Abstract
Purpose:
An inherent property of optical coherence tomography (OCT) well known to degrade image quality is speckle noise artifact. The technique of multiple B-scan averaging was previously shown to be a useful method for decreasing speckle noise in OCT, but the optimal number of multiple B-scan averaging was unclear. The purpose of this study is to determine the proper averaging number of spectral-domain optical coherence tomography (SD-OCT) image of retinal structure by using phantom eye model.
Methods:
A phantom eye model was made from TiO2-added silicone, and was scanned with the Spectralis OCT (Heidelberg Engineering). OCT images comprising 1, 5, 20, 60, 100 multiple B-scan averaged images were obtained and the mean gray value of the region of interest (ROI) in the several retinal phantom’s layers were evaluated. Objective differences in image quality were measured by comparing standard deviations (SD) of the reflectivity, the contrast-to-noise ratios (CNR) of each phantom's layers.
Results:
The SD of 100 times averaging (ART=100) was significantly less than either of ART=1 or ART=5 (P<0.05; post hoc Dunnett’s multiple comparison test), while no significant difference was observed against ART=20 in each layer. CNR of each layer was significantly larger in that of ART=100 than that of ART=1 or ART=5 (P<0.05; Dunnett’s post hoc test), while there was no significant difference between ART=100 and either of ART=20 or ART=60.
Conclusions:
We can quantify the image quality of SD-OCT on phantom eye model evaluated by SD and CNR, and no significant difference was found between ART=100 and ART=20. This result suggests that the optimal averaging number should be 20 or less.
Keywords: 551 imaging/image analysis: non-clinical •
630 optical properties