Abstract
Purpose :
Adaptive Optics Scanning Light Ophthalmoscopy (AOSLO) provides exceptional contrast of cone photoreceptor outer segments using reflectance confocal imaging due to the high reflectivity afforded by the cone geometry. Imaging modalities that provide contrast of other structures is more challenging, but nonconfocal split detection has proven effective at imaging inner segments and vasculature, and has been used to image Retinal Pigment Epithelium and ganglion cells. However, optimizing the detection geometry remains a largely brute force effort of trial and error. We hypothesize that measurement of the distribution of light in the focal plane at different retinal layers would provide valuable guidance for designing detector geometry to optimize contrast in retinal structure of interest.
Methods :
We previously demonstrated a custom microscope using a descanned imaging detector for ex vivo imaging that allowed exploration of image contrast by creating virtual nonconfocal geometries in post processing. While image sensors are far too slow to image the light distribution at each pixel in an AOSLO image in vivo, measurement of the statistical distribution of light is possible. A scientific CMOS operating at 13kHz frame rate (small ROI) was placed at the confocal image plane of an AOSLO using an integration time of 39us. A human subject was imaged with the AOSLO focused on photoreceptors (Fig. 1a) and inner retina (Fig. 1b) at 2 degrees temporal from the fovea. Each frame captured approximately half of the period of the 13kHz resonant scanner. Intensity thresholds were used to select only the frames corresponding to the pixels at the edge of the AOSLO image (Fig. 1c).
Results :
The average intensity (Fig. 1d-e) and coefficient of variation (Fig. 1f) are shown. The mean intensity peak drops when focused at the inner retina as expected and the distribution is broader. Since the the signal must vary as across pixels to produce image contrast, the coefficient of variation is used as an indication of contrast. C.V. is higher at the center when focused on photoreceptors, but when focused on the inner retina, the C.V. is higher at larger detector seperations.
Conclusions :
Measurement of the light distribution at the AOSLO focal plane is possible with a descanned CMOS detector and can provide guidance to optimize detection geometry for improved nonconfocal detection at different retinal layers.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.