Metasurface-assisted adaptive quantum phase contrast imaging

Xiaojing Feng; Juanzi He; Xingyu Liu; Xiaoshu Zhu; Yifan Zhou; Xinyang Feng; Shuming Wang

doi:10.3788/AI.2025.10014

1. Introduction

Conventional photodetectors measure only the light intensity ( $I \propto {| E |}^{2}$ ), thereby failing to capture the phase component ( $φ$ ) of the complex optical field ( $E = | E | e^{i φ}$ )^[1]. This limitation is especially pronounced in biological microscopy. Conventional bright-field microscopy relies on optical absorption differences within samples to generate imaging contrast. However, most biological cells exhibit weak light absorption, making it challenging to effectively characterize subcellular structures based on intensity variations alone. While fluorescence labeling and staining techniques improve imaging contrast, they inevitably induce phototoxicity, significantly perturbing native cellular physiology^[2–6]. To overcome this limitation, Zernike’s phase contrast microscopy technique^[7], proposed in 1942, ingeniously converts phase retardation ( $Δ φ$ ) into detectable intensity. The core principle is exploiting optical interference effects by introducing a phase shift between scattered and transmitted light from the sample, thereby transforming phase variations into measurable intensity changes. This innovation enabled label-free, high-contrast imaging of transparent biological specimens. While subsequent developments—including differential interference contrast (DIC) microscopy^[8,9] and spiral phase contrast microscopy^[10,11]—have improved phase sensitivity, these systems face challenges in miniaturization due to their reliance on bulky components (e.g., spatial light modulators and prisms), limiting their applications in portable or in vivo imaging^[12–14].

Recently, optical metasurfaces have attracted considerable research interest owing to their unique capabilities for miniaturization and system integration^[15–22]. As engineered two-dimensional platforms composed of subwavelength nanostructure arrays, metasurfaces enable precise and simultaneous control over multiple electromagnetic wave parameters through carefully designed meta-atom geometries, achieving highly versatile wavefront manipulation^[23]. Particularly noteworthy is the emerging integration of metasurfaces with microscopic imaging techniques, which has opened new avenues for developing compact, multifunctional phase imaging systems^[24–30]. Current metasurface-enabled quantitative phase imaging approaches, including implementations based on the transport of intensity equation (TIE)^[31,32], phase-shifting interferometry^[4,24,25], and digital holographic microscopy (DHM)^[33,34], demonstrate excellent capability for high-precision quantitative phase reconstruction and numerical contrast enhancement in transparent specimens. While these techniques excel at precise phase recovery for quantitative analysis, they are fundamentally optimized for numerical reconstruction rather than direct structural visualization of complex phase distributions. Their reliance on either computationally intensive post-processing algorithms or multi-frame acquisition protocols not only introduces fundamental limitations in temporal resolution and processing efficiency but also creates a technical gap in achieving high-contrast visualization of sample morphology. Although multimodal microscopy techniques^[30,35] show potential for enhanced imaging versatility, they suffer from limited resolution, low signal-to-noise ratios (SNRs), and photodamage risks to photosensitive specimens under high-intensity illumination. Furthermore, mechanical switching between imaging modes may lead to optical misalignment and reduced system stability, particularly in compact, enclosed, or sensitive environments. In such cases, non-contact, remotely controlled modulation offers a more stable and practical alternative.

To overcome these technical bottlenecks, the synergistic integration of quantum light sources with metasurfaces offers a solution for multimodal imaging^[36]. By harnessing nonclassical photon correlations to surpass the shot-noise limit, this approach enables high-contrast imaging under ultralow illumination intensities while qualitatively analyzing phase information^[37]. Although recent studies have demonstrated that quantum metasurface systems can achieve label-free imaging and multimodal switching^[38,39], a critical limitation persists: while integrating multiple imaging modalities, their phase contrast mode only supports single-contrast operation, with imaging performance heavily dependent on the match between sample phase characteristics and preset modulation parameters. Consequently, these systems exhibit high-contrast performance only for specific phase specimens and fail to accurately visualize the intricate phase distributions within complex biological samples^[39]. However, the imaging contrast is fundamentally governed by a dual-phase modulation mechanism: on the one hand, identical phase distributions exhibit markedly different contrast characteristics when subjected to distinct auxiliary phase modulations, and on the other hand, applying uniform phase modulation to different sample structures can generate substantially divergent imaging responses. This dual dependency restricts conventional systems employing single-phase modulation to fixed bright-field phase contrast modes, which are optimized only for specimens with specific phase profiles. When faced with the inherent phase heterogeneity of biological samples, such constrained modulation schemes fail to provide consistent, high-contrast imaging performance across the entire specimens^[27,40]. This critical technological limitation severely impedes the widespread biomedical application of phase contrast imaging applications. In our previous work, we achieved edge detection of pure phase objects (POs) with flexible switching among bright-field phase contrast, edge enhancement, and their superposition in POs^[41]. However, for transparent samples with arbitrary phase combinations, phase contrast imaging may provide limited resolution of the internal phase structures, highlighting a critical gap in current imaging capabilities.

