Early-Stage Detection of Psoriasis using Deep Learning Approach of Convolutional Neural Network and Generative Adversarial Network

In skin disease detection CASD (Computer aided skin diagnosis) plays major role. Non-invasive screening of Psoriasis plays a crucial role in dermatology. In this study Psoriasis dataset to predicts skin diseases by using multiple GAN (Generative Adversarial Network) models and Ensemble model. In GAN, traditional GAN and DEBLUR GAN are used and then trained 5 ensemble CNN (convolutional neural network) models on different train and test data. Comparative performance between traditional, DEBLUR and Ensemble Model is calculated using different performance metrics. Ensemble GAN CNN model outperforms than traditional augmentation CNN model and deblur GAN augmentation CNN model. Performance accuracy of 99.8737% is observed for proposed model.