Module 48 Section 2


	As mentioned in Section 1, spectral domain OCT technology is non-propietary, and many ophthalmic instrument companies are producing these next generation OCT instruments. Initially, there appears to be little difference between instruments regarding the process by which the scans are acquired. In other words, they seem to be gathering the same information. The differences are in the hardware and the images used for registration, and in the software post-processing of the images. In this section, spectral domain OCT technology will be discussed in general by describing the functionality of a specific instrument, the Opko/OTI spectral domain SLO/OCT. This particular instrument is highlighted for two reasons. First of all, because our office has purchased the instrument and we have daily access to it. Secondly, because the instrument is representative of all that the technology has to offer.

	Spectral Domain OCT Technology The physics underlying OCT technology is complicated and is beyond the scope of this module, but a simplified version of the difference between time domain and spectral domain technology can be seen in the schematics illustrated below. The OCT uses an interferometer that measures the time it takes for light to be reflected back from structures in the retina, as compared to the time it takes for light to be reflected back from a reference mirror. In time domain technology, the reference mirror is moving, slowing down the scanning rate. In spectral domain technology, the reference mirror is stationary, which speeds up the scanning process. The information that was provided by the moving reference mirror is replaced by employing a spectrometer on the detector side of the instrument.



	An example comparing time domain scans to spectral domain scans At this point, I would like to offer an example of a patient scanned with both the Zeiss Stratus OCT 3 and the Opko SLO/OCT. Having used the spectral domain technology now for about 9 months, it is difficult to go back to the Zeiss OCT 3. This does not mean that the Zeiss OCT 3 has become obsolete, although that day is coming. The Zeiss OCT 3 continues to be a useful clinical tool. In fact, because it was the only game in town for about 10 years, it has become a defacto standard of sorts. The eye that was scanned with both instruments has a macular hole and an epiretinal membrane. Below is a portion of a retinal thickness scan evaluation from the OCT 3, taken from a fast macular scan. The macular hole can be seen in the OCT image and there is a hint from the appearance of the left side of the scan that there may be an irregularity of the surface of the retina. We know from the image that the scan is passing through the macula, but the video image to the right is poor and it does not tell us where the scan is located.


	Below is an image captured with the OPKO SLO/OCT, using a "raster" scan. The scan has much more detail. The image at the bottom right is a fundus image that is "registered" to the OCT scan. The red line in the image tells us exactly where the scan is located on the retina. The image can be toggled so that the image of the retina is easier to see. The alternate image is shown next.





	Notice the detail in the fundus image to the left. This is an infrared image from the scanning laser ophthalmoscope. The presence of the epiretinal membrane is clearly indicated by the tortuosity of the retinal vessels. The red line tells us exactly where the OCT scan pictured to the right passes through.

	In the same amount of time (about 2 seconds) that it takes the OCT 3 to scan the 6 radial scans of the fast macular thickness scan, the spectral domain OCT scans 33 higher resolution scans that cover the entire posterior pole (raster scan). This is demonstrated using the two slide shows presented below. The first slide show runs through the six scans of the Zeiss OCT 3 fast macular thickness scan. the second slide show runs through the 33 raster scans of the Opko spectral domain instrument.

	Slide show 1 (FMT) will appear below. If you see a blank page, Java script may be blocked on your browser. If you see a "blocked content" security message, you will need to click the message and allow active controls on this page. The show runs automatically by default. You can scroll more quickly through the images manually, click the "stop" button and use the arrow keys to go forward or back.


	Slide show 2 (spectral domain raster) will appear below if you have allowed active controls on this page. Notice the detail in the retinal surface.

	We can see from this demonstration that the spectral domain OCT has greater image resolution and that the scan can cover more territory in the same amount of time. The OCT 3 is able to place 128 to 768 A-scans per scan pass. The spectral domain OCT (SD-OCT) can place 20,000 to 80, 000 A-scans per scan pass. In about 2 seconds, the OCT 3 places six 6mm lines scans in the macular area to measure retinal thickness/volume. In the same amount of time, The SD-OCT can scan a 10mm by 10mm block (raster) of the entire posterior pole from the superior arcade to the inferior arcade. Despite the marked improvement, we still have to remember that there is a trade-off between quality (resolution) and speed. The greatest resolution will still be found in the single line scan. If a large area is being scanned in short period of time, the thousands of A-scans will be spread over a larger area and each scan line in the block will be at a lower quality compared to a single line scan.

	The concept of capturing a block of OCT A-scans leads to 3-D rendering capabilities. In the image below, the software has created a 3-D topographical map, with the image of each underlying B-scan visible beneath the map (following image).





	On the computer screen, the 3-D topographical map can be rotated and viewed from different angles. The viewer can slice through each of the B-scan images by moving the mouse. You can get a sense of the capability by viewing the next slide show.

