Photoacoustic microscopy (PAM) utilizes tightly focused optical excitation and coaxial ultrasonics detection to directly image light-absorbing molecules with subcellular resolution, yielding detailed spectroscopic insights at the microscopic scale. These advantages enable label-free imaging of vasculature, pigments, lipids, and other exogenous chromophores, facilitating both functional and molecular assessments in vivo [1–5]. Multi-wavelength imaging in PAM is essential for capturing the distinct spectral signatures of biomolecules [6,7]. Multi-wavelength optical-resolution photoacoustic microscopy (MW-OR-PAM), a variant of PAM, can map the total concentration of hemoglobin, blood oxygen saturation (sO2), and blood flow speed [8,9]. MW-OR-PAM generally depends on laser sources providing rapid wavelength tuning, sufficient pulse energy, and consistent output stability to guarantee molecular spectroscopic measurements. However, conventional wavelength-tunable lasers, such as optical parametric oscillators (OPOs), dye lasers, or multi-laser combinations, remain costly, are technically complex, or have low repetition rates [10–13]. Although fiber-based supercontinuum (SC)-based laser sources have been explored as alternatives to MW-OR-PAM, their spectral output is predominantly concentrated in the longer near-infrared (NIR) region [14]. This makes them particularly suitable for in vivo imaging of lipids and other chromophores with characteristic absorption in that range. However, conventional optical imaging applications such as oxygen saturation mapping, hemoglobin quantification, and venous imaging, require excitation wavelengths in the visible spectrum, where SC sources are limited by reduced spectral power density and low energy efficiency after spectral filtering. Stimulated Raman scattering (SRS) in optical fibers provides a promising method for generating multi-wavelength sources in MW-OR-PAM, overcoming the limitations of wavelength-tunable lasers [15,16]. The sub-microsecond wavelength switching capability ranging from green (532 nm) to red wavelengths (620 nm) is particularly advantageous for imaging oxygen saturation dynamics within major vessels and capillary networks. Several research groups have validated the feasibility of fiber-based SRS lasers for functional and spectroscopic photoacoustic imaging in in vivo applications [17–20].

Despite their considerable promise, fiber-based SRS sources have intrinsic limitations when applied to MW-OR-PAM. First, expanding the accessible spectral range typically necessitates longer fiber length, which significantly reduces the optical damage threshold and thus limits achievable pump pulse energy. Consequently, this constraint severely attenuates the energy of red-shifted wavelengths, leading to insufficient photoacoustic signal strength in the red spectral region in MW-OR-PAM [18,19]. Second, chromophores such as hemoglobin exhibit absorption coefficients in the red spectral region that are one to two orders of magnitude lower than those in the green region, thereby requiring substantially higher laser pulse energies for effective in vivo MW-OR-PAM. However, the available energy at red wavelengths is inherently constrained by the limited efficiency of SRS conversion and the optical damage threshold of the fiber, thereby further restricting imaging performance within this spectral range [7,21].

Enhancing transducer sensitivity or simplifying the co-registration of optical and acoustic foci provides a direct route to improving the detection of weak photoacoustic signals. In the past decade, various probe architectures—including optical-acoustic combiners, ring-shaped focused ultrasonic transducers, and optically transparent designs—have been developed to increase acoustic collection efficiency and facilitate precise coaxial alignment of optical and acoustic foci. The optical-acoustic combiner, designed to achieve coaxial alignment between optical and acoustic beams, often involves complex fabrication procedures [22]. Hollow ring-shaped focused ultrasonic transducers allow optical excitation to pass through a central aperture, simplifying the optical-acoustic combiner design [23–25]. However, these transducers are often limited by a restricted aperture or low numerical aperture. MW-OR-PAM imaging requires a high numerical aperture for sensitivity and a long working distance for in vivo accessibility. Enlarging the transducer numerical aperture reduces acoustic bandwidth, potentially limiting the accuracy of functional measurements such as blood oxygen saturation [26]. The optically transparent transducers utilize advanced piezoelectric materials such as LiNbO3 or PIN-PMN-PT to improve sensitivity while eliminating the need for mechanical scanning. However, their bandwidth remains suboptimal, and limited thermal stability presents challenges for sustained and reliable operation in MW-OR-PAM [27–30]. State-of-the-art transparent transducers still face limitations in sensitivity within the red spectral region, potentially compromising the accuracy of blood oxygen saturation. The high sensitivity is critical for visualizing delicate tissues and low-absorbing chromophores while minimizing photobleaching, toxicity, and biological perturbation.

