The rising incidence of melanoma, the most common form of skin cancer, is characterized by the economic burden of skin cancer, which includes high costs for healthcare and lost productivity due to sickness. A blend of artificial intelligence and blockchain technology has resulted in a significant step forward within modern healthcare diagnostics for skin cancer. This research brings forth a novel approach to Decentralized Homomorphic Encryption using Transfer Learning for Skin Cancer (DHETL) that combines blockchain with transfer learning techniques to improve the accuracy of Convolutional Neural Networks (CNNs) in analyzing skin cancer. The DHETL model is a distributed system anchored at the convergence point of cryptography, machine learning, and blockchain for secure processing and analysis of medical images. Hence, particle swarm optimization (PSO) enhanced CNNs combined with Adam Max feature extractors are used in combination with dermatological image-based transfer learning from pre-trained models for skin cancer detection as proposed by the DHETL model located at the intersection of cryptography, machine learning, and blockchain. It also deals with limited labeled data in medical imaging through domain adaptation. Furthermore, the system incorporates homomorphic encryption to ensure patient information privacy and data security in the healthcare sector.