Abstract
Purpose:
to compare the diagnostic accuracy of optic nerve head (ONH) analysis using two different devices, confocal scanning laser ophthalmolscopy and spectral-domain optical coherence tomography and the agreement between tests.
Methods:
forty-six glaucomatous patients and 58 healthy subjects underwent a complete ophthalmological examination, visual field testing by standard automated perimetry (Humphrey Field Analyzer Carl Zeiss Meditec) and ONH assessment by confocal scanning laser ophthalmoscopy Heidelberg Retina Tomograph 3 (HRT3, Heidelberg Engineering, GmbH) and spectral-domain optical coherence tomography RTVue-100 (Optovue Inc.). HRT3 and RTVue-100 were performed in random order within the same session. Areas under the receiver operating characteristic curve, AUROCs, were calculated for ONH global parameters (rim area, RA, rim volume, RV, cup volume, CV, cup area, CA, Cup/Disc ratio, C/Dr) to compare their diagnostic accuracy. The differences of all global ONH parameters between the two devices were plotted against their means to evaluate their agreement (Bland-Altman plots). ONH sectorial analysis was considered to determine the sector with highest sensitivity considering normative classifications. The Kappa statistic was used to quantify and evaluate the agreement between sector analysis.
Results:
for both devices RA, RV, and C/Dr had good diagnostic accuracy (RTVue-100 AUROCs=0.89;0.92;0.88 and HRT3 AUROCs= 0.83; 0.84; 0.83) while CA had fair diagnostic accuracy (RTVue-100 AUROC=0.74 and HRT3 AUROC=0.76). HRT3 CV had fair diagnostic accuracy and RTVue-100 CV had very poor diagnostic accuracy. CV Bland-Altman plot showed the highest mean difference between the instruments. HRT3 infero-temporal (IT) sector had the highest sensitivity (80.43%) while for Rtvue-100 the supero-temporal (ST) sector had the highest sensitivity (76.1%). The agreement between all sectors was fair (ST, k=0.34, supero-nasal k=0.23, infero-nasal, k=0.21) while it was moderate for the IT (k=0.41).
Conclusions:
this study suggests that ONH global parameters have good diagnostic accuracy, except for CV RTVue-100 which is the only one not to be interchangeable between these two devices, while ONH sectorial analysis has good sensitivity but fair agreement between these instruments.