Advanced Skin Cancer Detection: A Deep Learning and Transfer Learning Framework for Melanoma Classification

Abstract

Early identification is essential for successful treatment of skin cancer, a widespread and possibly fatal illness. In recent years, models based on deep learning have shown promise in improving the accuracy of skin cancer detection. This research introduces an innovative method for predicting skin cancer by utilizing Convolutional Neural Networks (CNNs) and employing the DenseNet201 deep learning model for classification. The CNN model is utilized to extract significant features from the images, whilst the DenseNet201 is employed for classification. The proposed CNNs efficiently extracted hierarchical features from the skin images. The low complex structure and convolutional layers helped in identifying nuanced patterns and abnormalities in the input data, making the model effective instruments in detecting cancer at an early stage. The DenseNet201 design is well-known for its densely connected layers, which enable the reuse of features and improve the flow of gradients. The proposed modified DenseNet201 is very efficient in extracting highly useful features from images for the classification of skin cancer. The unique design of its architecture enhances the overall performance efficiency when compared to the existing techniques.