Free-moving-state microscopic imaging of cerebral oxygenation and hemodynamics with a photoacoustic fiberscope

doi:10.21203/rs.3.rs-3109380/v1

Download PDF

Article

Free-moving-state microscopic imaging of cerebral oxygenation and hemodynamics with a photoacoustic fiberscope

https://doi.org/10.21203/rs.3.rs-3109380/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 02 Jan, 2024

Read the published version in Light: Science & Applications →

You are reading this latest preprint version

We report the development of a head-mounted photoacoustic fiberscope for cerebral imaging in a freely behaving mouse. The 4.5-gram imaging probe has a 9-µm lateral resolution and 0.2-Hz frame rate over a 1.2-mm wide area. The probe can continuously monitor cerebral oxygenation and hemodynamic responses at single-vessel resolution, showing significantly different cerebrovascular responses to external stimuli under anesthesia and in the freely moving state. For example, when subjected to high-concentration CO₂ respiration, enhanced oxygenation to compensate for hypercapnia can be visualized due to cerebral regulation in the freely moving state. Comparative studies exhibit significantly weakened compensation capabilities in obese rodents. This new imaging modality can be used to study normal and diseased-state cerebrovascular functions and shows promise in neuroscience and microcirculation studies.

The brain uses nearly a quarter of the body’s oxygen intake to support neural activities, and adequate oxygen delivery is essential for normal brain functions^1,2. Oxygenation failure can induce cerebral hypoxia and may be fatal. Therefore, monitoring cerebral hemodynamics and oxygen metabolism is critical for evaluating brain function, detecting brain injury³, and improving cerebral outcomes for intensive care medicine⁴. Rodents, e.g., mice, are the most commonly used animal models for neuroscience studies because of the functional similarities between rodent and human brains and their smaller size. However, benchtop microscopes can image brain activities only under anesthesia or with the head fixed. In contrast, head-mounted microscopes are desirable for free-moving-state cerebral imaging because brain activities can differ significantly in stationary and moving states⁵. In addition, miniaturized microscopes can be used in neuroscience studies to investigate cognitive activity and social interactions. In recent years, lightweight, head-mounted wide-field fluorescence microscopes⁶ and two-photon microscopes (TPMs) have been developed to image cerebral neural activity in freely behaving mice⁷. With an additional fluorescent agent, a TPM can be used to simultaneously image neural activity and the associated vasodilation to study neurovascular coupling⁸. However, these microscopes are not suitable for imaging the cerebral oxygenation process.

Photoacoustic microscopy (PAM), which is based on the detection of ultrasound waves induced by hemoglobin absorption with a focused laser, is a new modality for high-resolution oxygenation imaging. By measuring the acoustic signals with multiple-wavelength excitation lasers, PAM can be used to quantify blood oxygen saturation (sO₂) and visualize the hemodynamics and oxygenation in target tissue, i.e., a mouse brain, at single-vessel resolution^9–14. The use of such microscopes for oxygenation imaging typically involves confocal alignment of lasers and ultrasound beams, rotational scanning with polygonal mirrors, and image stitching algorithms. These microscopes can be miniaturized and used on handheld probes; however, they are not suitable for wearable devices for small animals^15–19. A possible strategy to achieve a head-mounted PAM is to use an unfocused transducer. For example, a head-attached PAM based on an unfocused piezoelectric sensor or a disposable polymer-based optical sensor, can be repeatedly used to view cerebrovascular structures^20,21. Alternatively, single-pulse, wide-field cerebral imaging was achieved by using an ergodic relay cavity, in which the laser-induced ultrasound waves were reflected back and forth multiple times. The images were reconstructed based on the “fingerprint” waveform encoded in the temporal domain²². However, these imaging modalities are not capable of imaging oxygenation and cerebrovascular responses of animals in freely moving states.

Optical fiber technology offers versatile approaches for implementing miniaturized TPMs, optical-coherence tomography (OCT) catheters, and photoacoustic imaging devices. Here, we report a photoacoustic fiberscope for high-spatiotemporal-resolution cerebral imaging in a freely behaving mouse. The imaging probe uses two optical fibers for photoacoustic excitation and detection, and the probe weight is only 4.5 grams. The scanner driving and sensor interrogation modules can freely rotate in compliance with the animal motion, allowing continuous cerebral imaging over the whole wakening process from anesthesia to the freely moving state. The fiberscope has a 9-µm lateral resolution and 0.2-Hz frame rate, enabling the recording of cerebral responses to external stimuli at single-vessel resolution. The fiberscope can be used to quantify the oxygen saturation (sO₂), relative hemoglobin concentration (Hb), and vessel diameter associated with blood perfusion to assess cerebral oxygenation conditions. The proposed device shows promise in neuroscience and microcirculation studies due to its high detection sensitivity, high temporospatial resolution, and ability to image brain activity in the freely moving state.

