Quantifying Reflectivity
A new software system referred to as CORDA, developed at Diopsys, a medical instrumentation company in Pine Brook, N.J., performs a unique analysis of basic B-scan information output by any of the existing SD-OCT machines. It aims to provide a means to specify the type of changes taking place inside the retinal nerve fiber layer and potentially detect signs of glaucomatous damage earlier than current OCT algorithms.
Neuroophthalmologist Alberto Gonzalez, MD, is the inventor of the CORDA software. “We noted that different optic nerve edemas produce different reflectivity in OCT images,” he explains. “For example, if edema of the optic nerve is caused by an infectious disease, it tends to be more reflective than if it’s caused by an immunologic disease such as multiple sclerosis. That convinced me that it might be valuable to quantify the reflectivity of the OCT image.”
Dr. Gonzalez points out that different reflectivity in OCT images correlates with different components of the retinal nerve fiber layer. “The current OCT algorithms use reflectivity to find the boundaries of the different layers of the retina,” he says. “However, the structures inside the retinal nerve fiber layer each have a different index of reflectivity. Those differences can also be detected in the resulting OCT image.
“CORDA stands for Color Reflectivity Discretization Analysis,” he adds.
“Discretization is a known math calculation that allows us to group together sections of a continuous scale—the color scale in this instance—so we can quantify them for the purpose of analysis. Without this kind of grouping, quantifying the colors detected by the OCT machine would be impossible.”
Working with this technology in glaucoma patients, the team noted that the retinal nerve fiber layer changes in reflectivity as glaucoma progresses.
“During the glaucomatous process, ganglion cells die because of apoptosis,” he explains. “Unlike necrosis, apoptosis is characterized by a disappearance of structure, causing the reflectivity to decrease over time. By monitoring this, CORDA allows us to detect glaucomatous change and how fast it may occur.
“Current SD-OCT measurements of the thickness of the retinal nerve fiber layer are confounded by the presence of multiple tissues inside the layer,” he continues. “For example, it’s well known that there’s great variability in retinal blood vessels between individuals. Other structures such as glial cells and supportive tissues can also cause errors in your reading. Even vitreoretinal traction can alter retinal nerve fiber layer thickness. But CORDA can isolate what really represents axons inside the RNFL, thus telling you their condition without including the confounding factors.”
Supporting Data
Dr. Gonzalez says his group is currently preparing two papers regarding CORDA for publication. “One paper presents data regarding the reproducibility of CORDA’s analysis,” he says. “The second paper presents data showing that CORDA is able to detect damage in the structure of the RNFL more accurately than the current OCT algorithms.
“One of the studies we discuss in the second paper relates to CORDA’s ability to detect structural damage in glaucoma suspects,” he continues. “The standard algorithms and CORDA are both good at differentiating between healthy eyes and glaucomatous eyes, but CORDA does a much better job of detecting abnormalities in glaucoma suspects. This could have important clinical ramifications.
“Another finding with important implications has to do with the current idea that damage mostly occurs in the inferior and superior sectors of the RNFL, primarily affecting the magnocellular system,” he adds. “CORDA also found abnormalities in the temporal and nasal sectors. That supports the novel idea that the parvocellular system is also affected, and that furthermore, it may be affected before the magnocellular system.”
L. Jay Katz, MD, director of the Glaucoma Service at Wills Eye Hospital, and coworkers have collaborated with Dr. Gonzalez to examine the potential role of CORDA in glaucoma evaluation. “CORDA seemed to segregate retinal ganglion cells from glial support tissue, with a correlation between retinal ganglion cell thickness and functional testing on perimetry,” says Dr. Katz. “The potential ability to differentiate retinal ganglion cells and support glial tissue may be a huge step forward in early detection of glaucoma and disease progression. My enthusiasm for CORDA is high, based on the information generated thus far.”
|
Enhanced Vitreal Imaging
Another new approach, utilizing a prototype swept-source OCT system, is enabling enhanced 3-D imaging of the vitreous, revealing many potentially clinically useful details that were not visible previously. Two individuals working with the new approach, Jonathan Liu, a PhD candidate in Professor James Fujimoto’s group at the Massachusetts Institute of Technology, and Andre Witkin, MD, a vitreoretinal surgeon at the New England Eye Center and assistant professor of ophthalmology at Tufts University School of Medicine, explain.
“Our prototype system images at 100,000 A-scans per second and uses long wavelengths,” says Mr. Liu. “While spectral domain OCT detects signals using a spectrometer and camera, swept-source OCT uses photo-detectors with a swept laser, which are significantly more sensitive. With SD-OCT the sensitivity decreases across the imaging range; swept-source OCT maintains high sensitivity over a long range, producing a good signal in the vitreous, all the way to the choroid.”
Mr. Liu explains that their team combined this capability with two other innovations to allow unprecedented scanning of the vitreous. “An OCT signal has approximately 40 dB of dynamic range,” he says. “A computer monitor can only display about 256 gray levels, so OCT images are typically displayed with a logarithmic intensity scale in order to show their full dynamic range. However, this makes it hard to see subtle changes in regions of weaker signals, such as the vitreous. So we took the signals coming from the vitreous and displayed them using a scale that makes its features much more visible. We also applied motion-correction algorithms which enabled averaging multiple 3-D data sets. This, combined with the higher imaging speed of swept-source OCT, allows much clearer 3-D imaging of the vitreous.
“As we reported at this year’s ARVO meeting, this technology allows us to see the bursa premacularis, or posterior precortical vitreous pocket; the area of Martegiani, or Cloquet’s canal; the Bergmeister papilla; and posterior hyaloid detachment,” he continues. “We saw multiple granular opacities in different spaces in the vitreous, including the vitreous cortex, Cloquet’s canal and the posterior precortical vitreous pocket.”
Dr. Witkin notes that the development of new drugs that act at the vitreomacular interface, such as ocriplasmin, has renewed interest in the visualization of vitreomacular interface abnormalities. He adds that it’s possible to obtain cross-sectional images of the vitreous with SD-OCT technology by displaying orthoplanes from the 3-D volumes and adjusting the dynamic range and contrast of the images. “However, swept-source OCT has advantages, such as increased imaging speed and no signal drop-off over the imaging range, that make it the best for imaging the vitreous in 3-D, especially when combined with the other software enhancements mentioned previously,” he says. “One of our hopes is that if visualizing microscopic changes in the vitreous has significant clinical utility, manufacturers will be encouraged to make swept-source OCT a part of their commercial OCT product lines.” REVIEW
