Break It Down: What Are You Getting With SD-OCT?
Imagine a world, if you will, where diabetic macular edema is diagnosed only with contact lens biomicroscopy or recurrent exudation of a choroidal neovascular membrane requires repeated fluorescein angiography before an intravitreal injection. On a routine basis, this would be akin to an episode of the Twilight Zone. To the majority of us, optical coherence tomography (OCT) has been a vital tool throughout our residencies and fellowship training. You’ve ordered and interpreted more OCTs in your young career than you have visual fields, ultrasounds, and angiograms combined.
Just as with your personal computer or smartphone, advances in technology have made OCT systems better, faster, and relatively less expensive than early generation models used in our training programs. This article will focus on Fourier-domain, or spectral-domain, OCT (SD-OCT), the latest generation of commercially available OCT units with image resolution and imaging capabilities superior to earlier generation, time-domain OCT (TD-OCT) units.
Just as with any computer-based technology, your assessment as to what unit(s) to buy for your new practice or to upgrade in your existing multispecialty office should be based on understanding the technology, your intended use, ability to integrate into your existing imaging systems, and cost—not just the glitzy bells and whistles.
Why Go Spectral Domain?
OCT technology is based on the concepts of coherence and interferometry—light properties that allow the formation of interference patterns and detection of light echos. In a TD-OCT system, a superluminescent diode light source is split to a reference mirror and onto the retinal surface. The resultant reflected light waves (individual A-scans) are combined with the reference signal. The reflected waves are captured as sequential interference patterns and stored on a computer.1 The result is a 2-D, cross-sectional image with an axial resolution of approximately 10 µm, 400 pixels per A-scan, and scan velocity of 400 A-scans per second.
With SD-OCT, the mirror is replaced by a spectrometer and high resolution charge-coupled device camera. There is electronic, rather than mechanical, separation of the interference patterns that allows simultaneous rather than sequential analysis of all light echos. With SD-OCT images, the axial resolution is typically less than 7 µm, 2,048 pixels per A-scan, and a scan velocity of greater than 18,000 (up to 55,000) A-scans per second.2
So with SD-OCT, images are captured faster, with higher resolution, and with reduced motion artifact. In addition, there is an incredible amount of image data obtained at each session, allowing intricate 3-D reconstruction and better point-to-point registration for better reproducibility during follow-up scans. Furthermore, the density of scans over a given area is greater, resulting in far fewer imaging gaps throughout the posterior pole, making one less likely to miss subtle retinal architectural changes compared to TD-OCT.
Features to Evaluate
At the time of this printing, there are six US Food and Drug Administration-approved SD-OCT units on the US market, with another three units awaiting approval and available outside the United States. When determining which will be the best fit for your practice, there are several key features to compare (Table 1).
Axial resolution. Current SD-OCT units have optical axial resolutions ranging from 4 µm (HHP SDOCT, Bioptigen, Research Triangle Park, NC) to 7 µm (Spectralis, Heidelberg Engineering, Vista, CA), though the latter provides a 3.5 µm “digital” resolution. Higher resolution produces greater scan density resulting in greater definition of retinal structures and less motion artifact. When reviewing different images, it is helpful to turn off the automatic color function and look at the grayscale images. Similar to a multiple megapixel digital camera, it is difficult to find one unit that is superior to the rest; all provide superb detail of the retinal layers—far superior to any TD-OCT image (Figure 1).
Macular thickness. All units have the ability to calculate central retinal thickness. Numerous studies have attempted to correlate thickness measurements among different units, but the fact remains that each unit has a unique segmentation algorithm that defines the inner and outer retinal layers differently.3 This results in a subtle (within 30 to 70 μm) variation between certain units, but enough that treatment decisions may be incorrectly altered if these variations are not taken into consideration. Most have a normative database to compare your patient’s scan measurements and will calculate change in thickness between serial visits. The high resolution reduces image artifacts compared with TD units, and some units allow manual adjustment of the inner-outer retinal markers allowing for greater accuracy in thickness measurements.
