CAGAN: Classifier‐augmented generative adversarial networks for weakly‐supervised COVID‐19 lung lesion localisation
Abstract The Coronavirus Disease 2019 (COVID‐19) epidemic has constituted a Public Health Emergency of International Concern. Chest computed tomography (CT) can help early reveal abnormalities indicative of lung disease. Thus, accurate and automatic localisation of lung lesions is particularly impor...
Main Authors: | , , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Wiley
2024-02-01
|
Series: | IET Computer Vision |
Subjects: | |
Online Access: | https://doi.org/10.1049/cvi2.12216 |
_version_ | 1797320625112481792 |
---|---|
author | Xiaojie Li Xin Fei Zhe Yan Hongping Ren Canghong Shi Xian Zhang Imran Mumtaz Yong Luo Xi Wu |
author_facet | Xiaojie Li Xin Fei Zhe Yan Hongping Ren Canghong Shi Xian Zhang Imran Mumtaz Yong Luo Xi Wu |
author_sort | Xiaojie Li |
collection | DOAJ |
description | Abstract The Coronavirus Disease 2019 (COVID‐19) epidemic has constituted a Public Health Emergency of International Concern. Chest computed tomography (CT) can help early reveal abnormalities indicative of lung disease. Thus, accurate and automatic localisation of lung lesions is particularly important to assist physicians in rapid diagnosis of COVID‐19 patients. The authors propose a classifier‐augmented generative adversarial network framework for weakly supervised COVID‐19 lung lesion localisation. It consists of an abnormality map generator, discriminator and classifier. The generator aims to produce the abnormality feature map M to locate lesion regions and then constructs images of the pseudo‐healthy subjects by adding M to the input patient images. Besides constraining the generated images of healthy subjects with real distribution by the discriminator, a pre‐trained classifier is introduced to enhance the generated images of healthy subjects to possess similar feature representations with real healthy people in terms of high‐level semantic features. Moreover, an attention gate is employed in the generator to reduce the noise effect in the irrelevant regions of M. Experimental results on the COVID‐19 CT dataset show that the method is effective in capturing more lesion areas and generating less noise in unrelated areas, and it has significant advantages in terms of quantitative and qualitative results over existing methods. |
first_indexed | 2024-03-08T04:45:42Z |
format | Article |
id | doaj.art-5469ee21bf7a4e86b81763cc3ce89acc |
institution | Directory Open Access Journal |
issn | 1751-9632 1751-9640 |
language | English |
last_indexed | 2024-03-08T04:45:42Z |
publishDate | 2024-02-01 |
publisher | Wiley |
record_format | Article |
series | IET Computer Vision |
spelling | doaj.art-5469ee21bf7a4e86b81763cc3ce89acc2024-02-08T10:33:58ZengWileyIET Computer Vision1751-96321751-96402024-02-0118111410.1049/cvi2.12216CAGAN: Classifier‐augmented generative adversarial networks for weakly‐supervised COVID‐19 lung lesion localisationXiaojie Li0Xin Fei1Zhe Yan2Hongping Ren3Canghong Shi4Xian Zhang5Imran Mumtaz6Yong Luo7Xi Wu8Chengdu University of Information Technology Chengdu ChinaChengdu University of Information Technology Chengdu ChinaChengdu University of Information Technology Chengdu ChinaChengdu University of Information Technology Chengdu ChinaXihua University Chengdu ChinaChengdu University of Information Technology Chengdu ChinaUniversity of Agriculture Faisalabad Faisalabad PakistanWest China Hospital Sichuan University Chengdu ChinaChengdu University of Information Technology Chengdu ChinaAbstract The Coronavirus Disease 2019 (COVID‐19) epidemic has constituted a Public Health Emergency of International Concern. Chest computed tomography (CT) can help early reveal abnormalities indicative of lung disease. Thus, accurate and automatic localisation of lung lesions is particularly important to assist physicians in rapid diagnosis of COVID‐19 patients. The authors propose a classifier‐augmented generative adversarial network framework for weakly supervised COVID‐19 lung lesion localisation. It consists of an abnormality map generator, discriminator and classifier. The generator aims to produce the abnormality feature map M to locate lesion regions and then constructs images of the pseudo‐healthy subjects by adding M to the input patient images. Besides constraining the generated images of healthy subjects with real distribution by the discriminator, a pre‐trained classifier is introduced to enhance the generated images of healthy subjects to possess similar feature representations with real healthy people in terms of high‐level semantic features. Moreover, an attention gate is employed in the generator to reduce the noise effect in the irrelevant regions of M. Experimental results on the COVID‐19 CT dataset show that the method is effective in capturing more lesion areas and generating less noise in unrelated areas, and it has significant advantages in terms of quantitative and qualitative results over existing methods.https://doi.org/10.1049/cvi2.12216biomedical MRIcomputer graphicspatient diagnosis |
spellingShingle | Xiaojie Li Xin Fei Zhe Yan Hongping Ren Canghong Shi Xian Zhang Imran Mumtaz Yong Luo Xi Wu CAGAN: Classifier‐augmented generative adversarial networks for weakly‐supervised COVID‐19 lung lesion localisation IET Computer Vision biomedical MRI computer graphics patient diagnosis |
title | CAGAN: Classifier‐augmented generative adversarial networks for weakly‐supervised COVID‐19 lung lesion localisation |
title_full | CAGAN: Classifier‐augmented generative adversarial networks for weakly‐supervised COVID‐19 lung lesion localisation |
title_fullStr | CAGAN: Classifier‐augmented generative adversarial networks for weakly‐supervised COVID‐19 lung lesion localisation |
title_full_unstemmed | CAGAN: Classifier‐augmented generative adversarial networks for weakly‐supervised COVID‐19 lung lesion localisation |
title_short | CAGAN: Classifier‐augmented generative adversarial networks for weakly‐supervised COVID‐19 lung lesion localisation |
title_sort | cagan classifier augmented generative adversarial networks for weakly supervised covid 19 lung lesion localisation |
topic | biomedical MRI computer graphics patient diagnosis |
url | https://doi.org/10.1049/cvi2.12216 |
work_keys_str_mv | AT xiaojieli caganclassifieraugmentedgenerativeadversarialnetworksforweaklysupervisedcovid19lunglesionlocalisation AT xinfei caganclassifieraugmentedgenerativeadversarialnetworksforweaklysupervisedcovid19lunglesionlocalisation AT zheyan caganclassifieraugmentedgenerativeadversarialnetworksforweaklysupervisedcovid19lunglesionlocalisation AT hongpingren caganclassifieraugmentedgenerativeadversarialnetworksforweaklysupervisedcovid19lunglesionlocalisation AT canghongshi caganclassifieraugmentedgenerativeadversarialnetworksforweaklysupervisedcovid19lunglesionlocalisation AT xianzhang caganclassifieraugmentedgenerativeadversarialnetworksforweaklysupervisedcovid19lunglesionlocalisation AT imranmumtaz caganclassifieraugmentedgenerativeadversarialnetworksforweaklysupervisedcovid19lunglesionlocalisation AT yongluo caganclassifieraugmentedgenerativeadversarialnetworksforweaklysupervisedcovid19lunglesionlocalisation AT xiwu caganclassifieraugmentedgenerativeadversarialnetworksforweaklysupervisedcovid19lunglesionlocalisation |