The Image of Image Analyzers
December 7, 2005
Norma Devine, Editor
On Wednesday, December 7, 2005, Dr. Jonathan Myers, a glaucoma specialist at Wills, and the glaucoma chat group discussed “The Image of Image Analyzers.”
Moderator: Our topic is “The Image of Image Analyzers.” What are image analyzers?
Dr. Jonathan Myers: They are the newer machines for imaging the optic nerve. Many of you may have heard of the OCT, HRT, or GDx. OCT (optical coherence tomography) gives 3D images of thin optical “slices” of the nerve fiber layer and retina. HRT (Heidelberg Retina Tomography) performs confocal scanning of the surface of the optic nerve and retina. GDx (scanning laser polarimetry) provides a measure of the nerve fiber layer’s thickness across the retina.
P: Most of us here know we have glaucoma, so we’re more interested in detecting progression than in diagnosis. Are the new machines good at detecting progression?
Dr. Jonathan Myers: All of these machines are trying to analyze the optic nerve and nerve fiber layer to look for glaucomatous damage and progressive damage. However, there’s much more evidence on the efficacy of these machines to diagnose than there is regarding progression. Only a few studies have been published regarding progression. The best is by B. Chauhan in Nova Scotia.
In that study, 77 patients were followed for about 5 years. In that time, about a quarter got worse by visual field test and by HRT. Only 1 or 2 patients got worse by visual field test, but not by HRT. And about 40% got worse by HRT, but not visual field. That suggested to some that the HRT might be picking up earlier damage, but there’s no clear proof.
Moderator: How did these results compare with examination of the optic nerve by a doctor?
Dr. Jonathan Myers: Studies have shown that the optic nerve imagers usually, but not always, correlate to the doctor’s clinical findings. In cases where they differ from the doctor’s exam, either could be correct. Often the doctor and the machine are looking at two different aspects of the anatomy and physiology, so they may both be right, even if they seem to disagree.
P: Visual field tests are only as accurate as the patient taking the test. How accurate is an HRT?
Dr. Jonathan Myers: You are on to an important issue: reliability and reproducibility. We all know that visual field tests have a “human” component. The optic nerve imagers also have variability issues. Is the patient holding still? Is a cataract obscuring the best image? Is the technician the best technician? For this reason, there is also some variability in the nerve imagers. In major studies, the variability in the nerve imagers isn’t too bad. However, the variability of visual field testing also is often better than in the “real world.”
It is very important that the technician be very well trained for these machines, and that the doctor not “over-interpret” findings when the study quality is not optimal.
P: I’ve had two HRTs, and both reported many numbers. Are any especially relevant for glaucoma progression?
Dr. Jonathan Myers: The output of these machines has a variety of measures and indices, many compared to averages of “normal” patients. In the case of the HRT, these include measures of the optic disc size, shape, cup size and shape, rim area and volume, and topography profile. Additionally, in the HRT the patient’s values are compared in six sectors versus corresponding values from a normative database for patients of a similar age and race.
The GDx has similar measures, but derived from different aspects of the nerve fiber layer as it courses from the optic nerve to the outlying retina.
The OCT’s output is also similar to that. All of them have ways to present each new study’s findings in comparison to the last exam. Regarding this, the HRT’s analysis is the most sophisticated.
P: What tests, other than intraocular pressure (IOP) and visual fields, should a patient with diagnosed glaucoma be having and how often?
Dr. Jonathan Myers: The typical testing for glaucoma patients as recommended by the AAO (American Academy of Ophthalmology) includes frequent IOP checks, yearly visual field tests, and disc drawing and/or photos or imaging, and gonioscopy.
P: How much training do ophthalmologists receive in evaluating the output of the new devices and in using that information in diagnosing and managing the treatment of individual patients?
Dr. Jonathan Myers: Because these machines are new, most ophthalmologists have had little or no training in interpreting the output. The training doctors receive is often from the people who sell the machines, journal articles, or at national meetings and lectures.
