Title: Application of fuzzy logic on CT-scan images of COVID-19 patients

Authors: Fariha Noor; Md. Rashad Tanjim; Muhammad Jawadur Rahim; Md. Naimul Islam Suvon; Faria Karim Porna; Shabbir Ahmed; Md. Abdullah Al Kaioum; Rashedur M. Rahman

Addresses: Department of Electrical and Computer Engineering, North South University, Plot-15, Block-B, Bashundhara, Dhaka, Bangladesh ' Department of Electrical and Computer Engineering, North South University, Plot-15, Block-B, Bashundhara, Dhaka, Bangladesh ' Department of Electrical and Computer Engineering, North South University, Plot-15, Block-B, Bashundhara, Dhaka, Bangladesh ' Department of Electrical and Computer Engineering, North South University, Plot-15, Block-B, Bashundhara, Dhaka, Bangladesh ' Department of Electrical and Computer Engineering, North South University, Plot-15, Block-B, Bashundhara, Dhaka, Bangladesh ' Department of Electrical and Computer Engineering, North South University, Plot-15, Block-B, Bashundhara, Dhaka, Bangladesh ' Department of Electrical and Computer Engineering, North South University, Plot-15, Block-B, Bashundhara, Dhaka, Bangladesh ' Department of Electrical and Computer Engineering, North South University, Plot-15, Block-B, Bashundhara, Dhaka, Bangladesh

Abstract: Image processing is crucial in any image analysis to determine the problem. If it is a medical area, a suitable image processing method becomes even more imperative to get as accurate results as possible. Due to the widespread outbreak of coronavirus disease 2019 (COVID-19), an infectious respiratory disease, it has become quite urgent that a reliable method for identification of the disease is sought. In this paper, we have segmented images with two different techniques, fuzzy c-means, and k-means clustering. Our images include CT-scan data and X-rays of both two categories. The first being the COVID-19 infected patients; the other being a collection of normal persons, and viral pneumonia infected persons. Among the two clustering techniques, the k-means performed better. Later, we trained our CNN model with the segmented images and raw images. Interestingly, the segmented images of CT-scan, as well as X-rays, are performing well in CNN classification rather than raw images. After applying fuzzy edge detection, the segmentation was improved. The f1-score for our model is 91% and the support is 89%.

Keywords: medical image processing; fuzzy c-means clustering; image segmentation; convolutional neural network.

DOI: 10.1504/IJIIDS.2021.118561

International Journal of Intelligent Information and Database Systems, 2021 Vol.14 No.4, pp.333 - 348

Received: 29 Jun 2020
Accepted: 20 Oct 2020
Published online: 28 Oct 2021 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article