Skin cancer is one of the most common forms of cancer worldwide, and its earlydetection is essential for effective treatment and improved survival rates.Traditional deep learning models often require retraining from scratch when newdata becomes available, which increases computational cost and riskscatastrophic forgetting. To address this challenge, we propose a transformerbasedframework integrated with incremental learning for skin cancerclassification. The transformer architecture captures both local and globalcontextual features from dermoscopic images, while incremental learning allowsthe model to adapt continuously to newly available datasets without losingpreviously acquired knowledge. Experimental results demonstrate that theproposed approach achieves competitive accuracy and robustness compared toconventional deep learning models, while maintaining efficiency in handlingevolving datasets. This method shows potential for real-world deployment inclinical decision-support systems where continuous learning from diverse andgrowing datasets is critical.