P: Isn’t it easy for clinicians’ enthusiasm for the new devices to be influenced, not just by the science, but also by marketing and reimbursement?
Dr. Jonathan Myers: Clinicians, like all of us, are influenced by many things. Most put the patient’s welfare first, and are anxious to find new, easier ways to monitor, diagnose, and treat a disease like glaucoma. The machines are new and flashy, and there is an element of “keeping up with the Joneses” that pushes doctors to buy them. Add in that there is a reasonable reimbursement for the testing, and that there are companies aggressively marketing them with claims of great results, and you have a recipe for rapid adoption. The science behind the machines was lagging behind the marketing a few years ago, but it has caught up quite a bit now.
P: On the GDx, what does a blue area indicate?
Dr. Jonathan Myers: The GDx printout codes thicker and thinner regions of the nerve fiber layer with different colors. In general, thicker areas are orange, thinner areas blue. We expect that the nerve fiber layer will be thicker towards the top and bottom, and thinner at the sides of the optic nerve. This is the typical “hour glass” shape seen, as there is an orange spread above and below, roughly in the shape of an hour glass. A similar pattern is seen with the OCT and HRT, but the printout display is as a graph, not by color, so the thicker regions above and below show up as two humps corresponding to those areas (the “double hump” pattern).
P: What role do these new machines play in managing glaucoma?
Dr. Jonathan Myers: Many docs are now using the machines to supplement their examinations. The machines can confirm suspicions of pathology, and often may alert the doc to subtle missed clues to problems. So far, the machines have not been shown sufficient to replace the doctor’s careful examination of the optic nerve head, nor the visual field test. The machines do help to watch for subtle worsening (“progression”), which can be hard to pick up. It’s like watching grass grow.
P: Can you cite an example of a subtle clue that a doctor might miss, but a machine might detect?
Dr. Jonathan Myers: Sometimes, small areas of damage, such as a notch, or tiny defect, in just one area of the nerve, can be hard to appreciate on routine examination. The machine may pick this up, make it obvious on the printout, and allow the doctor to look back at the patient, and then notice the notch. On the other hand, a disc hemorrhage, a tiny spot of blood on the optic nerve, is also a sign that the glaucoma may be worsening. None of the current machines will detect disc hemorrhages, but the doctors can see them on careful examination of the nerve. That’s why the doctor and the machine complement each other.
P: So, therefore, the standard of necessary tests may soon be changing to include some of these new machines, but how necessary or useful are they?
Dr. Jonathan Myers: The standard of care is currently optic nerve drawing and photographs. We know that 20 years from now, a photograph will be helpful to look at. In 20 years, one or more of these machines may be ancient history, and so their printouts may be useless in the long, long term. Eventually, one or more of these machines will become the standard of care, I believe. We currently use them for most patients at the Glaucoma Service at Wills, and find them helpful. But, again, they do not replace our other tools.
P: Isn’t there also a subjective component to the HRT in that the clinician (not the technician, I assume) has to delineate the outline of the disc? Or do I have that wrong, and does the software determine or extrapolate the disc margin from the initial baseline scan?
Dr. Jonathan Myers: You are correct: the HRT I and HRT II require the doctor or technician to outline the disc. That influences the machine’s analysis of “normal” versus “glaucoma,” but does not influence its ability to monitor long-term progression. The HRT III (just released) apparently does not require outlining the disc. The OCT and GDx do not require outlining of the disc either.
P: If a person has tunnel vision in one eye, how accurate a reading can the doctor get with a visual field test? By tunnel vision I mean vision is clear straight ahead, but vision to the sides or up or down is distorted or double.
Dr. Jonathan Myers: With “tunnel vision” or a “central island,” it is important to adjust the visual field test parameters to get the most out of the test. It is useless to test the far periphery when it is already gone. More time and attention should be spent on the center. In these patients, watching the remaining central island is important, to help guard what is left.
