Early identification of umbilical blood flow restriction and maternal placental hypoperfusion with photoacoustic imaging

Abstract

In clinical practice, the prompt and accurate identification of acute fetal distress (AFD) is critical. Although cardiotocography and ultrasonography are the cornerstone clinical tools for fetal monitoring, they cannot quantitatively assess fetal cerebral hypoxia and carry inherent risks of underdiagnosis or false-positive interpretations. This study evaluates photoacoustic imaging (PAI) for early AFD detection. In mouse models, oxygen saturation (sO₂) in the fetal brain and placenta was measured using PAI, demonstrating a significant sO₂ decrease following placental blood flow obstruction - with more pronounced reductions observed in complete versus partial restriction cases. Crucially, a novel two-step PAI approach differentiated placental hypoperfusion from umbilical cord obstruction by analyzing distinct sO₂ patterns in both placenta and fetal brain tissues. This distinction is clinically vital, as placental and cord-related AFD require different urgent interventions. PAI’s ability to pinpoint the underlying cause highlights its potential for guiding precise treatment decisions.

1. Introduction

Acute fetal distress (AFD) presents a significant risk for neonatal hypoxic-ischemic encephalopathy, and can lead to permanent disabilities in children [1], [2]. Cardiotocography (CTG) is widely utilized to monitor changes in fetal heart rate and uterine contractions to assess fetal hypoxia, representing the conventional clinical practice for identifying AFD [3], [4]. While CTG has contributed to a decrease in neonatal mortality, it is associated with a false positive diagnosis, leading to unnecessary cesarean sections [5], [6], [7], [8], [9]. Diagnosing umbilical cord blood flow restriction resulting from conditions such as umbilical cord prolapse, especially occult prolapse presents considerable challenges [8], [9], [10], [11]. While ultrasound can detect umbilical cord abnormalities, it fails to directly assess fetal oxygenation and depends heavily on the operator’s skill [12]. Therefore, it is crucial to establish a protocol for the early and accurate diagnosis of AFD.

Photoacoustic imaging (PAI) is an emerging technology for accurately measuring tissue oxygen saturation (sO₂) by differentiating between oxygenated and deoxygenated hemoglobin concentrations [13], [14], [15]. Research conducted by Lawrence et al. and Liliya et al. has demonstrated that PAI can perform in vivo evaluations of placental oxygenation in a preeclamptic mouse model [16], [17], [18]. Tianqi et al. utilized PAI to show a significant decrease in blood flow to the fetal brain following acute prenatal ethanol exposure [19], [20]. These investigations highlight PAI’s substantial promise in the diagnosis of AFD. However, how to rapidly diagnose AFD with PAI has not yet been fully investigated.

In this study, we first investigate whether PAI can accurately detect the changes of sO₂ in the placenta and fetal brain at an early stage, when maternal blood flow is partially or totally obstructed. Subsequently, we examine whether PAI can localize abnormal fetal blood flow caused by blockages in maternal or umbilical cord circulation (Fig. 1). The purpose of this study was to provide a diagnostic protocol for rapid and accurate determination of AFD with PAI, thus offering an experimental basis for its broader clinical application.

Fig. 1.

Schematic representation of this study.

2. Materials and methods

2.1. Animal model and protocol

The study was approved by the Institutional Animal Care and Use Committee of Guangzhou Medical University (GY2021–151). The experimental procedure is divided into two parts (Fig. 2). Pregnant mice at gestational day (GD) > 15 were used for all imaging procedures. Mice were anesthetized with 5 % isoflurane for induction prior to surgical procedures. Protocol 1 aimed to assess the precision of PAI in detecting sO₂ changes within the placenta and fetal brain during the early stages (Fig. 2A). Eighteen mice were randomly divided into partial flow restriction (PFR), total flow restriction (TFR), and control groups. Throughout the procedures, anesthesia was maintained with 1.5 % isoflurane at the lowest effective dose to reduce fetal hemodynamic depression, as studies have shown stable fetal heart rate and blood pressure under this regimen [21], [22], [23]. The PFR group underwent uterine artery ligation while retaining partial placental blood flow via the ovarian artery. In contrast, the TFR group experienced complete ligation of all placenta supplying vessels [24], [25]. The control group underwent a sham operation without any surgical intervention following uterine exposure. The levels of sO₂ in the fetal brain and placenta were determined with PAI at the specific time until 120 min post-ligation. To corroborate the presence of hypoxic conditions, hematoxylin and eosin (H&E) staining was conducted.

Fig. 2.

The workflow of this study. (A) The PFR and TFR models were established. sO₂ in the fetal brain and placenta were determined with PAI at the specific time until 120 min post-ligation. H&E staining was performed to detect the signs of hypoxia. (B) The umbilical flow restriction and placental hypoperfusion models were established. sO₂ in the placenta and fetal brain were analyzed concurrently within 20 min following blood restriction. (C-F) Schematic of ligation site.

