Using Cirrus HD-OCT for the Management of Age-related Macular Degeneration

Learn about current applications and future capabilities.

By Philip J. Rosenfeld, MD, PhD

In 2001, the images obtained using the Stratus OCT 3 (Carl Zeiss Meditec, Dublin, Calif.) revolutionized the way physicians managed patients with macular diseases. Using the OCT 3 with breakthrough 3,072 A-scans within six diagonal B-scans centered on the fovea, we've been able to image most macular pathologies. However, this type of scanning pattern limited our ability to easily sample pathology outside the central 1 mm, therefore requiring multiple scans positioned outside the central macula. Moreover, the algorithm that measured retinal thickness was unreliable and prone to artifacts. Despite these limitations at the time, the technology was sufficient for most of our needs. Since then, our needs have changed, and our expectations have increased due to new spectral domain optical coherence tomography (SD-OCT) technology.

Now we have Cirrus HD-OCT (Carl Zeiss Meditec) with scanning patterns totaling 27,000 A-scans per second. The question heard most often involves the clinical use of the different scanning patterns: "What's the purpose of the scanning patterns, and when should we use them?"

I think of this the way I think of currency. As your salary increases, you find good use for the additional income. We've gotten an enormous raise in A-scans. Now I'm using them in ways I never imagined, and I wonder how I ever survived with just 3,072 A-scans. Here's some insight into the Cirrus HD-OCT scanning patterns and when and why you'll use them.

Cirrus HD-OCT Scan Patterns

Cirrus HD-OCT uses three major scanning patterns, each of which has different benefits and practical applications. There's the 200 A-scan × 200 B-scan macular cube with lower horizontal B-scan image quality used for volumetric and area analysis; the 512 A-scan × 128 B-scan macular cube that delivers an excellent, higher image quality B-scan representation of the macula; and the 5-line raster scan that delivers the highest B-scan image quality for a detailed analysis of the macula.

• Macular cube 200 × 200 combo: I use this cube scan on every patient to obtain proportionately accurate areas and volumes involving the macula. I like the 200 × 200 macular cube scanning strategy because both the A-scans and the B-scans are separated by 30 microns. As a result, I get a uniform sampling within a 6-mm × 6-mm area of the macula. If you want proportional volume and area measurements, and a retinal thickness map that's proportional in the X and Y dimensions, this cube of information is exceptional.

• Macular cube 512 × 128 combo: I use this scanning pattern on every patient as well, because the B-scans offer better image quality covering the macula. It's useful for qualitative analysis, and I can view the data on a monitor or with the new multislice printout, which soon will be available.

With a 6-mm cube based on 512 A-scans × 128 B-scans, you can obtain the scan image quality that's as good as or better than the standard OCT radial lines. But now, instead of six diagonal scans, you get 128 B-scans covering the macula. The A-scans are separated by 12 microns. Since the best you can get in transverse resolution is 20 microns, the separation of 12 microns is the best you will achieve no matter how many A-scans are used in the transverse dimension. The B-scans are separated by 47 microns, and a total of 128 high-quality B-scans are obtained in the central 6-mm × 6-mm area.

The new multislice printout will selectively show a sampling of the B-scans throughout the scanned macula, with the central macula being over-represented in the printout (Figure 1). In the central 1 mm of the macula — with every other B-scan shown and as the B-scans extend further from the fovea — every fourth and every eighth B-scan is displayed. However, the operator can adjust the sampling in the report. The printout will give you a beautiful representation of the entire macula, with the focus on the central macula. In summary, the standard default printout will include 11 B-scans on seven pages, so the complete 512 printout of the macula will provide an easy way to visualize this wealth of information. In the report, each B-scan will be accompanied by the laser scanning ophthalmoscopic (LSO) fundus image, the OCT fundus image and a retinal thickness map.

Figure 1. This multislice printout from Cirrus HD-OCT shows a sampling of the B-scans throughout the macula, emphasizing the central macula.

• 5-line raster: I use this scan on all my patients to get the highest quality image for detailed analysis. You can perform the 5-line raster scan horizontally, vertically or at any angle. The horizontal rasters give you 4,096 A-scans per B-scan that you can separate by 0.01 mm to 1.25 mm. And the length of the scan can vary from 3 mm to 9 mm. The resolution is the same, but the image quality is better than the 512 A-scan image.

