Skin lesion classification on dermatoscopic images using effective data augmentation and pre-trained deep learning approach

Skin cancer is a severe disease that is common and causes death if left untreated. When skin cancer is detected early through dermatoscopic imaging, the possibility of definitive treatment is very high. Although melanoma is one of the fatal types of skin cancer, early detection dramatically increases the chances of survival. There is a low morbidity rate and limited actual data to study this deadly disease. This is a significant handicap in the application of machine learning techniques. Accurate diagnosis is essential because of the similarity of some types of lesions. The accuracy of the diagnosis is related to the professional experience of the specialist. The development of rapid and successful computerized diagnostic systems for the diagnosis and classification of skin cancer has become increasingly important. Deep learning-based applications are especially new trend in the detection of diseases from medical images. In this study, an effective data augmentation and a pre-trained deep learning approach are proposed for skin lesion classification. A hybrid network model called the Inception-Resnet-v2 is proposed to classify skin cancer images. The main aim of this study is to increase the number of images in the dataset by applying the affine transformation technique (data augmentation) and analyzing its effect on the skin cancer classification system. The highest reported accuracy in this study with an augmented dataset is 95.09% for the Inception-Resnet-v2 model while the same model achieved 83.59% with the original dataset.