Artifacts, Pitfalls & Clinical Decision-Making
Projection artifacts, motion artifacts, segmentation errors. When a scan is unusable. The clinical decision framework for ordering OCTA and evaluating equipment.
Artifacts, Pitfalls & Clinical Decision-Making
A clinician who cannot recognize OCTA artifacts will misdiagnose pathology, over-refer, and order unnecessary follow-up. Artifacts are not edge cases — they appear in a meaningful proportion of every day's OCTA scans. Learning to identify them systematically is the difference between an OCTA that guides you and one that misleads you.
This module covers the major OCTA artifact categories, then transitions to the two decisions that determine whether your practice's OCTA investment succeeds: which patients to scan, and which equipment to choose.
Projection Artifacts and Removal
Projection artifact is the most clinically significant OCTA artifact because it directly mimics pathological flow in the outer retina — the same slab used for CNV detection. When light from superficial vessels (SCP, DCP) passes through the outer retina during acquisition, the motion of those vessels creates false decorrelation signal in the outer retina slab. The result is bright apparent "flow" in the outer retina where no vessels exist.
Identifying projection artifact:
- Spatial correspondence: Projection artifact follows the exact topography of the superficial vascular network. If the outer retina "flow" precisely matches the arcade pattern visible on the SCP slab — it is projection, not CNV.
- Pattern regularity: True CNV networks have irregular, branching neovascular loops. Projection artifact shows the clean, arcuate pattern of the normal superficial vasculature.
- B-scan correlation: True CNV on the outer retina slab has a B-scan correlate — RPE elevation, subretinal hyper-reflective material, or fluid. If the outer retina "flow" has no B-scan correlate, suspect artifact first.
Projection removal algorithms: Most modern OCTA platforms include projection artifact removal software. These algorithms subtract the superficial vascular signal from the outer retina slab and are effective but imperfect — always compare the raw and projection-removed images. Overcorrection can remove real CNV signal; undercorrection can leave false signal. Both the corrected and uncorrected views belong in a complete OCTA report.
| Artifact Type | Characteristic Pattern | How to Distinguish from Pathology |
|---|---|---|
| Projection (SCP to outer retina) | Bright arcade-following pattern in outer retina slab | Matches SCP exactly; no B-scan correlate; resolves with projection removal |
| Projection (DCP to choriocapillaris) | Dense capillary pattern in choriocapillaris slab | Follows DCP capillary distribution; no corresponding flow void |
| Shadow artifact | Dark (signal void) vertical streaks beneath opaque media or structures | Aligned with floater, hard exudate, or nevus located above the affected zone |
Motion Artifacts and Fixation Loss
OCTA acquires multiple B-scans at each position to calculate the decorrelation signal. Any eye movement during this process disrupts the calculation, creating characteristic motion artifacts visible on the en face image.
Blink artifacts: The most common motion artifact. Appears as a bright or dark horizontal stripe spanning the full width of the en face image at the scan position where the blink occurred. The stripe has sharp, clean horizontal edges — distinguishing it from disease-related signal changes. Blink artifacts in the foveal zone require rescan.
Saccade artifacts: Involuntary eye movements during scanning create a lateral shift that appears as a vertical or diagonal discontinuity in vessel continuity. Vessels that should run continuously appear offset at a line crossing the image. Saccade artifacts always require rescan; the degree of vessel displacement makes interpretation of the affected zone unreliable.
Fixation instability: Patients with dense central scotoma, nystagmus, or advanced AMD often cannot hold fixation reliably during the scan acquisition window. The resulting image has decreased vessel sharpness and diffuse apparent dropout that reflects poor image quality, not pathological non-perfusion. This is a critical distinction: do not diagnose capillary non-perfusion from a poorly fixated scan.
