Cholestatic liver diseases (CLD) are hepatobiliary conditions characterized by impaired bile formation, secretion, or flow [1], with a morbidity of 10.26% among hospitalized patients and strong prevalence to chronic liver disease. During disease progression, cholestasis serves as the predominant pathological feature and is a key determinant of disease severity [2]. Moreover, progressive cholestatic injury exacerbates hepatic deterioration, ultimately leading to end-stage liver disease characterized by severe cirrhosis and liver failure [3–5]. Therefore, early diagnosis and effective monitoring of cholestasis are essential to improving patient prognosis of CLD patients [6].

Currently, the early detection rate of cholestasis is constrained by the limitation of clinically used methods. Although serum biomarkers, such as alkaline phosphatase and γ-glutamyl transferase, provide convenient approach for CLD detection, the specificity and sensitivity remain contentious [7,8]. Meanwhile, imaging techniques, including ultrasound, computed tomography, and magnetic resonance cholangiopancreatography, contribute to the dominant proportion of CLD noninvasive diagnosis. However, diagnosis efficacy is limited by the compromised resolution for detecting the damages of the micro bile duct, which is the initial sign and important characteristic of early pathological changes of cholestasis. Therefore, it is urgent to develop new techniques for accurate and dynamic evaluation of micro intrahepatic bile duct structure and function, which is essential for the early detection of cholestasis.

Optical imaging is widely recognized for its exceptional spatiotemporal resolution, making it a powerful tool in biomedical research and clinical diagnostics [9,10]. Among its various modalities, photoacoustic imaging (PAI) and near-infrared-II (NIR-II, 1000–1700 nm) fluorescence imaging, which have emerged over the past decade, offer enhanced imaging precision and greater penetration depth in molecule tracing, anatomical visualization, and functional assessment in vivo, providing promising essentials in the imaging of the structure and excretion function of the micro bile duct and further facilitating early detection of cholestasis [11–14]. In our previous work, high resolution photoacoustic microscopy (PAM) has been developed to examine the microstructure and blood oxygen saturation (sO₂) of the liver in the diseased states of tyrosinemia, alcoholic liver [15,16]. Photoacoustic computed tomography (PACT) was applied for quantitative evaluation and real-time monitoring of the changes of liver function reserve via the clearance rate of exogeneous probes in vivo [16–18]. Near-infrared-II (NIR-II, 1000–1700 nm) fluorescence imaging can overcome the problems of strong tissue absorption and high scattering, offering superior spatiotemporal resolution, signal-to-noise ratio, and penetration depth. It has been successfully applied in liver surgical navigation when combined with indocyanine green (ICG) [19–21]. Moreover, the combination of the two modalities could provide more comprehensive information of the structural and functional disruption of micro bile ducts in cholestasis. However, to our best knowledge, neither the noninvasive PAI nor NIR-II fluorescence for evaluating micro bile duct structure and function has been reported.

In this study, we aimed to develop and validate a noninvasive imaging method using PAI and NIR-II fluorescence imaging to evaluate cholestasis severity in a CLD mouse model, revealing its impact on the liver microenvironment and hepatobiliary excretion function. Primary biliary cholangitis (PBC) causes progressive interlobular bile duct injury and inflammation [22,23]. In contrast, bile duct obstruction leads to rapid cholestasis [24]. In this study, we will use these two disease models to assess the severity of cholestasis. High-resolution PAM was employed to assess bile duct permeability with Evans blue (EB) and to evaluate hepatic lobular structure and sO₂ across diverse severity of cholestasis. Using ICG contrast enhancement, PACT was used to quantify liver function reserve, while NIR-II fluorescence imaging was utilized to map cholestasis distribution and monitor the dynamics of biliary excretion during disease progression. By analyzing EB penetration and ICG metabolism, this study offers insights into physiological indices and correlates cholestatic severity and imaging findings. Cross-scale optical imaging offers a preclinical approach for assessing cholestatic liver disease from structural and functional perspectives, thereby enhancing understanding of its underlying mechanisms.

