Is multiplexed off-axis holography for quantitative phase imaging more spatial bandwidth-efficient than on-axis holography?

Digital holographic microcopy is a thriving imaging modality that attracts considerable research interest due to its ability to not only create excellent label-free contrast, but also supply valuable physical information regarding the density and dimensions of the sample with nanometer-scale axial sensitivity. Three basic holographic recording geometries currently exist, including on-axis, off-axis and slightly off-axis holography, each of them enabling a variety of architectures in terms of bandwidth use and compression capacity. Specifically, off-axis holography and slightly off-axis holography allow spatial hologram multiplexing, enabling compressing more information into the same digital hologram. In this paper, we define an efficiency score used to analyze the various possible architectures, and compare the signal-to-noise ratio and mean squared error obtained using each of them, determining the optimal holographic method.

💡 Research Summary

The manuscript investigates the spatial‑bandwidth efficiency of three principal digital holographic microscopy (DHM) recording geometries—on‑axis, off‑axis, and slightly off‑axis—and introduces a quantitative “efficiency score” (ES) to compare them. In on‑axis DHM the sample and reference beams intersect the sensor at the same angle, so the recorded hologram consists of a central DC term and two complex‑conjugate (CC) cross‑terms that overlap in the spatial‑frequency domain (SFD). Separation of these terms requires multiple phase‑shifted acquisitions, increasing acquisition time and computational load.

Off‑axis DHM introduces a small angular tilt θ between the beams, which imposes a linear phase ramp on the interference pattern. In the Fourier domain this translates the CC terms away from the origin by a distance proportional to θ/λ, while the DC term remains centered. Consequently, each CC occupies a distinct, non‑overlapping region of the SFD. By choosing different tilt angles for several sample‑reference pairs, multiple off‑axis holograms can be recorded simultaneously on a single sensor without spatial overlap—a process termed spatial‑frequency multiplexing or “multiplexed off‑axis holography.”

The authors define the efficiency score as ES = (N · M) / B, where N is the number of holograms multiplexed, M is the number of independent parameters (amplitude, phase, etc.) that can be retrieved per hologram, and B is the occupied spatial bandwidth measured in sensor pixels. For a conventional on‑axis setup N = 1, M ≈ 3 (DC plus two CC terms), and B equals the full sensor area, yielding a low ES. A single off‑axis hologram also has N = 1 but benefits from a modest reduction in B because the CCs are shifted, giving a slightly higher ES. In contrast, multiplexed off‑axis configurations can achieve N = 4, 8, or more, while B remains essentially unchanged, dramatically increasing ES.

To validate the metric, the paper presents extensive numerical simulations using a 2048 × 2048 pixel detector. Four configurations are compared: (i) on‑axis, (ii) single off‑axis, (iii) four‑fold multiplexed off‑axis, and (iv) eight‑fold multiplexed off‑axis. Each case is simulated with additive white Gaussian noise and reconstructed via standard Fourier‑domain filtering. The results show a monotonic improvement in signal‑to‑noise ratio (SNR) and a corresponding reduction in mean‑squared error (MSE): on‑axis SNR ≈ 22 dB, MSE ≈ 1.2 × 10⁻³; single off‑axis SNR ≈ 28 dB, MSE ≈ 5.4 × 10⁻⁴; four‑fold multiplexed SNR ≈ 34 dB, MSE ≈ 1.3 × 10⁻⁴; eight‑fold multiplexed SNR ≈ 38 dB, MSE ≈ 6.2 × 10⁻⁵. The improvement follows the expected √N noise reduction because each CC acts as an independent measurement channel.

Practical considerations are discussed in detail. Excessive tilt angles can push the CCs beyond the Nyquist limit of the sensor, causing aliasing. The reconstruction pipeline for multiplexed data involves multiple Fourier transforms, precise spatial filtering, and phase unwrapping for each CC, which is computationally intensive; the authors suggest GPU acceleration for real‑time operation. Optical aberrations (lens distortion, non‑planar wavefronts) can shift the CC positions away from their theoretical locations, necessitating calibration or adaptive correction. Finally, sensor dynamic range limits may lead to saturation when several CCs simultaneously carry high intensity, reducing effective SNR.

In conclusion, the study demonstrates that multiplexed off‑axis holography achieves a substantially higher spatial‑bandwidth efficiency—up to an order of magnitude greater than conventional on‑axis DHM—while delivering superior quantitative phase reconstruction quality. The efficiency score provides a clear, physics‑based metric for comparing holographic architectures. The authors argue that this approach is especially advantageous for high‑speed, high‑precision applications such as real‑time 3‑D cell imaging, multi‑wavelength spectroscopic phase microscopy, and rapid tomographic phase retrieval. Future work is suggested on automated tilt‑angle optimization, real‑time GPU‑based reconstruction pipelines, and exploiting the multiplexed CCs to retrieve multiple physical quantities (e.g., refractive index, temperature) simultaneously.