Key Unique Features

Cirrus HD-OCT provides capabilities that no other instrument offers, including the most reliable boundary identification, segmentation and quantitation. The latter feature hinges on our ability to achieve the first two. That is, the OCT retinal thickness algorithms identify the boundaries of the internal limiting membrane (ILM) and the retinal pigment epithelium (RPE). The distance between these two boundaries determines the retinal thickness, and the more accurate the boundary identification, the more accurate the thickness quantitation.

Cirrus HD-OCT retinal segmentation algorithms are highly accurate and reproducible when compared with hand-drawn ILM-RPE boundaries, which are the gold standard.¹ Depending on the scan pattern you use to evaluate wet macular degeneration, the algorithms are 97.1% to 98.6% accurate for the ILM and 85.7% to 86.1% accurate for the RPE. The quantitation is far more reliable than the quantitation we had with the OCT 3.

Segmentation techniques make it possible for the data set to be displayed in layers, such as the ILM and RPE. These layers can be displayed individually and reconstructed into two- and three-dimensional images and thickness maps.

Cirrus HD-OCT Fundus Images

Another unique feature of Cirrus HD-OCT is the OCT fundus image. Think of it this way: A-scans make up a B-scan, B-scans make up the HD-OCT cube, and the fundus image is a projection of all of the summed reflectivity on one surface of the compiled B-scans. It's a virtual fundus image depicted on the top surface of the cube.² It's the projection of all the reflected coherent light condensed into an en face image.

What's more, not only do you get a great representation of the macula, but you can align the OCT fundus image with any other fundus image, such as a fundus photo, and you can achieve point-to-point correlation. In contrast to digital fundus cameras, which aren't calibrated, Cirrus HD-OCT gives you instant calibration of any fundus image because the OCT fundus image measures 6 mm × 6 mm.

What does the Cirrus HD-OCT fundus image show us? It reveals several types of pathology. For example, in a patient with geographic atrophy, you may see geographic atrophy on an HD-OCT fundus image that isn't visible with fundus photography (Figure 2). This occurs because HD-OCT shows the summation of the reflectivity, which is greater (brighter) in areas of geographic atrophy, and less (darker) where the RPE remains intact.

Figure 2. The Cirrus HD-OCT fundus image overlay on the LSO fundus image (left) clearly shows geographic atrophy. Compare to FAF (right).

We published a study³ that compared the areas of geographic atrophy depicted on Cirrus HD-OCT fundus images with the areas observed using fundus autofluorescence and found about a 3% difference in size.³ These two image modalities complement one another in that they measure geographic atrophy based on different principles and may reveal different progression rates. Autofluorescence depends on the presence of fluorophores, such as lipofuscin within the RPE, while HD-OCT relies on the presence of RPE and choriocapillaris to detect the boundaries of geographic atrophy. Both aspects of AMD merit further investigation. As a result, we find HD-OCT technology to be a useful adjunct in studying dry macular degeneration, and now we use it for all of our patients. In addition, because we know the calibration of the HD-OCT fundus image, we have instant measurements of the geographic atrophy.

Identifying Drusen

Using the Cirrus HD-OCT algorithm for boundary identification feature, you can accurately identify the ILM and the RPE, which enables you to locate the drusen. In fact, not only can you identify drusen, but you also can get point-to-point correlation in the two- and three-dimensional maps between the drusen on fundus photos and the drusen on HD-OCT maps. The maps also show how the retina has thinned over the drusen, as seen in a crater-pocked representation of the retinal thickness map (Figure 3).

Figure 3. The algorithm for boundary identification feature of Cirrus HD-OCT identifies the ILM and RPE, enabling you to locate drusen.

In addition, Giovanni Gregori, PhD, developed an algorithm that takes an interpolation of the normal RPE layer and subtracts it from the patient's real RPE, creating a difference map of the drusen. This correlates to the fundus image and will provide us with the ability to measure drusen areas and volumes. This feature is in development so it's not commercially available.