Female C57BL/6J mice (8 weeks old) were obtained from Guangdong Medical Laboratory Animal Centre (China) and maintained under specific pathogen-free conditions in ventilated cages. All experimental procedures were conducted in accordance with protocols approved by the Ethics Committee of Guangdong Provincial People’s Hospital (KY-N-2022-122). To establish a chronic intrahepatic cholestasis model of PBC, mice were intraperitoneally immunized with 100 μg of 2OA-BSA emulsified in 100 μL of complete Freund’s adjuvant (CFA, Sigma-Aldrich, USA) containing 10 mg/mL of Mycobacterium tuberculosis strain H37Ra. Subsequently, 100 ng of pertussis toxin dissolved in 100 μL of phosphate-buffered saline (PBS) was administered at the time of the first immunization and two days later. Booster injections of 2OA-BSA in 100 μL of incomplete Freund’s adjuvant (IFA, Sigma-Aldrich) were given at week 2, 4, 6, and 8 post-initial immunization. BDL was conducted to establish an acute extrahepatic cholestasis model. Mice were anesthetized with isoflurane, and a midline laparotomy was performed to expose the bile duct. Approximately 0.5 cm of the bile duct was carefully isolated using blunt ophthalmic forceps and ligated with surgical sutures. After verifying the absence of bleeding or organ damage, the abdominal cavity was closed in layers, and the skin was sutured. To assess disease progression at different stages, the models were divided into four groups: PBC-12W, PBC-24W (representing 12 and 24 weeks of primary biliary cholangitis induction, respectively) and BDL-1W, BDL-2W (representing 1 and 2 weeks after bile duct ligation, respectively). The selection of PBC-12W and PBC-24W is based on the time-dependent development of inflammation and bile duct injury, while the BDL model exhibits rapid cholestasis onset typically within 1–2 weeks post-surgery.

A commercial PAM instrument (G2, Inno Laser Co., Ltd., China) was equipped with series of multi-wavelength lasers, including 532 nm, 559 nm, and 750–840 nm, enabling high-resolution photoacoustic imaging. The system utilized an ultrasonic transducer with a central frequency of 50 MHz. Additionally, a PACT system (SIP-PACT, Union Photoacoustic Technologies Co., Ltd., China) was equipped with a laser with a 20 Hz repetition rate, supporting imaging at wavelengths of 680–950 nm, 1064 nm, and 1190–2600 nm for organ cross-section imaging. A 512-element ring array ultrasonic transducer with a central frequency of 5.5 MHz was used to acquire ultrasonic signals. A DPM system (Zhuhai Dipu Medical Technology Co., Ltd., China) was implemented with an InGaAs SWIR camera (Cheetah-640CL, Xenics, Belgium) and a lens (Nikon), and individual filters (950 nm, 1000 nm, and 1100 nm; Thorlabs, USA) were fixed in the front end of the lens to capture NIR-II fluorescence.

The self-written MATLAB R2021b (MathWorks Inc. South Natick, MA, USA) program and Origin 2016 were used for image processing, such as structural imaging, liver function reserve, permeability, and sO₂ calculation.

PACT image reconstruction and post-processing were conducted using MATLAB. To improve spatial accuracy, the raw data were reconstructed with a dual-speed-of-sound universal back-projection algorithm that accounted for acoustic velocity differences between biological tissues and the surrounding coupling medium [25]. Vascular features were further enhanced using a Hessian-based filtering technique to highlight fine microvascular structures. To assess dynamic contrast distribution, differential images were obtained by subtracting baseline images from those acquired after injection. Regions exhibiting signal intensities above the mean pixel value were identified as enhanced areas and displayed using pseudo-color mapping. These images were then superimposed onto anatomical reference maps acquired at 1064 nm.