Here, we present the first quantum-enabled bright–dark phase contrast imaging system supporting remote bright–dark phase imaging switching. This platform enables adaptive, high-contrast imaging of label-free transparent specimens with arbitrary phase compositions under ultralow photon flux. By leveraging the temporal correlations of entangled photon pairs, the system employs heralded imaging to suppress background noise, achieving a fourfold increase in contrast (from 0.22 to 0.81) compared to classical weak light imaging under identical illumination. This quantum illumination approach preserves spatial resolution and minimizes photodamage, offering a promising solution for long-term imaging of photosensitive live specimens. Crucially, in contrast to conventional systems that depend on mechanical switching components—which can cause optical misalignment or sample disturbance—our approach employs polarization entanglement to dynamically switch imaging modes without any physical adjustment. This ensures high-contrast performance in at least one imaging mode while enhancing system stability and operational flexibility. With the growing demand for non-contact control in constrained or sensitive environments—such as implantable microscopy^[42,43], live-brain imaging^[44], and space-limited diagnostic systems—our remotely switchable design offers a distinct advantage. Overall, this technology addresses the longstanding trade-off between contrast and adaptability in phase imaging and significantly expands the potential of quantum-enhanced microscopy in life sciences and clinical diagnostics.

2. Principle of Quantum Bright–Dark Phase Contrast Imaging

We first briefly review the basic principles of bright–dark phase contrast imaging based on classical continuous-wave (CW) collimated illumination. Subsequently, the method for remotely switching the image modes of the system following the utilization of the polarization multiplexed metasurface and quantum polarization entanglement source is delineated.

2.1. Bright–Dark Phase Contrast Imaging with a Metasurface

Phase contrast microscopy enables the visualization of transparent samples by converting subtle phase shifts induced by sample structures into measurable intensity variations. The technique exploits optical interference phenomena—when coherent illumination passes through a PO, the resulting scattered and unscattered wavefronts acquire different phase delays. By introducing a controlled phase shift between these wavefronts using specialized optical components, their subsequent interference produces a detectable amplitude contrast. This phase-to-amplitude transformation enables visualization of structural details that conventional bright-field microscopy cannot detect. Critically, the contrast can be optimized by precisely tuning the phase offset, enabling clear imaging of weakly scattering biological specimens without staining or labeling.

The key to bright–dark phase contrast imaging is the introduction of phase modulation on the Fourier plane. Based on this, we place the metasurface on the Fourier plane of the pure phase samples, and the mode switching of bright–dark contrast imaging can be realized by using the polarization-multiplexing metasurface to introduce different phases to the incident light under different polarizations. The incident light field is $E_{0}$ , and the light field becomes $E_{0} e^{i φ (x, y)}$ after passing through a purely phase sample with phase distribution $φ (x, y)$ . Placing the metasurface on the Fourier plane modulates the light field in different regions, i.e., modulates different parts of the frequency space information. Assuming that the part of the incident light field modulated by the metasurface on the Fourier plane is $E_{m}$ , the phase difference introduced is $ϕ_{m}$ , the unmodulated part accumulating phase on the Fourier plane is $ϕ_{n}$ , and the phase difference between the two parts is $Δ ϕ = ϕ_{m} - ϕ_{n}$ . Then, the phase samples are imaged by the $4 f$ system, and the light intensity distribution on the surface can be expressed as $I = {| (E_{0} e^{i φ (x, y)} - E_{m}) e^{i ϕ_{n}} + E_{m} e^{i ϕ_{m}} |}^{2} = {| (E_{0} e^{i φ (x, y)} - E_{m}) + E_{m} e^{i Δ ϕ} |}^{2} = {| E_{0} |}^{2} + 4 \sin^{2} \frac{Δ ϕ}{2} {| E_{m} |}^{2} + (e^{- i Δ ϕ} - 1) E_{0} e^{i φ (x, y)} E_{m}^{*} + (e^{i Δ ϕ} - 1) E_{0} e^{- i φ (x, y)} E_{m} .$ (1)

