A new hyperspectral camera can measure the oxygen level of the retinal tissues non-invasively. This may allow detection of retinal changes in patients with diabetes, and other retinal vascular diseases, before structural changes in the capillaries occur.
"Our ability to image the retina structurally has improved quite a bit in the last few years," said Dr. Fawzi, assistant professor of ophthalmology, Doheny Eye Institute, University of Southern California, Keck School of Medicine, Los Angeles. "We are able to look at retinal structure down to the level of the photoreceptors in very great detail. Ophthalmologists have all been faced with end-stage severely diabetic patients, who have substantial areas of capillary dropout and structural loss of capillaries with severe vision loss. We have no therapy to offer these patients. We would like to be able to image these patients non-invasively and detect early oxygen changes in the retina before capillary loss becomes severe and vision loss irreversible."
In light of this, she and her colleagues at Doheny and collaborators at the Jet Propulsion Laboratory of the National Aeronautical and Space Administration are developing a new hyperspectral oxygen camera that can obtain color images of the retina that detail the level of oxygen saturation of the tissues.
"We are interested in developing an instrument that can capture the entire image hyper-cube in one snapshot," she said. Using a two-dimensional diffractive element, the investigators can capture the hyper-cube from a fundus photograph using an attached charge-coupled device camera that can capture the dispersed spatial and spectral information from the fundus image obtained in a single image.
One snapshot can capture spatial and spectral data and the oximetry information can be extracted and analyzed offline. Three-wavelength oximetry software analysis is applied. The investigators tested the accuracy of the instrument based on prior data that there can be about a 30% difference in oximetry between the arteries and veins in the retina. The images obtained confirmed the difference between the structures, with the arteries having about 90% oxygen saturation and the veins about 70% oxygen saturation, Dr. Fawzi said.
Collaboration between Doheny and Reichert Inc., is resulting in a refinement of the software used in this process.
"We have now developed a very user-friendly software interface in which investigators can plug in the wavelength and in a few seconds the software demonstrates the oximetry. We have improved our background extraction algorithm, resulting in much more robust images compared with previously," she said.
In line with the expressed goal of detecting early changes in the retina before structural changes occur, Dr. Fawzi also showed images of the macula in patients with branch and central vein occlusions and diabetic retinopathy that showed capillary damage and zones of foveal avascularity. A concern associated with this technology was that the oxygen-rich choroid would overwhelm the technology; however, results were encouraging that the oximetry of the macula was indeed being captured.
She also was able to demonstrate the effect of treatment with bevacizumab (Avastin, Genentech) in a patient with central vein occlusion and that the oximetry increased after treatment, which may correlate with the immediate increase in vision after injection of the drug.
"This technology is a robust means to image retinal oximetry," Dr. Fawzi concluded. "It is rapid and noninvasive. Low oxygen levels correlated with ischemia/nonperfusion in retinal vascular disease. The hope is that imaging can detect changes before structural damage occurs and result in treatment algorithms before irreversible damage occurs."