A proprietary high-resolution spectral-domain optical coherence tomograph (Spectralis, Heidelberg Engineering) offers many advantages to clinicians, such as superior image quality, true multimodal imaging, and improved monitoring of disease progression, says one doctor.
"SD-OCT is the new standard for OCT image quality," said Dr. Walsh, assistant professor of ophthalmology, Keck School of Medicine and Doheny Eye Institute, Retina Division, University of Southern California, Los Angeles.
"With most SD-OCT instruments available commercially the temporal nerve fiber layer may be difficult to see, and sublaminae of the inner plexiform layer cannot be visualized. With [this SD-OCT unit], this is not the case. In fact, despite its quoted 7-µm axial resolution, images from [this instrument] rival or surpass current ultra-high-resolution SD-OCT systems," he said.
Dr. Walsh then raised the question about the need for real-time eye tracking for averaging.
"Active eye-tracking enables accurate volume B-scan averaging, which produces images that rival ultra-high-resolution OCT," he said.
Dr. Walsh said that real-time eye tracking may not be necessary for individual B-scans because some patients can keep their eyes steady for the fraction of a second it takes to acquire four, eight, or 16 B-scans.
"However, when it comes to the real power of SD-OCT, volume scanning, few patients can keep their eye[s] steady enough to capture hundreds of averaged B-scans. Therefore, for volume SD-OCT, real-time eye-tracking is a must." In Dr. Walsh's preliminary experiments, "the total amount of ocular movement in 1 to 2 seconds can exceed 500 µm in just 66 milliseconds, especially in patients with macular disease and poor fixation," he said.
It is now clear that SD-OCT outperforms time-domain (TD) OCT in almost all areas, he said.
Dr. Walsh also commented that, in his opinion, the optimal number of B scans to use for volume OCT is probably pathology-dependent. For example, to get a fairly good thickness map for macular edema, only sparse sampling may be required. On the other hand, detection of small features such as microaneurysms or pigmentary migration might require sampling in the hundreds of B scans. Dr. Walsh said he prefers more than 100 B scans but also is satisfied with 37 or 71 B scans on this device.
Alignment of B scans within a volume OCT scan, so-called intravisit registration, is required to align OCT data with other modalities, such as angiography and fundus autofluorescence.
Active eye-tracking, he said, allows registration with fundus images and comparison of the fundus between different visits (intervisit alignment). Passive eye-tracking also can be registered with fundus images, but comparison among different visits is harder. The absence of eye-tracking means that accurate registration with fundus images and accurate comparisons between visits are not available.
Accurate alignment of volume OCT scans between visits (intervisit alignment) requires good intravisit alignment.
In a 2008 study by Wolf-Schnurrbusch, et al., this proprietary SD-OCT device had the smallest measurable changes (3 µm) compared with a combination SD-OCT/scanning laser ophthalmoscope system (Ophthalmic Technologies), with a 7-µm change; another SD-OCT unit (Copernicus, Optopol Technology) and a TD-OCT unit (Stratus, Carl Zeiss Meditec), both with a 9-µm change; and another SD-OCT device (Cirrus, Carl Zeiss Meditec), with a 16-µm change.
For the optimal imaging results in his practice, Dr. Walsh explained his "dos and don'ts."
"I don't use retinal thickness maps because they're often wrong in my patients, of whom a large percentage have age-related macular degeneration. I only occasionally use three-dimensional renderings when counseling patients, because they don't show the clinical details I need to monitor my patients," he said. "I don't use printouts. And I don't use TD-OCT."
Dr. Walsh added, however, that he does look at every SD-OCT B-scan either on the machine (mouse scroll wheel) or via their Intranet (Heyex). He said he also monitors disease progression by qualitative comparisons of B scans from the same location.
Regarding the speed with which SD-OCT obtains scans, Dr. Walsh said that increased A-scan acquisition speed often comes at the price of decreased B-scan image quality and requires averaging to produce an acceptable image. "However, for clinical efficiency, the total amount of time that a patient spends being imaged is probably more important than the OCT acquisition speed and time," he said.
Dr. Walsh said he believes that real-time eye-tracking sets this particular SD-OCT system apart from its competitors.
"B-scan averaging with tracking in volume OCT mode provides superior image quality. Precise multimodal comparisons are possible since the location of every B-scan is known. Accurate intervisit comparisons can be made because B scans from the same locations can be compared. Since tracking and averaging somewhat compromise acquisition speed, an instrument's 'door-to-door' time must be balanced with the above-mentioned benefits," he concluded.