The photoacoustic fiberscope

Figure 1a shows a freely behaving mouse wearing an imaging probe, which was used to monitor cortical activity. The imaging probe uses two optical fibers for photoacoustic excitation and detection (Figs. 1b and c). The excitation laser beam is guided in the yellow-jacketed fiber (SMF-28e+, Corning, N.A. = 0.14), collimated with an achromatic lens (AC050-008-A, Thorlabs, N. A.=0.33 in air), reflected by a 45-degree tilted mirror, and focused in the brain tissue by using an additional conjugated lens. The hemoglobin in the cerebral vessels partially absorbs the laser pulses and generates ultrasound waves via the photoacoustic effect. A horizontally placed fiber optic sensor detects the laser-induced ultrasound waves. The blue-jacketed optical fiber guides the sensing light to the photodetector and signal demodulation module. Volumetric images are acquired via raster scanning of the laser beam using a microelectromechanical system (MEMS) scanner.

The imaging probe utilizes fiber optic sensors to detect the photoacoustic signals. A small fiber laser with a lasing frequency of approximately ω₀ = 2$\pi \times$195 THz is employed as the acoustically responsive element (see Methods). The laser cavity can be acoustically deformed, changing the lasing frequency by Δω₀$.$ The two orthogonally polarized laser modes have a 2-GHz frequency difference due to the intrinsic fiber birefringence. In the torsional-radial mode vibration state, the x- and y-polarized laser lights have identical frequency shifts but opposite signs, ±Δω₀. We beat the two laser modes at radio frequency to collect the acoustically induced frequency change, 2Δω₀. The radio-frequency beat signal responds to the ultrasound waves at S = 2.25 MHz/kPa over 30 MHz, which is mainly determined by the diameter, longitudinal and shear wave acoustic velocities and photo elastic coefficient of the optical fiber. The sensor can offer a noise-equivalent pressure (NEP) of 8 Pa.

Figures 1d and e show the laser-beam scanning and ultrasound detection schemes. The fiber optic sensor is immersed in an acoustic coupling medium. The sensor is placed 1.2 mm above the tissue surface, is parallel with the imaging plane, and detects ultrasound waves in a side-viewing manner. The sensor has a sensitive length of ~ 3 mm, which is defined by the two highly reflective Bragg reflectors. The ultrasound response has a cos(2θ) dependence, where θ denotes the incident angle of the ultrasound wave, offering a 60-degree view angle at half magnitude. By rotating the sensor, the overlap between the laser scanning and ultrasound detection areas can be maximized. The optical window for the excitation laser beam is sealed with a glass coverslip. The unfocused ultrasound detection reduces the optical alignment requirements, allowing an A-line scan rate of 100 kHz, a B-scan rate of 100 Hz, and a frame rate of 0.2 Hz over a 1.2 mm by 1.2 mm area. As a result, the imaging probe enables fast scanning, high optical resolution imaging of cerebral hemodynamics.

The headpiece has an aluminum baseplate (weight: 0.1 gram), which is permanently fixed on the mouse head over the cranial window (see Methods and Supplementary Figs. 1 and 2). The baseplate contains four permanent magnets in its corners, which are used to mount the imaging probe. The imaging probe was assembled by incorporating the optical fibers, lenses, and MEMS scanner before attaching the probe to the baseplate. This design is convenient for mounting the probe and guarantees imaging stability in freely behaving mice. Furthermore, this detachable design allows repeated imaging of the region of interest (ROI) based on the experimental requirements.

The experimental mouse wearing the imaging probe was placed in a cylindrical chamber that allowed free behavior (see Supplementary Fig. 3). We mounted a camera on the top of the chamber to track and record the motion of the mouse. The chamber has a gas inlet and an outlet at the bottom for hypercapnia and hypoxia experiments. The imaging probe is connected to the console through two optical fibers and an electrical wire. To prevent wire/fiber entanglements, the scanner driving and sensor interrogation unit was miniaturized, with an overall volume of 11×12×1.7 cm³, which includes the 980-nm semiconductor pump laser, erbium-doped fiber amplifier, photodetector, fiber polarizer, and optical isolator. A coaxial optical/electrical slip ring was used to transmit the excitation laser beam, supply power, and acquire electrical signals. As a result, this unit can rotate freely and is compliant with the motion of the experimental animal, allowing continuous cerebral imaging. The freely moving mouse can bend and vibrate the optical fibers in the experiment. However, the two orthogonally polarized laser beams of the sensing laser are highly correlated, and the heterodyne detection scheme utilizes common-noise cancelation to stabilize the sensor output during cerebral imaging. Moreover, the weight of the optical fibers is minimal and can be ignored. The lengths of the optical fibers were optimized to adapt to the size of the chamber.

For oxygenation imaging, we built a customized dual-colored laser source (see Methods and Supplementary Fig. 4). The 558-nm component is produced by the second-order Stokes wave generated by the stimulated Raman scattering in the highly nonlinear optical fiber. This component was combined with a 532-nm laser beam with the same repetition rate and an optimal pulse interval of 2.75 µs before being input into the imaging probe. A field-programmable gate array card (PXI-7852R, National Instruments) was used to synchronize the laser pulses, the MEMS scanner, and the data acquisition module. We use a LabVIEW program to acquire and process data. Figures 1f and g demonstrate photoacoustic images showing cerebral hemoglobin and sO₂ in the same region. Small veins, arteries, and capillaries can be visualized and distinguished in the images. Oxy- and deoxygenated hemoglobin have different absorption coefficients at these two wavelengths. Identical laser energies induce photoacoustic signals with different strengths PA₅₃₂ and PA₅₅₈ (Fig. 1h). As a result, oxygen saturation (sO₂), namely, the molecular ratio of oxygenated hemoglobin to total hemoglobin, can be quantitatively measured and imaged based on Eq. (1) (see Methods).