| Artifact Type | En Face Appearance | Action Required |
|---|---|---|
| Blink stripe | Horizontal bright or dark stripe; full image width; sharp edges | Rescan if stripe crosses fovea or optic nerve head zone |
| Saccade offset | Vessel discontinuity; diagonal shift line crossing image | Rescan required; use fixation assist device if available |
| Fixation instability | Global motion blur; diffuse vessel blurring; apparent patchy dropout | Improve fixation; use smaller scan area; shorter acquisition time |
| Microsaccades | Fine vessel zigzag pattern; slight lateral displacement of small vessels | Rescan with faster acquisition if critical zone affected; accept if peripheral only |
Segmentation Errors and Manual Correction
OCTA segmentation relies on the same automated boundary detection used for structural OCT. When that segmentation fails — which happens regularly in disease states — the wrong retinal depth is displayed on each slab. The pathology appears to be in the wrong layer, or disappears entirely.
Most common segmentation failures in OCTA context:
- ILM error at ERM: The ILM dips into the retina where ERM traction distorts it. All downstream slabs shift anteriorly — CNV-level flow may appear in the GCL slab instead of the outer retina slab. Always cross-reference with B-scan boundary overlays before interpreting.
- RPE mistrack at CNV: A Type 1 CNV elevates the RPE. If the algorithm tracks the apex of the elevation rather than the true RPE base, the outer retina slab boundaries shift, and the CNV network may be incorrectly assigned to the choriocapillaris layer.
- RPE mistrack at GA: Geographic atrophy causes RPE loss. The algorithm may skip to the choroidal-scleral interface, dramatically underestimating GA area and incorrectly placing choroidal flow in the outer retina position.
- High myopia segmentation error: In eyes with posterior staphyloma, the curved posterior pole confuses flat-plane algorithms, creating apparent outer retina flow artifacts in the staphyloma periphery.
When to perform manual correction: Manual correction is appropriate and expected when segmentation error produces clinically significant misinterpretation. The threshold: if the boundary error changes your clinical conclusion, correct it and re-export. Always document correction in the report ("Outer retina slab: manual RPE boundary correction applied at CNV site").
Signal Strength and Scan Quality Standards
Signal strength index (SSI) summarizes the overall light penetration and scan quality on a 0–10 scale. For structural OCT, a signal strength of 6 or above is generally acceptable. For OCTA, the threshold is higher — reliable flow quantification requires signal strength of 7 or above.
Why OCTA needs higher signal strength: The decorrelation calculation requires multiple high-quality B-scans at each position. Noise from weak signal creates false decorrelation — apparent non-perfusion appears where there is actually normal flow. Below SS 5, the noise floor exceeds the signal from small capillaries, making capillary-level OCTA analysis completely unreliable.
Pre-interpretation scan quality checklist:
- Signal strength ≥7 for OCTA interpretation (≥6 acceptable for structural-only analysis)
- No blink stripes crossing the macula or optic nerve head zone
- No saccade discontinuities through critical vessels
- Segmentation boundary overlay confirmed — check ILM and RPE tracking at the area of interest
- Projection removal confirmed — verify algorithm was applied before interpreting outer retina or choriocapillaris findings
When to Order OCTA: Clinical Decision Framework
OCTA adds the most clinical value in specific scenarios. Ordering OCTA for every patient regardless of indication generates artifact-heavy scans from poor cooperators, consumes chair time, and dilutes the technology's clinical signal. Use OCTA where it changes management.
| Clinical Scenario | What OCTA Adds | Changes Management? |
|---|---|---|
| AMD monitoring (dry, intermediate, or treated nAMD) | CNV detection, CNV activity quantification, GA choriocapillaris mapping | Yes — injection timing, referral threshold, monitoring interval |
| Unexplained visual loss with normal B-scan | Detects occult CNV, ischemic maculopathy, FAZ abnormality invisible on structural OCT | Yes — diagnosis revealed or excluded non-invasively |
| Moderate to severe NPDR monitoring | FAZ metrics, NPA quantification, early NVE detection | Yes — refines recall interval, earlier referral trigger for proliferative conversion |
| RVO follow-up | Perfusion trend, collateral formation, ischemic conversion monitoring | Yes — can reduce number of FA monitoring visits required |
| Glaucoma suspect with structural-functional discordance | Vessel density may detect early damage when RNFL and VF are equivocal | Adjunctive — supports decision, does not replace primary structural testing |
| Routine mild NPDR without DME | Limited incremental value over clinical exam and structural OCT | Marginal — use judiciously based on practice volume |
| Dense vitreous hemorrhage or significant media opacity | Signal quality too low for reliable interpretation | No — defer until media clarity improves |
Equipment Considerations: The $100,000 Decision
OCTA is available as an upgrade to most current-generation OCT platforms for approximately $100,000, on top of the baseline OCT investment. The practices that get the most value from OCTA are those that chose equipment matching their patient population and workflow — not those that bought based on a sales presentation.
