Stage 01 / 05
Sequential B-scan Acquisition
The OCT beam scans the exact same retinal cross-section multiple times (typically 2–8 repetitions).
Each repeated B-scan captures the reflectivity profile of every tissue layer. Notice how blood vessels
are present throughout the retinal layers — their contents will vary between scans.
Repeated B-scans at same retinal position
Vascular lumen (moving erythrocytes)
Highly reflective tissue boundaries
Low-reflectivity layers
Key concept: OCTA requires no dye injection. Unlike fluorescein angiography, it relies purely on
detecting differences between successive scans of the same location. The OCT hardware is identical —
only the acquisition strategy and post-processing algorithm differ.
Stage 02 / 05
Signal Comparison Across Scans
Each pixel is sampled across all 7 repeated B-scans (D1→D7). Static tissue produces a flat,
invariant intensity profile. Moving blood cells cause the OCT signal to oscillate — the larger
the flow velocity, the greater the inter-scan variation.
Static tissue pixel — flat signal
Vessel pixel — oscillating signal
Inter-scan variability is the core measurement. The variance (or decorrelation) between
repeated scans at the same pixel location directly encodes whether tissue is static or contains moving cells.
This is why eye movement artifacts — which shift all pixels simultaneously — must be carefully corrected.
Stage 03 / 05
Decorrelation Mapping
The computed decorrelation value at each pixel — ranging from 0 (identical signals = static) to 1
(maximally different signals = fast flow) — is color-encoded and overlaid on the structural B-scan.
This produces a flow map: vessels appear bright against a dark background of static tissue.
Decorrelation map overlaid on structural B-scan
High decorrelation → flow detected
Decorrelation ≈ 0 → static tissue
Decorrelation algorithms: Common implementations include SSADA (split-spectrum amplitude
decorrelation), optical microangiography (OMAG), and speckle variance OCT. Each trades off sensitivity
vs. specificity for slow capillary flow differently.
Stage 04 / 05
OCTA Volume Construction
Decorrelation maps from hundreds of B-scan positions are stacked to build a three-dimensional
OCTA data volume. Each axial position (A-line) carries both structural (OCT) and flow (OCTA) information
simultaneously — a key advantage over projection-only angiographic modalities.
3D OCTA data volume — rotate to inspect
Volume dimensions: A typical 6×6 mm macular OCTA cube on commercial devices uses
~300 A-lines × 300 B-scan positions × 2 repeated acquisitions per position. The resulting volume
contains depth-resolved vascular information enabling segmentation of individual plexuses.
Stage 05 / 05
En Face Angiogram — Vascular Projection
The OCTA volume is "flattened" by projecting (maximum or mean intensity) along the depth axis within
a defined retinal slab. This reveals the en face view of the vascular network — the superficial capillary
plexus shown here — with the foveal avascular zone (FAZ) clearly visible at center.
Superficial capillary plexus — en face projection
Foveal Avascular Zone (FAZ): The central ~300–500 μm around the fovea is physiologically
devoid of vessels, allowing unobstructed light transmission to the photoreceptors. Its area and
morphology are clinically significant in diabetic retinopathy, CRVO, and other macular diseases.
Slab segmentation: By choosing different depth boundaries, the same OCTA volume
can reveal distinct layers: superficial capillary plexus (SCP), deep capillary plexus (DCP),
outer retina (normally avascular), and choriocapillaris — each with distinct diagnostic value.