Keep in mind that drusen area and drusen volume aren't equivalent. The area of the drusen may be virtually the same between individuals with dry AMD, but the volume may be quite different. For future clinical trials, I'd propose placing more of an emphasis on drusen volume than drusen area.

In the past, I believed that drusen grew larger in area, then evolved into geographic atrophy, causing patients to lose vision, but we've learned that drusen volumes are quite dynamic while drusen areas may change little. Because the drusen maps are highly reproducible, you can use them to monitor drusen changes over time.

For example, we measured drusen in one patient during three visits over 6 months using Cirrus HD-OCT. The volume in the patient's right eye changed at 3 months and disappeared at 6 months. In contrast, only subtle changes were evident in the patient's left eye.

Measuring PED

Using the same modality of analysis we used for drusen, we can use Cirrus HD-OCT to measure pigment epithelial detachments (PED), which is particularly useful after anti-VEGF therapy.

When we observe a PED image, we can superimpose the HD-OCT fundus image on a fundus photo, obtain registration of the OCT B-scan and visualize the PED on the B-scan with fluid along its edges and under the retina, as depicted in the three-dimensional map (Figure 4).

Figure 4. The retinal thickness map (upper right) shows fluid on the edges of a pigment epithelial detachment (PED). The ILM and RPE segmentation maps display the PED in 3-D.

After treatment with bevacizumab (Avastin, Genentech), we saw changes in the area of the PED as we would with fluorescein angiography but, more importantly, we can follow changes in the volume of the PED. These B-scans and the volumetric analysis make it easier to follow and treat patients with VEGF inhibitors.

As a practical matter, it's important to emphasize the power of the Cirrus HD-OCT map. I've learned to look at the maps immediately when I receive the printout. For example, if I have a patient with a hemorrhagic PED, I can collect the baseline HD-OCT image, as well as the fluorescein and indocyanine green angiograms. The multislice printout shows the B-scans and the superimposed thickness map, so I can see the PED. As the patient receives bevacizumab treatments, the map will show the fluid resolution and guide future treatment decisions.

Macular Change Analysis

A new Cirrus HD-OCT feature that soon will become available is called macular change analysis, and it's going to make our lives even easier. Cirrus HD-OCT has excellent boundary identification, segmentation and quantitation. The macular change analysis software takes two HD-OCT fundus images and aligns them. This enables you to take the thickness map from one visit and subtract the thickness map from another visit, which results in a perfectly registered difference map.

The printout shows you the thickness maps from the two visits, along with the difference in thickness. As you move the horizontal B-scan on the first image, the B-scan moves accordingly on the second image. This permits the correlation of registered B-scans from one visit to the next. Over time, as the patient receives treatment, the difference map will show fluid changes. This is a useful feature, in addition to the high-resolution scans, particularly when using anti-VEGF therapy.

For example, I followed one patient with dry AMD in a clinical study. The Cirrus HD-OCT map showed me areas of geographic atrophy and dryness. Follow-up scans at one and a half months showed some activity. At two and a half months, the patient came in complaining of decreased vision, and the maps showed a little more fluid indicative of some diffuse occult leakage. I treated the patient; she returned, and the maps improved. The macular change analysis showed me the change, and I was able to scroll up and down on the map to precisely locate the fluid. The analysis gave me an accurate picture of the patient's condition at the beginning and end of treatment.

Power for the Future

In my view, Cirrus HD-OCT gives me one-stop shopping for dry and wet AMD. I still get fundus photographs, and I like autofluorescence, but I can see everything with Cirrus HD-OCT. I use a 200 × 200 macular cube when I want accurate volume and area measurements, which is most of the time. I use the 512 × 128 macular cube when I want a good representation of B-scans through the macula. Plus, I use the 5-line raster for high-definition images.

What I like about Cirrus HD-OCT is the quantitation and all of the different scanning strategies I can perform in a short period of time. The Cirrus lets me perform multiple scans, and it interprets the data in several ways. I've found that it quickly becomes indispensable.

My colleagues and I are incorporating all of these scan patterns in clinical trials as we move forward. I believe the device helps us manage our patients more successfully than ever before. OM

Dr. Rosenfeld is professor of ophthalmology at Bascom Palmer Eye Institute, University of Miami Miller School of Medicine.