If we make $E_{0} e^{i φ (x, y)} E_{m}^{*} = | E_{0} | | E_{m} | e^{i α}$ , where $α = φ (x, y)$ is the function of $x$ and $y$ , then the light intensity distribution is expressed as $I = {| E_{0} |}^{2} + 4 \sin^{2} \frac{Δ ϕ}{2} {| E_{m} |}^{2} + (e^{- i Δ ϕ} - 1) | E_{0} | | E_{m} | e^{i α} + (e^{i Δ ϕ} - 1) | E_{0} | | E_{m} | e^{- i α} = {| E_{0} |}^{2} + 4 \sin^{2} \frac{Δ ϕ}{2} {| E_{m} |}^{2} + 2 | E_{0} | | E_{m} | [\cos (Δ ϕ - 1) \cos α + \sin Δ ϕ \cos α] .$ (2)

Equation (2) reveals that the phase information at different positions on the phase sample is also a key factor contributing to intensity fluctuations in imaging. The imaging intensity is modulated by both the phase difference ( $Δ ϕ$ ) introduced by the metasurface and the phase distribution [ $α = φ (x, y)$ ] of the phase sample. When the phase difference introduced by the metasurface is fixed, different phase values on the sample lead to varying imaging outcomes. If the metasurface can only introduce a single-phase difference, it cannot ensure high-contrast imaging across all phase patterns. To overcome this limitation, we designed a polarization-multiplexed metasurface that brings distinct phase differences ( $Δ ϕ$ ) to different polarization states. By switching the incident light polarization from V to H, the phase difference is shifted by $π$ , causing the original bright spot to darken. This enables dynamic switching between bright-field and dark-field imaging modes, thereby achieving phase contrast imaging mode switching.

2.2. Principle of Quantum Remote Switching

We have successfully realized a remotely switchable quantum bright–dark phase contrast imaging scheme by utilizing polarization-entangled photon pairs. As shown in Fig. 1(a), the imaging system adopts a dual-optical-path design, including two functional modules: the imaging arm and the heralding arm. Among them, the polarization-entangled photons in the imaging arm are precisely modulated for sample illumination, while the herald arm realizes the remote switching control of the imaging mode through quantum state projection measurement. When the incident light from the imaging arm is $| H ⟩$ , the metasurface introduces an additional phase shift of $π / 2$ . Conversely, when the incident light is $| V ⟩$ , the additional phase shift is $3 π / 2$ . The light field modulated by the metasurface interferes with the unmodulated background light, forming a bright–dark difference phase contrast image, as shown in Fig. 1(b). In this study, we use a polarization-entangled state represented as $1 / \sqrt{2} (| H V ⟩ + | V H ⟩)$ as the illumination source, and when the input photon state in the heralding arm is set to $| V ⟩$ or $| H ⟩$ , the input photon state in the signal arm will also be $| H ⟩$ or $| V ⟩$ . This allows the heralding arm to act as a trigger for remote switching that determines the bright or dark phase imaging mode of the signal arm.

Figure 1.Schematic representation of a switchable quantum bright–dark phase contrast imaging system. (a) The metasurface enables bright–dark phase contrast imaging. The quantum state of a polarization-entangled photon pair is $1 / \sqrt{2} (| H V ⟩ + | V H ⟩)$ . Signal photons pass through the PO and are modulated by the metasurface (at the Fourier plane) for imaging. The imaging arm outputs a quantum superposition of bright–dark phase contrast images, while the heralding arm’s polarization selection remains undefined. QWP, quarter-wave plate; HWP, half-wave plate; PBS, polarization beam splitter. (b) The heralding arm remotely manipulates the bright–dark phase contrast imaging modes. When the polarization basis vector of the heralded photon is projected to $| H ⟩$ polarization, the image acquired by the intensified charge-coupled device (ICCD) camera in the imaging arm is a dark image; when the polarization basis vector of the heralding photon is projected to $| V ⟩$ polarization, the image is a bright image.