Hypercapnia experiment under anesthesia

We first investigated acute hypercapnia in the experimental animal under anesthesia and performed cerebral imaging with the photoacoustic fiberscope. We performed baseline imaging under normocapnia conditions (Fig. 2a). Then, we changed the respiratory gas from normal air (with 1.5% isoflurane anesthesia) to a 50%:50% air/CO₂ mixture to induce hypercapnia. A high CO₂ concentration was used to generate a quick cerebrovascular response (Supplementary Video 1). After imaging, we repeated the hypercapnia experiment and performed a cardiac blood analysis 50 seconds after the animal was presented with the air/CO₂ mixture. The measured CO₂ partial pressure of 100 ± 12 mmHg (normal condition: 40 mmHg) and oxygen partial pressure of 34 ± 8 mmHg (normal state: 90 mmHg) verified the occurrence of acute hypercapnia. As shown in Figs. 2b and c, hypercapnia dramatically increased the overall oxygen saturation (sO₂), while the relative hemoglobin concentration (Hb) decreased. After the hypercapnia experiment, the sO₂ and Hb levels recovered to normal levels (Fig. 2d).

We performed normocapnia–hypercapnia imaging with four mice and extracted the changes in the sO₂ and Hb levels and the diameters of the veins and arteries (Figs. 2e to g). Figure 2e shows that the averaged venous sO₂ level decreased from 92–64%, and the arterial sO₂ level decreased from 95–75% in the first 10 seconds before decreasing to a minimal value of 70% (see Supplementary Fig. 5 for the statistical results). Figure 2f exhibits that the Hb level decreased in both the arteries and veins by approximately 10%. Figure 2g shows the vasoconstriction in an artery (by 15%) and a vein (2.5%). sO₂ and Hb recovery took 50 seconds. These results are consistent with previous measurements of high-concentration CO₂ respiration in animals²³. Moreover, the oxygen extraction fraction (OEF), which is defined as the ratio of oxygen consumption to delivery in a tissue or organ, is a useful parameter for quantifying abnormal cerebrovascular oxygenation. Figure 2h shows the OEF variation in the hypercapnia experiment, which was calculated based on the sO₂ measurement results (see Methods). The OEF increased from 5–15% in the hypercapnia experiment due to insufficient oxygen delivery.

In comparison, we performed hypoxia experiments with anesthetized mice (Supplementary Fig. 6 and Supplementary Video 2). In this experiment, the oxygen fraction was reduced from ~ 21% of the respiratory air to approximately 10% by mixing air and nitrogen with the same volumes. This experiment lasted 100 seconds and was repeated with five mice. Hypoxia also induced a decrease in sO₂ from 90–70% in the veins and 95–80%in the arteries and an Hb decrease of approximately 7% (see Supplementary Fig. 7 for the statistical results). In addition, a maximum of 3% vasodilation was observed in the vein during the hypoxia experiment, and the OEF in the cerebral vessels changed from 4–8% during the experiment. The recorded hemodynamic response is consistent with the results of previous medical studies and the imaging results acquired using benchtop photoacoustic microscopy¹⁰. The above experimental results demonstrate that the proposed imaging system can visualize cerebrovascular responses with high spatial and temporal resolution. Furthermore, the OEF profiles differ in the hypoxia and hypercapnia experiments, suggesting that the photoacoustic fiberscope can distinguish different hemodynamic responses under various stimuli.

Cerebral imaging in the wakening process

Next, we placed an isoflurane-anesthetized, 8-week-old male BALB/c mouse wearing an imaging probe on its head in the chamber and continuously imaged its cerebral activity. Supplementary Video 3 shows the recorded mouse behavior during the period from anesthesia to freely moving and the real-time sO₂ and Hb photoacoustic images over 30 minutes. The video shows that the fiberscope can stably image cerebral hemodynamics in a freely behaving state, even when the mouse accidentally hits the chamber wall. Figure 3 demonstrates the motion trajectory of the mouse during wakening (Fig. 3a), anticlockwise movement (Fig. 3b), clockwise movement (Fig. 3c), and a short rest period in the chamber (Fig. 3d, see Methods for details about the motion trajectory tracking). The four photoacoustic images presented in Figs. 3a to d show that the sO₂ levels in the veins decreased when the mouse woke and moved freely. In the first 20 minutes, venous sO₂ decreased (Fig. 3e), arterial Hb increased (Fig. 3f) and vasoconstriction (Fig. 3g), and OEF increased (Fig. 3h) due to the increased oxygen metabolism in the wakening and freely moving states and the reduced isoflurane anesthesia. After 20 minutes, cerebral oxygenation recovered to a normal state.

