3

Cross-Sectional and En Face Visualization of Posterior Eye Circulations

Yali Jia, PhD; Bruno Lumbroso, MD; Simon S. Gao, PhD; David Huang, MD, PhD; and David J. Wilson, MD

Optical coherence tomography angiography (OCTA) has drawn significant clinical interest in the past couple of years. As a result, a number of supporting technologies have been developed to improve the ease of use to allow for rapid interpretation of collected volumetric data. These include methods to segment retinal layers for en face display and to generate multicolor composites of flow information for both cross-sectional and en face visualization. This chapter refers to OCTA performed using the split-spectrum amplitude-decorrelation angiography (SSADA) algorithm on a commercial spectral optical coherence tomography (OCT) system (RTVue XR Avanti, Optovue Inc). But the general principles are also applicable to other types of OCTA.

EN FACE VISUALIZATION OF SEGMENTED TISSUE SLABS

While OCT started as a predominantly cross-sectional imaging modality, OCTA was clinically used as an en face imaging modality from the start. This was enabled by initial work establishing the en face approach.^1–4 OCTA uses previously established techniques for automated segmentation of anatomic reference planes,^5–7 such as the inner limiting membrane (ILM) and outer boundaries of the inner plexiform layer (IPL), outer plexiform layer (OPL), and Bruch’s membrane (BM). Appropriate tissue layers or “slabs” can then be defined based on these reference planes. Accurate segmentation is important for clinical interpretation. In diseased eyes, pathologies such as drusen, intraretinal cysts, edema, or subretinal fluid can make automated segmentation less robust. Although significant improvements have been made,^8–10 expert manual correction is sometimes necessary. Software that aids or reduces the workload required for manual correction of volumetric data is beneficial.^10,11

En face presentation of OCTA helps clinicians recognize vascular patterns associated with various vascular abnormalities. En face angiograms are generated by summarizing the flow information within the depth range encompassed by relevant anatomic layers (slab), typically by taking the maximum or average decorrelation (representing flow) value. This projection process compresses the 3-dimensional (3D) information into several 2-dimensional (2D) images that can be more easily interpreted. En face presentation of these slabs can produce angiograms similar to fluorescein angiography or indocyanine green angiography. Using the segmentation of the ILM, outer boundary of the IPL, outer boundary of the OPL, retinal pigment epithelium (RPE), and BM, the following slabs can be visualized according to the current scheme that subdivides the retinal circulation into 2 plexuses and choroidal circulations into 2 slabs as well. At the end of this chapter, we also present new image processing algorithms that are able to show 3 distinct retinal vascular plexuses.

• Vitreous: normally avascular (above the ILM)
  
• Superficial retinal plexus: superficial portion of the retinal circulation (ILM to outer boundary of the IPL)
  
• Deep retinal plexus: deep portion of the retinal circulation (outer boundary of the IPL to outer boundary of the OPL)
  
• Inner retina: the combination of superficial and deep retinal plexuses (ILM to outer boundary of the OPL) constitute the normal retinal circulation
  
• Outer retina: normally avascular (outer boundary of the OPL to the RPE)
  
• Choriocapillaris: normally near confluent (10 to 20 μm below BM)
  
• Deeper choroid: larger choroidal vessels (more than 20 μm below BM)
  
• Choroid: the combination of choriocapillaris and deeper choroid
  
• Custom: user-defined slab that best highlights the vascular pathology

From the volumetric scan of a healthy eye (Figure 3-1A), the en face OCT angiogram above the ILM shows the normal, avascular vitreous (Figure 3-1B). The inner retina shows both large and small vessels in the superficial plexus (Figure 3-1C) and a fine capillary network in the deep plexus (Figure 3-1D) with no flow in the foveal avascular zone (FAZ). The outer retina should be avascular, but flow projection artifacts from the inner retina (Figure 3-1E) can be seen (Figure 3-1F). The flow projection artifact occurs because blood flow in the retinal vessels cast flickering shadows that cause OCT signal fluctuation in the layers below that is recognized as flow by the OCTA algorithm. Amplitude/magnitude/intensity-based, phase-based, and complex amplitude-based OCTA are all susceptible to similar flow projection artifacts. This shadowgraphic fluctuation is most noticeable in highly reflective layers such as the RPE and form a pattern that replicates the retinal circulation (see Figure 3-1F). The flow projection artifact can be suppressed by post-processing software (Figure 3-1G). The choriocapillaris shows nearly confluent flow (Figure 3-1H). Although artifactual projection of retinal vessels in the choriocapillaris is present, it is not very noticeable because the artifact is diffused by the overlying RPE and weaker than the choriocapillaris circulation. The lobular structures of the choriocapillaris are difficult to recognize because they are very dense at the macula and beyond the transverse spatial resolution of OCT, unless adaptive optics are used. The coarser lobules farther from the macula would be recognizable.¹² The deeper choroidal angiogram should show larger vessels but is more difficult to interpret because of flow projection from the choriocapillaris, shadowing, and fringe-washout artifacts (Figure 3-1I). Fringe washout occurs because high flow velocity (especially the axial component) mixes the phase of the interferometric signal within the integration time of the camera or photodetector in the OCT system.¹³ Together with shadowing, fringe washout can reduce OCT signal intensity below that needed for SSADA processing. Therefore, part of or entire large choroidal vessels can appear dark on en face OCTA. Finally, the volumetric data can be flattened using the RPE/BM segmentation to generate C-scan images (Figure 3-1K).

Stay updated, free articles. Join our Telegram channel

Join