Table 1. Performance of classification of CXRs dataset, DA using various feature extraction methods.

Feature*	Accuracy	Sensitivity	Specificity	Precision	F-score	GMean	AUC	CI#	SE#
Gabor1	0.865	0.808	0.923	0.913	0.857	0.864	0.888	0.769–0.958	0.0482
Gab2	0.865	0.885	0.846	0.852	0.868	0.865	0.919	0.809–0.976	0.0392
Gab3	0.712	0.654	0.769	0.739	0.694	0.709	0.771	0.633–0.876	0.0668
Gist1	0.942	0.961	0.923	0.925	0.943	0.942	0.957	0.861–0.994	0.0301
Gist2	0.885	0.846	0.923	0.917	0.880	0.884	0.905	0.792–0.969	0.0444
Gist3	0.827	0.808	0.846	0.840	0.824	0.827	0.868	0.746–0.946	0.0497
HOG1	0.865	0.846	0.885	0.880	0.863	0.865	0.904	0.790–0.968	0.0456
HOG2	0.846	0.769	0.923	0.909	0.833	0.843	0.880	0.760–0.954	0.0483
HOG3	0.865	0.846	0.885	0.880	0.863	0.865	0.907	0.793–0.970	0.0440
PHOG1	0.923	0.885	0.962	0.958	0.920	0.922	0.956	0.859–0.993	0.0294
PHOG2	0.827	0.769	0.885	0.870	0.816	0.825	0.859	0.735–0.940	0.0520
PHOG3	0.808	0.846	0.769	0.786	0.815	0.807	0.828	0.698–0.919	0.0577
GLCM2	0.808	0.769	0.846	0.833	0.800	0.807	0.859	0.735–0.940	0.0510
GLCM3	0.654	0.769	0.538	0.625	0.690	0.644	0.623	0.478–0.753	0.0803

*Features extracted from: ¹ indicates whole CXR, ² indicates manually segmented CXR and ³ indicates automatic segmented CXR.

^#CI refers to confidence interval at 95% at P-value <0.0001 whereas SE refers to standard error for AUC.