Abstract
Skin cancer is one of the most serious and rapidly increasing diseases worldwide, where early and accurate diagnosis is essential for effective treatment and improved patient survival. However, manual examination of dermoscopic images is time-consuming and highly dependent on dermatologists’ expertise. Variations in lesion shape, color, texture, illumination, and image noise make accurate diagnosis challenging. To overcome these limitations, this paper proposes an intelligent AI-empowered hybrid deep learning architecture for early detection, precise lesion segmentation, and accurate classification of skin cancer using high-resolution dermoscopic image analysis. The proposed framework integrates image preprocessing, lesion enhancement, segmentation, deep feature extraction, and hybrid classification techniques to improve diagnostic performance. Initially, preprocessing operations including noise reduction, contrast enhancement, and normalization are applied to improve image quality and lesion visibility. A U-Net-based deep segmentation network is employed to accurately detect and segment lesion regions from dermoscopic images. For classification, a hybrid deep learning model combining Convolutional Neural Networks (CNN), ResNet-50, and Long Short-Term Memory (LSTM) networks is utilized to extract spatial and sequential deep features for accurate skin cancer classification. Transfer learning and feature fusion techniques are further incorporated to enhance generalization capability and reduce computational complexity. The proposed model is evaluated on benchmark dermoscopic datasets using multiple performance metrics including accuracy, precision, recall, F1-score, Dice similarity coefficient, and area under the ROC curve (AUC). Experimental results demonstrate that the proposed framework achieves a segmentation accuracy of 97.8%, classification accuracy of 98.6%, precision of 98.1%, recall of 97.9%, F1-score of 98.0%, Dice coefficient of 97.5%, and AUC of 99.1%. Comparative analysis confirms that the proposed hybrid architecture outperforms conventional machine learning methods and existing deep learning models in terms of robustness, segmentation quality, and classification efficiency. The proposed intelligent system can serve as an effective computer-aided diagnostic tool for dermatologists and contribute toward the development of reliable AI-driven healthcare systems for automated skin cancer diagnosis.