Another alternative approach to enhancing the detection of weak signals in MW-OR-PAM involves the use of intravenously administered exogenous contrast agents [31] or topically applied optical clearing agents (OCAs) [32,33]. The applicability of contrast agents is limited by their biocompatibility and translational potential for widespread clinical use. Although conventional OCAs, such as glycerol, have been employed to improve imaging depth, their use is limited by potential tissue dehydration and the risk of irreversible tissue alteration [34,35]. Tartrazine, a widely used food dye approved by the U.S. Food and Drug Administration, has been shown to reversibly render skin optically transparent in vivo [34]. In the green region, particularly below 576 nm, transmission falls below 25%, indicating high absorption. Consequently, tartrazine is suboptimal for excitation at 532 nm. This limitation was also observed by Miller et al., who reported surface signal saturation—manifesting as a bright superficial band—when applying photoacoustic microscopy with 532 nm excitation [35]. Jia et al. demonstrated that photoacoustic signal enhancement at 532 nm excitation could be achieved using low concentrations of tartrazine in the mouse ear [36]. This observation is consistent with the strong absorption of tartrazine in the green spectral region, which limits its effectiveness under green-light excitation.

In this work, we developed an MW-OR-PAM system optimized for SRS-based optical excitation to broaden its spectral coverage and enhance imaging performance. The system incorporates two key innovations. First, we developed a large-aperture probe with a 9 mm diameter and a numerical aperture of 0.67, enabling flexible optical-acoustic co-focusing across a broad spectral range and enhancing the sensitivity for detecting weak signals under long-wavelength excitation. Second, we introduced tartrazine, a biocompatible tissue-clearing agent, to achieve reversible in vivo optical clearing at wavelengths above 600 nm, thereby enhancing the signal-to-noise ratio of photoacoustic imaging. These strategies markedly expand the functional imaging capabilities of MW-OR-PAM into the red spectral region, enabling deeper tissue visualization and broader spectroscopic access.

Figure 1(a) illustrates the schematic of the MW-OR-PAM. A nanosecond 532 nm laser (VPFL-G-20, Spectra-Physics) serves as the excitation source for generating multiple wavelengths through the SRS. The laser features an adjustable pulse width, which is configured to a fixed duration (7 ns) in this implementation to ensure optimal SRS performance. The 532 nm beam is split into two optical paths using a polarizing beam splitter (PBS), with the relative energy distribution controlled by a half-wave plate (HWP1, WPH10E-532, Thorlabs). One of the beams is launched into a 30 m polarization-maintaining single-mode fiber (PMSMF, HB-450-SC, Fibercore), where a sequence of Raman-shifted wavelengths ranging from 532 to 620 nm is generated via the SRS [Fig. 1(c)]. To maximize the Raman conversion efficiency, another HWP is placed before the fiber to align the beam polarization with the fiber’s principal axis. The resulting output wavelength is then selectively filtered using interchangeable optical filters. In parallel, the power of the second 532 nm beam is independently controlled by a neutral density filter (NDF, NDC-50S-3M, Thorlabs). The two optical paths are subsequently combined using a dichroic mirror (DM, DMLP550, Thorlabs) and coupled into a 2 m single-mode fiber. The output beam is focused through a pair of achromatic objectives, with the second lens (AC-127-025-A, Thorlabs) mounted on a custom-built, Z-axis adjustable holder. This design enables precise tuning of the optical focal plane across different excitation wavelengths shown in Fig. 1(c), facilitating accurate alignment with the acoustic focus and effectively mitigating chromatic aberration. The detected photoacoustic signals are amplified by 48 dB using two low-noise RF amplifiers (ZFL-500LN+, Mini-Circuits) and digitized with a high-speed data acquisition card (ATS9371, AlazarTech). Raster scanning of the acoustic-optical probe was performed using two motorized linear stages.

