Purchase this article with an account.
R. de Kinkelder, J. Kalkman, D. J. Faber, P. H. B. Kok, F. D. Verbraak, T. G. van Leeuwen; Post-acquisition Analysis of RNFL in SD-OCT (3DOCT-1000 of Topcon) Acquired Volumes. Invest. Ophthalmol. Vis. Sci. 2009;50(13):354.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
A novel algorithm for high accuracy volumetric measurements of the retinal nerve fiber layer (RNFL) thickness to monitor changes over time.
SD-OCT (Topcon 3DOCT-1000) volume scanning of the optic disk (128 x 512 a-scans, 6 x 6 mm, scan rate 27000 samples/s, axial resolution 5 micrometer) was done in one healthy volunteer. Ten volume scans were obtained with the optic nerve at a fixed location within the 3D volume. Using the segmentation of the RNFL done by Topcon 3DOCT-1000, the custom made algorithm determines the average RNFL thickness over a ring with an adjustable bandwidth, at a variable distance from the center of the ONH in all images. The average thickness and standard deviation as a function of angle around the optic disk were determined. The RNFL thickness was also calculated using the RNFL thickness program in the 3DOCT-1000 software. To compare both algorithms, the average RNFL thickness per segment (temporal-superior-nasal-inferior-temporal, TSNIT) was calculated at a fixed distance from the center of the optic disk (radius 1.7 mm, bandwidth 0.2 mm) with the custom made algorithm. Additionally, the variability of the overall average RNFL thickness, measured with the custom made algorithm, was determined for different observers.
The average RNFL thickness per segment (TSNIT) of ten scans calculated by the 3DOCT-1000 software showed a standard deviation of 2 to 8 % per segment. The average thickness calculated per segment by the custom made program showed similar results, 2 to 7% per segment. The accuracy of the custom made algorithm was determined; the intra observer variation for the average RNFL thickness was 0.2%, and was determined by calculating the RNFL thickness 10 times on the same data by the same observer. A same order of variability was found between different observers.
With the custom made algorithm we are able to calculate the RNFL thickness as a function of angle integrated over an adjustable bandwidth. The accuracy per segment was comparable to the current Topcon 3DOCT-1000 software. However, the new software enables to monitor local thickness changes over time and it enables determination of volumetric RNFL thickness.
This PDF is available to Subscribers Only