BMC Bioinformatics

Table 3 Accuracy of CNN models and standard deviation (SD) for classification of OCT images of AMD and DME in five independent runs of the experiments

From: Classification of age-related macular degeneration using convolutional-neural-network-based transfer learning

CNN-based model	Image set	Experiments
CNN-based model	Image set	1	2	3	4	5	Average accuracy	SD
Alexnet	Training set	0.9521	0.9479	0.953	0.9545	0.9508	0.9517	0.0025
Alexnet	Testing set	0.9576	0.9928	0.9752	0.969	0.9804	0.9750	0.0131
Googlenet	Training set	0.9527	0.9525	0.9536	0.9539	0.9527	0.9531	0.0006
Googlenet	Testing set	0.9845	0.9855	0.9793	0.9855	0.9845	0.9839	0.0026
VGG16	Training set	0.9592	0.9602	0.9602	0.9573	0.9602	0.9594	0.0013
VGG16	Testing set	0.9773	0.9959	0.9959	0.9783	0.9959	0.9887	0.0099
VGG19	Training set	0.9618	0.9593	0.961	0.9593	0.961	0.9605	0.0011
VGG19	Testing set	0.9938	0.9938	0.9948	0.9938	0.9948	0.9942	0.0005
Resnet18	Training set	0.9521	0.9509	0.9513	0.9507	0.9508	0.9512	0.0006
Resnet18	Testing set	0.9866	0.9886	0.9804	0.9866	0.9814	0.9847	0.0036
Resnet50	Training set	0.9565	0.9507	0.9568	0.9572	0.9568	0.9556	0.0028
Resnet50	Testing set	0.9917	0.9907	0.9897	0.9907	0.9917	0.9909	0.0008
Resnet101	Training set	0.9592	0.9584	0.9595	0.9595	0.9587	0.9591	0.0005
Resnet101	Testing set	0.9917	0.9917	0.9917	0.9917	0.9928	0.9919	0.0005

Back to article page

ISSN: 1471-2105

Contact us

General enquiries: journalsubmissions@springernature.com