This section examines the use of machine learning (ML) and deep learning (DL) methods for the classification and diagnosis of skin lesions using dermoscopic images. Existing techniques are grouped into three categories: handcrafted (HC) feature-based approaches, deep learning-based approaches, and hybrid models that fuse HC and DL features. Handcrafted methods rely on extracting features such as texture, color, shape, and borders, using support vector machines (SVM), neural networks (NN), and clustering techniques for classification. Examples include the use of histogram of oriented gradients (HOG), color histograms, and local binary patterns (LBP). Deep learning methods employ convolutional neural networks (CNNs) such as AlexNet, ResNet, DenseNet, and Inception, for automatic feature learning and classification. These models have been trained and evaluated using benchmark datasets like ISIC, PH2, and HAM10000, often with preprocessing techniques like noise and hair removal. Hybrid models combine the strengths of both approaches, using feature fusion, ensemble learning (e.g., Light GBM), and clinical rules such as the ABCD rule or seven-point checklist. The study also highlights recent advancements, including data augmentation, segmentation networks like U-net, and conditional random fields (CRFs) for contour refinement. Comparative results show DL methods often surpass traditional approaches and, in some cases, outperform expert dermatologists. The study underscores the potential of DL and hybrid systems to improve early detection, diagnostic accuracy, and clinical decision support, while emphasizing the need for continued research in model optimization and clinical integration.