diagnosing of skin lesions using deep convolutional neural network and support vector machines

The number of fatalities resulting from skin cancer has significantly increased over the past few years. early diagnosis is highly important for the quick treatment of skin cancer. computer-based dermoscopy analysis methods provide considerable information about the lesions that can be helpful to skin experts in the early detection of skin lesions. these computer-based diagnostic systems require image-processing algorithms to provide mathematical explanations of suspicious areas. convolutional neural network (cnn) as one of the deep learning algorithms has high scalability in interaction with big data, and can automatically extract key image features for classification and segmentation of images. in this study, a hybrid model consisting of deep learning and machine learning method is proposed to classify different types of skin lesions. in this model, at first, an input image is pre-processed to remove the negative effect of hairs on skin lesion detection and also to prepare it for applying to an efficient deep convolutional network employed as a feature extractor. then support vector machine (svm) is utilized as a classifier to detect and classify different types of skin lesions.