Eli Peli, M.Sc., O.D. - US grants

Image enhancement has the potential to provide elderly macular disease and cataract patients a better quality of life through the use and enjoyment of printed photos and television. Image enhancement is also a necessary component in the design of a new generation of electronic mobility aids. This study will first identify the best algorithm for the enhancement of still images and then formulate a real-time (video rate) version of the algorithm that can be implemented with current technology. Finally the effects of temporal factors such as image motion and eye movement will be evaluated;temporal processing will compensate for these effects. Improvement in recognition attained with various enhancement techniques will be evaluated for patients with central visual field loss or optical media opacities. Performance with filters designed to compensate for the patient's specific visual loss based on measured contrast sensitivity function will be compared with the improvement obtained with general heuristic enhancement techniques. Methods to improve the utilization of the limited dynamic range of the displays will be evaluated. Models for contrast perception and the perception of filtered images will be further developed and evaluated in the context of image enhancement. Patient's contrast sensitivity function will be measured to determine specific loss. Enhancement will compensate for loss in all spatial frequencies in one experiment and only for the highest visible frequency band in another. to utilize the limited range of luminances available on the display, partial compensation and saturation of the highest visible band will be used and evaluated. Patient recognition performance with the different enhancements will be evaluated with a criterion-free estimation methods using correlated receiver operating curve (ROC) analysis. The familiarity of the celebrities whose photos are presented to the patients will be verified with the patient's better eye after the experiment/ Patient's eye movement scanpath while examining faces will be compared with normal observes' scanpath using the scanning laser ophthalmoscope. The effect os image motion on enhanced images will be determined both by using full-face simple image motion and by measurement of patient's ability to correctly recognize changes in facial expression.

This proposal applies novel engineering approaches to the problems of low vision rehabilitation. We shall do this by building prototype devices based on solid theoretical foundations that, eventually, will become marketable rehabilitation products. The devices, designed and built with the help of our engineering partners, will be tested critically using diverse patient populations, with the help of the clinical partners to determine the effects on function and on the quality of life. We shall develop and test both optical and electronic devices that implement three specific engineering approaches aimed at restoring (at least in part) the important interplay of central (high-resolution) and peripheral (wide-field) vision. The three engineering approaches that we will explore are multiplexing dynamic control of display, and image enhancement. Also, we show that various combinations of these approaches are possible and likely to be beneficial. In our proposed assessment and testing we emphasize two approaches: a virtual environment for controlled and quantitative testing in the laboratory, and on-the-street evaluation for real-life determination of the effect and usefulness of the devices and techniques.

Project Summary / Abstract Severe peripheral field loss (PFL) sometimes called tunnel vision is a result of retinitis pigmentosa and related diseases as well as glaucoma. Currently there are no effective visual aids for these patients whose mobility is seriously affected by this impairment. The novel concept of vision multiplexing as a design principle for low vision devices has provided guidance for the development of numerous devices for visual field losses. Spatial multiplexing by shifting, combined with biocular multiplexing, has been a proven effective aid for homonymous hemianopia. This approach, known as EP Prisms, uses unilateral prisms, and is applied to the peripheral field where double vision is well tolerated, and has evolved into a successful optical treatment. In numerous studies half or more of patients fitted with the EP prisms continue to wear and use the device over a long term and reported improved mobility. For PFL, Trifield prism spectacles, a biocular multiplexing device that applied central visual confusion to patients with tunnel vision (e. g., due to RP) has proven much more difficult. Despite there being no alternative aids available, and despite it being effective for 25% of patients, it has not made inroads in the marketplace. It is very complicated to construct and most patient find the central double vision it imposes disturbing. In considering the successful and limiting aspects of both approaches, the main requirements of patients with severe PFL, and the dynamic of a walking situation, we have designed a new visual aid, Quadrafield lens (creating artificial peripheral visual islands). This is achieved by applying near peripheral field expansion while avoiding the central confusion and fields tradeoffs of prior designs. We also invented a new optical element, a multiplexing prism that can be applied to this and other devices. Multiplexing prisms provide simultaneous (monocularly) superimposed views of the shifted and direct see-through fields of view. The maintenance of the see through view permits treatment of monocular patients. It also permits fitting the prisms closer to the central field, as even with occasional encroachment, due to head movements, foveal vision is not completely blocked. In addition we have found a way to convert the regular vertical head bobbing motion into a lateral scanning through the prisms, implementing temporal multiplexing. Here we propose to implement, tune, and test the effectiveness of the multiplexing prisms as an aid for PFL, when used in the Quadrafield lens in laboratory studies and culminating in a multicenter randomized control clinical trial.

Individuals with visual field loss report collisions with other pedestrians or objects, tripping over obstacles, and are commonly not permitted to drive. All of these factors severely restrict their independence and quality of life. Visual field loss is common following brain injuries such as stroke, trauma, or tumors (hemianopic field loss, HFL) or it may be due to retinal diseases such as retinitis pigmentosa, choroideremia, and advanced glaucoma (peripheral field loss, PFL). Most visual aids developed for field expansion have had limited success. Prisms designed to shift portions of a scene from the blind field to the residual seeing field are the simplest, lightest, and most cost-effective devices for visual field loss patients. These prism devices create artificial visual islands that can help PFL and HFL patients detect and avoid collision risks. A pedestrian on a collision course will stay at a fixed position in the visual field of the patient. Our recent analysis found that the risk of a collision with other pedestrians is highest when the oncoming pedestrian approaches from an angle of 45. Conventional prism devices can shift images up to 30° but do not reach this area of peak collision risk. Further, the shifted images are distorted spatially (minified) and in color and have low contrast. When patients scan (look) toward the blind side the effective expansion benefit is limited to only 5° by current prism designs. Thus, the actual field expansion benefit of current devices falls below the best possible theoretical expectation. To overcome these limitations, we invented a new optical device, the ?multi-periscopic prism (MPP)?, which uses cascaded half-penta prisms (typically used in binoculars). Whereas conventional prisms use refraction, the MPP uses two reflections, resulting in a 45° image shift (improvement of 50% over current prisms) without the refraction effects of minification, color distortion, or contrast reduction. The MPP covers the peak collision risk eccentricity and permits 15° of effective eye scanning into the blind side (3 times wider than current prisms). We developed prototypes and preliminary configurations of this novel device to enable field expansion in HFL and PFL patients. This field expansion is intended to facilitate detection of pedestrian collisions when walking, or hazards at intersections when driving (HFL). We have proposed configurations of the device for PFL patients to allow for downward eye scanning and detection of tripping hazards. Here we propose to iteratively implement additional refinements, fine-tune, and test the effectiveness of the MPP as an aid for patients with HFL or PFL. This will begin with feasibility tests in the lab and culminate in a randomized controlled multicenter clinical trial. We will compare patients? pedestrian collision detection performance with the novel MPP devices and current prism devices and evaluate their device preferences. In the multicenter clinical trial, we will use an innovative virtual reality pedestrian collision detection test system that can be easily implemented at clinics using standard computers and large screen TVs. We will also conduct a lab test during the multi-center clinical trial to further study the efficacy of the MPP in HFL driving.