	Slide show 3 (spectral domain 3-D topographical map) will appear below if you have allowed active controls on this page. The show runs automatically by default. You can scroll more quickly through the images manually, click the "stop" button and use the arrow keys to go forward or back.

	The ability to gather "stacked" data with a SD-OCT instrument opens up many possibilities for viewing the information. Some viewing modes are more useful to the clinician than others. An individual scan from an eye with a macular hole and a sub-neurosensory detachment can be seen below. The same eye is imaged in the following animation as 3D tomography. This viewing mode perhaps has more visual interest than clinical significance.





	Spectral domain technology reduces scanning errors You will remember from section one that the Zeiss Stratus OCT 3 scanning of the macula has four major sources of error: Motion artifacts Failure to identify RNFL and RPE layers on low signal to noise ratio scans Interpolated data Lack of registration We will make a comparison to the SD-OCT for each of these sources of error to see if error has been reduced.

	1. Motion artifacts In the align process image below, the OCT 3 scanned image is the bottom image. You can see that there is a waviness that is characteristic of motion artifacts. The OCT 3 can perform a software enhancement called the "align process" that takes out the waviness (top image). The problem is that some morphological structures, such as the pigment epithelial detachment to the left, are distorted with the process (arrow).



	The same eye is scanned with the SD-OCT (image below). The speed of the scan (a fraction of a second) eliminates motion artifacts for this individual scan. However, for topographical map scans, which do take about 2 seconds of scan time, motion artifacts are still a factor, but to a lesser degree, as we will see in another example.



	2. Failure to identify RNFL and RPE layers on low signal to noise ratio scans This is still a problem, even with higher resolution scans. If the ocular media compromises the quality of the scan, then the image of the scan may be poor and the software will have difficulty identifying the layers of the retina correctly. Below is an example of a topography scan taken with the Opko instrument in which some of the scans were of poor quality. The map is color coded for retinal thickness, with the cooler colors being areas of less thickness and the warmer colors being areas of greater thickness. The top arrow points to a blue rectangular area within an area of green and yellow. This is a tip-off that there were scans in this area in which there was a software error in identifying the RNFL and/or RPE layers. A close examination of one of the scans in this area reveals that the RNFL identification line mistakenly dropped down to the RPE level (bottom arrow).



	3. Interpolated data As discussed in the first section, the OCT 3 uses only 6 scans in a radial pattern to produce the retinal thickness map. The data in between the six scans is interpolated, meaning a "best guess" is used to assign a number to the retinal thickness in the "in-between" areas. A raster scan is used to scan and create a topographical map of the posterior pole with the Opko SD-OCT. The area is scanned with 200 horizontal scans Interpolation is still used, but the area being interpolated is much smaller as a percentage of the total area. The area being scanned is approximately a 30 degree square, so the vertical interpolated area between scans is less than 2 tenths of a degree. A topographical map from the Opko SD-OCT is pictured below. Retinal thickness measurements can be viewed in a variety of formats. Individual scans that make up the raster can be viewed by sliding horizontal and vertical reference lines on the map.



	4. Lack of registration With the OCT 3, there is no registration of the scans with the fundus or fundus landmarks. This problem has been solved with the Opko SD-OCT by using the same light source for fundus imaging and for OCT scanning. For qualitative analysis, the exact position of the scan is indicated by the red line on the fundus image as shown on the image below.



	For quantitative analysis, the Opko SD-OCT is capable of registering topographical maps from two different visits and producing a difference map. The SD-OCT topographical map below is of the posterior pole of an eye with retinal edema prior to an intra-vitreal drug injection.



	The same eye was scanned one month post injection, with a dramatic reduction in retinal thickness evident on the topography scan as imaged below.



	The software can present the two visits side by side so that retinal thickness can be compared from visit to visit.



	The software can also produce a "difference" or "change" map from registering the two images, as shown below. The blue area represents the area that had a reduction in retinal thickness, with the color indicating the degree of the reduction on the scale to the left. Numerical values can also be produced, and the map can be viewed in 3D to give a "depth" indication of the magnitude of the change. Notice that the change map is askew. This is because there was an eye movement difference between the two scans and the software had to compensate during the registration process. Thus, motion artifacts have not been eliminated, even with the improved technology.




