OCT is "optical ultrasound": a low-coherence near-infrared source is split into a reference beam (bounced off a movable mirror) and a sample beam (sent into the tissue). Because the source's coherence length is tiny, the two beams only interfere when their round-trip path lengths match almost exactly — so as the reference mirror sweeps, the detector lights up only when it is scanning a real reflective interface inside the tissue.
I(z) ∝ Σ Rᵢ · exp( −(z − dᵢ)² / 2ℓc² )
z = reference-arm depth, dᵢ = true boundary depth
ℓc = coherence length (interference gate width)
- Reference-arm depth — mirror position; scanning it traces one A-scan (depth profile) at a single lateral point.
- Auto A-scan sweep — sweeps the mirror through the whole sample so every layer interface produces its own interference peak.
- Build cross-section — repeats the A-scan at many lateral positions and stitches the results into a 2D B-scan image, revealing the tissue's layered structure.
- Surface photo — a normal camera only catches light reflected from the outer surface; it cannot see anything below it, however many layers are underneath.
Axial resolution. The gate is Gaussian with standard deviation ℓc, so two interfaces are only separated if they lie farther apart than about the FWHM = 2√(2 ln 2)·ℓc ≈ 2.355 ℓc. Widen the coherence slider (a narrower-bandwidth source) and the thin layers merge into one blob; narrow it and each boundary appears as its own line. The "Layers resolved" readout is computed from the actual simulated A-scan (a dip of more than 15 % between neighbours), and the FWHM readout is cross-checked by scanning the gate numerically.
Honest simplifications. The depth axis is a scaled schematic: a real retina is only about 0.3 mm thick and real OCT axial resolution is 3-7 µm, so the numbers here (millimetres) keep the ratios, not the absolute size. Real systems also have attenuation with depth, speckle noise, refractive-index factors and (in modern Fourier-domain OCT) no moving mirror at all; this model shows the classic time-domain idea.
Real-world relevance: this exact technique is how ophthalmologists image the living retina non-invasively, resolving individual cell layers a few microns thick without ever touching the eye.