Protocol 2 sought to evaluate the ability of PAI to pinpoint restricted blood flow by selectively blocking umbilical or maternal blood circulation (Fig. 2B). Twelve mice were randomized into two groups: umbilical flow restriction and placental hypoperfusion. Mice were anesthetized with isoflurane (1 %–2 %), placed in a supine position on a thermostatic platform maintained at 37 °C, and continuously monitored for physiological parameters throughout the experiment. In the placental blood flow obstruction group, inadequate placental perfusion was simulated by dual ligation of the maternal blood vessels supplying a single placenta—specifically, the branches of the uterine arteries entering the placenta. In the umbilical cord restriction group, a 5-mm longitudinal incision was made in the uterine wall corresponding to the target embryo. The amniotic sac was carefully exposed, and the umbilical cord was gently retracted. The cord, which contains three vessels embedded within Wharton’s jelly, was either clamped with vascular forceps or tightly ligated with a 10–0 nylon suture to achieve blood flow restriction. PAI was performed at designated time points: before intervention (0 min) and at 2, 4, 6, 8, 10, and 20 min after intervention. The specific vascular ligation sites for Protocols 1 and 2 are depicted in Fig. 2C-F.

A total of 54 mice were included in the experimental procedures. For Protocol 1, the vast majority of surgeries were successfully completed, with 5 deaths due to anesthesia-related accidents, resulting in an exclusion rate of approximately 9.3 %. The final sample sizes included in the statistical analysis were as follows: Eighteen mice for placental imaging (6 each for the control, placental hypoperfusion, and umbilical blood flow restriction groups), and 18 mice for fetal brain imaging (6 each for the control, placental hypoperfusion, and umbilical blood flow restriction groups). No animal deaths occurred in Protocol 2. The final sample sizes included in the statistical analysis were: Six pregnant mice (corresponding to 6 fetuses) in the placental hypoperfusion group, six pregnant mice (6 fetuses) in the umbilical occlusion group, and 6 pregnant mice (6 fetuses) in the umbilical ligation group.

2.2. Two-dimensional ultrasonic and PAI of fetal brain and placenta

PAI was employed to investigate the temporal variations of sO₂ in the fetal mouse brain and placenta synchronously. These changes were analyzed using the Vevo 3100 Imaging System (FUJIFILM Visual Sonics, Toronto, Ontario, Canada). The system provides 50 μm axial and 110 μm transverse resolution with an imaging depth of 0–3 cm. Based on an Nd:YAG laser-pumped optical parametric oscillator (OPO) and second harmonic generator, it achieves multi-wavelength laser output (20 Hz pulse frequency) with 1 nm step size in the 680–970 nm range and < 0.4 s wavelength tuning speed for functional imaging. The peak laser energy is 26 mJ at 680 nm and 30 mJ at 970 nm, with energy density controlled below international laser safety standards.

The photoacoustic (PA) signal intensity was linearly related to the sum of oxyhemoglobin (HbO₂) and deoxyhemoglobin (Hb) concentrations, expressed as: PA ∝ (c_HbO₂ + c_Hb). This relationship indicates that at specific laser wavelengths, the PA signal intensity reflects the total hemoglobin concentration, forming the foundation for quantitative sO₂ measurements. sO₂ was calculated using the formula: sO₂ = c_HbO₂ / (c_HbO₂ + c_Hb) by differentiating the absorption characteristics of hemoglobin species at multiple wavelengths.

Utilizing PA/ultrasound dual-modal imaging at 750 nm and 850 nm, tissue sO₂ was evaluated by analyzing the concentrations of oxygenated and deoxygenated hemoglobin. The core assumptions of this algorithm are that hemoglobin serves as the primary absorber in tissue within this spectral range, and that the optical properties of the tissue (absorption and scattering) remain relatively stable during measurement. Actually, tissue sO₂ may be lower than that in blood vessels. In vivo PAI studies [26], [27] published by other research teams indicate similar sO₂ levels within tissues.

sO₂ maps were generated by combining ultrasound grayscale and PA images. The areas of interest within fetal brain horizontal sections and placental sagittal planes were delineated based on the ultrasound images, concurrent with the acquisition of tissue sO₂ levels using PAI. Offline quantitative analysis was performed using Vevo Lab Software 3.2.0 (FUJIFILM Visual Sonics).

In Protocol 1, the variations of sO₂ within the fetal mouse brain and placenta were determined prior to ligation (at 0 min) and subsequently at 4, 10, 20, 30, 60, and 120 min post-ligation. In Protocol 2, the sO₂ in the fetal mouse brain and placenta before (at 0 min) and at 2, 4, 6, 8, 10, 15 and 20 min following the ligation were examined. During the imaging procedures, mice were anesthetized with inhaled isoflurane at a concentration of 1 %-2 %.

2.3. Histopathological analysis

The animals were euthanized under anesthesia, and specimens of the fetal brain and placenta were collected for pathological analysis. These specimens were fixed, dehydrated, cleared and embedded before sectioning. After dewaxing and rehydration, the sections were stained with H&E staining and immunohistochemistry (IHC). IHC was performed to detect hypoxia markers Hypoxia inducible factor 1 subunit alpha (HIF-1α) and Vascular endothelial growth factor (VEGF) [28]. The specific procedures were as follows: Placental tissue and fetal brain samples were collected from the AFD model group and the normal control group, fixed in 4 % paraformaldehyde, embedded in paraffin, and sectioned into 4-μm-thick slices. Antigen retrieval was subsequently carried out, followed by overnight incubation at 4°C with primary antibodies (including anti-HIF-1α and anti-VEGF antibodies). The next day, Horseradish peroxidase (HRP)-conjugated secondary antibodies were applied, followed by Diaminobenzidine (DAB) development, hematoxylin counterstaining, and mounting with neutral gum. The expression levels of HIF-1α and VEGF were observed under a microscope and quantitatively analyzed using an image analysis system to determine the percentage of positive area and the mean optical density in all groups. A skilled pathologist, blinded to the study protocol, evaluated the pathological alterations.

