Research

Visual Optics and Retinal Imaging

Over the last few years, adaptive optics (AO) has enabled imaging at the cellular level in the living eye. This literally provides a window into the state of people's health, as well as opening up the possibility of earlier detection of retinal disease. AO-assisted imaging is able to monitor changes in retinal microstructure at a pre-clinical stage of disease, opening up the possibility of new diagnostic and treatment protocols to preserve the normal functioning of the retina. AO-assisted retinal imaging systems are available for use in the clinic, but they have not been widely adopted due to the lack of automatic tools to process the high-resolution images and to detect and track features of interest. In this project, we are collaborating with a team of ophthalmologists in Rome to develop these tools. As an example, many retinal conditions affect the photoreceptor cone mosaic. We have developed techniques (some based on astronomical image processing) to accurately identify cone photoreceptors and to track them among images taken at different times. We are investigating parameters of the photoreceptor mosaic (cone density, regularity, nearest-neighbour distance) and how they change over different stages of retinal diseases. Figure 1 shows an example of a high-resolution retinal image (on the right-hand side), with automatic identification of the cones (green dots) and automatic segmentation of the shadows of retinal blood vessels. The purpose of an ophthalmic adaptive optics system is to compensate for wavefront aberrations caused by a distorting ocular medium, which blurs the retinal field. AO systems are comprised of three main elements: a wavefront sensor that measures distortion as the light scatters off the retina and exits the eye, a wavefront corrector (adaptive mirror) that compensates for this distortion, and a control system to measure the distortion from the sensor and adjust the mirror shape for optimal correction.

Once the image of the retinal field is corrected by the AO system, the image processing is applied to enhance the level of detail visible for diagnostic of patient's health and early detection of retinal abnormality. Knowing the limitations of AO correction, it is vital to develop image processing techniques that allow obtaining useful biometric data from the image and use it in a clinical setting.

Adaptive Optics Vision Simulators

Enterprise Ireland has funded (0.3 M Euro International Research Grant) under a Eureka program the "For Your Eyes only" project, which aims at developing a customised surgical procedure and personally optimised intra-ocular lens design for cataract patients.The aging eye is usually affected by cataract when the crystalline lens becomes cloudy and needs to be replaced by an artificial (intra-ocular) lens that is transparent and has correct optical power to focus light exactly onto the retina. Figure 2 shows an aberrated eye with natural crystalline lens and the eye after cataract surgery with an implanted intra-ocular lens (IOL). A customised IOL design allows bringing most of light in one focus.

Optimising surgical outcomes is the main goal of the project; we modelled the corrected visionby mimicking the optical effect of the intra-ocular lens after implantation into the eye. The vision simulator system offers such possibility thanks to adaptive optics. The deformable mirror (DM)is the key component that can act as an intra-ocular lens with the patient specific shape or could negate the effect of the natural crystalline lens. For example to estimate spherical aberration (SA) of the crystalline lens, the subject's corneal topography and wavefront aberrations have to be measured and modelled with exact ray-tracing. Figure 3 shows the vision simulator system. The patient looks through the system at a target on the OLED microdisplay. This new method allows one to evaluate different optical designs of intraocular lenses subjectively for the patient prior to implantation surgery. Three types of single-piece IOL such as equi-spherical monofocal, aspheric monofocal, and diffractive bifocal were assembled into a model eye and evaluated using AO system. Comparative analysis of these IOL designs reveals the one that is optimal for the patient.

