Correction of the Eye's Aberrations using Adaptive Optics in a Confocal Laser Scanning Ophthalmoscope

Advances in retinal imaging have always been driven by the need for higher resolution images for the purpose of diagnosis of pathologies. However, resolution of retinal images is limited by the optical quality of the eye since any imaging system designed to image the living retina must include the eye's optics.

The aberrations introduced by the eye go beyond what an optometrist would check for and correct using spectacles or contact lenses, namely defocus and astigmatism. Higher order aberrations are also present. Though these aberrations affect visual performance considerably less than defocus and astigmatism they are still a cause for reduced resolution particularly if we are trying to image small structures on the retina, such as narrow blood vessels or even individual photoreceptor cells, or if we want to image a thin layer in the retina.

Confocal Microscopy and Ophthalmology

Confocal microscopy has been applied to ophthalmology to try to achieve the best possible resolution from retinal images. A confocal scanning ophthalmoscope offers improved lateral and depth resolution over conventional imaging techniques. However, a confocal scanning system is even more sensitive to the aberrations introduced by the eye than conventional imaging. This offers a bigger incentive to try and correct for the aberrations introduced by the eye.

 

Figure 1 shows a simple representation of a confocal setup where a point source is imaged onto the object by an objective lens and this image of the source is imaged further by a collector lens onto a point detector.

Figure 1 The confocal arrangement consists of a point source and a point detector

This configuration gives rise to improved lateral and depth resolution. However, it also makes the system more dependant on aberrations since the aberrations of both the objective and collector lenses will have an adverse effect on the resolution of the images obtained. In the eye, this effect can be understood by noting that on the way into the eye, the wave is being aberrated by the optics of the eye to form an aberrated point spread function (PSF) on the retina (the object). The outgoing light goes through the same optics a second time and hence is aberrated further. Thus any correction of the wavefront in a confocal system must correct both the incoming and the outgoing beams.

Adaptive Optics

Correcting for the eye's aberrations, however, cannot be done using static correction since the eye's aberrations are changing with time. For this reason the system we are currently working on is designed to correct for the eye's aberrations dynamically using Adaptive Optics, which is a technique in which a wavefront sensor is use to determine the shape of the wavefront and a control system gathers this data and uses it to drive a deformable mirror which corrects the wavefront.

Before an aberrated wavefront can be corrected it has to be measured to find out what aberrations are present. The wavefront sensor used in our system is a Shack-Hartmann sensor. The sensor consists of a 2-D array of lenslets which sample the local gradient of the wavefront in the x- and y- axes as shown in figure 2.

Figure 2 Shack-Hartmann wavefront sensor. An array of lenslets samples the local gradient of the incident wavefront and produces spots on a CCD camera displaced from the spots that would be obtained for a plane wavefront

The data, which in its raw form is a series of frames from the CCD camera consisting of the spots produced by the lenslet array, is fed to our control system where it is processed and used to drive a correcting device in a closed-loop. The correcting device being used is a 37-channel membrane deformable mirror. The control system sets the mirror surface to a shape which matches closely the aberrated wavefront (with half the amplitude) so that the reflected wavefront is as close to a plane wave as possible.

Laser Scanning Adaptive Ophthalmoscope

Figure 3 shows schematically the setup of the Laser Scanning Adaptive Ophthalmoscope (LSAO). The source used is a 633 nm laser which is shaped by the deformable mirror on its incoming path. A pair of scanning mirrors scan the laser beam into an x-y raster onto the retina. Light is back-scattered back from the retina following the same path as the incoming beam. The outgoing beam is corrected again by the deformable mirror. Some of the returning light is sampled onto the wavefront sensor and the remaining is incident on a pinhole and photodetector to produce an image of the retina.

Figure 3 Schematic representation of the Laser Scanning Adaptive Ophthalmoscope set up in our labs

Figure 4 shows a selection of retinal images obtained from this system with and without AO correction. The images are taken from a selection of sizes and at different magnifications.

Figure 4 Retinal images before and after AO correction. Images represent retinal patches of size (a) 2.0 deg x 1.4deg, (b) 2.0 deg x 1.4 deg, (c) 4.0 deg x 4.0 deg, (d) 1.0 deg x 0.7 deg and (e) 0.8 deg x 0.6 deg.

The improvement in image contrast and resolution can be observed from the images themselves; however they are best illustrated by figures 5 and 6. Figure 5 is a plot of the power spectra of one of the retina images with and without AO. These power spectra show that the AO-corrected image has higher spatial frequency contributions, indicating a higher lateral resolution. The case shown in this figure has a 35% improvement.

Figure 5 The power spectra of retinal images taken without and with AO correction. The red lines indicate the extent of the spread in the power spectra, with the shift towards higher spatial frequencies for the lateral horizontal direction indicated with red arrows.

Figure 6 is a histogram of the pixel values of the same set of images. The wider spread of the histogram for the AO-corrected image indicates an increase in contrast of the image; the example shown shows a near doubling of the width of the histogram.

Figure 6 Histograms of the pixel values from 0 to 255 of a retinal image taken without and with AO correction.

Investigators on this project are Steve Gruppetta and Carl Paterson at Imperial College London and Chris Dainty at NUI Galway

For further information contact Steve Gruppetta (s.gruppetta@imperial.ac.uk)

Last update 27/02/2004