Skin Cancer Classification Using Pre-trained CNNs: A Transfer Learning Approach Addressing Imbalanced Data Challenges

Skin cancer is a leading cause of cancer-related deaths. Automatic and accurate classification of skin lesions is important, as it can significantly improve patient recovery from this deadly disease. Generally, dermatologists detect skin cancer. However, the decision might vary from dermatologist to dermatologist. So in this work, we present a novel deep learning-based approach for the classification of skin cancer. Our proposed method leverages transfer learning from pre-trained convolutional neural networks (CNNs) and employs advanced data augmentation techniques to address the challenges posed by limited and imbalanced labeled data. To further enhance the performance, the model incorporates fine-tuning and a carefully designed architecture tailored to the specific characteristics of dermoscopic images. Extensive experiments were conducted to evaluate the proposed method, and the results consistently demonstrate its ability to accurately differentiate between multiple classes of skin lesions, including melanoma, basal cell carcinoma, and benign nevi. The proposed model shows comparable performance with existing works, offering a more accurate and efficient solution for skin cancer diagnosis. This approach provides a valuable tool for healthcare professionals, potentially aiding in early detection and improving the overall effectiveness of skin cancer treatment.