Accelerated and precise skin cancer detection through an enhanced machine learning pipeline for improved diagnostic accuracy

Unrepaired DNA damage in skin cells causes mutations leading to skin cancer, a highly aggressive malignancy. This study proposes a machine learning (ML)-based framework for accurate and automated skin cancer detection, integrating EfficientNetV2L for advanced feature extraction and LightGBM (LGBM) for gradient boosting. The ensemble model effectively classifies benign and malignant skin lesions, leveraging EfficientNetV2L's feature extraction capabilities and LGBM's computational efficiency. A dataset comprising 3,297 images of benign and malignant skin cancer classes was used for training. Data augmentation techniques were applied to enhance dataset reliability. The proposed training pipeline, optimized for modularity and performance, demonstrated significant improvements in accuracy and computational efficiency over state-of-the-art methods. The model achieved a training accuracy of 99.57% and a validation accuracy of 99.93%. Using 5-fold cross-validation, it recorded a test accuracy of 99.90% in the fifth fold. For benign cases, the model achieved a precision of 0.99, recall of 0.98, and F1-score of 0.98. Similarly, malignant cases achieved a precision of 0.98, recall of 0.98, and F1-score of 0.98. The ROC-AUC scores for both classes were 0.98, further validating the system's reliability. These results highlight the robustness and effectiveness of the EfficientNetV2L-LGBM framework in skin cancer classification, offering a reliable and scalable solution for early detection. This approach demonstrates significant potential in advancing diagnostic systems for improved patient outcomes.