Download full size

View all figures

2.3 Sample Design and Characterization

The low phase gradient sample used in this method is shown in Fig. 2(a), which features a flower with phase shifts in the color-coded regions all differing by $π / 5$ , with phases of 0, $π / 5$ , and $2 π / 5$ from the flower core to the outermost petals, phases of $3 π / 5$ and $4 π / 5$ for the stems and leaves, respectively, and a blue background with a phase of $π$ . As illustrated in Fig. 2(b), the unit structure of the metasurface is depicted schematically. The substrate is made of silicon dioxide, on which the amorphous silicon pillars are 190 nm in length, 110 nm in width, and 390 nm in height, with a periodicity of 350 nm and an overall size of 35 µm × 35 µm. Figure 2(c) presents a scanning electron microscope (SEM) image of the metasurface. We use commercial software finite difference time domain (FDTD) to calculate the phase change of incident light after passing through individual nanopillars, thereby determining the phase and amplitude modulation. Applying periodic boundary conditions along the $x$ and $y$ directions and a perfectly matched layer (PML) along the $z$ direction, we scan the length and width of the nanopillars to obtain the additional phase and transmittance for horizontal and vertical directions of the nanopillars. Based on this, we can find a set of suitable sizes of the unit structure to be arranged to obtain the polarization-multiplexing metasurface we need so that different phase modulations can be obtained after the incident of different polarized light. The size of the metasurface directly affects the phase contrast imaging performance. Through MATLAB simulations of various dimensions, we optimized the metasurface size to 35 µm, achieving high-contrast bright–dark phase contrast imaging.

Figure 2.Characterization of experimental samples. (a) Phase distribution of the PO. Color-coded regions represent distinct phase shifts: yellow core (0), light pink first petal layer ( $π / 5$ ), dark pink second petal layer ( $2 π / 5$ ), light green stems ( $3 π / 5$ ), dark green leaves ( $4 π / 5$ ), and blue background ( $π$ ). All phase shifts are quantized in $π / 5$ increments (see the Supplement 1, Sec. 2). (b) Schematic diagram of the metasurface. The metasurface has a size of 35 µm × 35 µm and a structural period of 350 nm, exhibiting a periodic arrangement of amorphous silicon pillars with dimensions of 190 nm in length, 110 nm in width, and 390 nm in height. (c) Top-view SEM image of the metasurface.

Download full size

View all figures

3. Results

3.1. Experimental Setup

The experimental setup is shown in Fig. 3(a). The pump source is a high intensity pulsed laser that produces a focused 390 nm femtosecond laser beam with very high instantaneous power for efficient excitation of spontaneous parametric down-conversion (SPDC). The pump light is focused by a convex lens into two beam-like BBO crystals arranged in a sandwich configuration and generates a pair of photons in the state ${| H ⟩}_{1} {| V ⟩}_{2}$ through spontaneous parametric down-conversion in the first layer. The photon state transformed to ${| V ⟩}_{1} {| H ⟩}_{2}$ after passing through the half-wave plate, followed by another pump light in the second layer. The pump light then excites another pair of parametric photons in the ${| H ⟩}_{3} {| V ⟩}_{4}$ photon state in the second crystal layer, and both pairs of parametric photons have a wavelength of 780 nm. Each pair of parametric photons is reflected by the triangular prism to the heralding arm (left) and the signal arm (right), respectively. Since the two photons that are divided into the same arm come from different pairs of parametric photons, they are generated at different temporal and spatial positions. In addition, different polarized light propagates in different directions and at different speeds in spatially anisotropic crystals. Therefore, the lithium niobate crystal ( ${LiNbO}_{3}$ ) and the yttrium vanadate crystals ( ${YVO}_{4}$ ) are placed in each of the heralding and signal arms to compensate for the temporal and spatial walk-off of the parametric light to achieve the indistinguishability of the photons. Simultaneously, a lens positioned between the crystals is used to collimate the optical path. At this point, the two-photon state interferes, resulting in the final output parametric light state $1 / \sqrt{2} ({| H ⟩}_{i} {| V ⟩}_{s} + {| V ⟩}_{i} | {| H ⟩}_{s}$ , where the subscripts $i$ and $s$ denote the position of the photon in the heralding and signal arms, respectively.