Hypercapnia experiment in freely behaving mice

Next, we imaged hypercapnia-induced cerebral responses in a freely moving mouse (Supplementary Video 4). The experiment was performed over 100 seconds. Figure 4 shows the sO₂ and Hb photoacoustic images captured during the baseline (Fig. 4a), hypercapnia (Figs. 4b, c), and normocapnia states (Figs. 4d, e). The imaging results show that the overall sO₂ level increased slightly when CO₂ was injected into the chamber and subsequently decreased after the experiment. The mouse could move freely in the chamber during the hypercapnia experiment, as shown by the recorded motion trajectory in Fig. 4f. We repeated the imaging experiment with four mice to record the temporal variations in the individual hemodynamic parameters. Figures 4g to j demonstrate that the sO₂ level increased (by 5% in the artery and 4% in the vein, Fig. 4b), Hb levels slightly increased (Fig. 4h), vasodilation decreased (by 5% in the vein and 2% in the artery, Fig. 4i), and the OEF decreased by 4% (Fig. 4j). The results suggest that the cerebral vessels tend to deliver more oxygen to compensate for hypercapnia. This oxygenation enhancement induced by CO₂ respiration was not observed in the anesthetized mice. As shown in the motion trajectory in Fig. 4f, this compensation process is accompanied by fast mouse locomotion (frames in c and d). As the hypercapnia continued, the arterial and veinous sO₂ levels decreased by 15% and 14% (beginning at approximately 150 seconds), and the total Hb level decreased by approximately 5%. The OEF gradually increased over the next 300 seconds, corresponding to insufficient blood oxygen delivery, which is consistent with previous observations in human experiments conducted under hypoxic conditions^24,25.

Hyperlipidemia is known as one of the major causes of vascular dysfunction^26,27. It can slow blood flow, cause cerebral hypoperfusion, and induce chronic brain hypoxia^27–30. Here, we also imaged the cerebrovascular responses in freely moving obese (hyperlipidemia) mice (Fig. 5 and Supplementary Video 5, see Methods for details). The photoacoustic images shown in Figs. 5a to e demonstrate that the overall sO₂ levels tended to remain stable (Fig. 5b) and then decreased by approximately 10% during the experiment (Fig. 5c). In addition, the Hb levels decreased by approximately 10% in the arteries and 5% in the veins (Fig. 5h). The vessel diameter increased by 2% in the vein and decreased by 2% in the artery (Fig. 5i). After the experiment, the sO₂ levels in the cerebral veins remained low. The tracked motion trajectory presented in Fig. 5f shows that the obese mice can freely move in the experiment as well as the healthy mice. We selected three periods and compared the cerebral responses of the healthy and obese mice, including period #1 (115 to 135 s, at the beginning of the hypercapnia experiment), period #2 (170 to 190 s, at the end of the experiment), and period #3 (270–290 s, after the experiment). The statistical results shown in Figs. 5k to n were calculated based on the measured results presented in Figs. 4 and 5 over the three selected periods. The results suggest that oxygenation enhancement occurred in both the healthy and obese groups. However, the induced oxygenation was significantly stronger in the control group than in the obese group (Figs. 5k and l). The sO₂ level changed only slightly in response to the hypercapnia experiment (P = 0.0013 in artery and P = 0.0193 in vein for 15–35 seconds). At the end of the experiment, a notable decrease in the sO₂ and Hb levels was observed in the obese group, while normal oxygenation levels were maintained in the control group. The Hb levels in the veins and arteries decreased significantly 70–90 seconds after gas injection.

The OEF decrease in the experiment was hardly observable in the obese group. In addition, the sO₂ level decreased at 120 seconds (20 seconds after CO₂ injection), which was earlier than in the control group, and the artery diameter decreased by 4%. The different hypercapnia-induced responses in obese and healthy mice can be understood by comparing the results shown in Figs. 4 and 5. Additionally, the baseline OEF values in the obese mice were higher than those in the healthy mice.

In conclusion, we developed a photoacoustic fiberscope to visualize cerebral hemodynamics and oxygen metabolism in freely behaving rodents at single-vessel resolution. This imaging modality can be used to visualize the different cerebrovascular responses of anesthetized and freely behaving mice. In the freely behaving mice, the oxygen supply is immediately enhanced after high-CO₂-level respiration, which was not observed in the anesthetized mice. Furthermore, the results of comparative studies showed that the oxygen compensation capability is significantly weakened in obese mice compared to that in control mice.

The cerebrovascular responses of head-fixed mice have been studied by using multiple technological means. For example, the cerebral oxygen partial pressure (pO₂) has been measured using Clark-type polarographic electrodes³¹. Additionally, spectroscopic imaging has been applied to map cerebral sO₂ levels by collecting reflectance scattering light from brain tissue. However, this method is limited by insufficient spatial resolution, and microcirculation cannot be observed^32,33. By using an oxygen-sensitive fluorescent probe, a two-photon phosphorescent lifetime microscope (2PLM) can be used to measure and image the partial pressure in vessels by recording the fluorescence decay time. However, accurate oxygenation quantification requires averaging multiple measurements, and this method is limited in terms of its temporal resolution^32,34,35. In contrast, the proposed photoacoustic fiberscope is advantageous due to its light weight, high temporospatial resolution and oxygenation measurement capability. The fiberscope can be used to elucidate cerebral hemodynamics and metabolism and can potentially be applied to detect cerebral dysfunction, study therapy responses, and monitor brain functional development. For example, cerebral vessels can dynamically regulate blood flow and oxygen supply differently in veins and arteries. The response magnitude and temporal process of Hb and sO₂ changes are of great interest to understanding this response mechanism. Such changes can occur in seconds and can be measured at high temporal resolution by the photoacoustic fiberscope.