2.4. The blood gas analysis procedure

First, a mouse model is used to establish placental hypoxia models with varying degrees of blood flow restriction and different durations. Subsequently, immediately after PAI, the placenta is collected for blood sampling and blood gas analysis to ensure stable blood sO₂. A portable blood gas analyzer is then used to measure blood sO₂ parameters. Finally, the blood gas data from each model are recorded, and mean values along with standard deviations are calculated for comparative analysis with the PAI results [29].

2.5. Statistical analysis

Statistical analysis was conducted using IBM SPSS Statistics (26.0) software. Differences in sO₂ among subgroups (Control, PFR, TFR) were evaluated using the Kruskal-Wallis test. Comparisons of sO₂ between the hippocampus and the entire brain, as well as between the maternal side of the placenta and the entire placenta, were performed using the Wilcoxon signed-rank sum test. For the placental hypoperfusion and fetal umbilical flow constriction, differences in sO₂ between the placenta and fetal brain were examined with the Wilcoxon signed-rank sum test. A significance level of P < 0.05 was considered statistically significant.

2.6. ROI definition and blinding design

Given the anatomical heterogeneity of the placenta and fetus as well as individual differences, an experienced operator manually delineated the regions of interest (ROIs) based on anatomical landmarks visible in the ultrasound images. The ROIs were carefully outlined along structurally discernible boundaries of the placental tissue and fetal brain.

To ensure analytical objectivity, a triple-blind design was implemented throughout the study: the operator was blinded to group assignments and experimental objectives; all images were saved with random codes and stripped of group information; and statisticians independently analyzed only de-identified data to minimize human bias.

3. Results

3.1. Progressive sO₂ decrease in the fetal brain during ischemia

sO₂ data for PFR, TFR, and control groups were obtained before blood flow restriction (at 0 min) and at subsequent intervals of 4, 10, 20, 30, 60, and 120 min post-restriction. Map of sO₂ demonstrated a significant decrease in PFR and TFR groups over time, contrasting with the control group which maintained stable sO₂ levels (Fig. 3A). Quantitative assessment of sO₂ in the fetal brain confirmed these observations (Fig. 3B). Initially, there were no significant differences in sO₂ among the three groups (52.07 ± 2.33, 52.13 ± 2.48, and 50.97 ± 1.85, respectively; P > 0.05). However, at the 4-minute post-restriction, significant disparities emerged (control: 52.07 ± 1.71; PFR: 47.60 ± 2.48; TFR: 44.56 ± 2.22; P < 0.05). Both imaging and quantitative data revealed a more rapid and marked reduction in sO₂ for the TFR group compared to the PFR group. This decline was most pronounced within the first 10 min, plateauing thereafter until 120 min. The specific measured sO₂ Average and sO₂ Average Total values are provided in detail in Table S1. Histopathological analysis indicated that brain tissue edema became increasingly evident 10, 30, and 120 min after ligation of vessels in both PFR and TFR groups as shown in Fig. S1. In the TFR group, edema was more pronounced, accompanied by hippocampal necrosis after 30 min post-ligation.

Fig. 3.

Gradual decrease in fetal brain sO₂ due to placental blood flow restriction. (A) Representative ultrasound (US) and PA images of fetal brain from Control, PFR, and TFR groups. Images exhibited before (0 min) and at 10, 30, and 120 min following the onset of blood flow restriction. The delineated regions of interest according to ultrasound images are indicated by yellow dashed lines. (B) Line graphs illustrating the temporal changes in sO₂ within the fetal mouse brain before (0 min) and following blood flow restriction at intervals of 4, 10, 20, 30, 60, and 120 min. Compared to 0 min, *P < 0.05. Compared to control group, ^#P < 0.05. Scale bars represent 1 mm.

3.2. The hippocampus in the fetal brain is vulnerable to ischemia

Figs. 4A and 4B depicted two-dimensional grayscale images of the mouse fetal brain, with the hippocampus indicated by arrows. The hippocampal identification on these images was confirmed by the pathological specimens presented in Figs. 4C and D, where the hippocampus was circumscribed by yellow circles. PAI revealed no significant differences in sO₂ levels between the hippocampus and adjacent brain prior to blood flow restriction (Figs. 4E and F). Both the PFR and TFR groups exhibited significantly lower sO₂ levels in the hippocampus compared to other regions of the fetal brain at 10, 30, and 120 min post-blood flow restriction (Fig. 4G-L). This observation was corroborated by quantitative analysis of PAI, as demonstrated in Figs. 4M and 4N. Before surgery, there was no significant statistical difference between the hippocampus and the whole brain, with a difference of 1.41 ± 1.64 in the PFR group and −0.32 ± 1.41 in the TFR group. Postoperatively, the hippocampal sO₂ decreased more rapidly and exhibited lower levels compared to the overall brain. A significant increase in the disparity of sO₂ between the hippocampus and the whole brain was observed in both groups (P < 0.05). Specifically, the mean discrepancies in the PFR group were 5.27 ± 0.74, 4.48 ± 0.90, and 6.64 ± 1.67 at 10, 30, and 120 min respectively (Fig. 4O). Meanwhile, the TFR group demonstrated mean discrepancies of 3.42 ± 0.83, 3.08 ± 0.39, and 4.29 ± 1.74 at corresponding intervals (Fig. 4P).

