Deep Learning-Based Automated Classification of Skin Lesions Using CNN and Computer Vision

Skin cancer, particularly melanoma, continues to pose a significant public health challenge worldwide due to its high prevalence and mortality rate. Early and accurate detection of skin lesions is essential for improving clinical outcomes. This paper proposes a two pipeline diagnostic model in which deep learning approaches are combined with conventional machine learning methods to improve the classification of skin lesions on the HAM10000 dataset. The first pipeline utilizes the AlexNet Convolutional Neural Network (CNN) to classify images in an end-to-end system and reaches a classification accuracy of 97.18%. The second pipeline uses a Support Vector Machine (SVM) classifier with handcrafted features that were extracted using region-based segmentation and ABCD (Asymmetry, Border, Color, Diameter) analysis with 93.75% accuracy. By comparing the performance of these two approaches, the study evaluates the trade-offs between feature-driven and deep learning models, offering insights into their suitability for automated melanoma screening and clinical decision support systems.