The recently reported head-mounted photoacoustic microscopes detect laser-induced ultrasound waves using piezoelectrical transducers¹⁷. However, the tradeoff between the transducer size, detection sensitivity, and field of view hinders the oxygenation imaging capability of such devices. The proposed photoacoustic fiberscope uses a fiber optic sensor for high-sensitivity ultrasound detection, addressing this limitation. Compared with piezoelectric detection systems, optical ultrasound sensors can translate acoustic displacement into phase or intensity modulations based on the light signal, significantly amplifying the detection sensitivity³⁶. Moreover, the proposed system uses a lightweight headpiece, reducing the requirements for precise optical and acoustic alignment.

The proposed fiberscope showed superior imaging capabilities; however, the current imaging instrumentation has some limitations. We aim to improve the system in terms of the following aspects. First, the weight of the headpiece can be reduced by using polymer lenses instead of glass lenses, and the metal mounters can be replaced with lighter materials. Second, the field of view is limited by the maximum scanning angle of the MEMS mirror and the field of view of the object lens. The current imaging system uses achromatic doublets with an aperture of only a 1.4 mm. A compound lens with a flat field and low aberration could be introduced to extend the field of view. Third, the fiberscope can be employed to obtain blood flow rate measurements based on Doppler detection or time-correlation measurements.

a. Animal preparation

All animal procedures were conducted based on the “Guiding principles in the care and use of animals” (GB/T 35892 − 2018, China) and were approved by the laboratory animal ethics committee of Guangzhou Huateng Biomedical Technology (IACUC: HTSW220304). We fed C57BL/6 mice (4 weeks old, 15–25 g in weight) either a control diet (n = 10, five males and five females) or a high-fat diet (n = 30, fifteen males and fifteen females). The mice were housed in individual cages in an environment with a 12-hour light-dark cycle at a temperature of 25°C with food and water ad libitum. We measured the total cholesterol (TC), triglyceride (TG), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C) via blood tests with a lipid panel every four weeks (see Supplementary Fig. 8). We measured the body weight every two weeks. After 8 weeks of feeding, the healthy and obese groups were used for cerebral imaging. The illumination fluence of the excitation laser in the cerebral tissue was in a safe range, following the American National Standard Institute (ANSI) safety standard.

b. Hypercapnia experiment under anesthesia

We performed systemic hypoxia experiments with 1.5% isoflurane anesthesia. We placed the experimental animal on a heating pad to maintain a 37-degree body temperature. Hypercapnia was induced by changing the inspiratory air from normal to a 50:50 air/CO₂ mixture. After 100 seconds of baseline imaging, we induced hypercapnia over 50 seconds and imaged the resultant changes in the cerebral vascular structure and blood sO₂ levels.

c. Hypercapnia experiment with freely moving mice

We performed hypercapnia experiments with freely moving mice in a gas chamber with a diameter of 40 cm and a height of 30 cm. The chamber has a gas input and an output at the bottom. After 100 seconds of baseline imaging, we pumped the air/CO₂ mixture at 20 L/min to induce hypercapnia through the inlet. Then, normal air was pumped at 10 L/min to replace the air/CO₂ mixture to recover to normal conditions. Standard atmospheric pressure was maintained in the tank.

d. Preparation for cerebral imaging

We created a transcranial window under 1.5% isoflurane anesthesia. The mouse head was fixed in a stereotaxic apparatus with a head fixture device and positioner for precise positioning. First, the scalp was shaved with an electric razor, and a 1.5 cm incision was made along the middle line. Then, the scalp was removed to expose the whole skull, and the periosteum on the skull was removed using a scalpel and scissors. The skull of the region of interest (ROI) was then partially removed using a surgical dental drill, during which saline flushing was performed periodically to prevent overheating and bleeding. After window preparation, saline was used as a medium between the imaging probe and the mouse head.

For freely moving state imaging, we performed the following additional steps. First, the skin incision was treated with a tissue adhesive (3M Vetbond) after the surface of the skull was dried. Then, the ROI was identified, and a customized holder was fixed to the skull with cyanoacrylate (Loctite 454). Afterward, dental adhesive resin cement (Sun Medical, Super-Bond C&B) was deployed to fill the remaining exposed skull and fix the holder. Finally, ultrasound gel was used to fill the gap between the tissue surface and the sensor. The head-mounted imaging device was fixed to the holder via magnetic force using four magnets. We adjusted the focal length of the laser beam before imaging to optimize the imaging performance (see Supplementary Figs. 1 and 2 for details).

e. Tracking the motion trajectory of the mouse

We recorded the motion trajectories of the freely moving mice with a video camera. The camera was placed on the top of the chamber, and a wide-angle lens was used to cover the entire motion area of the mouse. The chamber floor has a significant contrast in color to the mouse, and the animals were located via computer vision software. The 1920 × 1080 pixel videos were recorded at 30 frames per second. We used an OpenCV-based channel and spatial reliability tracking (CSRT) tracker to plot the motion trajectory. We selected the mouse head and the head-mounted microscope as the tracking targets. In the first frame, to initialize the tracker, the algorithm reads and learns the geometric and colorimetric features of the selected target. In the following frames, the target is found based on the pretrained features, and its position in (x, y) coordinates is recorded. Finally, the recorded coordinates in each frame are read out to plot the motion trajectory.

