Examining The Effect Of Pre-processed Covid-19 Images On Classification Performance Using Deep Learning Methods

Abstract

In recent years, researchers have been using different artificial intelligence models to process x-ray images and make a determination about the patient's condition. Pre-processing is applied to medical images by many researchers. In this way, researchers know that the results they will obtain will be better and that their study results will be more accepted in the literature. As with all other medical images, pre-processing of Covid-19 images is generally done to obtain better classification results. In this study, some preprocessing was done with Covid-19 images. Experimental studies were performed using the ResNet18 deep learning model. According to experimental studies carried out on non pre-processed images, an average accuracy of 0.85206% was obtained in the test processes, while an accuracy rate of 0.93086% was obtained in the test processes obtained from pre-processed images. It was observed that better results were obtained by processing pre-processed images with the same model.