Do different spectral domain OCT hardwares measure the same? Comparison of retinal thickness using third-party software

Abstract

PurposeSpectral-domain optical coherence tomographies (OCTs) from different companies do not give identical retinal thicknesses. The purpose of this study was to evaluate if differences in thickness when using a spectral domain Cirrus OCT or a Heidelberg Spectralis are due to hardware differences, or if they are caused by the segmentation algorithms.MethodsThirty-seven healthy eyes were examined within the same session with a Cirrus OCT and a Spectralis OCT, the latter using averaged B-scans. Scans from similar positions and passing the fovea were analyzed by custom-made software. Thickness was analyzed at the fovea, the central 1-mm line and the 6-mm line.ResultsWhen Cirrus and Spectralis scans were analyzed with the same software, the retinal thickness at the foveal center was 225.92 μm (SD 17.0) using the Cirrus and 228.70 μm (SD 18.4) using the Spectralis; the difference of 2.78 μm was not significant (p = 0.055). For the central 1 mm, the difference was 1.78 μm (p = 0.0414), and for all points out to 6 mm, the Spectralis retinal thickness was also significantly larger than the Cirrus thickness (p = 0.0052), though the mean difference was only 1.85 μm. Also for the RPE_OScomplex, Spectralis measured a greater thickness than did Cirrus, with a mean of 3.32 μm (p < 0.0001) for all points.ConclusionThe retinal thicknesses from the Cirrus and from the Spectralis differed by 14 μm with the standard software of the instruments, and by less than 3 μm when analyzed with the same custom-made software, indicating that the major differences between the two SD-OCT systems are due to differences in their built-in software algorithms.

Parafoveal thinning of inner retina is associated with visual dysfunction in Lewy body diseases

“…Another important factor that might underline the discrepancies between OCT studies in iPD relates to the retinal areas and layers for which macular measures were obtained . For example, the RTVue‐100 OCT (Optovue Inc., Fremont, CA, USA) calculates the average thickness of the inner retinal layers (retinal nerve fiber layer, GCL, and IPL) within foveola‐centered 5‐mm, 6‐mm, or 7‐mm diameter circular grids, whereas the Spectralis OCT provides the average thickness per layer in each sector of the 1‐, 3‐, and 6‐mm ETDRS grid, and the Cirrus OCT (Carl Zeiss Meditec AG, Jena, Germany) computes GCL and IPL thickness as a single measure and provides the mean values of the 6 sectors within an elliptical annulus around the fovea.…”

“…Another important factor that might underline the discrepancies between OCT studies in iPD relates to the retinal areas and layers for which macular measures were obtained. 41 For example, the RTVue-100 OCT (Optovue Inc., Fremont, CA, USA) calculates the average thickness of the inner retinal layers (retinal nerve fiber layer, GCL, and IPL) within foveola-centered 5-mm, 6-mm, or 7-mm diameter circular grids, whereas the Spectralis OCT provides the average thickness per Univariate regression analysis of optical coherence tomography measurements revealed GCIPL thickness variables within the central 3 mm of the macula as significant predictors of visual disability in Parkinson's disease patients, after correcting P values for multiple comparisons with the Bonferroni method. Concretely, from the overlapping optical coherence tomography variables, the mean GCIPL thickness in a 1-to 3-mm ring was found to be the most relevant variable.…”

Glaucoma: Hot Topics in Pharmacology

“…However due to the slow progression of the disease and the technologies recent emergence, there are no long term studies demonstrating SD-OCT's ability to detect glaucoma in its early stages. Furthermore studies have shown readout variability between SD-OCT devices as each companies inbuilt software systems define the outer retinal border at different levels (78). For example Cirrus HD OCT delineates the outer retinal border as the level of interdigitation between outer segments and retinal pigment epithelium (RPE) whilst the Heidelberg Spectralis defines it as level of the RPE-Bruch's membrane complex (79).…”

Comparison of the Iowa Reference Algorithm to the Heidelberg Spectralis optical coherence tomography segmentation algorithm

“…The accuracy of different segmentation algorithms in surface detection varies and this is important to consider when evaluating retinal thickness measurements . Different SD‐OCT devices can produce different thickness readings, but the majority of the retinal thickness differences seen between different SD‐OCT systems seems to be more related to the software algorithm used rather than the SD‐OCT hardware . Previous studies showed that both IRA and SD‐OCT device‐specific software exhibit high reproducibility of retinal thickness measurements, although in certain parameters IRA may provide lower variability and greater reproducibility than the device‐dependent algorithms .…”