🧬 biology

Depth-multiplexing spectral domain OCT for full eye length imaging with a single modulation unit

This paper presents a cost-effective, single-modulation-unit depth-multiplexing spectral domain OCT system that enables simultaneous full eye length imaging and myopia monitoring by computationally demixing signals from multiple depths, offering a viable alternative to expensive swept-source OCT.

Original authors: Guanghan Meng, Xue Dong, Andrew Zhang, Fabio Feroldi, Austin Roorda, Laura Waller

Published 2026-02-17

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Guanghan Meng, Xue Dong, Andrew Zhang, Fabio Feroldi, Austin Roorda, Laura Waller

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Imagine your eye is a long, dark tunnel. To understand if someone is getting nearsighted (myopia), doctors need to measure the exact length of this tunnel from the front (cornea) to the back (retina).

Currently, the "gold standard" tool for this is a high-tech machine called Swept-Source OCT. Think of this like a Ferrari: it's incredibly fast and accurate, but it costs a fortune, so only big hospitals can afford it.

There is a cheaper alternative called Spectral Domain (SD) OCT. This is more like a reliable family sedan. It's affordable and great for most jobs, but it has a major limitation: it can only see a short distance into the tunnel. If the eye is too long (which is common in myopia), the camera can't see the back wall.

The Problem: The "Short-Sighted" Camera

To see the whole eye with the cheap camera, previous scientists tried to build a "super-camera" by attaching three different lenses or three different light modulators to the machine.

The Analogy: Imagine trying to take a photo of a whole mountain range. Instead of one camera, you tape three cameras together, each pointing at a different part of the mountain.
The Downside: This makes the machine bulky, complicated, and expensive again. It defeats the purpose of using the cheap camera in the first place.

The Solution: The "Magic Prism" Trick

The researchers at UC Berkeley came up with a clever, low-cost solution. They figured out how to use one single camera and one single moving mirror to see the front, middle, and back of the eye all at once.

Here is how they did it, using a simple analogy:

1. The "Color-Coded" Echo

Imagine you are in a canyon shouting "Hello!"

Normally, the echo comes back at the same pitch.
In this new system, the researchers use a special mirror that wiggles slightly as it scans. This wiggle changes the "pitch" (frequency) of the light returning from different parts of the eye.
The Analogy: It's like shouting "Hello" to the front of the canyon, "Hello" to the middle, and "Hello" to the back, but giving each shout a slightly different musical note.
- The Cornea (front) sings a Low Note.
- The Pupil/Lens (middle) sings a Medium Note.
- The Retina (back) sings a High Note.

2. The "Digital DJ" (Computational Demixing)

When the light bounces back, the camera hears a jumbled mess of all three notes at once. It looks like static noise.

The Analogy: Imagine a DJ mixing three songs together. To the listener, it's just a loud, confusing noise. But if you have a special software (the "DJ"), you can isolate the bass, the vocals, and the drums.
The researchers use a computer algorithm to act as this DJ. It separates the "Low Note" (front of the eye) from the "High Note" (back of the eye). Suddenly, the jumbled noise turns into three clear, separate images stacked on top of each other.

Why This Matters

One Tool, Three Views: They managed to see the entire length of the eye using just one cheap camera and one moving mirror, instead of three expensive setups.
No Light Loss: Because they split the light in the reference arm (the "mirror" side) rather than the eye side, the light hitting the patient's eye is bright and clear, ensuring a high-quality image.
Real-World Test: They tested this on a real human with myopia. The machine successfully measured the distance from the cornea to the retina, matching the results of the expensive "Ferrari" machines.

The Bottom Line

This paper is about taking a cheap, limited tool and upgrading it with a clever software trick and a simple mirror adjustment. It turns a "short-sighted" camera into a "long-distance" viewer.

Why should you care?
If this technology becomes standard, clinics in small towns or lower-income communities could finally afford high-quality eye length measurements. This means doctors could track myopia progression earlier and cheaper, potentially saving children from getting worse nearsightedness without needing a multi-million dollar machine.

1. Problem Statement

Clinical Need: Axial eye length measurement is critical for tracking myopia progression and early detection of ocular changes.
Current Limitations:
- Swept-Source OCT (SS-OCT): The current gold standard for full-eye imaging but is prohibitively expensive due to high-cost light sources and high-bandwidth data acquisition systems, limiting accessibility in lower-income communities.
- Spectral Domain OCT (SD-OCT): A more affordable alternative, but it suffers from a limited imaging depth range (typically insufficient to capture the entire eye length from cornea to retina in a single scan).
- Existing Depth-Multiplexing (DM) Solutions: Previous methods to extend SD-OCT depth range typically require multiple light modulation units, interferometers, or detectors to image different depths simultaneously. This increases system complexity, bulk, and cost, negating the affordability advantage of SD-OCT.