Fig. 4.

Differential sO₂ levels between the hippocampus and whole fetal brain in PFR and TFR groups. (A, B) Ultrasonographic images of the fetal brain in PFR and TFR groups with arrows highlighting the hippocampus. (C, D) Histopathological analysis shows hippocampus edema and necrosis at 120 min post-ligation signs of hypoxia. (E-L) Representative sO₂ maps of brain before blood flow restriction (0 min), and at 10, 30, and 120 min following vascular ligation in both PFR and TFR groups. Hippocampus regions and the extent of the whole brain are demarcated by yellow dashed-line circles. (M, N) The quantitative analyses for the PFR and TFR groups at the indicated time points. Compared to the whole brain group, *P < 0.05. (O, P) The differential in sO₂ between the hippocampus and the overall brain in both PFR and TFR group. Compared to 0 min, ^#P < 0.05. Scale bars represent 1 mm.

3.3. Progressive decline in sO₂ during placental blood flow obstruction

sO₂ levels decreased gradually over time in response to placental blood flow obstruction in both the PFR and TFR groups at 0, 4, 10, 20, 30, 60, and 120 min post-ligation. In contrast, the control group exhibited minimal fluctuation, as shown in Fig. 5A. Quantitative analysis revealed relatively stable sO₂ values (Fig. 5B) in the control group, ranging from 58.12 ± 1.31–57.01 ± 1.26. Conversely, both the PFR and TFR groups experienced significant decreases, ranging from 57.16 ± 1.02–49.17 ± 2.11 and from 57.78 ± 1.25–43.52 ± 1.94, respectively. Notably, sO₂ levels in the PFR and TFR groups declined sharply within 10 min post-ligation and then remained relatively stable until 120 min, with a more pronounced decrement observed in the TFR group compared to the PFR group. The specific measured sO₂ Average and sO₂ Average Total values are provided in detail in Table S1. Additionally, H&E staining of placenta revealed an accumulation of edema over time in both groups (Fig. S2).

Fig. 5.

The temporal variations in placental sO₂ in response to varying extents of blood flow restriction. (A) Ultrasound (US) and photoacoustic images of mouse placenta for three groups: control, PFR and TFR, before blood flow restriction (0 min) and at 10 min, 30 min, and 120 min post-restriction. (B) Line graphs show sO₂ variations in mouse placenta at the specified time. Compared to 0 min, *P < 0.05. Compared to control group, ^#P < 0.05. Scale bars represent 1 mm.

3.4. The maternal side of the placenta is more susceptible to blood flow obstruction

Fig. 6A-B illustrated two-dimensional grayscale images of placenta, with arrows highlighting the maternal side. The histologic structure of the placenta was further confirmed by H&E staining (Fig. 6C-D). Before ligation, the maternal side of the placenta exhibited the highest sO₂ levels within the placenta (Fig. 6E-F). Following ligation, both the PFR and TFR groups experienced a notable decline in sO₂ levels across the entire placenta, with a more rapid decline observed on the maternal side at 10, 30, and 120-minute intervals (Fig. 6G-L). Quantitative analysis of PAI results corroborated these observations (Fig. 6M-N). Prior to surgery, sO₂ levels on the maternal side of the placenta were significantly higher than those of the entire placenta in both the PFR and TFR groups, with differences of 4.29 ± 2.69 in PFR and 7.54 ± 3.61 in TFR group. However, the sharper decline in sO₂ on the maternal side after surgery led to the elimination of significant sO₂ discrepancies between the maternal side and the rest of the placenta. The differences in sO₂ between the maternal side and the entire placenta at 10, 30, and 120 min after surgery were −3.73 ± 8.07−2.78 ± 6.84, and −3.98 ± 6.27 for the PFR group, and −2.26 ± 5.49−3.20 ± 5.47, and −5.02 ± 4.74 for the TFR group (Fig. 6O-P).

Fig. 6.

Variations in sO₂ between the Maternal Side and the Entire Placenta. (A, B) Representative the ultrasound images of placenta in PFR group and TFR group, demarcated by a green dashed line. (C, D) H&E staining suggests the edema of placenta and histopathologic confirmation of the morphology of the maternal side of the placenta. (E-L) sO₂ maps of placental pre-blood flow restriction (0 min), and at 10, 30 and 120-minutes post-restriction in the PFR and TFR group. (M, N) The quantitative analyses for the PFR and TFR groups at the indicated time points. Compared to the whole placenta group, *P < 0.05. (O, P) The differential in sO₂ between the maternal side and the overall placenta in both PFR and TFR groups. Arrows highlighting the maternal side of placenta. Compared to 0 min, ^#P < 0.05. Scale bars represent 1 mm.