f. Imaging data analysis and statistics

The Hb and sO₂ photoacoustic images were obtained as follows. Based on the recorded amplitude of the 532-nm excited photoacoustic signal (PA₅₃₂) at each pixel, Hb images were created by using the maximum-amplitude projection (MAP) method. The sO₂ values in each pixel were then calculated based on the measured amplitudes excited at 532 nm (PA₅₃₂) and 558 nm (PA₅₅₈) as

$$\text{s}{\text{O}}_{2}=\frac{\frac{{PA}_{558}}{{PA}_{532}}{\mu }_{532}^{Hb}-{\mu }_{558}^{Hb}}{\frac{{PA}_{558}}{{PA}_{532}}\left({\mu }_{532}^{Hb}-{\mu }_{532}^{Hb{O}_{2}}\right)-\left({\mu }_{558}^{Hb}-{\mu }_{558}^{Hb{O}_{2}}\right)}$$

where ${\mu }_{532}^{Hb}=40584$, ${\mu }_{558}^{Hb}=54164$, ${\mu }_{532}^{Hb{O}_{2}}=43876$, and ${\mu }_{558}^{Hb{O}_{2}}=33456$ (cm⁻¹∙mole⁻¹) are the absorption coefficients.

For the statistical analysis, we created a vessel segmentation algorithm to identify the arteries and veins in each frame. First, the arteries and veins were manually identified in the first frame based on the sO₂ level. Then, a vessel mask was applied to remove the arteries and veins with diameters smaller than 40 µm in the subsequent frames. Then, the average sO₂ and relative Hb levels were calculated based on the mean value of the filtered results. After performing a 10-fold interpolation based on the images, we conducted Gaussian fitting in the direction perpendicular to the blood vessels. Next, we quantified the full width at half maximum (FWHM) as the local vessel width. A matching algorithm based on normalized cross-correlation (NCC) was used to correct the mask for each frame to compensate for the possible deformation or offset of the target vessels. Finally, the OEF was calculated based on the ratio of the oxygen saturation difference between the arteries and veins: $\frac{{s}_{a}{O}_{2}-{s}_{v}{O}_{2}}{{s}_{a}{O}_{2}}$. The acquired hemodynamic parameters were extracted from each imaging frame, and their variations are plotted in Figs. 2 to 5.

We calculated the fractional changes of the recorded relative Hb and diameter variations with respect to the baseline values, as shown in Figs. 2 to 5. We used two-way ANOVA to determine significance, and all statistical data are presented with the s.e.m., with P < 0.05 considered significant.

g. Dual-wavelength pulsed laser source

We built a dual-wavelength pulsed laser source for multispectral photoacoustic imaging. The 532-nm laser light was emitted from a nanosecond laser (VPFL-G-HE-30, Spectra-Physics), and the 558-nm laser beam was obtained by optically pumping a single-mode fiber (HB450-SC, Fibercore) with another 532-nm nanosecond laser. The stimulated Raman scattering (SRS) effect induces energy conversion for the first-order Stokes wave at 545 nm and the second-order wave at 558 nm. We used a bandpass filter (ZET561/10x, Chroma) to filter out the 558-nm component. The first half-wave plate (GCL-060411, Daheng Optics) was used to adjust the light intensity of the Raman pump path with a polarizing beam splitter cube (PBS101, Thorlabs). We coupled the pump laser beams into the optical fiber through a collimator (F220FC-532, Thorlabs). We employed a second half-wave plate to adjust the laser polarization state to achieve a maximal efficiency greater than 30% in the 532 nm-to-558 nm wavelength conversion. The maximal pulse energy at 558 nm can reach 2.0 µJ. Two neutral density filters (NDC-50C-4 M, Thorlabs) were used to adjust the intensities of the 532-nm and 558-nm components. A dichroic mirror (ET550lp, Chroma) was applied to combine the two-color laser beams. The laser beam was coupled into a single-mode fiber and delivered to the imaging probe.

h. Ultrasound detection with the fiber optic sensor

We used a fiber optic sensor to detect laser-induced ultrasound waves. We ultraviolet-inscribed two highly reflective index gratings (4 mm each in length and 3 mm in separation) to form a high-finesse cavity in the erbium/ytterbium codoped core of an optical fiber (EY305, CorActive). This sensor works in laser mode, unlike the passive optical resonators used in previous works. The rare-earth dopant offers optical gain, and the two reflectors provide optical feedback. A 980-nm low-coherence semiconductor laser pumped the cavity to lase at a single longitudinal mode at 1530 nm with a kilohertz linewidth. The incident ultrasound wave can deform the sensor and induce a lasing-frequency variation.

The acoustically induced laser frequency change is read out via heterodyne optical detection³⁶. The sensor laser naturally has two orthogonally polarized modes with slightly different lasing frequencies. The two laser beams pass through a 45-degree polarizer and generate a beat signal at the high-speed photodetector. Their beat signal falls in the radio frequency range (approximately 2 GHz). The ultrasound wave-induced identical frequency variations to the x- and y-polarized laser beams have opposite signs. As a result, the acoustically induced lasing frequency change can be measured in the radio frequency range.

