Covid-19 Image Classification With Image Enhancement And Transfer Learning

Anonymous

Keywords: Image Classification, Image Enhancement, Transfer Learning and Covid-19

TL;DR: Image enhancement impact on transfer learning models for Covid-19 image classification

Abstract: Traditional methods and clinical practice of medical image classification have reached their limit regarding performance, making it difficult to improve through normal means. The emergence of deep neural networks such as the convolutional neural network (CNN) have proven to be an effective method on varying image classification tasks. To improve the performance of these methods, we propose to use image enhancement techniques to improve the quality and perception of medical images and to reduce inherent noise. We propose to use two spatial domain image enhancement techniques as a preprocessing step, median filter, a filtering technique, and contrast limited adaptive histogram equalization, a contrast enhancement technique, on Covid19 CT scans before processing them through pretrained transfer learning algorithms. Through this experiment, we will demonstrate the effect of image enhancement on classifying and predicting Covid19 images and diagnosis.