This paper presents a novel Grid-Based Structural and Dimensional Explainable Deep Convolutional Neural Network (GBSD-EDCNN) for accurate and interpretable skin cancer classification. The proposed model addresses the challenges of complexity, low reproducibility, and explainability in AI-based skin cancer diagnosis. It employs adaptive thresholding for region of interest (ROI) extraction, VGG-16 for hierarchical feature extraction, and a grid structure to capture spatial relationships within lesions. An Adaptive Intelligent Coney Optimization (AICO) algorithm is used to optimize hyperparameters and enhance the model's performance. The model was trained and tested using the ISIC and MNIST datasets, achieving accuracy and CSI values of 0.96 and 0.97, respectively, with minimal FPR and FNR. The AICO self-feature selected ECNN model outperforms existing techniques in terms of accuracy, interpretability, and robustness, aiding clinicians in early diagnosis and treatment of skin cancer.This paper presents a novel Grid-Based Structural and Dimensional Explainable Deep Convolutional Neural Network (GBSD-EDCNN) for accurate and interpretable skin cancer classification. The proposed model addresses the challenges of complexity, low reproducibility, and explainability in AI-based skin cancer diagnosis. It employs adaptive thresholding for region of interest (ROI) extraction, VGG-16 for hierarchical feature extraction, and a grid structure to capture spatial relationships within lesions. An Adaptive Intelligent Coney Optimization (AICO) algorithm is used to optimize hyperparameters and enhance the model's performance. The model was trained and tested using the ISIC and MNIST datasets, achieving accuracy and CSI values of 0.96 and 0.97, respectively, with minimal FPR and FNR. The AICO self-feature selected ECNN model outperforms existing techniques in terms of accuracy, interpretability, and robustness, aiding clinicians in early diagnosis and treatment of skin cancer.