Multi-Modal Skin Lesion Classification Based on Dermoscopic Images and Meta-Data

Skin cancer is one of the most dangerous diseases, it can be deadly if not treated at an early stage. The intervention of deep learning has made a significant contribution in the field of dermatology medicine. In this work, we applied deep learning methods represented by Two dimensional Convolutional Neural Network and Multi-layer Perceptron on uni-modal and multi-modal data while taking into consideration two classes (malignant and benign). Additionally, due to the imbalanced data-set, we transformed images into a vector to solve this problem by applying SMOTE method, which is very important in classification problems.