super-resolution imaging of retina

Even we can eliminate the optical aberrations of eye using adaptive optics technique, we still can not resolve individual cones at foveal region where the highest acuity of vision takes place. The reason is the pupil size at 6-8 mm imposes an upper limit of numerical aperture. Equivalently speaking, the limited pupil will cut the high spatial frequency information associated with the retinal structure. Super-resolution method provides a way to break this limit by shifting the high spatial frequency into the low spatial frequency so that it can pass through the eye pupil. Traditional method is structure illumination onto the object, where a sinusoidal intensity pattern is illuminated onto the object. This means this sinusoidal pattern is multiplied by the object field. Reflected light from this synthetic field is propagated to the pupil plane, which corresponds to a Fourier transformation in mathematical principle. Then the delta function related to the Fourier transformation of the sinusoidal pattern will be convolved with Fourier transformation of the object field. Invoking convolution theorem, you can see how the high spatial frequency sneak through the pupil. Usually nine patterns are needed to achieve structure illumination imaging, which is difficult to realize for in vivo retinal imaging because of eye movement. In this project, a virtual structure illumination scanning laser microscopy is adopted to realized high spatial frequency reconstruction. The trick is using point scanning to replace wide illumination. The point illumination pattern serves as the trick of helping the high spatial frequency to pass the pupil. Two-dimensional camera has to be used to keep the high frequency component to be reconstructed.

• C Liu*, Y Zhi*, B Wang, D Thapa, Y Chen, M Alam, Y Lu, X Yao, "In Vivo Super-Resolution Retinal Imaging Through Virtually Structured Detection", Journal of Biomedical Optics (Letter) 21 (12), 120502 (2016).