Deep learning based analyses of computed tomography (CT) images contribute to automated diagnosis
of COVID-19, and ensemble learning may commonly provide a better solution. Here, we proposed an ensemble
learning method that integrates several component neural networks to jointly diagnose COVID-19. Two ensemble
strategies are considered: the output scores of all component models that are combined with the weights adjusted
adaptively by cost function back propagation; voting strategy. A database containing 8 347 CT slices of COVID-
19, common pneumonia and normal subjects was used as training and testing sets. Results show that the novel
method can reach a high accuracy of 99.37% (recall: 0.998 1; precision: 0.989 3), with an increase of about 7%
in comparison to single-component models. And the average test accuracy is 95.62% (recall: 0.958 7; precision:
0.955 9), with a corresponding increase of 5.2%. Compared with several latest deep learning models on the identical
test set, our method made an accuracy improvement up to 10.88%. The proposed method may be a promising
solution for the diagnosis of COVID-19.
WANG Zhiming(王志明), DONG Jingjing (董静静), ZHANG Junpeng∗ (张军鹏)
. Multi-Model Ensemble Deep Learning Method to Diagnose COVID-19 Using Chest Computed Tomography Images[J]. Journal of Shanghai Jiaotong University(Science), 2022
, 27(1)
: 70
-80
.
DOI: 10.1007/s12204-021-2392-3
[1] AI T, YANG Z L, HOU H Y, et al. Correlation ofchest CT and RT-PCR testing for coronavirus disease2019 (COVID-19) in China: A report of 1014 cases [J].Radiology, 2020, 296(2): E32-E40.
[2] ZHANG N R, WANG L L, DENG X Q, et al. Recentadvances in the detection of respiratory virus infectionin humans [J]. Journal of Medical Virology, 2020,92(4): 408-417.
[3] HUANG C L, WANG Y M, LI X W, et al. Clinical featuresof patients infected with 2019 novel coronavirusin Wuhan, China [J]. The Lancet, 2020, 395(10223):497-506.
[4] CHUNG M, BERNHEIM A, MEI X Y, et al. CT imagingfeatures of 2019 novel coronavirus (2019-nCoV) [J].Radiology, 2020, 295(1): 202-207.
[5] ISMAEL A M, S窫NG¨UR A. Deep learning approachesfor COVID-19 detection based on chest X-ray images[J]. Expert Systems With Applications, 2021, 164:114054.
[6] OH Y, PARK S, YE J C. Deep learning COVID-19features on CXR using limited training data sets [J].IEEE Transactions on Medical Imaging, 2020, 39(8):2688-2700.
[7] LI L, QIN L, XU Z, et al. Using artificial intelligence todetect COVID-19 and community-acquired pneumoniabased on pulmonary CT: Evaluation of the diagnosticaccuracy [J]. Radiology, 2020, 296(2): E65-E71.
[8] RAHIMZADEH M, ATTAR A, SAKHAEI S M. Afully automated deep learning-based network for detectingCOVID-19 from a new and large lung CT scandataset [J]. Biomedical Signal Processing and Control,2021, 68: 102588.
[9] SONG Y, ZHENG S J, LI L, et al. Deep learning enablesaccurate diagnosis of novel coronavirus (COVID-19) with CT images [J]. IEEE/ACM Transactionson Computational Biology and Bioinformatics, 5361,PP(99): 1.
[10] BAI H X, WANG R, XIONG Z, et al. Artificial intelligenceaugmentation of radiologist performance in distinguishingCOVID-19 from pneumonia of other originat chest CT [J]. Radiology, 2021, 299(1): E225.
[11] SHI W Q, TONG L, ZHU Y D, et al. COVID-19 automaticdiagnosis with radiographic imaging: Explainableattention transfer deep neural networks [J]. IEEEJournal of Biomedical and Health Informatics, 2021,25(7): 2376-2387.
[12] LI J P, ZHAO G M, TAO Y L, et al. Multi-task contrastivelearning for automatic CT and X-ray diagnosisof COVID-19 [J]. Pattern Recognition, 2021, 114:107848.
[13] QIAN X L, FU H Z, SHI W Y, et al. M3 Lung-Sys: A deep learning system for multi-class lung pneumoniascreening from CT imaging [J]. IEEE Journalof Biomedical and Health Informatics, 2020, 24(12):3539-3550.
[14] ZHANG K, LIU X, SHEN J, et al. Clinically applicableAI system for accurate diagnosis, quantitativemeasurements, and prognosis of COVID-19 pneumoniausing computed tomography [J]. Cell, 2020, 181(6):1423-1433.
[15] POLIKAR R. Ensemble based systems in decisionmaking [J]. IEEE Circuits and Systems Magazine,2006, 6(3): 21-45.
[16] FOLINO F, FOLINO G, GUARASCIO M, et al. Onlearning effective ensembles of deep neural networksfor intrusion detection [J]. Information Fusion, 2021,72: 48-69.
[17] HANSEN L K, SALAMON P. Neural network ensembles[J]. IEEE Transactions on Pattern Analysis andMachine Intelligence, 1990, 12(10): 993-1001.
[18] GUNRAJ H, WANG L, WONG A. COVIDNeT-Ct: Atailored deep convolutional neural network design fordetection of COVID-19 cases from chest CT images[J]. Frontiers in Medicine, 2020, 7: 608525.
[19] ZHAO J Y, HE X H, YANG X Y, et al. COVIDCT-dataset: A CT scan dataset about COVID-19 [EB/OL]. [2021-01-07]. https://arxiv.org/abs/2003.13865.
[20] SIMONYAN K, ZISSERMAN A. Very deep convolutionalnetworks for large-scale image recognition[EB/OL]. [2021-01-07]. https://arxiv.org/abs/1409.1556.
[21] HE KM, ZHANG X Y, REN S Q, et al. Deep residuallearning for image recognition [C]//2016 IEEE Conferenceon Computer Vision and Pattern Recognition(CVPR). Las Vegas, NV: IEEE, 2016: 770-778.
[22] HUANG G, LIU Z, VAN DER MAATEN L, et al.Densely connected convolutional networks [C]//2017IEEE Conference on Computer Vision and PatternRecognition (CVPR). Honolulu, HI: IEEE, 2017: 2261-2269.
[23] HUANG G, LI Y X, PLEISS G, et al. Snapshot Ensembles:Train 1, get M for free [EB/OL]. [2021-01-07].https://arxiv.org/abs/1704.00109.
