Abstract
Purpose: :
Adaptive optics (AO) retinal imaging of rodent eyes should yield better image quality than human eyes, due to their superior numerical aperture. In practice image quality is actually much worse. This may be due to the high power and large dioptric thickness of rodent retina, which could invalidate several assumptions commonly made in AO imaging. We investigated the validity of these assumptions in a schematic model of the rat eye.
Methods: :
The Hughes rat (F ~ 305 D) and Liou-Brennan human (F ~ 60 D) schematic eyes were modeled in ZEMAX. The AO corrector was a phase plate in the exit pupil. A 650 nm AO beacon was focused on the most inner retinal layer. Zernike coefficients were optimized by brute force to achieve best AO correction.We investigated residual wavefront error under the 3 common manipulations described below. Calculations were made on-axis and 5, 8, 10 and 15° off-axis. In each case the camera plane was moved to maximize calculated image quality.
Results: :
i) Varying the image plane through the entire retinal thickness (rat = 170 µm, human = 250 µm): The rat eye could be made diffraction-limited only for eccentricities less than 10°. Performance was primarily limited by residual astigmatism and coma. The human eye remained diffraction-limited to 40°.ii) Displacing the AO corrector from the pupil plane: The rat eye could be made diffraction-limited only when the corrector was less than about 0.5 mm from the exit pupil, due to induced spherical aberration. The human eye could be made diffraction-limited with corrector positioning error greater than 100 mm.iii) Varying imaging wavelength from 475-650 nm: Residual spherical aberration increased with the square of the difference in wavelength between sensing and imaging light. The rat eye could be made diffraction-limited for wavelength differences less than 75 nm on-axis and less than 25 nm at 8° off-axis, compared to the human eye which remained diffraction-limited over the full 175 nm range.
Conclusions: :
It is typically assumed in AO imaging that changes in wavelength and plane of interest can be affected by simply altering defocus or camera position. It is also assumed that error in location of the corrector relative to the exit pupil may be a few millimeters without compromising image quality. Our analysis shows that these assumptions are indeed valid for human eye imaging, but not for the much higher powered rodent eye.
Keywords: aberrations • optical properties • imaging/image analysis: non-clinical