vessel is slower than the scan time of B-scans in rapid succession, then the B-scans acquired at that location would display no differences, and thus the flow would not be detected. The normal interscan time of consecutive B-scans on SD-OCTA devices is approximately 5 ms, while that of the vertical cavity surface-emitting laser (VCSEL) SS-OCTA prototype is approximately 1.5 ms. Faster scanning speeds of SS-OCTA allow devices to capture more B-scans in rapid succession at a particular location without substantially increasing imaging time. With more B-scans available at a particular location, non-consecutive B-scans can be compared. For example, instead of assessment of the decorrelation signal between consecutive B-scans, the differences between alternate (every other) B-scans, now with an interscan time of approximately 3 ms (doubled from the approx. 1.5 ms of consecutive scans), can be compared [8, 16, 40]. This technique is used in VISTA, which allows for the detection of slower flow speeds by varying the time between consecutively acquired B-scans at the same location, thereby obtaining a different decorrelation signal compared to that generated by the traditional consecutive B-scan interscan time. Consequently, VISTA has been able to decrease the standard OCTA threshold of the slowest detectable flow, as well as vary the fastest discernible flow.
Fig. 2. a SD-OCTA image of a 92-year-old female with advanced dry AMD with GA. The region of GA (yellow contour) depicts a well-defined border between normal CC around the GA lesion and loss of CC within the lesion itself. Underlying larger choroidal vessels, which have likely migrated upwards, are clearly visualized. b The corresponding structural B-scan shows increased choroidal light penetration in this region due to RPE loss.
VISTA has helped improve our understanding of GA by allowing for better visualization of slow flow in the CC, which may have previously been seen as an area of no flow. Moult et al. [16] have demonstrated that choroidal vessels not visible by regular OCTA become visible after the application of VISTA, indicating that areas around GA may involve flow impairment, rather than complete loss of flow. However, CC alterations within the GA lesion tend to be primarily atrophic, while CC changes at the periphery of GA and beyond appear to be flow reduction.
Of note, the CC alterations that extend beyond the border of GA are often difficult to visualize with SD-OCTA, as the RPE is still intact at the periphery. Conversely, within the GA lesion, the atrophied RPE no longer attenuates the SD-OCT signal. It has been suggested that at the periphery of a GA lesion, abnormalities of the RPE and photoreceptors and drusenoid deposits are linked to an increase in area of GA [31]. These findings, combined with the peripheral microvascular changes detectable with VISTA, suggest that CC flow alterations may occur sooner than previously thought and precede the overlying structural atrophy well visualized with OCT.
While retinal vascular changes do not occur in early AMD, as in more advanced stages, where there is atrophy of the retinal layers such as in GA, changes in the retinal vasculature have also been demonstrated. Thinning of the inner and outer retina and reduced flow in the superficial plexus in intermediate AMD patients precede the development of GA [42]. These are thought to be because of a functional and structural atrophy of the retina, with the RPE being the last structure that is intact.
Other morphologies associated with higher risk of progression to GA include RPD [43]. RPD were first described in 1990 as yellow macular deposits that did not fluoresce on FA, but demonstrated increased visibility under blue light [39, 44]. As a distinct phenotype of AMD, RPD are an independent risk factor for the progression of AMD and are associated with worse visual outcomes even at an earlier stage [39]. Studies have linked RPD to choroidal thinning [28, 43]. Additionally, compared to eyes with a similar pattern of drusen, eyes with RPD have demonstrated greater CC non-perfusion [43]. Unsurprisingly, CC and choroidal alterations are a key factor in RPD pathogenesis as well, with the same questions being raised about which vascular/structural changes precede the others [43].
The techniques of microperimetry and minimum intensity OCT may assist in predicting where the GA lesion will progress [45]. Microperimetry, functional testing of the retina, detects photoreceptor abnormalities before GA develops in that region. These early visual deficits associated with early photoreceptor changes suggest that perhaps it is even the photoreceptors that precede the other structural changes, namely in the CC and RPE, associated with GA [46]. Minimum intensity OCT visualizes photoreceptor disruption with a different principle. Minimum intensity projections are created by finding the minimum, or darkest intensity, along each A-scan between the ILM and RPE. Stetson et al. [47] found that, after analysis of the minimum OCT intensities surrounding a GA lesion, GA tended to grow into areas of increased minimum intensity, rather than uniformly around the lesion itself. Overall, studies analyzing changes in the CC, photoreceptors, and RPE at the margins of GA demonstrate that much remains to be discovered about the development GA and, subsequently, prediction of GA progression.
Limitations of OCTA
Despite the valuable insight OCTA has offered into the pathogenesis and features of dry AMD, it is still an evolving technique that is not without its limitations. OCTA images are prone to degradation by motion and projection artifacts [48]. Projection artifacts occur when superficial retinal blood vessels scatter light that is reflected by other deeper reflective layers. As a result, during segmentation, larger superficial retinal blood vessels may falsely appear in deeper layers, obscuring visualization of true vascular pathology. Since OCTA analysis of dry AMD is focused on visualization of the choroid and deeper retinal layers, projection artifacts must be kept in mind during image analysis to avoid inaccurate detection of flow. Furthermore, as discussed earlier, drusen can cast shadows underneath, which may be falsely interpreted as decreased flow in the underlying CC. However, analysis coupled with en face OCT intensity images can assist in avoiding such error.
Additional artifact may be generated due to the underlying larger choroidal vessels as well. With CC atrophy, choroidal vessels are more readily visualized, but also may migrate upwards, closer to Bruch’s membrane, occupying the generated void. These vessels may confound the OCTA image, demonstrating CC flow when there is actually CC flow impairment or atrophy.
Despite its limitations, OCTA has expanded upon our previous knowledge of dry AMD, to allow further insight into the development and progression of this pathology. With future developments offering artifact-removal algorithms, a wider field of imaging, longer wavelength light sources for increased depth-penetration, faster imaging speeds, and an improved VISTA technique, OCTA remains a promising technique for detecting and monitoring retinal and choroidal changes in dry AMD.
Conflict of Interest Statement
The authors have no conflicts of interest to declare.
References