By David Marchand (Matrox Electronic Systems, Ltd.) and Alex Wade (Department of Visual Science, Institute of Ophthalmology)

Combining commercially available imaging technology with a cLSO (Confocal Scanning Laser Ophthalmoscope), researchers at the Department of Visual Science, Institute of Ophthalmology in London, UK have developed a system to automate the capture and processing of video images of the human retina. Images of sub-retinal deposits are captured using an auto-fluorescence signal rather than a reflectance signal to allow visualization of deposits previously unseen by other imaging methods. By integrating a frame grabber and imaging library with a cLSO, a 10-fold decrease in the amount of time it took to collect patient data was achieved along with a dramatic increase in data precision. Researchers at the Institute are using the system to study a deposit's ability to auto-fluoresce, and the use of this characteristic as an indication of retinal and choroidal pathologies.

It has been discovered that some deposits in the eye, when illuminated by a specific wavelength of light, have an inherent ability to "fluoresce automatically" without the aid of a dye. The researchers at the Institute are studying the relationship between auto-fluorescence and specific deposits like lipofuscin, a fatty substance that collects in the retina. Lipofuscin's role in the normal, aging and diseased eye has sparked much interest, since many retinal diseases are believed to be associated with a build-up or absence of lipofuscin in the retina. Since researchers had no means of imaging lipofuscin's in vivo characteristics, it was necessary to develop a system that could provided the best possible images for interpretation.

When capturing images using a fluorescent signal rather than a reflective one, image information was found to be weaker and moderate to high levels of noise was evident as random speckles or snow. To use these images effectively in diagnosis, processing was required to remove the unwanted noise and heighten the weak fluorescent signal. A prototype of the system involved a tedious method of 'hand-made' image processing. To remove the noise, a precision image alignment was performed using a technique of rapidly alternating the display between one raw image and a reference image and manually aligning the images to minimize the subjective flicker. This was a slow method; 32 images took an experienced operator at least 15 minutes, effectively limiting the quality of the final (auto-fluorescence) image and the number of images that could be processed. To solve this problem, the Institute developed their new cLSO system to automatically create a noise free picture of the macula from a series of raw video frames.

cSLO auto-fluorescence has been shown to be increased in diseases known to accumulate lipofuscin (e.g. Stargardt's disease, Best's Disease), and absent in conditions of retinal atrophy (e.g. atrophic age-related macular degeneration, retinitis pigmentosa). Determined to a combination of thermal noise from the LSO photo-diode and stray photons from ambient light.

The frame grabber provides high transfer rates of up to 60MB/sec to the PC. Thirty-two images are acquired from the video source and transferred to RAM. A convolution operation is used to remove all unwanted high-frequency noise. To align the images, first a model is defined. Upon grabbing each image, a search is performed to find the model. Alignment is then performed on each image based on the location of the model. Once the thirty-two images have been precisely aligned, they are then averaged to generate the final image. In total, this process is reported to take an average of 90 seconds. Depending on how much low-pass filtering needs to be performed on the individual images, noise reduction can take anywhere from 3 minutes for high-noise images to as little as 30 seconds for low-noise images. Since the frame grabber can provide high resolution capture, the developers have increased resolution to 768 pixels x 576 lines, over double the earlier system's 256 pixels x 256 lines.

The Institute, funded for their research by Fight for Sight and Marks and Spencer and assisted with a LSO donation from Carl Zeiss, is currently using the system for in-house research only. However, there are plans to use their system for regular clinical work in hospitals soon. Preliminary results from the Institute's system has demonstrated the importance in monitoring patients over time for any changes in size and location of auto-fluorescent sub-retinal deposits. The Institute's ongoing studies in both inherited and acquired disorders has shown altered auto-fluorescence patterns with time, as well as illustrating that deposits may be displayed differently in inherited versus acquired disorders. The Institute will be doing further research of the diagnostic and clinical utility of this new in vivo technique of demonstrating auto-fluorescence of deposits within the retina.

For more information, contact our Media Relations Team.