Dr Graham Lakkis
BScOptom GradCertOcTher FACO
Lakkis Optometry, Keilor East VIC
Nidek RS-3000 Advance OCT Angiography
Designs For Vision
OCT Angiography (OCT-A) is a newly developed, non-invasive technique to image blood flow in the retinal vessels and small capillaries without the need for injection of fluorescein dye. As an early adopter of OCT-A 12 months ago, I have experienced a learning curve on interpreting the results and in determining which eye diseases are best managed with the extra angiographic information.
The technology of OCT-A is still in its infancy, and there have been multiple software and hardware upgrades from Nidek in the past year that have improved acquisition speed, scan resolution and data analysis of the information.
What has changed in Nidek OCT-A? How has it helped in diagnosing and managing patients?
Imaging speed through better eye tracking
OCT-A operates by scanning the same area of the fundus multiple times, subtracting any stationary structures that are present in all the images and leaving only images of the moving blood within the vessel walls. The Nidek RS-3000 Advance OCT uses advanced eye tracking in three planes (horizontal, vertical, torsional) to maintain the same scanning position.
The earlier version would stop scanning if the eye lost fixation, and would resume scanning only once the eye had refixated. This was a problem in patients with central vision loss such as centre-involving diabetic macular oedema and exudative AMD who had poor central visual acuity and were unable to fixate steadily on the fixation cross. The newest upgrade allows the infrared OCT beam to follow the tracking (cSLO) beam whenever the patient moves their eye without having to pause the scan, and allows the same retinal position to be scanned even in a moving eye. This has allowed us to image patients more rapidly, three to four times more quickly, and to capture images in eyes that were too unsteady with the earlier version of software.
Those patients with reduced central vision often benefit greatly from the extra information generated by OCT-A. (Figure 1)
Figure 1. A: Previous version. Patient with advanced glaucoma affecting central vision. After three minutes, only one-third of the optic nerve head scan was completed due to frequent fixation losses, so eye tracking was switched off for the remainder of the scan, resulting in unusable information. B: Upgraded version. Another patient with advanced glaucomatous central vision loss rapidly and successfully scanned with upgraded eye tracking algorithm. Except for the region from 8:30 to 11:00 o’clock, all the disc capillaries have disappeared due to glaucoma.
Improved resolution through image oversampling
With conventional OCT line scans, repeating the scan multiple times in the same location improves resolution by eliminating speckle noise from the images. The early version of OCT-A was able to oversample each scan line only twice. Version 2 software can now oversample eight times. This has achieved a remarkable improvement in capillary detail and resolution. (Figure 2)
Figure 2. A. x2 oversampling and B. x8 oversampling. Significantly improved image quality and capillary resolution with the increased image oversampling allows better determination of the foveolar avascular zone and detecting other areas of capillary drop-out in patients with diabetes and retinal vein occlusions.
One of the benefits of conventional OCT is the ability to quantify retinal parameters such as ganglion cell complex thickness, retinal oedema, and nerve fibre layer distribution while comparing these parameters to a normative database. OCT-A has been lacking in this area of analytics but is starting to improve. Nidek AngioScan Phase 2 now allows measurement of the foveolar avascular zone (FAZ) area so that worsening capillary non-perfusion can be quantified by comparison to the patient’s own baseline. (Figure 3)
Figure 3. A: En face OCT image of the macula in a patient with type 2 diabetes showing numerous microaneurysms. B: OCT-A image of the same eye with quantification of the FAZ area in mm2 (in centre). The area of capillary drop-out can be monitored for change in the future.
Dynamic assessment of images
Properly interpreting OCT-A images requires an active and dynamic visualisation of the retinal vasculature that is not easily achieved by looking only at single images of the retina. To navigate between the retinal layers, earlier versions of the software required input of numerical values to move the image anteriorly and posteriorly through the retina and to select various image slab sizes. The newest software employs a much more intuitive system where the boundary lines on the OCT image can be clicked and dragged anteriorly and posteriorly while the adjacent OCT-A image updates in real time, and the slab itself can also be moved around by grabbing between the boundary lines with the computer mouse. (Figure 4)
Figure 4. The lower line (white arrow) on the OCT image is clicked and dragged up and down as the adjacent OCT-A image updates instantly allowing dynamic visualisation of the retinal vasculature. The space between the two lines (between two white arrows) indicates the ‘slab’ size; the slab can be moved by clicking between the lines and dragging the whole slab up and down.
During OCT-A scanning, the superficial retinal blood vessels reflect their image onto the deeper retinal layers, causing vascular projection artefacts. The outer retina normally houses the photoreceptors/RPE but does not contain any blood vessels. Choroidal neovascularisation can invade the outer retina in exudative (wet) AMD so it is important to detect any blood vessels in this area. Projection artefacts can make this difficult, so Nidek version 2 software has the ability to automatically remove the artefacts, allowing a more accurate representation of the outer retina. (Figure 5).
Figure 5. A: 112-micron slab located at the outer nuclear layer (ONL) shows an OCT-A image containing numerous blood vessels even though the ONL normally contains no vasculature. B: After application of projection artefact removal, the same slab is now relatively free of blood vessels, indicating most of the vessel images were artefacts
The Nidek OCT-A produces cube scans of 3 x 3 mm up to 9 x 9 mm in multiple size increments; however, increased scan area results in a loss of resolution as the A-scans are spread over a wider area. To counteract this, the new version 2 software takes multiple high-resolution scans and stitches them together automatically into a montage of up to 9 x 12 mm without loss of capillary detail. This is now almost equivalent to the area of fundus imaged by traditional 45-degree fluorescein angiography cameras but in much greater detail.
CNVM flow mode
One of the most devastating causes of central vision loss is exudative macular degeneration where abnormal choroidal vessels break through Bruch’s membrane into the sub-RPE and/or sub-retinal space. This neovascularisation leaks fluid and exudate into the retina, resulting in profound vision loss. OCT and OCT-A have a vital role in detecting these choroidal neovascular membranes (CNVM) so that anti-VEGF therapy can be promptly initiated.
Figure 6. A: Horizontal OCT of the macula showing central retinal thickening with disorganisation of the retinal layers, cystoid macular oedema and subretinal fluid. There is an outer retinal hyper-reflective mass suggestive of CNVM. B: Depth colour-coded OCT-A image of the same eye. Superficial capillaries are coloured red and yellow while the deep capillary plexus is in dark blue. The presence or absence of CNVM is difficult to determine. C: CNVM Flow image of the same eye clearly showing the presence of two areas of CNVM in the outer retina denoted in yellow. A diagnosis of exudative AMD can be made.
Even with dynamic assessment of the vascular layers it can be difficult to detect CNVM as the retinal layers are often disorganised, oedematous, and full of scattered haemorrhage and exudates. Nidek AngioScan Phase 2 now has ‘CNVM Flow’ mode where the presence of CNVM in the outer retina (which is normally avascular) is automatically detected and colour-coded to enhance visibility. The effectiveness of anti-VEGF therapy in shrinking the size of the CNVM can be more easily determined by comparing future CNVM Flow images to baseline. (Figures 6 and 7).
Figure 7. A: Horizontal OCT of the macula showing a very large retinal pigment epithelial detachment (PED) with a smaller area of subretinal fluid laterally. There is no hyper-reflective mass in the outer retina. B: CNVM Flow reveals an extensive choroidal neovascular net (yellow) containing very large vascular trunks, located in the region of the PED causing the chronic leak in fluid. A diagnosis of exudative AMD can be made.