3.5. Detection of abnormal blood flow sites by PAI

To assess the effectiveness of PAI in identifying abnormal blood flow locations, AFD models of fetal umbilical flow constriction and placental hypoperfusion were established. In the umbilical flow constriction mice, placenta and fetal brain were delineated based on two-dimensional ultrasound images (Fig. 7A, G and M). The levels of sO₂ decreased only in the fetal brain rather than in the placenta (Fig. 7H-L, N-R). Correspondingly, ultrasound imaging (USI) of the placental hypoperfusion model identified both placenta and fetal brain (Fig. 7 S), with sO2 maps indicating a decrease in both tissues (Fig. 7T-X).

Fig. 7.

Changes in fetal brain and placental blood sO₂ under conditions of restricted umbilical cord blood flow and placental hypoperfusion. (A–F): Anesthesia sham surgery control group; (G–L): Umbilical cord blood flow restriction model established using vascular clamping; (M–R): Blood flow restriction model established using umbilical cord ligation; (S–X): Maternal placental hypoperfusion model. Each panel displays representative ultrasound and photoacoustic imaging at specific time points post-ligation. Scale bar: 1 mm.

Quantitative analysis revealed a decline in brain sO₂ from 52.70 ± 2.03–39.97 ± 3.25 (occlusion group), 48.57 ± 2.66–39.09 ± 1.86 (ligation group) whereas trivially altered placental sO₂ from 57.79 ± 1.74–55.75 ± 1.05 (occlusion group), 55.36 ± 3.16–54.60 ± 2.24 (ligation group) in the umbilical flow restriction group 20 min post-surgery (Fig. 8A, C). The placental hypoperfusion model, however, showed a concomitant reduction of sO₂ levels within both brain and placenta, with measurements dropping from 57.36 ± 2.49–46.19 ± 3.63 in brain, and from 51.12 ± 3.39–41.93 ± 3.75 in placenta, respectively, 20 min post-surgery (Fig. 8B). Notably, the sO₂ discrepancy between placenta and fetal brain in the umbilical flow restriction group widened from 5.09 ± 1.55–15.78 ± 3.79 (occlusion group), 6.49 ± 2.94–15.12 ± 2.86 (ligation group). However, in the placental hypoperfusion scenario, the disparity exhibited minor variation, ranging from 6.24 ± 2.70–4.27 ± 4.39 (Fig. 8E).

Fig. 8.

Quantitative results of blood sO₂ over time. (A) sO₂ change curve during umbilical blood flow restriction by vascular clamping. (B) sO₂ change curve during umbilical blood flow restriction by suture ligation. (C) sO₂ change curve in the maternal placental hypoperfusion model. (D) sO₂ change curve in the sham-operated control group. (E) Differential analysis of sO₂ changes over time in placental and fetal brain tissues.Data are expressed as mean ± standard deviation.

In this study, 1.5 % isoflurane anesthesia was employed. References [30], [31], [32] indicate that this dosage exerts minimal effects on placental and cerebral blood flow in pregnant rats. To exclude potential interference of anesthesia on SO₂ distribution, we designed a sham surgery comparison experiment following isoflurane anesthesia. Results indicate that SO₂ levels in placental and fetal brain tissues exhibited stable trends under anesthesia (Figs. 7A-F and 8D). No significant fluctuations in the SO₂ difference were observed between placental and fetal brain tissues in the control group (Fig. 8E). These findings exclude blood flow redistribution as a major confounding factor. The consistent anesthesia protocol employed in this study ensures the reproducibility and reliability of experimental data.

3.6. Detecting the hypoxic state of placental tissue and brain tissue by IHC and blood gas analysis

The hypoxic state of placental tissue and brain tissue was further validated by IHC staining. Both the PFR and TFR groups showed significantly increased expression of VEGF and HIF-1α compared to the normal control group (Fig. 9A). In placental tissues, the mean optical density of VEGF, a hypoxia-related marker, was significantly elevated in both the PFR group (45.26 ± 6.47) and the TFR group (54.89 ± 4.44) compared to the normal control group (29.19 ± 6.76) in the Fig. 9B. Similarly, the VEGF-positive area percentage was increased in the PFR group (44.59 ± 2.62 %) and the TFR group (47.98 ± 5.88 %) relative to the control group (30.42 ± 3.79 %) in the Fig. 9C. The mean optical density of HIF-1α was also higher in the PFR group (47.67 ± 7.16) and the TFR group (49.95 ± 12.59) than in the normal group (20.78 ± 2.24), accompanied by increases in the positive area percentage to 42.06 ± 3.70 % and 42.63 ± 5.33 %, respectively, compared to 21.18 ± 3.56 % in the controls (Fig. 9D, E).

Fig. 9.

Immunohistochemical analysis of VEGF and HIF-1α in placental tissue. (A) Representative IHC images. (B, D) Quantitative results: (B) Mean gray value of VEGF and (D) Mean gray value of HIF-1α. (C, E) Quantitative results: (C) Positive area (%) of VEGF and (E) Positive area (%) of HIF-1α.