The acoustic-optical probe in MW-OR-PAM adopts a custom-built high-sensitivity transducer [Fig. 1(b)]. A 9-μm-thick P(VDF-TrFE) film is sandwiched between a planoconvex lens (LA1207-A, Thorlabs) and a custom-fabricated wide-aperture acoustic lens with a high numerical aperture of 0.67, designed to enhance acoustic collection efficiency. The central region of the film features a 3-mm-diameter hollow aperture, rendering the entire probe optically transparent with a transmittance exceeding 90%. This configuration enables efficient optical transmission through the acoustic stack and substantially simplifies the co-registration of acoustic and optical foci. Compared with transparent ultrasound transducers, our probe features a high numerical aperture (0.67) and a broad one-way acoustic bandwidth (98.94%), which significantly enhance the axial resolution and detection sensitivity.

By adjusting the pulse energy of the pump laser, the SRS-based wavelength shifter generates discrete output wavelengths spanning from 532 to 620 nm. Owing to the spectral versatility of SRS, multiple laser configurations can be tailored to address specific imaging needs [Fig. 1(c)]. A 532/558 nm dual-wavelength setup was implemented for vascular structural imaging and oxygen saturation mapping, capitalizing on the high optical absorption contrast of hemoglobin in the green spectral region. In comparison, a 604/620 nm red-wavelength configuration was developed to enable deep vascular imaging and contrast-enhanced applications, facilitated by using tissue-transparent molecules and the optical clearing strategy introduced in this study [Fig. 1(d)]. The reduced scattering at longer wavelengths allows for greater imaging depth and improved sensitivity within the red spectral region.

According to the Kramers–Kronig relations, its strong absorption in the blue–green spectrum alters the dispersion profile of the surrounding medium, effectively reducing the refractive index mismatch between water and lipids at red wavelengths. This reduction in optical scattering could significantly enhance the photoacoustic imaging depth in MW-OR-PAM, particularly in the red spectral region above 600 nm generated by SRS. A 0.6 mol/L tartrazine solution was selected in this study [Fig. 1(d)]. Figure 1(e) shows normal-incidence light transmittance as a function of the available SRS wavelength in MW-OR-PAM. We selected the 604/620 nm as the optical transparency window for in vivo imaging.

We evaluate the imaging performance of the MW-OR-PAM system at representative green (532 nm) and red wavelengths (604 nm), including the imaging depth and spatial resolution. The imaging depths were assessed by inserting a black human hair obliquely into freshly excised chicken breast tissue. The maximum penetration depths at 532 nm and 604 nm were measured to be 0.75 mm and 0.81 mm, respectively [Figs. 2(a) and 2(b)]. Lateral resolutions were quantified using a 1951 United States Air Force (USAF) resolution target (RTS3AB-P, Lbtek, China). Amplitude profiles were extracted across pattern edges to derive the edge spread function (ESF), which was then differentiated to obtain the line spread function (LSF). The full widths at half-maximum (FWHM) of the LSF yielded lateral resolutions of 3.8 μm at 532 nm and 5.7 μm at 604 nm [Figs. 2(c) and 2(d)]. To assess axial resolutions, A-line signals were acquired from the resolution target and processed via Hilbert transformation. The resulting FWHMs of the envelope profiles were 33.9 μm and 38.5 μm at 532 nm and 604 nm, respectively [Figs. 2(e) and 2(f)], consistent with theoretical predictions based on the system’s acoustic bandwidth [22].

To further evaluate spatial resolution, we performed a qualitative assessment using Groups 6 and 7 of the USAF resolution target. Line profiles were extracted to determine the minimum distinguishable feature size, defined as line sets exhibiting contrast above half the maximum signal. As shown in Figs. 2(g), 2(i) and 2(h), 2(j), the smallest resolvable features were 2.76 μm at 532 nm and 4.38 μm at 604 nm, demonstrating subcellular imaging capability. These results are consistent with quantitative measurements, providing a comprehensive validation of the system’s spatial resolution performance.

To evaluate system robustness, we imaged a leaf-skeleton phantom that closely mimics microvascular architecture. As shown in Figs. 2(k) and 2(l), the reconstructed photoacoustic image clearly resolved the intricate vein structures, demonstrating the system’s high-resolution imaging capability and structural fidelity in complex biological analogs.

