Abstract
Purpose :
Speckle is an interference effect, caused by the mutual interference of coherent or partially-coherent wavefronts. In imaging applications where light sources with a high degree of temporal coherence (such as lasers) are used, speckle results in a “grainy” effect that degrades image contrast. We present an approach to mitigate the problem of contrast-degrading speckle by combining data from a fast sequence (or “burst”) of partially-uncorrelated infrared (IR) images to calculate a single output IR image with significantly reduced speckle.
Methods :
In this study, widefield fundus images were acquired on nine subjects (N=9 eyes) using a prototype slit-scanning device with a 790nm laser diode source. Natural eye motions, which the human eye consistently makes (e.g., drift, saccades) even while fixating, lead to small changes in orientation that result in an image with a slightly changed speckle pattern. In order to reduce the speckle, a sequence of 10 fundus images acquired over the course of 1 second were registered and averaged.
Results :
For a total of nine subjects, mean speckle contrast was compared for the single-capture images and the averaged registered sequence of images. The percent reduction of mean speckle contrast of 34.8±14.5% was statistically significant (p<0.01).
Conclusions :
In laser-based retinal imaging systems, the motion of the eye during image acquisition can be exploited to suppress unwanted speckle, thereby improving image contrast. Averaging a sequence of frames would not significantly increase patient exam time, yet could boost contrast, lending to improved image quality. Laser diodes can provide high radiant power with low cost and compact size, but speckle limits their suitability for imaging. We showed that speckle averaging can provide a simple means to mitigate this limitation in a 790nm widefield fundus imaging system.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.