U-Net Based Chest X-ray Segmentation with Ensemble Classification for Covid-19 and Pneumonia

Authors

DOI:

https://doi.org/10.3991/ijoe.v18i07.30807

Keywords:

Segmentation, U-Net, classification, CNN, chest X-rays

Abstract


Respiratory diseases have been known to be a main cause of death worldwide. Pneumonia and Covid-19 are two of the dominant diseases. Several deep learning based studies are available in the literature that classifies infection conditions in chest X-ray images. In addition, image segmentation has been also applied to obtain promising results in deep learning approaches. This paper focuses on using a modified version of the U-Net architecture to conduct segmentation on chest X-rays and then use segmented images for classification to assess the impact on the performance. We achieved an Intersection over Union of 93.53% with the proposed modified U-Net architecture and achieved 99.83% accuracy on segmentation aided ensemble classification.

Author Biographies

K. A. S. H. Kumarasinghe, University of Moratuwa

Ms. K.A.S.H. Kumarasinghe is a B.Sc. Eng. undergraduate in the Department of Computer Science and Engineering at the University of Moratuwa, Sri Lanka. Her main research interests are computer vision and deep learning.          

S. L. Kolonne, University of Moratuwa

Ms. S. L. Kolonne is a B.Sc. Eng. undergraduate in the Department of Computer Science and Engineering at the University of Moratuwa, Sri Lanka. Her main research interests are computer vision and deep learning.

K. C. M. Fernando, University of Moratuwa

Ms. K.C.M. Fernando is a B.Sc. Eng. undergraduate in the Department of Computer Science and Engineering at the University of Moratuwa, Sri Lanka. Her main research interests are computer vision and deep learning.           

D. Meedeniya, University of Moratuwa

Prof. D. Meedeniya is a Professor in Computer Science and Engineering at the University of Moratuwa, Sri Lanka. She holds a PhD in Computer Science from the University of St Andrews, United Kingdom. She is the director of the Bio-Health Informatics group at her department and engages in many collaborative research. She is a co-author of 100+ publications in indexed journals, peer-reviewed conferences and international book chapters. She serves as a reviewer, program committee and editorial team member in many international conferences and journals. Her main research interests are Software modelling and design, Bio-Health Informatics, Deep Learning and Technology-enhanced learning. She is a Fellow of HEA (UK), MIET, MIEEE, Member of ACM and a Chartered Engineer registered at EC (UK).

Covr iJOE Vol. 18 No. 07 (2022)

Downloads

Published

2022-06-14

How to Cite

Kumarasinghe, H., Kolonne, S., Fernando, C., & Meedeniya, D. (2022). U-Net Based Chest X-ray Segmentation with Ensemble Classification for Covid-19 and Pneumonia. International Journal of Online and Biomedical Engineering (iJOE), 18(07), pp. 161–175. https://doi.org/10.3991/ijoe.v18i07.30807

Issue

Vol. 18 No. 07 (2022)

Section

Papers

License

Copyright (c) 2022 K. A. S. H. Kumarasinghe, S. L. Kolonne, K. C. M. Fernando, D. Meedeniya

Creative Commons License

This work is licensed under a Creative Commons Attribution 4.0 International License.

The submitting author warrants that the submission is original and that she/he is the author of the submission together with the named co-authors; to the extend the submission incorporates text passages, figures, data or other material from the work of others, the submitting author has obtained any necessary permission.
Articles in this journal are published under the Creative Commons Attribution Licence (CC-BY What does this mean?). This is to get more legal certainty about what readers can do with published articles, and thus a wider dissemination and archiving, which in turn makes publishing with this journal more valuable for you, the authors.
By submitting an article the author grants to this journal the non-exclusive right to publish it. The author retains the copyright and the publishing rights for his article without any restrictions.