Purchase this article with an account.
C. A. Briceno, R. R. Pappuru, Y. Ouyang, A. C. Walsh, S. R. Sadda; Clinical Significance of Frame Averaging With Spectral Domain OCT. Invest. Ophthalmol. Vis. Sci. 2010;51(13):1779.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The ability to average multiple scans and suppress speckle noise artifact is a potentially important advantage of spectral domain optical coherence tomography (SDOCT), but the amount of averaging required to produce clinically significant benefits has not been established. In this study, we evaluate the benefits of B-scan averaging by assessing the visualization of outer retinal structures of interest.
We collected OCT data from the Cirrus HD-OCT (ver 4.5) from 35 eyes of 35 consecutive patients referred to the imaging unit for retinal disease. All patients underwent 3 scan acquisitions: (a) 1024 x 5 high-definition (HD) B-scan raster without averaging, (b) 1024 x 5 with 4x averaging, and (c) a single B-scan (1024 x 1) through the foveal center but with 20x averaging. For each acquisition type, the central B-scan was used to assess visualization of four outer retinal features of clinical value: (a) external limiting membrane (ELM), (b) inner segment - outer segment (IS-OS) junction, (c) retinal pigment epithelium (RPE), and (d) choroid. Quality of visualization of each feature was graded in a masked fashion (as to scan type) according to three parameters (ability to identify, brightness, and continuity), each of which were rated on a 0-3 qualitative scale (0 being worst and 3 best). Quality Sores (QS) for the three scan types were compared using Chi square test for difference.
Quality scores are shown in Table1. QS consistently improved with 4X averaging and improved slightly further with 20x. The most significant benefits were observed fro brightness scores. For ELM identification, 91.4% of 20x cases achieved the highest QS score ("3") on all parameters compared with 82.9% of 4x cases, and only 54.3% of non-averaged cases. Amongst the various structures, averaging appeared to have the least beneficial effect for assessment of the RPE and IS-OS junction.
Retinal structures of clinical importance, in particular the ELM, appear to be better visualized with averaging of Cirrus OCT B-scans. Benefits of oversampling are apparent with averaging of even relatively few frames, with less incremental improvement with higher degrees of oversampling.
This PDF is available to Subscribers Only