Abstract
Purpose :
To evaluate through-focus axial intensity profiles of the retina using line-confocal imaging and a novel 3-phase structured light imaging (SLI) technique.
Methods :
Non-mydriatic fundus images of 10 normal subjects aged 50±15 yr were obtained at 20 fps with 860 nm illumination over a 37 deg field using the Digital Light Ophthalmoscope (DLO). Through-focus axial scans were obtained using a motorized stage with 0.2 D steps over a ±8.6 D emmetrope range centered at each subject’s best corrected focus. The DLO obtains line-confocal images by synchronizing a sequence of illumination lines to the rolling shutter read-out of a CMOS sensor. By adjusting the DLO’s temporal synchronization, the image intensity can be spatially modulated at a specific frequency and phase delay to perform SLI and generate images with reduced unwanted out-of-focus scattered light. Through-focus line-confocal imaging and SLI were performed with 102 and 406 uW time-averaged power and an aperture width of 564 and 56 um at the retina, respectively. A multiply scattered light image was collected at each focus location and used to attenuate out of focus SLI frames. Axial intensity profiles were generated using a least-squares 4th order polynomial fit after registering the images and down-sampling to 64x64 regions of interest. The central 32x32 axial profiles were analyzed for each subject according to their full-width-half-max (FWHM), max/min intensity ratio and anterior/posterior slope ratio. Axial profiles were removed from the data sets when an intensity peak was not well-defined.
Results :
The SLI images were dimmer and contained significantly less unwanted scattered light from out-of-focus planes than the line-confocal images. Between the line-confocal and SLI images, the mean axial profile slope ratio and FWHM did not vary significantly (slope ratio: -1.8±1.3 vs. 1.0±0.05; p=0.07, FWHM: 53±26 vs. 50±99; p=0.57). The mean profile peak ratio (max/min) did vary significantly (1.115±0.002 vs. 1.648±0.075; p=0.0001) and the percentage of axial profiles removed from the sets for line-confocal and SLI images were 13.60±1.53% and 1.62±0.05%, respectively.
Conclusions :
The axial intensity profile peaks were found to be better defined, as measured by the max/min axial intensity, when using SLI instead of line-confocal imaging. A better-defined peak is expected to improve axial sectioning and the visualization of retinal abnormalities.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.