in expediting accurate identification, thereby enhancing patient care, reducing healthcare costs, and bolstering survival rates among individuals affected by skin cancer.