Abstract
Purpose: :
To compare retinal thickness OCT metrics obtained using manufacturer supplied software to corresponding measurements made after import of the images into a reading centre's customised software platform.
Methods: :
OCT scans from 4 acquisition systems (Heidelberg Spectralis HRA OCT, Zeiss Stratus OCT, Optovue RTVue, and Optopol SOCT Copernicus HR), which are in common use were selected from a bank of 600 sets of images in a reading centre. These images were acquired from the IVAN study which was approved by the UK’s national ethics committee. Images were subjected to measurements using the manufactures own review software and also uploaded onto a server and viewed on standardised monitors using a customised software package iP OptoMize [Digital Healthcare, Cambridge UK]. Thickness measurements were made in duplicate at the fovea on a single line scan where the fovea was clearly visible. The following were measured: Neuroretinal foveal thickness (NFT), thickness of the outer high reflectivity band (OHRB) which represents the outer components of the photoreceptor mosaic, RPE and choriocapillaris, height of a pigment epithelial detachment (PED) and subretinal fluid (SRF) if present. The differences between measurements by these two methods were compared for each parameter. As a PED was absent in all of the cases scanned by the Optopol and Stratus, the results were not applicable across systems and therefore this parameter was not included in the analysis. Also only 4 of 16 eyes had SRF present and this parameter was also excluded.
Conclusions: :
No statistically significant differences were observed in the thickness metrics recorded using a standardised reading centre platform when compared to that acquired using the manufacturer supplied software. The mean differences were less than 6 µm which is the axial resolution of spectral domain OCT devices, and thus could represent operator generated measurement error.
Keywords: imaging/image analysis: non-clinical • image processing