Microvascular morphology plays a critical role in tissue regeneration and disease progression, including processes such as wound healing, inflammation, and tumorigenesis [37–41]. To evaluate the in vivo imaging performance of the MW-OR-PAM system, we conducted experiments in healthy female ICR mice (5–6 weeks old) for skin microvasculature, oxygen saturation (sO2), and brain imaging. All procedures were approved by the Animal Ethics Committee of the Suzhou Institute of Biomedical Engineering and Technology. During imaging, mice were anesthetized with 1.5% (volume fraction) vaporized isoflurane delivered via inhalation. We employed 532 nm excitation for microvascular imaging and dual-wavelength excitation at 532/558 nm combined with spectral unmixing for functional imaging. Leveraging the system’s multi-wavelength capability and high sensitivity, MW-OR-PAM enabled visualization of total hemoglobin concentration [Fig. 3(b)] and sO2 [Fig. 3(c)]. The resulting images revealed a well-defined microvascular network, allowing clear differentiation between arteries and veins. We further applied MW-OR-PAM to brain imaging, where the photoacoustic images clearly resolved vascular networks at multiple depths, highlighting its potential for neuroscience applications [Figs. 3(e) and 3(f)]. All images were acquired over a 5mm×5mm field of view using excitation energies far below 100 nJ, well within the safety limitation defined by the American National Standards Institute (ANSI, 20mJ/cm2).

Optical scattering in biological tissue arises primarily from refractive index mismatches between aqueous-based components and lipid- or protein-rich structures. An ideal OCA should reduce this scattering without inducing biocompatibility concerns or irreversible tissue alterations. However, few OCAs fulfill these criteria. For instance, conventional agents such as glycerol can improve optical penetration but are associated with tissue dehydration and structural disruption, limiting their suitability for reversible and clinically applicable tissue clearing [42].

Recently, the water-soluble dye tartrazine has been reported to induce in vivo tissue transparency by increasing optical transmission by more than 40% in the red spectral region (>600nm), through modulation of the refractive index of aqueous components, as predicted by the Lorentz oscillator model and Kramers–Kronig relations [Figs. 1(d) and 1(e)]. Building on this mechanism, we implemented an SRS-generated 604/620 nm wavelength configuration to enable deep vascular imaging and contrast-enhanced applications.

To evaluate the effectiveness of tissue-clearing molecules in MW-OR-PAM, we conducted a series of tissue phantoms and in vivo imaging experiments in five- to six-week-old female mice. To specifically assess improvements in imaging depth, phantom studies were carried out using freshly excised chicken breast tissue containing obliquely embedded pencil leads. As shown in Fig. 5 (Appendix A), tartrazine treatment at red wavelengths increased the maximum photoacoustic imaging depth from 0.81 mm to 1.89 mm, corresponding to a 2.3-fold enhancement. An experimental workflow for ear imaging is illustrated in Fig. 4(a). Prior to imaging, tartrazine powder (T818650, Macklin) was dissolved in deionized water to a final concentration of 0.6 mol/L and incubated at 40°C for 10 min to ensure full dissolution. To enhance dye penetration, the stratum corneum of the mouse ear was gently abraded using a suspension of pumice particles applied with a cotton swab for 30 min. The tartrazine solution was then topically applied and massaged into the ear for approximately 15 min until maximal optical transparency was observed. After imaging, the treated area was rinsed with water, and the tissue gradually returned to its original appearance within two days. Imaging was performed at each stage to independently assess the effects of clearing and recovery on vascular visualization. Representative photographs and corresponding MW-OR-PAM images are presented in Fig. 4(b), demonstrating the enhanced optical transparency achieved through the tartrazine treatment.

