Medical Image Recognition Method for Respiratory System Viruses Diagnosis based on a Proposed Combined Neural Network System

Abstract

— Respiratory viral detection is a confusing and time-consuming task of constantly looking at clinical pictures of patients. So, there is a need to develop and improve the respiratory case prediction model as soon as possible to control the spread of Respiratory viruses. Today advanced machine learning methods diagnose viruses such as Corona viruses that can be effectively classified. This paper proposes a scanning model based on using a combination of CNN (Convolutional Neural Network) and PNN (Probabilistic Neural Network) to classify images of Corona viruses. The proposed combined network, (PCN) system. The images used are resized, and the light was adjusted in a way that reflects the size of the Plaque damage and highlighted by converting the image to Hue Saturation Value (HSV). Essential distance information in an image is filtered, leaving important features of the image information. The PCN system uses the Convolutional Neural Network to calculate the dependent factors and passes the results to the Probabilistic Neural Network, and link the features of intermediate stages with the combined network to predict segmentation. As a result, PCN system achieves 100% accuracy.