Table 3.

Area under the curve (AUC) for differentiation of UMI or non-UMI groups

P area	Area	Standard error	P	Approaching 95% confidence interval
				Lower limit	Upper limit
2SD Parea−DL	0.859	0.066	< 0.05	0.730	0.988
2SD Parea−O	0.824	0.073	< 0.05	0.681	0.967
3SD Parea−DL	0.887	0.057	< 0.05	0.775	0.998
3SD Parea−O	0.840	0.069	< 0.05	0.705	0.975
4SD Parea−DL	0.855	0.066	< 0.05	0.725	0.986
4SD Parea−O	0.781	0.084	< 0.05	0.616	0.947
5SD Parea−DL	0.891	0.056	< 0.05	0.781	0.999
5SD Parea−O	0.781	0.085	< 0.05	0.615	0.947
FWHM Parea−DL	0.797	0.079	< 0.05	0.642	0.951
FWHM Parea−O	0.797	0.079	< 0.05	0.643	0.951

Note: SD, standard deviation; P_area−DL, percentage of myocardial enhancement area with deep learning late gadolinium enhancement; P_area−O, percentage of myocardial enhancement area with original late gadolinium enhancement; 2, 3, 4, and 5SD threshold methods, mean P_area respectively adding 2, 3, 4, and 5 times of standard deviation of P_area as the threshold for myocardial enhancement area; FWHM, full width at half maximum