Figure 3.Experimental setup and characterization of an entangled light source. (a) Experimental setup. A 390 nm pulsed ultraviolet laser was focused through a lens onto two sandwich-structured beta barium borate (BBO) crystals to generate entangled photon pairs. Each pair of entangled photons is indistinguishable after passing through the time-compensated and space-compensated crystals. The QHP combination is used to selectively detect heralding photons, which subsequently reach the SPAD trigger ICCD. The signal photons are modulated by the metasurface and then finally collected for imaging by the ICCD. The QHP combination comprised a QWP, an HWP, and a PBS. FC, fiber coupler; BPF, bandpass filter; YV, yttrium vanadate crystal ( ${YVO}_{4}$ ) for space compensation; LN, lithium niobate crystal ( ${LiNbO}_{3}$ ) for time compensation; PO, phase object. (b) Sinusoidal fit of the data when the HWP in the signal arm is fixed at 0° (red) and 22.5° (blue). (c) Real and imaginary parts of the density matrix of the quantum state.

Download full size

View all figures

A long-pass filter and a narrow-band filter are placed in both the heralding and signal arms to eliminate undesirable fundamental frequency light. In the heralding arm, the heralding photons first pass through a QHP combination (a quarter-wave plate and a half-wave plate) and a polarization beam splitter (PBS). The PBS selects the photon polarization and projects the state onto a specific polarization basis, enabling switching between bright-field and dark-field imaging. Subsequently, the single-photon avalanche diode (SPAD) converts the received photons into electrical signals, triggering an ICCD camera in digital delay and gate (DDG) mode to capture the signal photons. Due to the time delay in this process, to ensure that the photons captured by the ICCD camera are from the same pair as those detected by the SPAD, a 20-m-long optical fiber and the nanosecond-level electronic delay built into the ICCD are used to compensate for the time difference. Considering the polarization dephasing effect of single-mode optical fiber, a set of HWP and QWP is added after the fiber to correct the polarization state of the exiting photons. When the flip-flop mirror is upturned into the optical path, it directs the signal photons emerging from the fiber into the SPAD to test the quality of the quantum state. The QWP and HWP are then adjusted by observing the coincidence under different polarization basis vectors until the polarization state is completely corrected.

After preparing a high-quality quantum state, we lower the reflector into position, directing signal photons into the imaging optical path. The signal beam, with a 1 mm spot diameter, passes through a beam reduction system before illuminating the PO. The resulting light is then amplified by an objective lens (20 mm focal length) and a 100 mm lens. Following interaction with the PO, the signal light traverses a $4 f$ system where a metasurface at the Fourier plane modulates both spatial spectrum and photon polarization states. The processed light is finally collected by the ICCD for imaging.

3.2. Characterization of Quantum Light Sources

The brightness and fidelity of the polarization-entangled light source are critical factors for imaging quality. Prior to imaging experiments, we first characterized the source performance. At 73 mW pump power, we measured single-photon count rates of 110 k/s (the heralding arm) and 210 k/s (the signal arm). The coincidence counts (C) exhibited strong polarization correlation: reaching a maximum of 25 k/s for orthogonal polarization bases (H/V or V/H), and dropping to 0.6 k/s for parallel bases (H/H or V/V). The measured heralding efficiency reached 22.73%.

For a comprehensive assessment of the prepared quantum state, we performed polarization interference measurements and quantum tomography on the constructed polarization-entangled source. By fixing the HWP of the signal arm at 0° or 22.5° while rotating the HWP of the heralding arm in 10° increments, we obtained polarization interference fringes with coincidence counts (1 s integration time) showing high-visibility sinusoidal modulation [Fig. 3(b)], providing further confirmation of entanglement. The polarization visibility, defined as $V = \frac{C_{\max} - C_{\min}}{C_{\max} + C_{\min}}$ , was calculated to be $9 5.3 % \pm 0.1 %$ . According to the theoretical relationship between Bell’s inequality and quantum correlations, any local hidden variable theory imposes an upper bound of $S \leq 2$ in the CHSH inequality. In our experiment, we performed a complete Bell test using standard analyzer settings and obtained a Bell parameter of $S = 2.662 \pm 0.006$ , clearly exceeding the classical limit^[45,46]. This significant violation provides direct and quantitative evidence of quantum entanglement, confirming the nonclassical nature of our photon source (see the Supplement 1, Sec. 1 for details).

