A Comparative Analysis of Deep Learning Algorithms for Skin Cancer Detection

In the current world, machine learning and computing techniques are being used in every industry. One of these involves the healthcare and medical industry. Already due to intensive research and analysis huge leaps have been taken in strengthening the existing system. One area in the research is Skin Cancer Detection. Skin Cancer is diagnosed when damaged DNA cells start to multiply uncontrollably. This makes the detection in the preliminary stages highly essential. This is an area where machine learning or deep learning algorithms can help in the development of models which accurately predict the onset of diseases in early stages. Some research has been done in this area but high accuracy results haven’t been produced. Therefore, the focus of this research work is to compare between four deep learning techniques RNN (Recurrent Neural Networks), CNN (Convolutional Neural Networks), ResNet50, and Xception to conclude which algorithm works the finest in successfully predicting the disease with high accuracy. The dataset used for the purpose of this research is the HAM10000 which is available on Kaggle. The empirical results demonstrate that the accuracy by using CNN, RNN, ResNet50, and Xception was 72%, 69%, 79%, and 93% respectively with Xception performing the best.