Abstract
Purpose :
Confocal adaptive optics imaging provides high contrast, cellular scale, images of the retina. However, confocality limits information from the retina to primarily singly scattered light. Imaging with multiply scattered light can provide exquisite detail on otherwise difficult to see retinal features that are not as evident in confocal images. Designs for capturing multiply scattered light have been limited to a limited range of aperture designs, yet in principle it should be possible to optimize apertures for specific retinal features. Here we present the design and initial tests of a confocal adaptive optics SLO (AOSLO) that uses a programmable micromirror array to dynamically adjust the division of multiply scattered light to different detectors.
Methods :
We designed and built an AOSLO using a supercontinuum laser (Fianium), a woofer tweeter AO subsystem, and a steerable front end. The major difference from our previous systems is the light detection channels. We use a reflective confocal aperture that directs singly scattered light to an avalanche photodiode (APD). The surround of the point spread function is magnified by 50% and re-imaged onto a micro-mirror array (TI- DLP 6500). Each micro-mirror is 10 microns (~ 0.15 Airy disc diameters). The array then directs light, pixel by pixel, to one of two detectors. Apertures were programmable from a MATLAB program that allowed changing aperture parameters during imaging from an initial set of 5 parameterized shapes. Shapes were then uploaded to the micro-mirror controller.
Results :
Images were obtained on 3 eyes. Confocal images had high contrast and resolution (Figure 1). Multiply scattered light images showed expected features, with the exception of small flickers affecting < 1% of the image pixels. Efficiency of the programmable array was less than our system which uses movable apertures, but in line with design efficiency for the micro-mirrors (> 80%). In practice the multiply scattered light images were bright and did not require increased retinal irradiance.
Conclusions :
A programmable array system for dividing a single point scanned SLO into multiple channels and to alter that division of light on the fly holds promise for matching the imaging system to the detection of specific retinal features. The current system has some drawbacks. These arise from the purely complementary nature of the two multiply scattered light images, and the addition of a small amount of pixel noise.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.