Computational Optics

In 2015, the Applied Optics group became associated with the new Centre for Cognitive, Connected and Computational Imaging (C3I), which has been established in the College of Engineering and Informatics, NUIG. The centre works with industry to develop novel solutions for Consumer Imaging. Seed funding has been obtained from Science Foundation Ireland, which together with funding from the company FotoNation (with R&D division based in Galway), provides for 8 PhD students and 2 post-docs. Two of the PhD students are supervised by staff from the Applied Optics group. One is working on the concept optimising the design of arrays of cameras on smartphones (supervised by Dr Devaney). These can be used for depth estimation and 3D imaging, and applications such as the detection of facial expression. The other PhD student is doing research on novel optical designs for smartphone cameras (supervised by Dr Goncharov) using adaptive optics. One of the concepts includes "smart optics" that is imperfections in a camera lens are corrected by an adaptive element using the image sharpening criterion. The cost of making a perfect camera lens can be reduced if one includes such an element with an adjustable shape to match typical errors in low-cost camera lenses.

The other concept is based on the camera lens that adapts its imaging function differently when operating in visible and near-infrared (NIR) light while maintaining high spatial resolution. The application of this lens concept is primarily aimed at iris imaging and recognition. As a robust method of person authentication, iris biometrics is making its way into consumer devices such as smartphones. Current iris image acquisition devices typically work under a controlled environment and constrained acquisition conditions (e.g. airports). The new camera lens is being developed to provide iris biometrics for unconstrained, hand-held devices such as smartphones or laptops. This device is equipped with a single image sensor with both visible and NIR sensing capabilities. The device is analysed in terms of its optical properties and iris imaging capabilities. The goal is to develop a new optical system and sensor that could combine iris biometrics in NIR with conventional front camera functions such as video call and the capture of selfie in one compact camera module. Figure 1 shows the iPhone 6 plus lens and the new lens for front facing the camera with iris recognition capabilities. The coaxial adaptive lens design allows simultaneous imaging of objects in the visible and near-infrared light on a single sensor with constant angular magnification and spatial resolution. Visible light passes through the central zone of the lens while NIR light passes through the whole unobstructed aperture. Images of iris and background are focused sharply on the sensor optimised to detect both visible and NIR light. The patent application for the lens design has been submitted.

Optics for Astronomy and Space Science

In 2015, the European Space Agency awarded a 1M Euro contract to the Applied Optics group NUI to develop a prototype active optics system. Dr Nicholas Devaney and Dr Alexander Goncharov will design and build a functioning active optics system suitable for application to space telescopes. Part of the work will be subcontracted to the Fraunhofer Institute for Applied Optics and fine Mechanice (http://www.iof.fraunhofer.de).

Space telescopes can provide exquisite images of the cosmos. A prime example is the Hubble Space Telescope, which has been operating since 1990, and the James Webb Space Telescope will be launched by NASA in 2018. Astronomers are already planning the following generation of large space telescopes, which may have mirror diameters in the range 10-15m. The main problem with sending large telescopes into space is the huge cost of launching so much weight. In order to tackle this, engineers have developed ultra-thin mirrors, which weigh a fraction of what a normal mirror would weigh. However, such thin mirrors are inherently 'floppy' and the result is a severe blurring of the images. In addition to this, thermal effects can cause time-varying aberrations. A solution to this problem is 'active optics'. An active optics system measures and corrects for the misalignments in the telescope optics. The measurements may be based on the quality of the images being recorded and the optical elements are simply adjusted until a sharp image is obtained. More precise control can be obtained using specialised sensors, called 'wavefront sensors'. These are specially designed to measure the deviation of the light waves from their ideal shape, i.e. the shape that would give a sharp image. In addition, a 'deformable mirror' can be used to correct the light beam.

This technology is similar to Adaptive Optics, which was developed for telescopes on the ground to correct for blur caused by atmospheric turbulence, and has found uses in ophthalmology and microscopy. The Applied Optics group has extensive experience in applying adaptive optics to each of these fields. We have proposed a baseline 8-m diameter space telescope (named 'Hypatia'), see figure 1, and we are currently developing the preliminary design for the telescope's active optics system. The following stage will be to build the laboratory prototype to demonstrate the extremely accurate wavefront sensing and the stability required by deformable mirror M4. This project offers great potential for future involvement in Space Optics projects and inspiration for our graduates.