Medical image classification is enhanced by deep learning models which improve generalization over traditional methods. This paper proposes a hybrid model combining CNN, Naïve Bayes, SVM, and MLP to integrate CNN-extracted high-level features with selected conventional features. CNN is trained supervisedly to convert raw pixels into feature vectors, conventional features are extracted, and a neural network fuses these for classification. Datasets HIS2828 and ISIC2017 achieve 97% and 96% accuracy, surpassing current methods. © 2022 Elsevier B.V., All rights reserved.