An ICG clearance test was performed to assess liver function reserve. Relative photoacoustic signal intensity (RI) was calculated to reflect the ICG concentration change as follows: RI(t)=([PA(t)−PA(0)]/[PA(0)],(1)where PA(t) represents post-injection time-varying photoacoustic signal value, and PA(0) is the basis value before injection. The time course of liver RI to reflect the change of ICG concentration was fitted using an exponential decay model [26], RI(t)={0,t<t0A[1−e−α(t−t0)]q⋅e−β(t−t0),t≥t0,(2)where A is the maximum value of the RI, α is the rate of contrast uptake (s−1), β is the rate of contrast washout (s−1), q is a parameter related to the slope of the early uptake, and t0 is the rise time point (s). The maximum peak time Tmax and the half-life period T1/2 of ICG concentration were calculated from the fitted curve.

HbO2 and Hb are the two major endogenous absorbers, which provide strong photoacoustic signals for sO₂ measurement. Basically, the blood absorption coefficient under two wavelengths can be expressed as follows [27]: $$\begin{gathered} μ(λ1)=εHb(λ1)CHb+εHbO2(λ1)CHbO2, \\ μ(λ2)=εHb(λ2)CHb+εHbO2(λ2)CHbO2. \end{gathered}$$(3)

In the above equations, CHb and CHbO2 represent the Hb and HbO2 content, respectively, and εHb(λ1), εHb(λ2), εHbO2(λ1), and εHbO2(λ2) are the extinction coefficients of the Hb and HbO2 at wavelengths λ1 and λ2. Assuming the photoacoustic signal intensity is proportional to the absorption coefficient of the tissue, the sO₂ can be calculated as follows [28]: $$\begin{gathered} CHb=εHbO2(λ2)A(λ1)−εHbO2(λ1)A(λ2)εHb(λ1)εHbO2(λ2)−εHb(λ2)εHbO2(λ1), \\ CHbO2=εHb(λ1)A(λ2)−εHb(λ2)A(λ1)εHb(λ1)εHbO2(λ2)−εHb(λ2)εHbO2(λ1), \\ sO2=CHbO2CHb+CHbO2. \end{gathered}$$(4)

In this equation, A(λ1) and A(λ2) represent the photoacoustic signal intensity obtained at wavelengths λ1 and λ2, respectively.

Liver tissues were fixed in 10% paraformaldehyde, embedded in paraffin, and sectioned. Hematoxylin and eosin (H&E) staining was performed to examine the basic pathological structures of the liver. Additionally, CK-19 antibody (Servicebio, China) was used to specifically label bile duct epithelial cells, enabling the evaluation of bile duct morphology and pathological alterations.

For liver structural imaging in mice with cholestatic liver disease, mice were anesthetized with 2% isoflurane, and a 5 mm abdominal incision was made to expose the liver lobe. Structural and functional information of hepatic lobules was measured using dual wavelengths of 532 nm and 559 nm. The imaging parameters were set as follows: a scanning step size of 3 μm and a laser frequency of 1 kHz.

For the intrahepatic bile duct permeability experiment, mice were also anesthetized with 2% isoflurane, and the liver lobes were exposed via a 5 mm abdominal incision. The common bile duct was bluntly dissected, and an indwelling needle was inserted into the bile duct before imaging. A syringe pump was used to inject 60 μL of 1% EB solution at a rate of 10 μL/s. The imaging of hepatic lobules was performed at a wavelength of 532 nm, and dynamic changes in EB signals were monitored over 30 min. The scanning parameters were consistent with those described above.

C57BL/6J mice were injected with 100 μL of ICG at a dose of 0.5 mg/kg body weight. NIR-II imaging was performed to capture fluorescence images at various time points before and after injection. A 1000 nm long-pass filter was used for fluorescence imaging, with an 808 nm excitation laser and an exposure time of 50 ms. Data analysis was performed using MATLAB software.

