Diagnosing Covid-19 and Pneumonia from Chest CT-Scan and X-Ray Images Using Deep Learning Technique

ABSTRACT

The outbreak of Covid-19 (COV-19) has become one of the global severe public health issues. It is an ongoing global pandemic that has spread rapidly. It not only affected human beings but also paralyzed most industries. It is very critical to diagnose COV-19 & Pneumonia (PNA) because both are having the same sign & symptoms that correlate to the most extent. The (RT-PCR) Reverse Transcription- Polymerise Chain Reaction widely used official screening method for detection of COV-19, while radio-logical imaging CT Scan (Computed Tomography) of human chests is used for detection of PNA as well as for COV-19 diagnosis also. The method for detecting COV-19 & PNA using CXR (Chest X-ray) & CT scans is too time consuming for an expert while the result accuracy is less. In this proposed project the work is based on the transfer learning model utilizing the Deep learning module for COV-19 & PNA solicitation from radiological images of the patient suffering from PNA & COV-19 which plays critical role in early detection & classification of COV-19 & PNA. In the present work multi-model deep learning modules are used for the image detection are as follows DenseNet 121, MobileNet, Inception V3, ResNet 50, & VGG 16. In this study, the data set contains X-Ray & CT-Scan Images which had been collected from CORD-19 & PNA Data sets. To evaluate the effect of dataset size based on the performance of multi-modal deep learning modules, to train the proposed module of deep learning using both the original & augmented dataset, the result were quite promising for DenseNet-121 for CT Scan with 97% accuracy while VGG 16 for CXR images with 99% accuracy. © 2022 IEEE.