Comparison of manual and artificial intelligence-automated choroidal thickness segmentation of optical coherence tomography imaging in myopic adults

Table 3 Agreement and correlation between triton DRI-OCT and PLEX Elite 9000 for manual-segmented choroidal thickness measurement

Subjects	n	LOA		Proportion of outliers (n [%])	Fixed bias	95% CI	P value	Proportional bias (β)	P value	ICC ^a	95% CI	P value	r	P value
Subjects	n	Lower bound LOA (95% CI)	Upper bound LOA (95% CI)	Proportion of outliers (n [%])	Fixed bias	95% CI	P value	Proportional bias (β)	P value	ICC ^a	95% CI	P value	r	P value
All subjects	456	− 66.501 (− 70.045 to − 62.957)	21.775 (18.231 to 25.320)	23 (5.0)	− 22.363	− 24.435 to − 20.291	< 0.001	− 0.020	0.126	0.930	0.602 to 0.974	< 0.001	0.964	< 0.001
AL < 26 mm	188	− 77.698 (− 84.515 to − 70.882)	30.783 (23.967 to 37.600)	7 (3.7)	− 23.457	− 27.439 to − 19.476	< 0.001	− 0.019	0.442	0.908	0.639 to 0.961	< 0.001	0.944	< 0.001
AL ≥ 26 mm	268	− 57.005 (− 60.723 to − 53.287)	13.815 (10.097 to 17.533)	18 (6.7)	− 21.595	− 23.768 to − 19.422	< 0.001	− 0.017	0.273	0.928	0.416 to 0.976	< 0.001	0.969	< 0.001
ChT < 300 µm	375	− 63.906 (− 67.666 to − 60.145)	20.972 (17.211 to 24.733)	14 (3.7)	− 21.467	− 23.665 to − 19.268	< 0.001	0.001	0.946	0.874	0.423 to 0.951	< 0.001	0.932	< 0.001
ChT ≥ 300 µm	81	− 77.295 (− 87.126 to − 67.464)	24.270 (14.439 to 34.102)	4 (4.9)	− 26.512	− 32.241 to − 20.783	< 0.001	− 0.049	0.421	0.759	0.175 to 0.905	< 0.001	0.866	< 0.001

P values in bold indicate statistical significance
n = number of eyes; LOA = limits of agreement; β = unstandardized coefficient; r = Pearson correlation coefficient; ICC = intraclass correlation coefficient; AL = axial length; ChT = choroidal thickness
^aICC was calculated based on two-way mixed effects, absolute agreement model

ISSN: 2326-0254