Throughout the experiment, each mouse was maintained under 2% isoflurane anesthesia and positioned upright in the center of the imaging chamber. Prior to imaging, an indwelling needle was inserted into the tail vein. Using a syringe pump (D107886, KD Scientific Inc., USA), 200 μL of ICG solution (5 mg/kg body weight) was delivered at a rate of 0.2 mL/s. Before injection, images were acquired at 1064 nm and 760 nm to obtain cross-sectional structural images of the liver. Following injection, the laser wavelength was set to 760 nm for cross-sectional liver imaging, and photoacoustic signals were continuously collected from the liver.

Statistical analysis was performed using SPSS 27 software (IBM, USA). Statistical significance was assessed using a two-tailed unpaired Student’s t-test. For comparisons between each experimental group and the control group, multiple comparison correction was performed using Dunnett’s multiple comparisons test, as facilitated by the statistical software. The annotations ns, *, **, and *** indicate no statistical significance, p<0.05, p<0.01, and p<0.001, respectively (n=3 per group). A p-value<0.05 was considered statistically significant. Data were presented as mean±standard deviation (SD).

The schematic diagram of the project is shown in Fig. 1(a). The cross-scale optical imaging platform comprised three components: PAM, PACT, and the NIR-II fluorescence imaging system. With a motor step of 3 μm, PAM achieved a lateral resolution of approximately 5 μm at 532 nm. PACT provided a lateral resolution of about 125 μm at 780 nm, with a penetration depth of approximately 5 cm in small animals. The NIR-II fluorescence imaging system achieved a lateral resolution of around 56 μm. This imaging platform enabled the analysis of liver microstructure changes, sO₂ levels, permeability, and metabolic dynamics. By integrating multiple imaging modalities, it facilitated the identification of distinct liver regions. Image processing was performed using MATLAB software to extract both structural and functional information. Key physiological relevance to cholestatic liver disease was analyzed, including hepatic lobule microstructure (e.g., vessel density and average vessel diameter), sO₂ levels, permeability, and metabolic dynamics (e.g., biliary excretion and liver function reserve). Collectively, these parameters provided a comprehensive assessment of bile duct injury and its impact on cholestasis.

The results of H&E and CK-19 immunohistochemical staining are shown in Fig. 1(b). Normal liver tissue exhibited an intact structure, with hepatic lobules arranged in a regular polygonal pattern. The bile ducts surrounding the portal vein were well-defined, and CK-19-positive signals were uniformly distributed and confined to bile duct epithelial cells, with no signs of bile duct hyperplasia or damage. Following induction of PBC using 2-octynoic acid-bovine serum albumin conjugate (2OA-BSA), H&E staining revealed marked T lymphocyte infiltration around the damaged bile ducts in the portal vein region. Some bile ducts exhibited atrophy, blurring, or complete disappearance, and a few epithelioid granulomas were observed in the liver parenchyma. As the disease progressed, T lymphocyte infiltration became more pronounced. CK-19 immunohistochemical staining showed significant bile duct hyperplasia, with an increased number of intrahepatic bile ducts confined to the portal regions. An acute extrahepatic cholestasis model was established using bile duct ligation (BDL). H&E staining demonstrated marked bile duct dilation and bile stasis, accompanied by extensive hepatocyte necrosis in the bile-stasis regions and inflammatory cell infiltration. CK-19 immunohistochemical staining further confirmed bile duct hyperplasia.