In brain tissue, IHC staining also revealed an increase in the expression levels of VEGF and HIF-1α (Fig. 10A). In the brain and hippocampal regions, VEGF mean optical density was 42.92 ± 13.57 in the PFR group and 46.77 ± 5.44 in the TFR group, both higher than the control group (36.11 ± 12.36) (Fig. 10B). The positive area percentage for VEGF was elevated in the TFR group (66.01 ± 5.63 %) compared to controls (34.42 ± 12.23 %), while the PFR group showed intermediate values (45.29 ± 9.85 %) (Fig. 10C). For HIF-1α, both mean optical density and positive area percentage showed minimal variation among groups in brain and hippocampal regions, with values of 25.40 ± 6.86 (PFR), 27.25 ± 4.99 (TFR), and 23.98 ± 6.04 (control) for mean optical density, and 27.74 ± 8.29 % (PFR), 30.82 ± 2.29 % (TFR), and 25.95 ± 6.79 % (control) for positive area percentage (Fig. 10D, E).

Fig. 10.

Immunohistochemical analysis of VEGF and HIF-1α in brain tissue. (A) Representative IHC images. (B–C) Quantitative results of VEGF: (B) Positive area (%) and (C) Mean gray value. (D–E) Quantitative results of HIF-1α: (D) Positive area (%) and (E) Mean gray value.

To validate the quantitative accuracy of sO₂ measurements via PAI [33], we concurrently conducted blood gas analysis experiments to detect blood sO₂ concentrations in mice at 0, 10, and 30 min (under PFR and TFR treatments) (Fig. S3A). Concurrently, PAI-derived sO₂ imaging data were acquired at corresponding time points (Fig. S3B), and PAI-measured sO₂ levels in blood were quantitatively analyzed (Fig. S3C). Results demonstrated a significant linear correlation between PAI-measured sO₂ and blood gas analysis values (r = 0.7704, P = 0.0008) (Fig. S3D), indicating that PAI serves as a quantitative hypoxia indicator in this model and validating the accuracy of PAI-derived SO₂ values.

3.7. More results for validating the potential of PAI

Dual-modality ultrasound and PAI of the placenta was performed on six subjects under both invasive (direct uterine surface imaging) and non-invasive (transabdominal imaging) conditions (Fig. S4A). Statistical analysis of the imaging results showed no significant differences between the two groups (Fig. S4B), indicating that invasive manipulation did not compromise measurement reliability. (2) To address the disparity between the imaging depth in mouse models (≤10 mm) and human placental depth, chicken breast tissue was superimposed to extend the imaging depth to 21.5 mm and 27 mm. At these depths, the hyperoxic pattern on the maternal side of the placenta remained clearly discernible (Fig. S5A), with no significant differences observed in sO₂ values compared to the superficial state (Fig. S5B, p > 0.05). (3) To simulate a baseline-free diagnostic scenario, we conducted baseline sO₂ measurements on 30 mice. Fig. S6A visually demonstrates the hypoxic status of each group, while Fig. S6B confirms that all sO₂ values follow a normal distribution. Quantitative analysis (Fig. S6C) reveals a 95 % confidence interval ranging from 57.66 % to 60.97 %. Six hypoxic mice (marked in red) exhibited sO₂ values outside this range, successfully simulating the identification of hypoxic conditions under baseline-free diagnostic circumstances; (4) Imaging of 1.2 mm copper wires at depths of 15, 19, 21.5, and 27 mm showed clear structural visualization at all depths (Fig. S7); (5) The blue dashed lines in Fig. S8A indicate the analysis path for signal-to-noise ratio evaluation. Further analysis demonstrates that the PA signals from the copper wire at three depth positions (1_1, 1_2, and 1_3) all exhibit high signal-to-noise ratios. The gray value distribution curves corresponding to the red line position clearly delineate the copper wire's boundaries. While the signal-to-noise ratio decreases with increasing imaging depth (position 1_4, Fig. S8B), the copper wire's imaging width maintains excellent resolution quality. This indicates that the system maintains stable and reliable detection capability even in deeper tissue regions.

Comparative analysis revealed that PAI and Doppler spectroscopy yielded highly consistent findings. The control group showed no significant changes, while both blood flow restriction groups exhibited decreased sO₂, with a more pronounced effect under complete restriction (Fig. S9, Table S1). Corresponding Doppler spectra showed hemodynamic signs of hypoxia (e.g., decreased PSV, increased RI) (Fig. S10, Table S1), which appeared earlier in the complete restriction group. Integrating these modalities provides a comprehensive diagnosis of fetal hypoxia from both hemodynamic and sO₂ perspectives, improving accuracy and mutual validation.

While spectral Doppler revealed non-specific hemodynamic deterioration (decreased PSV/RI leading to flow disappearance) in both umbilical cord compression and placental hypoperfusion groups, it failed to differentiate between them (Fig. S11, Table S2). In contrast, PAI distinguished the two conditions within 4 min based on unique sO₂ maps, enabling early and specific diagnosis (Fig. 7, Fig. 8).

4. Discussion

AFD represents a critical obstetric emergency characterized by sudden-onset fetal hypoxia that requires immediate intervention to prevent severe complications including hypoxic-ischemic encephalopathy or fetal demise. While fetal heart monitoring and USI remain the most widely used diagnostic modalities for AFD, these conventional methods exhibit significant limitations in detecting umbilical cord compression and quantifying the severity of fetal hypoxia [34], [35]. This study introduces PAI as a transformative diagnostic modality that overcomes these limitations through real-time, quantitative mapping of tissue oxygenation in both fetal and placental tissues.

