Hybrid Deep Learning and Machine Learning Approach for Skin Cancer Classification Using Dermoscopic Images

Skin cancer is one of the most prevalent forms of cancer worldwide, and early detection plays a crucial role in improving survival rates. Manual diagnosis of skin lesions using dermoscopic images is time-consuming and prone to misclassification due to the visual similarity among different lesion types. Recent advancements in artificial intelligence and deep learning have enabled automated skin lesion classification systems that assist dermatologists in diagnosis. In this work, a hybrid deep learning and machine learning approach is proposed for multi-class skin cancer classification. The HAM10000 dataset containing dermoscopic images of seven different skin lesion categories is used for training and evaluation. Deep features are extracted using the EfficientNetB0 convolutional neural network model. The extracted features are further optimized using Principal Component Analysis (PCA) to reduce dimensionality and improve computational efficiency. The optimized feature vectors are classified using the XGBoost machine learning algorithm. The proposed hybrid model achieves an accuracy of approximately 78–80% on the test dataset. Additionally, a Flask-based web application is developed to allow users to upload dermoscopic images and obtain classification results along with confidence scores. The proposed system demonstrates the potential of combining deep learning and machine learning techniques for assisting dermatologists in early skin cancer detection.