Ocular motion tracking. This feature allows a more exact point-to-point registration of fundus landmarks, ensuring that scans will encompass the same retina region in the same eye over different scan sessions. The presumed advantage is that specific abnormalities can be imaged reproducibly to observe changes over time, helping in diagnosis or in monitoring of treatment effects. Even without this feature, the high scan resolution probably will not miss most macular pathologies, and manual registration options would allow you to reproduce scans in a particular region if needed.
Fundus image registration. The OCT unit may have a true or pseudo-fundus camera built into the unit. Compared with TD units, which had live video feeds that were often very grainy images, SD units can provide a crisp view to directly compare points on the OCT with fundus red-free, autofluoresence, angiogram, or indocyanine green images. Some units have a scanning laser ophthalmoscope or digital fundus cameras, providing detailed images, while others use a sum voxel OCT fundus representation that has lower but adequate resolution fundus details. With the fundus image on the same screen, one can decide if the scan is capturing the fundus abnormality in question (Figure 2).
Software and accessibility. A user-friendly interface is important if your practice utilizes different staff members to obtain OCT images. I find that most units are easy to use, but not all interfaces are intuitive. All systems will have their own viewer where you will be able to perform real-time image processing. These functions are fantastic but require good understanding of your unit’s capabilities and likely some extra time. In a busy clinic, you may consider setting up default views. Some units, such as the Cirrus HD OCT4000 (Carl Zeiss Meditec, Dublin, CA), allow segmentation of specific retinal layers (so called C-scan) that can provide a topographic view in isolation or very pretty 3-D projections (Figure 3) that can be rotated on screen.
If you are using electronic medical record or image management software, be sure the SD-OCT device is compatible. Viewing prior TD-OCT images may be a challenge if you switch platforms to another company. For example, the 3D-OCT 1000 (Topcon) has a built-in Stratus (Carl Zeiss Meditec) TD OCT reader. SD-OCT images are also very large and require appropriate backup and storage protocols.
Footprint. A practical consideration is the size and portability of the unit as it fits in your office. Some units have built-in computer interfaces, making them smaller, while others require a separate computer terminal that extends across a wider desk. The Bioptigen scanning head is hand-held, allowing imaging of infants and adjustment of the scanning angle to image the far retinal periphery.4
Cost. As great as a gadget may look or feel, purchasing a $60,000 to $120,000 toy that may give you essentially the same information as your current technology requires real-world consideration. The base price of an SD-OCT unit is greater than a TD unit, but there are some value added factors to consider. With an SD-OCT unit, your patients are being scanned with higher resolution, which may equate to greater accuracy in detecting retinal pathology. This may result in earlier detection of treatable conditions, earlier initiation of therapy, and potentially fewer (though more efficient) visits per patient.
For your patients and practice, having an SD-OCT image may carry a certain “wow” factor because the images are so detailed. This may enhance your ability to counsel your patient or referring doctor. Multispeciality groups may want to consider SD-OCT units that have specific glaucoma or anterior segment assessment features, allowing the initial cost to be recouped faster across your group as more physicians utilize the unit.
Add-on features such as fundus cameras or angiography capability will increase the cost but may allow more efficient imaging of patients and eliminate, in some cases, the need for multiple imaging devices. This may be a plus in a satellite office or smaller work area.
SD-OCT units with sub-7 µm resolution are the current standard in commercially available retinal imaging devices—until the next best thing arrives (and it will). However, as retinal physicians, we can all appreciate the vivid anatomic details seen with these SD units and the potential benefits for patient care. If you are considering a purchase of any new device, try it out first! Contact the sales reps and arrange and in-office demonstration or visit the booth at the next meeting you attend. NRMD
Acknowledgements: Michael P. Kelly, Duke Eye Imaging, Durham, NC, and Gregory C. Hoffmeyer, Carl Zeiss Meditec, Dublin, CA, assisted in the preparation of this manuscript.
Prithvi Mruthyunjaya, MD, is an Assistant Professor of Ophthalmology, Vitreoretinal Surgery, and Ocular Oncology at Duke Eye Center in Durham, NC, and is a New Retina MD Editorial Board member. Dr. Mruthyunjaya has no financial interest or conflict of interest with the imaging technologies discussed. He performs non-sponsored clinical research with the Spectralis, Cirrus, Stratus and Bioptigen imaging systems. He can be reached via e-mail at email@example.com.