As an important tool in investigating neural activities, brain imaging allows us to directly observe in vivo psychological processes and cognitive activities [1]. In most experiments, the use of anesthetized and head-restrained animals is common and mature, but inevitably affected by neural inhibition and restricted behavior [2–4]. Hence, carrying out brain imaging in freely moving animals is a promising solution.

Miniaturization of imaging devices is an absolute prerequisite for brain studies using freely moving animals. Prior investigations have fully explored the miniaturization of various optical imaging modalities, such as two-photon microscopy [5,6], fluorescence microscopy [7,8], laser speckle imaging [9,10], etc. [11,12]. These miniaturized probes have provided a number of fundamental insights into the truth of cellular activity and how neuronal subpopulations operate during behaviors. Besides cellular and neural activities, hemodynamics is also vital in the investigation of brain activities [13]. In comparison with pure optical imaging techniques, optical resolution photoacoustic microscopy (ORPAM) is a more powerful tool for investigating hemodynamics [14–16]. Owning to its unique working principle based on label free imaging of optical absorption, both structural and functional information of vascular networks can be revealed [17–19].

Unlike conventional optical microscopes, in ORPAM, the low-scattered ultrasonic wave is excited and detected instead of the high-scattered optical photon in biological tissues to guarantee a deeper penetration depth. To excite and record the acoustic wave, separation and integration of acoustic and optical paths are commonly required, making it more difficult to be miniaturized compared with pure optical imaging modalities. In our previous study, we have successfully achieved a wearable photoacoustic (PA) microscope in freely moving rats [20]. However, the weight and size of the ORPAM probe for rats are still unacceptable for mice. In addition, to avoid motion artifacts and achieve stable image quality, the probe was permanently fixed on the rat brain, preventing it from use in long-term brain studies. Among mammalian species, the mouse is a pre-eminent animal model for biomedical studies due to its various pathological and genetic features. Thus, developing an ultralight miniaturized ORPAM probe capable of long-term imaging in freely moving mice is of particular importance.

In this Letter, we extended our work from freely moving rats to mice. A miniaturized ORPAM probe consisting of an optical fiber, a miniaturized optical scanner, and a small piezoelectric ultrasonic transducer was presented. Compared with previous wearable imaging probes for rats, we reduce the weight from 8 g to 1.8 g and increase the field of view (FOV) from ${1.2}\;{\rm mm} \times {1.2}\;{\rm mm}$ to ${3}\;{\rm mm} \times {3}\;{\rm mm}$. In addition, the current imaging probe is detachable, making it accessible to longitudinal monitoring of the mouse. Through in vivo experiments, we prove the feasibility of long-term, stable, and high-quality imaging in freely moving mice using this probe.

Figure 1(a) shows the configuration of the ORPAM imaging system. The excitation laser beam is emitted from a 532 nm pulsed laser (GLPM-10, IPG, USA) with a repetition rate of 50 kHz, a single pulse energy of 20 µJ, and a duration of 2 ns. Before being coupled into the single mode fiber (SMF), the laser beam is reshaped through an iris (ID12, Thorlabs, USA) and filtered via a customized spatial optical filter consisting of two lenses and a 25 µm high-power pinhole (P25, Thorlabs, USA). A customized optical and electrical rotatory joint is employed to prevent extensive twisting of the optical fiber and electrical wires during the experiment. The imaging probe contains miniaturized optical components for collimation and focusing, a micro-electro-mechanical-system (MEMS) scanner for raster scanning, and a customized small piezoelectric ultrasonic detector for PA signal detection.

 

Fig. 1. Configuration of the ORPAM system and schematic of the miniaturized imaging probe for brain imaging in freely moving mice. (a) The setup of the imaging system. Obj, objective lens; SMF, single mode fiber; RJ, optical and electrical rotatory joint; C, cable; Amp, amplifier; FG, functional generator. (b) The detailed configuration of the probe. CG, cover glass. (c) The probe 3D rendering. (d) Photographs of a well assembled imaging probe and a C57 mouse wearing the imaging probe. The inserted photo shows the weight of the probe. See Visualization 1.

Download Full Size | PDF

Figure 1(b) shows the detailed internal structure of the probe. The fiber tip is fixed in a 1.25 mm diameter ceramic ferrule. An aspheric lens (#83-616, Edmund, USA) collimates the laser beam emitted from the fiber tip. A plano-convex lens (#45-272, Edmund, USA) with a focal length of 12 mm focuses the collimated laser beam. A ${3}\;{\rm mm} \times {3}\;{\rm mm} \times {3}\;{\rm mm}$ prism reflects the focused laser beam to the MEMS scanner with a mirror diameter of 1 mm. It can achieve the maximal optical scanning angle of $\pm {10}^\circ$ with the maximal driving voltage of 4 V. The MEMS scanner then reflects and scans the laser beam to propagate through a water cube with a thin cover glass tilted with an angle of 45° to illuminate the tissue surface. The cube allows the full transmission of the laser beam and partial reflection of the generated acoustic waves [21]. Scanning control is implemented by a function generator (FG) card (PCI-6733, National instruments, USA). By changing the driving voltages of four actuators, the MEMS mirror can achieve a two-dimensional raster scanning of the laser beam on the tissue surface. The induced PA signals are reflected by the 45° tilted cover glass and detected by a customized ultrasonic transducer with a center frequency of 10 MHz, a bandwidth of 80%, and an aperture size of 3 mm in diameter. The signals are amplified with a gain of ${\sim}{66}\;{\rm dB}$ and acquired by a high-speed data acquisition (DAQ) card (ATS9350, Alazartech, Canada) at a sampling rate of ${\sim}{125}\;{\rm MS/s}$. In the study, we used a laser power of 20 mW at a repetition rate of 50 kHz, resulting in single pulse energy of 400 nJ, to perform the in vivo experiments. All of the experimental results showed that there was no obvious damage to the mouse brain after long-term continuous recording.