We used a multifunction reconfigurable I/O module (NI PXIe-7852R) to generate a synchronized trigger signal. In addition, a vector signal transceiver (NI PXIe-5646R) was used to record the waveform of the frequency-downshifted signal generated by mixing the beat signal with a local reference microwave source. The phase variation and the frequency shift were acquired via I/Q demodulation. The sampling rate was 125 MHz, which allows an acquisition bandwidth of 62.5 MHz. In addition, we used the integrated FPGA in the transceiver to calculate the sO₂ level and display the hemoglobin and sO₂ images in real time. All the image sequences and pulse energies of the 558-nm laser component were saved in a high-speed disk array.

i. blood gas analysis

To perform a blood gas analysis with a mouse, we anesthetized the experimental animal with 1.5% isoflurane, obtained a small blood sample from the mouse by using a cardiac puncture technique, and then analyze the blood sample using a blood gas analyzer.

Acknowledgments

Y.L. acknowledges the financial support from the National Natural Science Foundation of China (62275104). L. J. acknowledges the financial support from the National Natural Science Foundation of China (62122031). B.-O. G. acknowledges the financial support from the National Natural Science Foundation of China (62135006 and 61860206002), the Local Innovative and Research Teams Project of Guangdong Pearl River Talents Program (2019BT02X105).

Author contributions

Y. L., L.J. and B.-O.G. conceived the project and supervised the research. X. Z., Y.L., X. W., H. L., and X. B. prepared the sample and performed the experiments. X. Z., Y. L., and L.J. contributed to data analysis. Y.L. performed theoretical analysis. Y.L., L.J., and X. Z. prepared the manuscript. B.-O.G. participated in the discussion and manuscript writing.

Competing interests

All authors declare no competing interests.

Data availability

The data that support the findings of this study are available from the authors upon reasonable request.

Code availability

The authors declare that for data collection the commercially available software from LabVIEW 2015 (National Instruments, USA) and Matlab 2018b (Matlab, Mathworks, USA) were used. Data analysis was conducted in Matlab using its built-in functions. The image algorithm has been patented and is available upon discretion from the corresponding author.

