Widefield optical coherence tomography by electro-optical modulation

Abstract

Optical coherence tomography (OCT) is a unique imaging modality capable of axial sectioning with a resolution of only a few microns. Its ability to image with high resolution deep within tissue makes it ideal for material inspection, dentistry, and, in particular, ophthalmology. Widefield retinal imaging has garnered increasing clinical interest for the detection of numerous retinal diseases. However, real-time applications in clinical practice demand the contrast of swept-source OCT at scan speeds that limit their depth range. The curvature of typical samples, such as teeth, corneas, or retinas, thus restricts the field-of-view of fast OCT systems. Novel high-speed swept sources are expected to further improve the scan rate; however, not without exacerbating the already severe trade-off in depth range. Here, we show how, without the need for mechanical repositioning, harmonic images can be rapidly synthesized at any depth. This is achieved by opto-electronic modulation of a single-frequency swept source laser in tandem with tailored numerical dispersion compensation. We demonstrate experimentally how real-time imaging of highly-curved samples is enabled by extending the effective depth-range 8-fold. Even at the scan speed of a 400 kHz swept source, harmonic OCT enables widefield retinal imaging.

Fig. 1.

Harmonic Optical Coherence Tomography. a. Harmonic OCT system layout: BS1, 80:20 beamsplitter; BS2, 50:50 beamsplitter; FL, focusing lens; GS, galvo-scanner; EOM, electro-optical modulator; BPD, balanced photo-detector; ADC, analog-to-digital converter. b. Conventional OCT imaging, notice we cannot image the entire retina due to the curvature of the surface. c. After de-tuning the reference arm length to move the real image out of the accessible region, the optical phase of the reference arm is rapidly modulated, this leads to generation of sideband harmonics above and below the frequency of the real image, only the low frequency sideband is imaged, whilst other images are filtered out. d. The phase modulation frequency is varied as a function of lateral position, we can use this principle to cancel out the curvature of the retina, and therefore keep the entire surface in frame during acquisition. e. The curvature of the image is reconstructed post-acquisition, the entire retina can be recovered this way, alternatively, we can say that the imaging depth limit has been adjusted to conform to the curvature of the retina.

Fig. 2.

Detailed system diagram for harmonic OCT using electro-optical modulation. The reference arm contains the electro-optical modulator (EOM) for fast depth adjustments, as well as a slow mechanical delay stage for coarse adjustments of the optical path difference prior to the experiment. a. Sample arm configuration for imaging of the model eye. b. Sample arm configuration for imaging of the grape. An objective lens substitutes the ocular lens. LPF: low-pass filter, BD: balanced detector, FC: fiber coupler, VCO: voltage-controlled oscillator, AMP: amplifier, IO: input/output.

Fig. 3.

Variability of the sweep-rate. a. The angular acceleration, i.e., the rate of change of angular frequency with time, used for simulations is on the order of $1.2\times {10}^{20}{\text{rad/s}}^2\approx 20\text{EHz/s}$ . This was estimated by recording the k-clock signal from the swept source. b. Simulated A-scans showing a single reflector. The broadening of the A-line is a result of the non-linear frequency sweep, despite linear-in-k sampling.

Fig. 4.

Numerical dispersion compensation by Pseudo Wigner-Ville Distribution (PWVD) analysis. a. PWVD of simulated interferogram sampled in time, and corresponding A-scan. Note the phase error is depth-dependent. b. PWVD of simulated interferogram sampled in wavenumber and corresponding A-scan. With the correct sampling, the phase error is no longer depth-dependent. c. Cross-correlation of PWVD, showing translation of the A-scan along depth. Both the theoretical and numerically-extracted ridge are shown. d. PWVD of interferogram after numerical compensation using extracted ridge, and corresponding A-scan. Note the apparent sample distance is no longer dependent on wavenumber.

Fig. 5.

Experimental demonstration of effectiveness of numerical dispersion compensation for Harmonic OCT. a. Pseudo Wigner-Ville distribution of an A-line, we can see the position of the reflection move with the sweep, indicating a phase error in the interferogram. The red dashed line denotes the band-pass filtered A-scan used for cross-correlation. Each column was normalised individually for visual clarity. b. Cross-correlation of selected band-pass filtered A-scan with the full Wigner-Ville spectrum, the apparent displacement $\mathrm{\Delta }z$ of the sample is shown with a red dashed-line. c. B-scan of harmonic image prior to dispersion compensation, the blurring is primarily due to the non-linear laser sweep. d. B-scan after dispersion compensation has been applied. Scale bars are $1\text{mm}$ .

Fig. 6.

Assessment of long axial range shifting capability of the opto-electronic reference arm. a. B-scans taken at increasing values of $z_s-z_r$ show how high image contrast is maintained beyond the centimeter range. It decreases gradually until it is lost at around $40\text{mm}$ . b. The SNR as a function of the path difference $z_s-z_r$ . The first point corresponds to the conventional OCT image. Up to $15\text{mm}$ , the signal-to-noise relative to the conventional image remains above the -3 dB threshold. Scale bars are $1\text{mm}$ .

Fig. 7.

Experimental demonstration of the flattening of a convex surface (a grape). a. Conventional OCT image. b. High-contrast, flattened, harmonic image, acquired by electro-optically tracking the surface and compensating numerically for phase errors. Scale bars are $1\text{mm}$ .

Fig. 8.

Experimental demonstration of electro-optical-digital flattening of the concave retina. a. The Harmonic image without fast depth-range extension. The field-of-view width without overlap is restricted to about $3\text{mm}$ near its center. Overlap with the complex conjugate image prevents imaging beyond a depth of approximately $1\text{mm}$ , yet the curved retina spans about $12.5\text{mm}\times 4\text{mm}$ . b. By tuning the frequency offset, the A-scan position is adjusted to cancel out the curvature of the sample and disentangle the desired image from the conjugate artifact. c. Harmonic image following numerical curvature reconstruction. Although the complex conjugate image can be computationally removed prior to reconstruction, it is purposely left in to show that it does not overlap with the true image. The conjugate ambiguity has been resolved without resorting to full-range techniques.