	RNFL Analysis You may recall from Section 1 that the Zeiss Stratus OCT 3 scanning of the retinal nerve fiber layer around the optic nerve head has three major sources of error: Motion artifacts Failure to identify the upper and lower borders of the RNFL on low signal to noise ratio scans Lack of registration RNFL layer identification has improved with SD-OCT because of generally improved signal to noise scanning with the new technology. The dramatic increase in scanning speed of the SD-OCT has greatly reduced motion artifacts for a single circular scan around the optic nerve head. However, a more significant motion artifact with the Zeiss OCT 3 occurs when there is eye movement between the time that the circle is placed by the operator around the optic nerve head and the moment at which the scan is acquired. Many times the circle is no longer centered around the optic nerve head when the scan is captured. Also, the operator cannot be sure of the position of the circle because the video image captured is not registered to the scan that is captured. Because the same light source is used for the fundus (SLO) image and for OCT scanning, the Opko SLO/SD-OCT offers a unique solution to these problems. This instrument is able to track the movements of the eye and keep the scanning circle in the correct position during the scanning process. Upon re-examination, the instrument can place the circle in the same position relative to the optic nerve head as it was in the previous examination. In the screenshot below, the software has frozen a fundus image of the optic nerve. The operator can now move the blue scanning circle to center it on the frozen image. Upon clicking "OK" the instrument will track the eye and keep the circle in place during the scanning process.



	Eye tracking and microperimetry The ability to track eye movements also gives the Opko SLO/SD-OCT instrument the ability to perform microperimetry. The instrument is also able to register the microperimetry pattern with the OCT topographical map. This gives information to the clinician and researcher that has never before been available. Retinal sensitivity can be measured and retinal morphology can be observed in registration for any spot or area in the posterior pole. Microperimetry is simply visual field testing with some unique and valuable capabilities as compared to visual testing with a bowl type perimeter such as the Humphrey Field Analyzer (HFA). Microperimetry dates back to the early 1990s with the old Rodenstock SLO. It was mainly used as a research tool and as a tool to guide low vision rehabilitation. Today, microperimetry is available in only two commercially available instruments. The Nidek MP-1 is a an excellent dedicated microperimetry instrument. The Opko SLO/SD-OCT is capable of microperimetry as well as SD-OCT scanning. Microperimetry compared to bowl type perimetery: The microperimeter projects the stimulus directly onto the retina. With bowl type perimetry, the patient observes the reflected stimulus as it is projected onto the surface of the bowl. The microperimeter tracks the movements of the eye and projects the stimulus onto same spot on the retina for each stimulus presentation. This eliminates fixation loss error which is a major source of error with bowl type perimeters. Tracking and registration allow the user to repeat field exams with the pattern being automatically projected onto the same area of the retina that was previously tested. The microperimeter pattern is registered to an image of the fundus, so that retinal sensitivity can be paired with the corresponding structures of the retinal surface. Bowl type perimeters do not have this feature. The Opko microperimeter is capable of registering the retinal sensitivity pattern to a topographical OCT map so that the morphological structure underlying the tested area can be studied. The microperimeter tracks fixation and creates a fixation map. This gives the clinician more information about the quality of vision. Better vision corresponds with better fixation. The fixation map is also useful in guiding low vision rehabilitation. Both commercially available microperimetry instruments use background illumination that is considerably less bright than the standard background illumination used in bowl type perimetry testing. This low background type of testing may reveal scotomas at an earlier stage in a disease process as compared to field testing with a brighter background level.
	Below is a results screen from the Opko microperimeter. The test pattern on the left is registered to an infra-red image of the fundus. Threshold decibel values are given for each spot tested. A higher decibel value means better retinal sensitivity, with a scale from 0 (brightest stimulus) to 20 (dimmest stimulus). The threshold is defined as being that level of light intensity that the patient responds to 50% of the time. A 4-2-1 threshold strategy is used. For more information on threshold testing strategy, see Module 15. On the right of the screen are fixation maps, which give graphical representations and numerical measures of how well the patient was able to maintain fixation on the internal fixation cross.



	Below is a retinal sensitivity/retinal thickness comparison map. Retinal sensitivity is given for each spot tested with the microperimeter. The pattern is registered to a retinal topography map which gives retinal thickness values within each square of the map.




	Below is pictured a screenshot of the Opko microperimetry mode. The features are as follows: Arrow 1: A live video image of the patients eye, which aids in maintaining alignment. Arrow 2: A live video image of the stimulus pattern as projected onto the retina. Arrow 3: The fixation tracking map. Dots are continuously plotted to represent the patient's fixation relative to the center of the fixation cross. Arrow 4: The microperimetry exam history. By clicking on a previous exam, the exact exam can be repeated in the same position on the retina.





	Back to top Go to Section 3

	Module 48
Home	Catalog	Contact us
	Go to Section 3	i

	Module 48:	Spectral Domain Optical Coherence Tomography
	Section 2:	The Advances of Spectral Domain Technology

	Contents:	Spectral Domain OCT Technology Example comparing time domain to spectral domain Spectral domain technology reduces scanning errors Motion artifacts Failure to identify RNFL and RPE layers Interpolated data Lack of registration RNFL Analysis Eye tracking and microperimetry