Skip to main content
. 2022 Apr 7;5:827299. doi: 10.3389/frai.2022.827299

Table 1.

Accuracy, sensitivity, and specificity of TB-Net on the test data from the multi-national patient cohort.

Architecture Accuracy Sensitivity Specificity
(%) (%) (%)
CheXNet (Rajpurkar et al., 2017) 99.42 100 98.85
EfficientNetB0 (Tan and Le, 2019) 98.99 99.42 98.56
NASNetMobile (Zoph et al., 2018) 99.28 98.84 99.71
TB-Net 99.86 100 99.70

Better performance metric in bold.