Comparative Analysis of Deep Learning Models for Early Skin Cancer Detection Using 3D Total Body Photography

Skin cancer presents a significant public health challenge, with early detection being crucial in mitigating its life-threatening implications. This study explores the application of deep learning models for early skin cancer detection using 3D Total Body Photography (TBP) images from the ISIC dataset. The dataset includes single-lesion crops from a large and diverse population, enabling comprehensive algorithm training. The study evaluates four models: EfficientNet-B0, CNN, ResNet50, and Inception V3. Among these, EfficientNet-B0 achieved the highest accuracy of 98%, underscoring the importance of model selection in skin cancer diagnostics. The results emphasize the potential of deep learning models in aiding early detection and reducing mortality from skin cancer.