High-resolution PAM images [Fig. 2(a)] demonstrated that in a healthy liver, the hepatic lobule exhibited a polygonal shape in the transverse plane. Hepatic sinusoids radiate around the central vein, forming a well-defined vascular network due to the strong photoacoustic signals generated by hemoglobin. However, in PBC mice, the hepatic vascular network appeared less distinct, accompanied by abnormal dilation of larger blood vessels. Despite this, the overall structure of the hepatic sinusoids remained largely intact, suggesting that PBC primarily induces vascular dilation without severe destruction of hepatic sinusoids. In contrast, BDL mice exhibited more pronounced vascular structural disorganization, characterized by uneven photoacoustic signal distribution and void areas observed in certain portal vein regions. To validate these findings, Gaussian fitting was applied to the photoacoustic signal intensity profile along the white dashed line for typical large blood vessels shown in Fig. 2(a). Quantitative analysis [Fig. 2(b)] revealed that the diameters of large blood vessels in PBC and BDL mice were significantly larger than those in the control group and progressively increased with disease severity. Statistical analysis of the mean vessel diameter, vascular binarization [Fig. 2(c)], and vascular skeleton extraction [Fig. 2(d)] further confirmed a significant enlargement of mean vessel diameter in PBC and BDL mice (control: 15.1±0.8μm; PBC-12W: 24.6±3.2μm; PBC-24W: 38.5±7.5μm; BDL-1W: 26.1±3.8μm; BDL-2W: 37.6±6.0μm). Meanwhile, vascular area density (control: 61.6%±1.5%; PBC-12W: 38.7%±5.6%; PBC-24W: 25.4%±2.9%; BDL-1W: 52.8%±3.3%; BDL-2W: 43.6%±2.7%) and total vascular length (control: 37.7±0.8μm; PBC-12W: 23.3±1.4μm; PBC-24W: 18.2±1.8μm; BDL-1W: 31.4±1.5μm; BDL-2W: 25.2±1.0μm) showed a gradual decreasing trend, particularly in the PBC-24W and BDL-2W groups. To evaluate the degree of hypoxia in liver tissue, we mapped the spatial distribution of tissue sO₂ using dual-wavelength PAI at 532 nm and 559 nm combined with a spectral unmixing method [Fig. 2(e)]. Quantitative analysis [Fig. 2(i)] revealed a significant reduction in sO₂ levels in hepatic lobules of cholestatic mice compared to healthy mice, with a further decline as the disease progressed (control: 96.5%±0.9%; PBC-12W: 91.8%±0.5%; PBC-24W: 87.6%±1.1%; BDL-1W: 88.1%±2.0%; BDL-2W: 84.3%±1.6%). This phenomenon is attributed to the microvascular system serving as a key channel for oxygen and nutrient delivery to liver tissue. The destruction of the hepatic lobular microvascular structure severely impaired oxygen transport, exacerbating local tissue hypoxia as cholestasis progressed. This persistent hypoxic state not only reflected pathological changes in hepatic sinusoids but also highlighted the long-term damage to the hepatic vascular microenvironment caused by cholestasis.

In healthy liver tissue, the tight junctions of bile duct epithelial cells were highly intact, forming a robust barrier that prevented bile component reflux or exosmosis, thereby maintaining the structural and functional stability of the bile duct. However, in cholestatic liver disease, both PBC and BDL cause structural disruption of bile duct epithelial cells, impairing their barrier function. To evaluate these permeability changes, we applied high-resolution PAM combined with retrograde EB injection into the common bile duct, allowing us to dynamically monitor its distribution in PBC-12W and BDL-2W mice [Fig. 3(a)]. In the control group, EB signals were punctuated and regularly distributed, mainly concentrated around the portal vein, consistent with the anatomical localization of bile ducts in the portal vein region. Following injection, the EB signal disappeared rapidly, indicating that a healthy bile duct barrier effectively restricts EB molecules to the liver parenchyma and facilitates rapid excretion into the gallbladder. In contrast, the EB distribution pattern in PBC and BDL mice differed significantly from the control group, with aggregated signals indicating EB leakage from the damaged biliary barrier into the hepatic parenchyma. Over time, EB signal intensity increased and gradually diffused, forming a diffuse clump signal with a prolonged retention time compared to the control group, suggesting increased biliary permeability.