Furthermore, we performed complete quantum state tomography to reconstruct the two-photon density matrix. Figure 3(c) presents both the real and imaginary components of the experimentally reconstructed density matrix $\hat{ρ}$ . We quantified the source quality through state fidelity, where $F = ⟨ Ψ | \hat{ρ} | Ψ ⟩$ , $| Ψ = 1 / \sqrt{2} (| H V ⟩ + | V H ⟩)$ represents the ideal Bell state. The measured fidelity of 95.1% confirms our source produces highly pure entangled photon pairs suitable for quantum imaging applications.

3.3. Higher Contrast than Classical Imaging at Equal Photon Flux and Background Noise

Based on the successful preparation of a high-fidelity polarization-entangled source and the systematic characterization of the electromagnetic control mechanism and imaging principles of the metasurface, this study demonstrates the superior performance of our metasurface-integrated quantum imaging system compared to conventional imaging approaches. Specifically, by exploiting the temporal correlations of photon pairs and two-photon coincidence detection techniques, quantum imaging can effectively suppress background noise while maintaining identical photon flux. This approach significantly enhances imaging contrast, thereby providing an innovative solution to overcome the technical limitations of conventional optical imaging systems.

To demonstrate high-contrast imaging advantages achieved by combining quantum light sources with metasurfaces, we acquired comparative images under both classical imaging mode and metasurface-free quantum imaging conditions. Figure 4(a) shows a classic image obtained in continuous collection mode, where the ICCD was internally triggered with the camera gate held open throughout acquisition. In this mode, all incident signal photons are captured along with substantial ambient noise photons. The resulting accumulation of uncorrelated background photons overwhelms the signal, obscuring any discernible image features. However, the entangled photon pairs generated by utilizing spontaneous parametric down-conversion have strong correlation properties in the time domain and can be imaged by secondary correlation through the externally triggered mode of the ICCD. When the SPAD detects a heralding photon, it triggers the time gate of the ICCD at the signal side to open briefly. In the experiment, we set the acquisition time to 20 min and the time width of the gate to 5 ns so that only twin signal photons captured in the time gate during the acquisition time accumulate to form an image, thus significantly filtering out noise and realizing high-contrast imaging. Figures 4(b)–4(d) show a bright phase contrast image and a dark phase contrast image without metasurface modulation, respectively, acquired by the DDG mode of the ICCD under the illumination of the quantum light source. Figures 4(e)–4(h) show the intensity maps of the light field varying along the dashed line in Figs. 4(a)–4(d). Comparing the classical imaging effect, it can be seen that even the DDG modes without metasurface modulation have shielded a lot of noise and have shown imaging effects, but the contrast is very low. Only with the modulation of the metasurface can we realize the high-contrast bright and dark phase contrast imaging effect, so the light field modulation ability of the metasurface plays a key role in the imaging scheme. Based on the definition of contrast ratio $V = \frac{C_{\max} - C_{\min}}{C_{\max} + C_{\min}}$ , the imaging contrast ratios of Figs. 4(a)–4(d) are calculated to be 0.22, 0.27, 0.75, and 0.81, respectively. Therefore, quantum imaging has the advantage of higher contrast under the same imaging conditions.

Figure 4.Imaging comparison between classical and quantum light sources. Under identical photon flux illumination conditions, phase contrast imaging using a quantum entangled source achieves high contrast compared to classical weak light sources, while high-contrast imaging can only be realized with the incorporation of metasurface modulation. (a) Bright phase contrast imaging with metasurface modulation under classical weak light illumination. (b) Bright phase contrast imaging without metasurface modulation under quantum light illumination. (c) Bright phase contrast imaging with metasurface modulation under quantum light illumination. (d) Dark phase contrast imaging without metasurface modulation under quantum light illumination. (e)–(h) Intensity maps at the white dashed lines in (a)–(d).

Download full size

View all figures

3.4. Remote Switching of Imaging Modes

Quantum imaging not only has the advantage of high-contrast imaging as described in the previous section, but it also can utilize the polarization entanglement property of entangled photon pairs to realize the function of remotely switching the imaging modes of the imaging system, which improves the flexibility of this system. As shown in Fig. 5, these are the images obtained by projecting the signal and heralding arms on different polarization bases, where the horizontal (vertical) axis indicates the projected polarization state in the signal (heralding) arm, and 1200 frames were acquired for each image, with an exposure time of 1 s per frame. The image in the orange border indicates that the projected polarization states of the signal arm and the heralding arm are orthogonally polarized bases when $1 / \sqrt{2} (| H V ⟩ + | V H ⟩)$ is satisfied to realize the bright–dark phase contrast image. The green border indicates that the bright–dark phase contrast image cannot be realized when the projected polarization states of the signal arm and the heralding arm are the same. Figure 5(a) demonstrates remotely switched quantum bright–dark phase contrast imaging of the low phase gradient sample.