Tartrazine-treated images exhibited markedly enhanced visualization of the capillary network compared to baseline and pumice-treated conditions. A substantial increase in the number of detectable microvessels was observed, with previously indistinct venules and arterioles becoming clearly distinguishable. As illustrated in Fig. 4(b), tartrazine treatment yields an average 1.6-fold increase in signal compared to the untreated condition (Fig. 6 in Appendix A). The recovery image acquired two days post treatment revealed vascular structures consistent with pre-treatment conditions, confirming the reversible and biocompatible nature of tartrazine as an effective optical clearing agent. Transcranial brain and thigh imaging results under identical experimental parameters and excitation wavelengths are shown in Figs. 4(c) and 4(d). In both cases, tartrazine treatment significantly improved vascular visibility relative to untreated controls, underscoring the potential of spectrally selective absorbing of dye molecules to enhance optical penetration in scattering tissues. These findings highlight the promise of tartrazine-assisted clearing in advancing depth-resolved photoacoustic imaging and facilitating clinical translation in fields such as dermatology. Notably, while tartrazine treatment significantly enhanced vascular visualization in the ear and thigh, the improvement in transcranial brain imaging was comparatively modest. This is likely due to the optical scattering and absorption properties of the skull, which are not substantially affected by the tissue-clearing mechanism, thereby limiting the degree of enhancement achievable through optical clearing alone in transcranial applications.

We developed an SRS-based multi-wavelength optical-resolution photoacoustic microscopy (MW-OR-PAM) system with sub-microsecond wavelength switching capability across a broad spectral range spanning from green (532 nm) to red (620 nm). To support high-sensitivity detection across this range, we engineered a custom acoustic-optical probe featuring a large aperture (9 mm diameter), high numerical aperture (0.67), broad acoustic bandwidth (98.94%), and tunable optical focus, enabling precise optical-acoustic co-alignment. Furthermore, to address the challenges of reduced absorption and scattering at longer wavelengths, we introduced a biocompatible, tissue-transparent dye to achieve reversible in vivo optical clearing, significantly enhancing imaging sensitivity and depth in the red spectral region. Compared with conventional imaging at wavelengths above 600 nm, the application of tissue-clearing molecules resulted in improved image details and the signal-to-noise ratio (SNR) in vivo, as demonstrated in mouse ear, transcranial brain, and thigh imaging.

While we successfully demonstrated enhanced imaging performance at wavelengths above 600 nm, several aspects remain for further optimization. The current noise-equivalent pressure (NEP) is estimated to be 0.11mPaHz−1 [43], and future improvements may be achieved by optimizing the acoustic impedance matching between the P(VDF-TrFE) film and the acoustic lens, potentially increasing both bandwidth and sensitivity. Enhancing imaging speed would further enable the capture of dynamic physiological processes. To further advance imaging capabilities, ultrafast MW-OR-PAM implementations leveraging polygon scanners or voice coil actuators could substantially accelerate scanning throughput [25]. In parallel, the integration of deep-learning-based reconstruction with sparse sampling strategies may offer enhanced frame rates and imaging depths while preserving spatial resolution [44]. Furthermore, considering the complementary spectral characteristics of SRS- and SC-based light sources, their hybridization presents a compelling route toward a unified multi-wavelength OR-PAM platform with continuous spectral coverage from the visible to the near-infrared region.

Tissue optical clearing has demonstrated substantial improvements in photoacoustic imaging of soft tissues, but its effectiveness in transcranial imaging remains limited due to the inherent optical scattering and absorption of the skull. Future advances may be achieved by integrating optimized acoustic-optical probe designs, improved transcranial imaging strategies, and more efficient in vivo optical clearing approaches, collectively enabling new breakthroughs in deep-brain photoacoustic imaging. Additionally, expanding the accessible wavelength range of MW-OR-PAM, combined with the use of tissue-transparent dyes, may further extend its applicability in the long-wavelength regime for deeper and more versatile in vivo imaging. Although tartrazine is FDA-approved for dietary use and is generally regarded as safe at low concentrations, its long-term biocompatibility under conditions involving repeated administration or elevated doses—particularly in the context of optical modulation in large-animal models and deep-tissue applications—remains insufficiently characterized. Current evidence suggests rapid renal clearance and minimal systemic accumulation; however, further studies will be essential to comprehensively assess its biosafety in chronic imaging paradigms and high-dose protocols targeting deep-seated tissues.

In summary, enhancing the imaging depth, spectral range, and acquisition speed of OR-PAM is key to expanding its applicability across biomedical scenarios. Coupled with its intrinsic label-free contrast, MW-OR-PAM enables high-resolution mapping of capillary-level structures and functional parameters such as blood oxygenation. These capabilities hold strong potential for precise tumor localization, margin assessment, and ultimately, the clinical translation of photoacoustic microscopy in image-guided diagnosis and intervention.