2. Methodology

The authors propose a novel Depth-Multiplexing Spectral Domain OCT (DM-SD-OCT) system that achieves full-eye imaging using a single interferometer and a single light modulation unit.

Core Principle: The system utilizes a computational imaging approach where distinct phase ramps are encoded onto reference beams targeting different depths (cornea, pupil, retina).
Hardware Setup:
- Reference Arm: Three reference arms are directed onto a single galvanometer scanner (reference scanner) at different off-center positions.
- Modulation: As the galvanometer scans, the off-center incidence of the beams creates a linear Optical Path Length (OPL) change, resulting in distinct phase ramps along the transverse axis of the B-scan for each reference arm.
- Optical Components: The system uses a low-cost line scan camera (Teledyne Dalsa Spyder3 1k), a holographic grating, and off-the-shelf optics. A lenslet array (1.0mm x 1.4mm pitch) reflects light with different OPLs back to the correct path.
- Beam Configuration: Three beams are separated by 1.0mm on the scanner ( $s_1=0.5$ mm, $s_2=1.5$ mm, $s_3=2.5$ mm) to target the retina, lens, and cornea respectively.
Signal Processing (Demixing):
- The distinct phase ramps shift the spectral centers of the signals from different depths into separate frequency bands within the Fourier domain of the B-scan.
- A Hilbert Transform (HT) is applied to the Fourier-domain signal along the lateral-scanning direction to recover amplitude and phase components.
- Frequency Demixing: By cropping specific spectral regions (e.g., $\pi/6$ , $\pi/2$ , $5\pi/6$ ) corresponding to the phase shifts of each arm, the contributions from the three depths are computationally separated and reconstructed into distinct images.

3. Key Contributions

Single-Unit Modulation: Unlike previous DM-SD-OCT methods requiring multiple modulators or detectors, this approach uses a single galvanometer scanner to encode depth information for multiple arms simultaneously.
Three-Depth Imaging: The system successfully demonstrates simultaneous acquisition and demixing of signals from three distinct depths (cornea, pupil, retina), an improvement over previous methods that typically handled only two depths.
Cost-Effectiveness: By eliminating the need for multiple expensive detectors or complex multi-interferometer setups, the method offers a low-cost solution compatible with standard SD-OCT hardware.
Power Efficiency: Light division occurs in the reference arm, ensuring that the signal returning from the sample arm does not suffer from light loss, maintaining high signal-to-noise ratios (SNR).

4. Results

System Characterization:
- Imaging Range: Achieved a 1.4 mm imaging range within a 6dB signal fall-off (spectral resolution 0.11 nm).
- Depth Targets: Successfully centered reference arms at 0 mm (cornea), 4.531 mm (lens/pupil), and 36.512 mm (retina).
- Performance: Achieved a peak Signal-to-Noise Ratio (SNR) > 90 dB at all three depths with a 300 kHz A-scan rate. Axial resolution was measured at 14.4 $\mu$ m.
In Vivo Validation:
- Tested on a myopic adult subject.
- Successfully reconstructed images of the cornea, pupil, and retina simultaneously.
- Accuracy: Measurements of eye structures (axial length, anterior chamber depth, etc.) showed strong agreement with the clinical gold standard, the Zeiss IOL Master 700.
Phase Ramp Validation: The system generated expected phase ramps of 0.542, 1.63, and 2.71 rad/line for the three arms, shifting spectral centers to $\pi/6$ , $\pi/2$ , and $5\pi/6$ respectively, enabling clean separation.

5. Significance

Clinical Impact: This technology provides a cost-effective, portable, and accessible alternative to SS-OCT for comprehensive myopia management. It enables full-eye length measurements in community healthcare centers and lower-income settings where high-end SS-OCT is unavailable.
Scalability: The method is applicable to any existing SD-OCT device where extended depth is required, without necessitating a complete hardware overhaul.
Future Potential: The authors suggest that the current free-space optics setup can be miniaturized using fiber-based modules. Furthermore, the distinct sparsity of eye structures suggests that deep learning algorithms could eventually replace the frequency demixing process, potentially simplifying the workflow further.

In conclusion, this work bridges the gap between the affordability of SD-OCT and the full-depth capabilities of SS-OCT, offering a practical solution for global myopia monitoring.