KMID : 1142220220170020193
|
|
Regulatory Research on Food, Drug & Cosmetic 2022 Volume.17 No. 2 p.193 ~ p.198
|
|
A Study of a Chest X-ray Image-based COVID-19 Diagnostic Model using the End-to-End Ensemble Method of Convolutional Neural Networks
|
|
Kim Chae-Hyeon
Kim Ka-Young Kim Sung-Min
|
|
Abstract
|
|
|
Since the outbreak of Coronavirus Disease 2019(COVID-19), it has rapidly spread and significantly influenced on the global health. In order to prevent rapid spread as much as possible, we propose an end-to-end ensemble method to detect COVID-19 in chest X-ray images based on convolutional neural networks as a method for diagnosing COVID-19. First of all, to reduce the class imbalance problem, the dataset was composed of 442 normal images and 442 COVID-19 images based on the two public databases. The proposed model is made up of a feature extractor and a classifier. The feature extractor was designed as an ensemble structures by using DenseNet201, VGG19, ResNet50, Xception, and lnceptionV3. As a result, the proposed model improved classification performance compared to the single model. In conclusion, the End-to-End ensemble model confirmed that the proposed ensemble method is an useful method to improve the performance of the COVID- 19 diagnostic model. The model using the End-to-End ensemble learning method can be applied to the screening stage to help prevent the spread of infectious diseases.
|
|
KEYWORD
|
|
Deep learning, Covid-19, Chest X-ray image, CNN, End-to-End Ensemble Learning
|
|
FullTexts / Linksout information
|
|
|
|
Listed journal information
|
|
|
|