Deep ensemble learning for skin lesions classification with convolutional neural network

<span lang="EN-US">One type of skin cancer that is considered a malignant tumor melanoma. Such dangerous disease can cause lot death in the world. The early detection lesions becomes an important task diagnosis cancer. Recently, machine learning paradigm emerged known as deep (DL) utilized for classification. However, some previous studies by using seven class images diagnostic classification based on single DL approach with CNNs architecture does not produce satisfying performance. allows development medical image analysis system improving performance, such convolutional neural networks (DCNNs) method. In this study, we propose ensemble combines three DCNNs architectures Inception V3, ResNet V2 and DenseNet 201 performance terms accuracy, sensitivity, specificity, precision, F1-score. Seven classes dermoscopy categories are 10015 from well-known HAM10000 dataset. proposed model produces good 97.23% 90.12% 97.73% 82.01% 85.01% F1-Score. This method gives promising results classifying diagnosis.</span>