Figure 5.Remote switching of quantum bright–dark phase contrast imaging functions. (a) Bright–dark phase contrast images of a low phase gradient sample. The number of acquisition frames is 1200 with an exposure time of 1 s per frame. (b) Quantum bright–dark phase contrast image of onion epidermis. The number of frames acquired is 3600. The horizontal axis represents the projected polarization state in the signal arm, while the vertical axis represents the projected polarization state in the heralding arm. The images displayed in the coordinate system show those acquired by the ICCD camera under different polarization states and processed to remove noise.

Download full size

View all figures

In biomedical imaging, conventional bright-field microscopes are challenged by the low contrast and homogenization of transparent biological samples, while conventional staining methods may affect cellular activity, and this system proposes a new solution to this dilemma. Figure 5(b) demonstrates quantum bright and dark phase contrast imaging of onion epidermal cells. When the polarization basis vectors of the two paths are projected to $| H V ⟩$ or $| V H ⟩$ , the ICCD can acquire two high-contrast images of onion epidermal cells in either bright or dark phase contrast, and the switching of the two modes only needs to be executed at the heralding arm, which realizes the remote manipulation function of image acquisition. The technique exploits the nonclassical properties of quantum-entangled light sources to enable high-contrast imaging under very low light conditions while avoiding phototoxicity to cell samples. This makes the introduction of quantum bright–dark phase contrast imaging techniques particularly important. By integrating remote switching functionality, quantum bright–dark phase contrast imaging not only offers higher phase resolution and sensitivity, but also enables dynamic visualization of biological processes at the microscopic scale under non-invasive and contact-free conditions. This advantage is particularly critical in application scenarios where mechanical intervention is infeasible, such as in in vivo imaging systems or enclosed microscale environments^[42–44]. As a result, this approach offers excellent integrability and scalability, and it holds strong potential for applications in miniaturized integrated imaging systems, in vivo neural imaging platforms, and remote imaging setups under extreme environmental conditions.

4. Discussion and Conclusion

Compared with existing quantitative phase imaging techniques^[24,25,28], our imaging system can obtain high-contrast images without requiring multi-frame image acquisition and complex post-computational processing, thereby reducing computational costs. Furthermore, our scheme demonstrates superior temporal performance, and its remote switching capability enables on-demand mode switching while maintaining mechanical stability. These features make it particularly suited for stringent clinical applications requiring high-efficiency and high-contrast imaging. These technical advantages may translate to significant practical value in clinical diagnostics.

In summary, this study experimentally demonstrates a novel optical imaging system based on a quantum polarization entangled source and a polarization multiplexed metasurface. The system achieves adaptive high-contrast label-free imaging of transparent samples with arbitrary phase combinations under low photon flux illumination, demonstrating both the flexibility and highly integrated design of the platform. Specifically, we employ an ICCD camera to detect twin-photon signals and innovatively utilize the quantum polarization-entangled source as a remote switching trigger, enabling dynamic switching between quantum bright–dark phase contrast imaging modes. This design allows imaging mode switching without requiring any optical path adjustment between the sample and metasurface, thereby ensuring system stability while guaranteeing acquisition of at least one high-contrast imaging mode. Notably, our approach is particularly well-suited for applications requiring low photon flux. Experimental results confirm that the system maintains high-contrast imaging under external noise. With these advantages, the system overcomes the constraints of conventional phase contrast imaging, demonstrating strong potential for urgent clinical demands for integrated, adaptive, high-contrast, and remotely accessible imaging of biological cells, as well as providing a powerful tool for biomedical research and demonstrating broad potential for future applications in biological imaging, life science studies, and remotely controlled integrated imaging systems.

Category: Research Article

Received: Apr. 26, 2025

Accepted: Aug. 19, 2025

Published Online: Sep. 23, 2025

The Author Email: Shuming Wang (wangshuming@nju.edu.cn)

DOI:10.3788/AI.2025.10014