| Factor | Why It Matters | Questions to Ask the Vendor |
|---|---|---|
| Acquisition speed (A-scans/sec) | Faster acquisition means fewer motion artifacts — critical for elderly patients and those with poor fixation | What A-scan rate? What is the acquisition time for a 6×6mm OCTA cube? |
| Widefield capability | 12×12mm OCTA extends NVE detection for DR and RVO perfusion mapping to the periphery | What scan sizes are offered? Is widefield OCTA available as a standard feature? |
| Projection artifact removal quality | Outer retina interpretation requires effective removal — quality varies significantly between platforms | Ask the vendor to demonstrate outer retina slab with and without removal on the same patient scan |
| Automated quantification | FAZ metrics, vessel density maps, NPA quantification — these are the data your management protocols will use visit after visit | Which metrics are generated automatically? How are they displayed and tracked longitudinally? |
| Software upgrade path | OCTA analysis software is evolving rapidly — today's platform will have significantly better tools in 2–3 years | What is the software update policy? Are new analysis features included in maintenance or add-on cost? |
| Staff training burden | More complex platforms have higher training burden — a real factor for practices with technician turnover | What is typical time to technician proficiency? What training support is included with purchase? |
| Workflow integration | Same-manufacturer OCTA minimizes retraining if your practice already has that brand's OCT | Does it integrate with your existing EHR and imaging archive without a separate workstation? |
The Maestro2 context for this decision: The Maestro2's automated acquisition — the system aligns and captures with minimal technician input — makes it uniquely suitable for practices where technician time is constrained or where backup staff need to operate the instrument reliably. Simultaneous color fundus photography at every visit eliminates a separate fundus camera acquisition step. That workflow efficiency has real dollar value in a busy practice over thousands of annual patient visits.
Trade-in programs and financing options from OCT vendors can significantly reduce the out-of-pocket investment for practices with older OCT units. The acquisition math looks very different when you are crediting a current instrument versus buying from zero investment.
- Patient volume math: At current Medicare rates (~$56.93 per CPT 92137), approximately 1,757 billable OCTA studies recover a $100,000 equipment cost. For most practices with an AMD and DR patient population, systematic OCTA use across 3–4 years makes this achievable — evaluate your patient volume and payer mix carefully before committing.
- Clinical competence first: You have completed all three tiers of SlabED. You now understand what OCTA shows, what it can miss, and how to use it to change management decisions. That competence — not the equipment — is what creates sustainable clinical and financial value.
- Next step: Model your specific patient volume and payor mix against the Maestro2 upgrade investment. The ROI analysis is the conversation that closes the decision with confidence.
You have now completed all three tiers of SlabED. You understand the technology, the vascular anatomy, the FAZ, CNV detection, disease-specific management protocols, artifact recognition, clinical decision frameworks, and equipment evaluation criteria. The optometrists building OCTA competency now are positioning their practices for the next decade of retinal care.
- OCTA technology and vascular layer anatomy ✓
- FAZ analysis and clinical metrics ✓
- CNV types, treatment response, and OCTA vs FA ✓
- Disease-specific OCTA management protocols ✓
- Artifact recognition and systematic scan quality review ✓
- Clinical decision framework and equipment evaluation criteria ✓
Sign up to track your progress and access all lessons.
Create Free Account