Journal
IMAGING SCIENCE JOURNAL
Volume -, Issue -, Pages -Publisher
TAYLOR & FRANCIS LTD
DOI: 10.1080/13682199.2023.2178094
Keywords
COVID-19; trigonometric function; deep convolutional neural networks; chest X-Rays; metaheuristic algorithms; optimization; diagnostic; image processing
Categories
Ask authors/readers for more resources
This study proposes using a novel Trigonometric Function (TF) for training fully connected layers to improve the overall accuracy of the COVID-19 detection model. The designed model achieves competitive results on the COVID-Xray-5k dataset and utilizes the class activation map theory to detect potentially infected areas by the Covid-19 virus.
With growing demands for diagnosing COVID-19 definite cases, employing radiological images, i.e., the chest X-ray, is becoming challenging. Deep Convolutional Neural Networks (DCNN) propose effective automated models to detect COVID_19 positive cases. In order to improve the total accuracy, this paper proposes using the novel Trigonometric Function (TF) instead of the existing gradient descendent-based training method for training fully connected layers to have a COVID-19 detector with parallel implementation ability. The designed model gets then benchmarked on a verified dataset denominated COVID-Xray-5k. The results get investigated by qualified research with classic DCNN, BWC, and MSAD. The results confirm that the produced detector can present competitive results compared to the benchmark detection models. The paper also examines the class activation map theory to detect the areas probably infected by the Covid-19 virus. As experts confirm, the obtained results get correlated with the clinical recognitions.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available