To further analyze the spatial distribution of EB penetration, photoacoustic images at different time points post-injection were processed by subtracting pre-injection images, generating differential photoacoustic images [Fig. 3(b)]. The differential images clearly illustrated the spatial distribution of EB signals and their dynamic changes over time. For quantitative analysis, the 1 min post-injection image was selected as the baseline, the differential images were normalized, and the change curves of the photoacoustic signal intensity enhancement [Fig. 3(c)] and relative diffusion area [Fig. 3(d)] were calculated over time. The results showed that in the control group, EB signals were confined to the capillary bile ducts, with the diffusion area and signal intensity rapidly decreasing. EB clearance was nearly completed within 30 min. In PBC mice, EB signals exhibited mild diffusion due to the loosening of bile duct epithelial tight junctions, decreasing after 10 min, with 39.2% residual signal intensity at 30 min. In BDL mice, EB penetration was more pronounced, with substantial permeation into the intercellular matrix. The photoacoustic signal diffusion area continued expanding 10 min post-injection, and the EB signal intensity was 1.5 times higher in BDL mice than in PBC mice at 30 min, indicating that BDL-induced biliary barrier disruption was more severe. These results suggest that EB penetration assays can effectively quantify bile duct epithelial tight junction integrity and assess changes in biliary barrier function, which are critical indicators of bile duct permeability.

Cholestasis is a chronic liver disease characterized by impaired bile flow, leading to bile acid accumulation and elevated systemic bile acid concentrations. After intravenous administration, ICG binds to plasma proteins, is selectively taken up by hepatocytes during the first pass, and is excreted directly into bile [29]. Thus, ICG transport serves as a reliable method to evaluate biliary excretion kinetics [30]. As shown in Figs. 4(a)–4(e), following intravenous tail injection of ICG, the NIR-II fluorescence signal in the liver increased rapidly, reached a peak, and then gradually decreased. In the control group, the fluorescence signal returned to baseline within approximately 90 min, indicating efficient bile excretion. In contrast, PBC and BDL mice exhibited significant fluorescence retention and delayed signal attenuation. Quantitative analysis of liver fluorescence signal intensity 420 min post-injection showed that the fluorescence intensity in the control group was near baseline, indicating unobstructed bile excretion. In contrast, the fluorescence signals in the PBC-12W and PBC-24W groups were 3.8% and 12.9%, respectively, significantly higher than those in the control group, reflecting impaired bile excretion caused by bile duct structural and functional damage. The fluorescence signals in the BDL group remained high at 420 min, at 51.9% and 74.9%, respectively, indicating severe cholestasis and ICG retention in liver tissue, directly caused by the physical blockade of bile excretion channels and increased biliary pressure.

Furthermore, we dynamically monitored ICG transport in the livers of healthy and PBC mice [Figs. 5(a)–5(d)]. Following intravenous injection of ICG, fluorescent signals appeared in the liver within 2 s, with prominent and continuous vascular signals visible at 4 s, at which point the fluorescence signal in the liver parenchyma was lower than that in the blood vessels. As hepatocytes took up intravascular ICG, the vascular fluorescence signal slowed, and some liver parenchyma began to show fluorescence. In the control group, the fluorescence signal in the liver parenchyma reached equilibrium with the peripheral vascular signal at 11 s, while in the PBC-12W group, equilibrium occurred at 18 s due to impaired hepatocyte uptake. In BDL-2W mice, dynamic fluorescence changes in the cholestatic region over time are shown in Fig. 5(e). Hepatocyte necrosis caused by biliary toxicity reduced ICG uptake in these regions, resulting in fluorescence signals lower than those in surrounding tissues 1 min post-injection. However, as ICG was excreted with bile, fluorescence signals in this region gradually increased, starting from the periphery and moving inward after approximately 1 h [Fig. 5(f)], while surrounding fluorescence signals gradually decreased. Fluorescence signals in the cholestatic region were significantly higher than in normal tissue 10 h post-injection. This indicated that although the cholestatic region lacked hepatocyte uptake function, bile produced in other liver regions flew into these areas, increasing the local cholestasis burden and expanding the accumulation area. Additionally, fluorescence images of BDL-1W and BDL-2W mice at 7 h post-injection were compared [Fig. 5(g)]. Cholestasis appeared as high-fluorescence spots on the liver, reflecting the distribution and severity of cholestasis. Due to common bile duct ligation, cholestasis regions of varying severity were observed throughout the liver. Quantitative analysis showed that the average area of a single cholestasis region in mice after 1 and 2 weeks of BDL was approximately (1.38±0.48)×10−2mm2 and (8.61±0.96)×10−2mm2, respectively [Fig. 5(h)]. Additionally, the cholestasis area density increased significantly in the BDL-2W group, reaching 2.55 times that of the BDL-1W mice [Fig. 5(i)]. These results demonstrate that as cholestasis worsens, the affected area becomes larger and denser, causing further liver damage and accelerating disease progression. NIR-II fluorescence imaging provides a robust visualization method for studying bile excretion pathways and their spatial and temporal distributions.