That raises a point regarding the imagers. Generally, they are better in analyzing early to mid-stage glaucoma. In advanced glaucoma, when most of the optic nerve is damaged, there is less to image, and the image analyzers are much less useful in detecting progressive damage. For that matter, with advanced disc damage, an image analyzer is not needed to know there is glaucoma damage.
P: At what percentage of vision loss are field tests no longer useful?
Dr. Jonathan Myers: When the patient does not see at least several points consistently, then the field test cannot be used to look for progressive change, and will not be helpful. That may occur at different amounts of vision loss, depending on the exact clinical situation. Usually, it is when the vision is down to less than 20/400, or a field of much less than 10 degrees.
P: Isn’t it true that even measurement of intraocular pressure can vary from doctor to doctor, on the same day, on the same machine?
Dr. Jonathan Myers: The variability between two experienced doctors using calibrated tonometers should be within 1 or at most 2 mm Hg of pressure. The variability in the pressure from morning to evening may be much more than that for a given patient, so the time of day matters a lot, too. At this time, we don’t have a way to measure pressure continuously or frequently, which would help a lot.
P: My HRT printouts list measurement changes for parts of the optic nerve. Some are negative numbers; some are positive. Additionally, the figures are very small — hundredths or thousands of a millimeter. Are such tiny shifts meaningful or consequential?
Dr. Jonathan Myers: For most of the parameters on any of these machines, a change of more than 10% of the value (for example, 0.1 mm2 out of 1 mm2 or 0.02 mm2 out of 0.2 mm2) is necessary before you can be sure that the change is real and not just fluctuation. Almost any change of significance deserves a repeated test to confirm. Small changes are usually just “noise” in the system — movement, opacities (such as cataracts or tear film issues), technician issues, etc.
P: I would think a visual field test that did not require the patient to push a button, but somehow could measure a field without any patient input would lead to far more accurate testing.
Dr. Jonathan Myers: The Acumap is a new approach to field testing that looks at brain waves, like an EEG, to measure a patient’s ability to see lights. We are working with the Acumap to see how it compares to the standard visual field. There are a lot of technical challenges — skull thickness, patients daydreaming – – that can affect the test. So, our early results suggest similar findings to fields, but we’re not certain yet if it can replace fields. We do know that even though it takes about twice as long as a standard field test, patients like it much more.
P: How we would all welcome something like the Acumap and the new tonometer that measures IOP through the closed eyelid! No numbing drops needed in the eyes.
Dr. Jonathan Myers: I agree. However, we’ve found the through-the-eyelid tonometer to be sadly inaccurate.
P: How is gonioscopy useful?
Dr. Jonathan Myers: Gonioscopy is crucial to the diagnosis and classification of glaucoma. Knowing the angle anatomy is crucial to knowing the proper treatment for a given patient. Over time and with treatments, the gonioscopic findings may change, and so we usually repeat gonioscopy every one to several years.
P: How are floaters and vitreous detachments detected?
Dr. Jonathan Myers: Vitreous floaters and detachments can usually be seen and evaluated at the slit lamp microscope by the clinician. There are not good ways to image vitreous detachments, and the treatment is limited to observation and monitoring for any sign of a retinal detachment.
P: Even though I have an HRT yearly, my doctor still refuses to base much of his evaluation on its numbers. He insists that the software is still evolving, and its real utility will be proven in retrospect. Which aspect of the HRT software needs to be improved with successive generations of software to make it a more robust objective test?
Dr. Jonathan Myers: We don’t know much about the long-term performance (greater than 5 years) of any of these instruments in comparison to traditional measures. A 10- year study of a large number (more than 100) of patients comparing progressive field loss, image analyzer damage, clinician examinations, and other measures would be hugely helpful. Some clinicians worry that some of the early damage found with these machines may not turn out to correlate with other measures over time. Studies suggest that the analyzers will be more and more helpful, but your doctor isn’t wrong. We just are not sure yet.
Moderator: Thank you, Dr. Myers. Great answers.
Dr. Jonathan Myers: A pleasure to be with you all. Happy, healthy New Year to all! Goodnight.