This document describes using convolutional neural networks (CNNs) and residual neural networks (ResNets) to detect COVID-19 from chest X-rays and CT scans. CNNs were found to achieve 80% accuracy on X-rays and 72% accuracy on CT scans, outperforming ResNets which achieved around 50% accuracy. The document introduces CNNs and ResNets, describes their architectures and how they were applied to classify medical images as COVID-19 positive or normal. Evaluation metrics showed CNNs were better able to precisely and recall COVID-19 cases compared to ResNets. In conclusion, CNNs were determined to be the better algorithm for this medical image classification task.

1. DETECTION OF COVID19 FROM CHEST
X-RAY AND CT SCAN USING CNN
Submitted to:
Mr. Dharmender Nishad (Assistant Professor)
Project Guide

2. Table of content
• Introduction
• Description and Working
• Result
• Conclusion
• References

3. INTRODUCTION

4. Motivation of our project
• Coronavirus affect on society.
• Difficulty in testing Covid 19.
• Duplicate Testing Kits..

5. Our solution
The use of Deep Neural Networks such as CNN/ResNet etc. As AI is
progressing, We can use medical Images such as X-Ray and CT Scan
images to Identify.
Fig. 1 X-Ray Image of Covid positive person
Fig. 2 CT Scan Image of Covid positive person

6. Description and
Working

7. WHAT IS CNN?
Convolution Neural Network (CNN) is a mixture of math and biology
commonly used in image processing.
How it Works?
Fig. 3 Working of CNN layers

8. The three layers of CNN:
• Input Layer
Data is Provided to this layer
• Hidden Layer
All the Computation on data takes place here
• Classification Layer
Finally, this layer is used to classify the data
Fig. 4 How CNN process Images

9. What is ResNet and Why we used it in Parallel with
CNN
Residual Neural Network ResNet are also a Deep learning Image Processing Neural
Network.
Purpose of Using ResNet
• Supposed to be more accurate
• Faster Computation

10. ResNet Architecture
Fig. 5 ResNet Basic Architecture

11. Results

12. CNN on X-Ray Dataset
Precision Recall F1-Score Support
Covid 0.71 1.08 0.83 5
Normal 1.00 0.08 0.75 5
Accuracy 0.80 10
MacroAvg. 0.86 0.80 0.79 10
Weighted Avg. 0.80 0.80 0.79 10
Fig. 6 Training Loss and accuracy on dataset

13. CNN on CT Scan Dataset
Precision Recall F1-Score Support
Covid 0.66 0.84 0.74 70
Normal 0.82 0.61 0.70 80
Accuracy 0.72 150
MacroAvg. 0.74 0.73 0.72 150
Weighted Avg. 0.74 0.72 0.72 150
Fig. 7 Training Loss and accuracy on dataset

14. ResNet on X-Ray Dataset
Precision Recall F1-Score Support
Covid 0.50 1.00 0.67 5
Normal 1.00 0.08 0.75 5
Accuracy 0.50 10
MacroAvg. 0.25 0.50 0.33 10
Weighted Avg. 0.25 0.50 0.33 10
Fig. 8 Training Loss and accuracy on dataset

15. ResNet on CT Scan Dataset
Precision Recall F1-Score Support
Covid 0.59 0.59 0.59 70
Normal 0.64 0.65 0.65 80
Accuracy 0.62 150
MacroAvg. 0.62 0.62 0.62 150
Weighted Avg. 0.62 0.62 0.62 150
Fig. 9 Training Loss and accuracy on dataset

16. Conclusion

17. Is CNN best Image processing algorithm?
In our study we have used two algorithms, CNN and ResNet.
CNN – 80% accuracy
ResNet - ~50% accuracy
Conclusion: CNN is best among these two

18. References

19. References
1. P. Sun, X. Lu, C. Xu, W. Sun, and B. Pan (2020) “Understanding of COVID-19 based on current evidence,” Journal of Medical Virology, vol.
92, no. 6, pp. 548–551.
2. D. J. Cennimo (2020), “Coronavirus disease 2019 (COVID-19) clinical presentation,” vol. 8, pp. 101489–101499.
3. B. Kayalibay, G. Jensen, and P. van der Smagt (2017) “CNN-based segmentation of medical imaging objects.
4. N. Asada, K. Doi, H. MacMahon et al (1990) “Potential usefulness of an artificial neural network for differential diagnosis of interstitial lung
diseases: pilot study,” Radiology, vol. 177, no. 3, pp. 857
5. S. Katsuragawa and K. Doi (2007), “Computer-aided diagnosis in chest radiography,” Computerized Medical Imaging and Graphics, vol. 31,
no. 4-5, pp. 212–223.
6. S. Minaee, R. Kafieh, M. Sonka, S. Yazdani, and G. Jamalipour Soufi (2020) “Deep-covid: predicting covid-19 from chest x-ray images using
deep transfer learning,” Medical Image Analysis, vol. 65, Article ID 10179
7. A. S. Lundervold and A. Lundervold (2019) “An overview of deep learning in medical imaging with a focus on MRI,” Zeitschrift für
Medizinische Physik, vol. 29, no. 2, pp. 102–127.
8. Zeitschrift für Medizinische Physik, (2019) “An overview of deep learning in medical imaging with an emphasis on MRI”, vol. 29, no. 2, pp.
102–127.
9. Rouhi, M. Jafari, S. Kasaei, and P. Keshavarzian (2015) “Benign and malignant breast tumours classification based on region growing and
CNN segmentation,” Expert Systems with Applications, vol. 42, no. 3, pp. 990–1002.

20. THANKYOU