Hall, C. N., Klein-Flügge, M. C., Howarth, C. & Attwell, D. Oxidative phosphorylation, not glycolysis, powers presynaptic and postsynaptic mechanisms underlying brain information processing. Journal of Neuroscience 32, 8940-8951 (2012).
Raichle, M. E. & Mintun, M. A. Brain work and brain imaging. Annu. Rev. Neurosci. 29, 449-476 (2006).
Taran, S., Pelosi, P. & Robba, C. Optimizing oxygen delivery to the injured brain. Current Opinion in Critical Care 28, 145-156 (2022).
Jain, I. H. et al. Hypoxia as a therapy for mitochondrial disease. Science 352, 54-61 (2016).
Gao, Y.-R. et al. Time to wake up: Studying neurovascular coupling and brain-wide circuit function in the un-anesthetized animal. Neuroimage 153, 382-398 (2017).
Ghosh, K. K. et al. Miniaturized integration of a fluorescence microscope. Nature methods 8, 871-878 (2011).
Zong, W. et al. Fast high-resolution miniature two-photon microscopy for brain imaging in freely behaving mice. Nature methods 14, 713-719 (2017).
Chow, B. W. et al. Caveolae in CNS arterioles mediate neurovascular coupling. Nature 579, 106-110 (2020).
Yao, J. et al. High-speed label-free functional photoacoustic microscopy of mouse brain in action. Nature methods 12, 407-410 (2015).
Zhu, X. et al. Real-time whole-brain imaging of hemodynamics and oxygenation at micro-vessel resolution with ultrafast wide-field photoacoustic microscopy. Light: Science & Applications 11, 138 (2022).
Taboada, C. et al. Glassfrogs conceal blood in their liver to maintain transparency. Science 378, 1315-1320 (2022).
Wang, L., Maslov, K. & Wang, L. V. Single-cell label-free photoacoustic flowoxigraphy in vivo. Proceedings of the National Academy of Sciences 110, 5759-5764 (2013).
Wang, Y. et al. High-speed multi-parametric photoacoustic microscopy of cerebral hemodynamic and metabolic responses to acute hemodilution. Optics Letters 47, 1988-1991 (2022).
He, Y., Shi, J., Maslov, K. I., Cao, R. & Wang, L. V. Wave of single-impulse-stimulated fast initial dip in single vessels of mouse brains imaged by high-speed functional photoacoustic microscopy. Journal of Biomedical Optics 25, 066501-066501 (2020).
Park, K. et al. Handheld photoacoustic microscopy probe. Scientific reports 7, 1-15 (2017).
Chen, J., Zhang, Y., Zhu, J., Tang, X. & Wang, L. Freehand scanning photoacoustic microscopy with simultaneous localization and mapping. Photoacoustics 28, 100411 (2022).
Chen, Q. et al. Ultracompact high-resolution photoacoustic microscopy. Optics letters 43, 1615-1618 (2018).
Lin, L. et al. Handheld optical-resolution photoacoustic microscopy. Journal of biomedical optics 22, 041002-041002 (2017).
Jin, T., Guo, H., Jiang, H., Ke, B. & Xi, L. Portable optical resolution photoacoustic microscopy (pORPAM) for human oral imaging. Optics letters 42, 4434-4437 (2017).
Guo, H., Chen, Q., Qin, W., Qi, W. & Xi, L. Detachable head-mounted photoacoustic microscope in freely moving mice. Optics letters 46, 6055-6058 (2021).
Li, H. et al. Disposable ultrasound-sensing chronic cranial window by soft nanoimprinting lithography. Nature communications 10, 4277 (2019).
Li, Y. et al. Snapshot photoacoustic topography through an ergodic relay for high-throughput imaging of optical absorption. Nature Photonics 14, 164-170 (2020).
Ikeda, N., Takahashi, H., Umetsu, K. & Suzuki, T. The course of respiration and circulation in death by carbon dioxide poisoning. Forensic science international 41, 93-99 (1989).
Lewis, N. C., Messinger, L., Monteleone, B. & Ainslie, P. N. Effect of acute hypoxia on regional cerebral blood flow: effect of sympathetic nerve activity. Journal of applied physiology 116, 1189-1196 (2014).
Morris, L. E., Flück, D., Ainslie, P. N. & McManus, A. M. Cerebrovascular and ventilatory responses to acute normobaric hypoxia in girls and women. Physiological Reports 5, e13372 (2017).
Boillot, A. et al. Obesity and the microvasculature: a systematic review and meta-analysis. PloS one 8, e52708 (2013).
Sorop, O. et al. The microcirculation: a key player in obesity-associated cardiovascular disease. Cardiovascular research 113, 1035-1045 (2017).
M Dorrance, A., Matin, N. & W Pires, P. The effects of obesity on the cerebral vasculature. Current vascular pharmacology 12, 462-472 (2014).
Knight, S. P. et al. Obesity is associated with reduced cerebral blood flow–modified by physical activity. Neurobiology of Aging 105, 35-47 (2021).
Amen, D. G., Wu, J., George, N. & Newberg, A. Patterns of regional cerebral blood flow as a function of obesity in adults. Journal of Alzheimer's Disease 77, 1331-1337 (2020).
Bentley, W. J., Li, J. M., Snyder, A. Z., Raichle, M. E. & Snyder, L. H. Oxygen level and LFP in task-positive and task-negative areas: Bridging BOLD fMRI and electrophysiology. Cerebral cortex 26, 346-357 (2016).
Zhang, Q. et al. Cerebral oxygenation during locomotion is modulated by respiration. Nature Communications 10, 5515 (2019).
Ma, Y. et al. Wide-field optical mapping of neural activity and brain haemodynamics: considerations and novel approaches. Philosophical Transactions of the Royal Society B: Biological Sciences 371, 20150360 (2016).
Lyons, D. G., Parpaleix, A., Roche, M. & Charpak, S. Mapping oxygen concentration in the awake mouse brain. Elife 5, e12024 (2016).
Esipova, T. V. et al. Oxyphor 2P: a high-performance probe for deep-tissue longitudinal oxygen imaging. Cell Metabolism 29, 736-744. e737 (2019).
Liang, Y., Sun, H., Cheng, L., Jin, L. & Guan, B.-O. High spatiotemporal resolution optoacoustic sensing with photothermally induced acoustic vibrations in optical fibres. Nature Communications 12, 4139 (2021).

(Not answered)

2.Supplementaryinformation20230626.docx
3.SupplementaryVideoforsubmission.zip
Supplementary videos

Download PDF

Journal Publication

published 02 Jan, 2024

Read the published version in Light: Science & Applications →

Editorial decision: revise
21 Aug, 2023
Review #5 received at journal
14 Aug, 2023
Review #1 received at journal
10 Aug, 2023
Reviewer #5 agreed at journal
07 Aug, 2023
Review #2 received at journal
07 Aug, 2023
Reviewer #4 agreed at journal
04 Aug, 2023
Review #3 received at journal
04 Aug, 2023
Reviewer #3 agreed at journal
27 Jul, 2023
Reviewer #2 agreed at journal
24 Jul, 2023
Reviewer #1 agreed at journal
21 Jul, 2023
Reviewers invited by journal
17 Jul, 2023
Submission checks completed at journal
03 Jul, 2023
First submitted to journal
26 Jun, 2023
Editor assigned by journal
26 Jun, 2023

You are reading this latest preprint version

Free-moving-state microscopic imaging of cerebral oxygenation and hemodynamics with a photoacoustic fiberscope

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Results

The photoacoustic fiberscope

Hypercapnia experiment under anesthesia

Cerebral imaging in the wakening process

Hypercapnia experiment in freely behaving mice

Discussion

Materials and methods

a. Animal preparation

b. Hypercapnia experiment under anesthesia

c. Hypercapnia experiment with freely moving mice

d. Preparation for cerebral imaging

e. Tracking the motion trajectory of the mouse

f. Imaging data analysis and statistics

g. Dual-wavelength pulsed laser source

h. Ultrasound detection with the fiber optic sensor

i. blood gas analysis

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1