Convolutional Neural Network with Bidirectional Gated Recurrent Unit for Classification of Psoriasis using Dermatological Images

Psoriasis is a common chronic inflammatory skin disease that causes physical and psychological burden to patients. However, various type of dermatological image blend in the original skin due to overfitting and unwanted noise occur in the image classify. This paper, proposed Convolutional Neural Network (CNN) with a Bidirectional Gated Recurrent Unit (Bi-GRU) combined to performed in the dermatological image classification efficiently capture the disease achieve better accuracy. Initially data obtained from the Dermnet NZ dataset and pre-processing utilized the min-max normalization is efficiently handle the uniformed feature vector and remove unwanted noise. The segmentation involved Otsu thresholding technique is utilized to create a mask image then applied it grayscale image and handle the overfitting issue then lesion segmented form normal skin region. the feature extraction using VGG 16 technique is efficiently extracted the high level feature. The classification utilized CNN with Bi-GRU process the feature sequences capturing dependencies across the feature dimensions, learning the pattern efficiently and achieve better performance. performance. The performance of DL model evaluated with various metrics such as accuracy, precision, recall and f1-score metrics on Dermnet NZ dataset.