Cholestasis profoundly alters the liver metabolic function. As a clinical contrast agent, ICG serves as a critical indicator of liver function reserve. PACT, with its advantage of deep tissue imaging, enables macroscopic visualization of liver structures. In this study, we assessed the overall metabolic function of the liver in cholestasis using the PACT technique enhanced by ICG contrast. Liver structural profiles were obtained under 1064 nm laser irradiation, while the distribution of ICG was visualized under 760 nm laser irradiation. As illustrated in Fig. 6(a), photoacoustic signals from liver slices recorded after ICG injection provided data for liver function reserve assessment. The color-coded signals represented the time-dependent distribution of ICG in the liver. Liver function reserve curves [Fig. 6(b)] demonstrated that cholestasis significantly impairs liver function, particularly in terms of transport and excretion capacity. Quantitative analysis [Fig. 6(c)] revealed that the Tmax in the PBC group was markedly longer than that in the control group, with the BDL group showing an even greater prolongation (control group: 8.52±0.33min; PBC group: 10.58±0.71min; BDL group: 63.01±2.23min). Additionally, the T1/2 in the PBC group was slightly longer than in the control group (control: 20.14±1.09min; PBC: 26.96±3.01min), showing an exponential attenuation in the excretion process. However, the BDL group exhibited sustained high photoacoustic signals at 120 min post-injection, reflecting a severe impairment in bile excretion function due to biliary hypertension. These findings indicated that as cholestasis worsens, the liver’s capacity to uptake, transport, and excrete ICG and other substances is progressively diminished, leading to significant impairment of overall metabolic function. Noninvasive PACT provides an in situ tool for real-time monitoring of liver function reserve in cholestatic diseases.

Cholestasis, characterized by bile flow obstruction and bile acid accumulation in the liver caused by biliary obstruction, is generally classified into intrahepatic and extrahepatic cholestasis [31]. Extrahepatic cholestasis is characterized by lesions or obstructions in larger intrahepatic ducts. Intrahepatic cholestasis involves dysfunction in bile synthesis, secretion, and excretion, often due to lesions in smaller bile ducts or bile capillaries [32]. Current imaging techniques for diagnosing cholestatic diseases are limited in resolution and often focus only on larger bile ducts. The present study established and validated a cross-scale optical imaging platform to evaluate the structural and functional alterations in CLD. By integrating high-resolution PAM, PACT, and NIR-II fluorescence imaging, we provided a comprehensive, noninvasive method for assessing bile duct injury, hepatic microvascular changes, and biliary excretion dynamics.

In this study, two models of CLD were established. The first model used 2OA-BSA to induce PBC, serving as a representation of intrahepatic cholestasis. The second model of acute extrahepatic cholestasis was successfully established by using the BDL method, which is the standard model for studying extrahepatic obstructive cholestasis. Both H&E staining and CK-19 immunostaining strongly validated the successful establishment of PBC and BDL model of cholestasis, which were used as typical representatives of intrahepatic and extrahepatic cholestasis, respectively, for subsequent experiments [33–35].

