Kumar Sahoo Saroj, Houssein Essam H, Premkumar M, Kumar Saha Apu, Emam Marwa M
Department of Mathematics, National Institute of Technology Agartala, Tripura 799046, India.
Faculty of Computers and Information, Minia University, Minia, Egypt.
Expert Syst Appl. 2023 Oct 1;227:120367. doi: 10.1016/j.eswa.2023.120367. Epub 2023 May 6.
The COVID-19 is one of the most significant obstacles that humanity is now facing. The use of computed tomography (CT) images is one method that can be utilized to recognize COVID-19 in early stage. In this study, an upgraded variant of Moth flame optimization algorithm (Es-MFO) is presented by considering a nonlinear self-adaptive parameter and a mathematical principle based on the Fibonacci approach method to achieve a higher level of accuracy in the classification of COVID-19 CT images. The proposed Es-MFO algorithm is evaluated using nineteen different basic benchmark functions, thirty and fifty dimensional IEEE CEC'2017 test functions, and compared the proficiency with a variety of other fundamental optimization techniques as well as MFO variants. Moreover, the suggested Es-MFO algorithm's robustness and durability has been evaluated with tests including the Friedman rank test and the Wilcoxon rank test, as well as a convergence analysis and a diversity analysis. Furthermore, the proposed Es-MFO algorithm resolves three CEC2020 engineering design problems to examine the problem-solving ability of the proposed method. The proposed Es-MFO algorithm is then used to solve the COVID-19 CT image segmentation problem using multi-level thresholding with the help of Otsu's method. Comparison results of the suggested Es-MFO with basic and MFO variants proved the superiority of the newly developed algorithm.
新冠病毒病是人类目前面临的最重大障碍之一。使用计算机断层扫描（CT）图像是一种可用于早期识别新冠病毒病的方法。在本研究中，通过考虑非线性自适应参数和基于斐波那契逼近法的数学原理，提出了蛾火焰优化算法的一种升级变体（Es-MFO），以在新冠病毒病CT图像分类中实现更高的准确率。使用19个不同的基本基准函数、30维和50维的IEEE CEC'2017测试函数对所提出的Es-MFO算法进行评估，并将其性能与多种其他基本优化技术以及MFO变体进行比较。此外，通过弗里德曼秩检验和威尔科克森秩检验等测试，以及收敛性分析和多样性分析，对所提出的Es-MFO算法的鲁棒性和耐久性进行了评估。此外，所提出的Es-MFO算法解决了三个CEC2020工程设计问题，以检验所提方法的问题解决能力。然后，借助大津法，使用所提出的Es-MFO算法通过多级阈值处理来解决新冠病毒病CT图像分割问题。所提出的Es-MFO与基本算法和MFO变体的比较结果证明了新开发算法的优越性。
