DERMATOSCOPIC PIGMENTED SKIN TYPE CLASSIFICATION USING CNN WITH TRANSFER LEARNING

Abstract

Skin cancer is among the most prevalent and life-threatening diseases globally, where
early and accurate diagnosis plays a crucial role in effective treatment. Dermatoscopy
has become an essential technique for examining pigmented skin lesions, enabling
dermatologists to identify various skin conditions. However, manual analysis of
dermatoscopic images is often time-consuming and highly dependent on clinical
expertise, which may lead to variability in diagnosis.
To address these challenges, this study presents an automated approach for
dermatoscopic pigmented skin lesion classification using Convolutional Neural
Networks (CNN) combined with transfer learning. The proposed system utilizes the
HAM10000 dataset, which consists of diverse dermatoscopic images representing
multiple categories of skin lesions. Image preprocessing and data augmentation
techniques are applied to enhance data quality and improve model generalization.
A pre-trained deep learning model, ResNet50, is employed as the feature extractor,
leveraging transfer learning to capture complex visual patterns effectively. Additional
fully connected layers are integrated into the architecture to classify images into seven
distinct skin lesion categories, including melanoma, melanocytic nevi, basal cell
carcinoma, and others