Through controlled animal experiments, this study has demonstrated PAI's unique capability to precisely monitor dynamic sO₂ changes in fetal brain and placental tissues during induced hypoxic events (Fig. 3, Fig. 5). Several key findings emerged from this investigation: First, and most notably, our study provides the first experimental evidence using PAI to demonstrate that the hippocampus is the most severely affected brain region during fetal hypoxia [36]. This pioneering finding showed the fetal hippocampus exhibited the most pronounced sO₂ decline during hypoxic episodes (Fig. 4), establishing it as a highly sensitive early biomarker for AFD. Second, the maternal side of the placenta, while normally maintaining the highest oxygenation levels within the structure, exhibited the most significant sO₂ reduction during hypoperfusion events (Fig. 6).

Critically, the PAI data in this study reveal two distinct pathophysiological patterns: placental insufficiency manifests as synchronized declines in oxygenation levels in both the placenta and fetal brain, indicating global oxygen exchange dysfunction, while umbilical flow restriction demonstrates a characteristic dissociation pattern-maintained placental oxygenation with acutely deteriorating fetal oxygenation. These differentiable characteristic patterns provide a basis for precise etiological diagnosis and guide targeted interventions: urgent delivery is required for placental hypoperfusion, whereas for umbilical cord compression, obstruction relief through maternal repositioning or amnioinfusion should be implemented while simultaneously preparing for cesarean section.

These physiological insights enabled the development of a novel two-step diagnostic algorithm: (1) initial detection of hypoxia through hippocampal sO₂ assessment, followed by (2) differentiation between umbilical cord compression (characterized by decreased brain sO₂ with preserved placental sO₂) and maternal circulatory insufficiency (manifested by concurrent decreases in both brain and placental sO₂) (Fig. 7, Fig. 8). This etiological distinction is clinically paramount as it guides fundamentally different management strategies.

Based on literature reports regarding oxygenation distribution and photoacoustic signal characteristics across placental regions [17], we analyzed the placental photoacoustic oxygen distribution map (Fig S12A). The image clearly reveals the oxygenation status in the decidual zone (D), junctional zone (JZ), and placental labyrinth zone (L), confirming the presence of two typical highly oxygenated regions in the normal placenta: the labyrinth zone and the maternal surface (Fig. S12). These findings are highly consistent with those reported by Yamaleyeva et al. (2017, FASEB J). Quantitative results (Fig. S12B) further demonstrated extremely significant differences (P < 0.0001) in photoacoustic sO₂ values among the regions. In our experiments, oxygenation changes on the placental maternal surface were particularly pronounced, therefore we focused on investigating the alterations in sO₂ in this specific region. Additionally, placental blood oxygenation on the maternal side is influenced by microcirculation and metabolic oxygen consumption, resulting in local oxygen distribution that differs from arterial blood oxygenation. This phenomenon has been documented in relevant literature [37], [38], [39].

Based on Fig. S13, hypoxia reduced sO₂ in both placental and fetal brain tissues. However, while placental HbT avr showed no clear trend, fetal brain HbT avr significantly increased in PFR and TFR groups, indicating a hypoxia-induced compensatory rise in hemoglobin concentration to maintain oxygen supply. This brain-specific response highlights a protective mechanism and offers a potential imaging marker for early detection of fetal cerebral hypoxia.

A significant systematic bias must be considered: a 30 % change in the directly measured SO₂ corresponds to only an approximately 18 % change in the PAI SO₂. We acknowledge that in vivo PAI SO₂ measurement exhibits systematic bias and limited absolute quantitative accuracy, primarily due to tissue complexity. Therefore, our data are interpreted as semi-quantitative. Their core value lies in effectively revealing relative trends, differences between groups, and variations across different tissues, thereby serving dynamic physiological assessment.

The clinical translation of PAI shows strong potential, particularly regarding system miniaturization and practical implementation. A 3–7 MHz flexible array transducer could optimally balance imaging depth and resolution while maintaining compatibility with existing obstetric ultrasound workflows. Based on commercially available hardware, the system remains cost-effective for resource-limited settings. The transabdominal imaging approach aligns with conventional ultrasound protocols, and real-time sO₂ algorithms are expected to enhance diagnostic efficiency. However, further validation is needed regarding probe stability across varying patient morphologies and gestational stages, along with enhanced operator training and multicenter clinical trials. While demonstrating solid foundations for clinical adoption, widespread application will require careful consideration of equipment performance, operational standards, and clinical needs.

The PAI/USI hybrid imaging system represents a significant technological advancement by synergistically combining the functional oxygenation data provided by PAI with the superior anatomical resolution of USI [40]. This dual-modality approach enables simultaneous localization of hypoxic regions and detailed structural evaluation, providing clinicians with comprehensive diagnostic information [41].

Zhu et al. established transvaginal PAI as clinically effective for pelvic diagnostics, demonstrating clear ovarian visualization with a 4.5 cm tissue penetration depth [42]. In clinical practice, the combined transvaginal and transabdominal ultrasound approach provided extensive visualization of both fetal brain and placental anatomical structures [43]. The integrated ultrasound guidance system, combined with PAI's penetration capacity, may potentially enable assessment of unilateral fetal brain and most placental tissue sO₂, positioning PAI as a promising, innovative modality for advanced fetal monitoring, although additional clinical validation studies are needed to fully confirm its diagnostic utility.

