DETECTION OF COVID-19 IN X-RAY IMAGES USING DEEP LEARNING

Abstract

The COVID19 outbreak has had such a negative impact on people's health and well-being all around the world. The pandemic has developed into one of the major humanitarian catastrophes in modern human history due to its devastating consequences on lives and livelihood. As a result, a comprehensive strategy to battle the pandemic is required. Due to the limitation of reverse transcription-polymerase chain reaction (RT-PCR) kits, it was impossible to test every patient with a respiratory disease. In the fight against COVID-19, rapid testing of infected patients is critical, and radiological evaluation utilizing chest Xrays is one of the most effective and successful screening procedures. The patient's chest X-ray scans revealed various abnormalities that are characteristic of a COVID19 infection. In this study, we have used a publicly available dataset from Kaggle. The 9034 chest Xrays utilized in this study was categorized into five categories (Pneumonia, Normal, COVID19, Tuberculosis and Pneumothorax). To perform COVID19 diagnosis based on chest Xray pictures, we suggest combining convolutional processes CNN and dilated CNN and building a unique deep neural network with two inputs and one output. To extract features from the chest x-rays, we used two deep neural networks in parallel. The proposed neural network has an 82.53% accuracy rate in detecting Covid19 or Pneumonia or Normal cases or Tuberculosis or Pneumothorax. The x-ray scans of covid-19 patients can be utilized to diagnosis in areas where expert radiologists are not present, led to advances AI algorithms.