Compared with conventional imaging modalities that primarily focus on large bile ducts, PAM enabled micrometer-scale visualization of hepatic lobular structures, including bile duct permeability and microvascular integrity [36,37]. EB, a commonly used azo dye with high affinity for plasma albumin, is frequently used as a tracer to evaluate the integrity of the blood-brain barrier in neuroscience studies. In this study, PAM effectively captured structural alterations in cholestatic models, revealing a reduction in microvessels in both PBC and BDL groups. The spatial distribution of sO₂ in lobular tissue reflected the progression of hypoxia during cholestasis, consistent with previous findings that liver injury leads to vascular disorganization and localized hypoxia early in the disease course [38,39]. In healthy mice, bile ducts exhibited intact tight junctions, maintaining bile excretion without leakage. However, in PBC and BDL models, PAM imaging revealed significant bile duct disruption, characterized by altered permeability and bile leakage. This was further supported by EB assays, which demonstrated prolonged dye retention in cholestatic liver tissue. These findings align with histological analyses, where CK-19 staining confirmed bile duct hyperplasia and structural alterations in diseased states.

NIR-II fluorescence imaging is a wide-field, high-resolution optical in vivo imaging method. Compared with visible light and the NIR-I window, tissue light absorption in the NIR-II window is significantly increased, reducing scattering background in two-dimensional planar imaging and minimizing autofluorescence, which collectively enhances imaging quality [40,41]. ICG, an FDA-approved imaging agent, is widely utilized in hepatobiliary diagnostics and therapeutics due to its liver-specific uptake and biliary excretion [42]. The bile duct serves as the channel for the excretion of endogenous bile and certain exogenous drugs, such as ICG. The transport of ICG from plasma to bile involves at least three steps: uptake across the hepatocyte sinusoidal (basolateral) membrane, intracellular passage, and excretion through the canalicular (apical) membrane [43]. NIR-II fluorescence imaging allowed real-time, whole-liver mapping of biliary excretion dynamics. In healthy livers, ICG fluorescence rapidly decreased, indicating efficient bile excretion. In contrast, PBC and BDL models exhibited prolonged fluorescence retention, with focal high-intensity regions corresponding to cholestatic lesions. Notably, BDL mice showed a significant increase in cholestatic area density, emphasizing the progressive impact of bile duct ligation on hepatic function.

Furthermore, PACT facilitated macroscopic, in situ quantification of liver function reserve. By tracking ICG clearance, we observed that cholestasis significantly delayed biliary excretion, with the BDL group exhibiting the most severe impairment. These results suggest that bile duct obstruction leads to progressive hepatic dysfunction and impaired metabolic clearance, key characteristics of CLD progression.

Our study has some limitations which must be considered. The penetration depth of PAM is restricted to superficial liver regions, limiting its ability to assess deeper bile ducts. Additionally, while ICG-based PACT enables functional assessment of biliary excretion, it cannot directly visualize bile duct structure without contrast agents. Furthermore, while our murine models replicate key aspects of cholestatic disease, translation to human applications requires further validation in preclinical and clinical settings.

In summary, the integration of photoacoustic and NIR-II fluorescence imaging modalities facilitates a comprehensive evaluation of both vascular architecture and bile duct function within the liver. This multimodal platform evaluates the integrity of the interlobular bile duct barrier and liver function reserve and enables in vivo assessment of hepatic vascular structures and bile duct functionality. It allows for dynamic monitoring of hepatobiliary excretion, providing spatial visualization of cholestatic regions within the liver. The integrated imaging offers critical insights into the pathophysiology of cholestasis and underscores the potential of cross-scale optical imaging for therapeutic evaluation and mechanistic investigation in preclinical settings.