With an imaging depth of several centimeters, PAI has been successfully applied to clinical imaging of organs such as the breast, brain, and liver. Breast imaging studies have demonstrated that PAI can capture vascular and blood oxygenation functional information at tissue thicknesses comparable to that of a late-term placenta, meeting clinical requirements [44]. In brain imaging, PAI enables visualization of vascular and functional structures several centimeters beneath the scalp, supporting non-invasive diagnosis of brain disorders [45]. For liver imaging, PAI can stage liver fibrosis and accurately reflect tissue pathological states [46]. Overall, owing to its deep tissue penetration capability and functional imaging advantages, PAI exhibits broad potential for clinical application. Furthermore, in the field of functional imaging for disease diagnosis, multimodal PAI/USI imaging systems have achieved important research results in the diagnosis of rheumatoid arthritis [47]. These recent translational achievements have further strengthened our confidence in the future clinical translation of our research.

This study robustly established the diagnostic efficacy of PAI for blood flow-related AFD, while it still had several limitations. The study focused specifically on umbilical cord compression and maternal hypoperfusion scenarios, while AFD in clinical practice may result from diverse etiologies including fetal arrhythmias or placental abruption. Additionally, the transition from animal models to human applications will require careful optimization of imaging protocols and further technical development. Future research directions should include: (1) expansion to other AFD pathophysiology models, (2) development of standardized clinical imaging protocols, and (3) validation studies in human pregnancies.

5. Conclusion

PAI represents a paradigm-shifting advancement in AFD diagnosis, offering two critical capabilities: (1) early hypoxia detection through sensitive hippocampal sO₂ monitoring, and (2) precise etiology differentiation via placental sO₂ analysis. The developed two-step diagnostic algorithm significantly enhances diagnostic accuracy while providing time-sensitive pathophysiological information crucial for clinical decision-making. These advancements position PAI as a potentially transformative tool in obstetric care that could substantially improve outcomes in AFD situations.

CRediT authorship contribution statement

Qiufang Ouyang: Supervision, Investigation. Mengyu Zhou: Resources, Investigation. Luo Zhijia: Resources. Fan Meng: Supervision. Luting Zhang: Writing – original draft, Methodology, Data curation. Min Chen: Writing – review & editing. Zhen Yuan: Supervision, Investigation. Jian Zhang: Writing – review & editing, Project administration, Funding acquisition. Zongjie Weng: Supervision, Investigation.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Biographies

Luting Zhang is currently a sonographer at the Fujian Provincial Maternity and Child Health Hospital. Her main research interest focuses on the application of photoacoustic imaging in disease diagnosis.

Mengyu Zhou received her B.S. in Biomedical mechanical engineering and automation from Shihezi University in 2020. Then she continued his studies at Guangzhou Medical University for a Master's degree in Biomedical Engineering. Her research interests are in Application of photoacoustic technology in clinical medicine.

Qiufang Ouyang, Ph.D., Associate Chief Physician, Director of the Ultrasound Department at the Second People's Hospital of Fujian University of Traditional Chinese Medicine, and Senior Visiting Scholar at the University of Glasgow. She specializes in obstetric ultrasound, abdominal ultrasound, and cardiovascular ultrasound diagnostics. In her research work, she focuses on multimodal ultrasound and ultrasound radiomics. She has led six national and provincial-level research projects and has been selected as a "High-Level Talent (Category C)" in Fujian Province.

Fan Meng received her B.S. in the Pharmaceutical Preparation Engineer Excellence Program of China Pharmaceutical University in 2022. Then, she continued his studies at Guangzhou Medical University for a Master's in Biomedical Engineering. Her research interest is visualization of photoacoustic imaging in vivo.

Zhen Yuan is a professor in the Faculty of Health Sciences at University of Macau. His main research areas are as follows: Neuroscience and neuroimaging; Biomedical optics including functional near infrared spectroscopy, photoacoustic tomography/optoacoustic microscopy, optical coherence tomography.

Min Chen is a Professor at the Fetal Medicine Center, Third Affiliated Hospital of Guangzhou Medical University. He holds a PhD in Obstetrics and Fetal Medicine from the University of Hong Kong and has published over 70 SCI papers. He is also an Assistant Professor at the Chinese University of Hong Kong and a member of ISUOG, IROFAN, and FMF. Dr. Chen serves on the editorial board of Ultrasound in Obstetrics and Gynecology.

Zongjie Weng is director of the Ultrasound Department at Fujian Maternal and Child Health Hospital, is a senior expert in prenatal ultrasound diagnosis. With extensive clinical experience, he specializes in fetal anomaly screening and has dedicated his career to techniques in early diagnosis of fetal abnormalities. His main research focus is the early diagnosis of fetal malformations.

Jian Zhang is a professor in the School of Biomedical Engineering at Guangzhou Medical University. He received his Ph. D. degree in optics from South China Normal University. He is a senior member of the Chinese Society of Biomedical Engineering. His main research field is developing optical imaging technology, and applying them in the clinical medicine.

Appendix A. Supplementary material

Supplementary material

Data availability

Data will be made available on request.