Figure 1(c) shows the three-dimensional (3D) rendering of the image probe. In order to realize the detachable feature of the probe from the mouse head smoothly and accurately, we designed three feet outside the shell of the probe. Three magnets are attached to the tips of the feet. A mounting base is also designed, with three magnets corresponding to the feet on the probe. The mounting base is attached to the mouse skull by tissue adhesive and dental powder to position and immobilize the probe. The outer size of the probe is ${12}\;{\rm mm} \times {6}\;{\rm mm} \times {20}\;{\rm mm}$. Figure 1(d) exhibits the photograph of a well assembled imaging probe and a C57 mouse wearing this probe. The inserted photo shows that the weight of the probe is less than 1.8 g, which is light enough for an adult mouse. Besides, Visualization 1 shows the behavior of the mouse wearing this probe.

We first evaluated the key parameters of this probe, including FOV and lateral and axial resolutions. To measure the FOV of this probe, we successively scanned the cerebral cortex of a C57 mouse using the proposed probe and a previously reported rotatory scanning ORPAM (RS-ORPAM) system [22]. With a known FOV of the RS-ORPAM system, we can estimate the FOV of this probe. The maximum amplitude projection (MAP) images of the mouse brain acquired by the probe and the RS-ORPAM system are shown in Figs. 2(a) and 2(b), respectively. The FOV of the RS-ORPAM system is a 10 mm diameter circular area, and the corresponding FOV of the probe is illustrated by a dashed box. Comparing these two MAP images, we may estimate the FOV of this probe to be ${\sim}{3}\;{\rm mm} \times {3}\;{\rm mm}$. To evaluate the lateral resolution, we imaged a sharp edge of a surgical blade and obtained the edge spread function (ESF) curve, as shown in Fig. 2(c). The black dots and curve in Fig. 2(c) represent the normalized raw and fitted cross sectional profiles, respectively. The red curve in Fig. 2(c) shows the derived line spread function (LSF) with an estimated lateral resolution of 2.8 µm. Theoretically, the lateral resolution is approximately 2.5 µm with a numerical aperture (NA) of 0.13, which is slightly better than the experimental value. The major reason for this discrepancy is probably caused by the mismatch of the optical index inside the water cube. To evaluate the axial resolution of the probe, we derived the envelope of a PA signal generated by a single carbon fiber. The envelope is obtained by calculating the absolute value of the Hilbert-transformed PA signal waveform, as shown in Fig. 2(d). By measuring the full width at half-maximum (FWHM) of the Hilbert-transformed PA signals, the axial resolution of the system is estimated to be 165 µm and agrees well with the theoretical value.

 

Fig. 2. Performance of the miniaturized imaging probe. (a) The MAP image of a C57 mouse brain acquired by the miniaturized probe. (b) The MAP image of the same C57 mouse brain acquired by the previously reported RS-ORPAM system with a circular FOV of 10 mm in diameter. (c) The ESF and derived LSF of a sharp surgical blade. The inserted image is the MAP image of the surgical blade. The position of the profile is denoted by an orange dashed line. (d) The original PA signal of a single carbon fiber and its Hilbert-transformed waveform before and after Gaussian fitting. The inserted image is the MAP image of carbon fibers buried in agar phantom. The scale bar is 1 mm.

Download Full Size | PDF

We then conducted in vivo experiments as a further evaluation of this probe. We first anesthetized and maintained a C57 mouse by using the isoflurane (concentration: 2% Vol; gas velocity: 0.4 L/min), then depilated its brain to avoid the influence of hair. After that, we removed the scalp while keeping the skull intact and carefully attached the mounting base to its skull by using tissue adhesive and dental powder. The imaging probe was then fixed on the mounting base by magnets. To ensure ultrasonic transmission, the gap between the imaging probe and the mounting base was filled with medical ultrasonic gel. After the mouse was fully awake, we started our experiment to record a series of cortical images of the mouse in a large transparent polymethyl methacrylate (PMMA) barrel. All of the experimental procedures were approved by the ethics committee at the Southern University of Science and Technology (SUSTech), and all animals were sacrificed using the SUSTech-approved standard procedure after the experiment.

We carried out longitudinal imaging of the cerebral cortex in a freely moving mouse for 40 min to test the short-term stability of this probe. Considering the tradeoff between scanning speed and FOV of the MEMS mirror, we adjusted the scanning parameters to balance the acquisition time and image quality. In this experiment, we acquired 500 A-lines within a B-scan and a total number of 500 B-scans to form a MAP image with a size of ${500} \times {500}$ pixels that takes 5 s. Figure 3(a) shows the MAP images acquired at the 0th min. Figure 3(b) represents the overlapped MAP in the 0th and 40th min. The result demonstrates that there are only a few discrepancies, which are marked by the white dashed boxes. In the remaining part, the first and last images have a perfect match, and no obvious differences are observed. The curve in Fig. 3(c) quantifies the correlation of these MAP images over the entire acquisition time. We can see that the correlation of the entire acquisition series is higher than 90% and has a correlation variation of ${\lt} {10}\%$, which demonstrates that the miniaturized probe acquires high-quality ORPAM images with high stability and spatial resolution in the brain of freely moving mice. Visualization 2 displays these MAP images during the acquisition time.

 

Fig. 3. In vivo short-term freely moving imaging experimental results of C57 mice. (a) The MAP image of the continuous scanning experiment at the 0 min. (b) The overlapped first and last MAP image of the acquisition series. The white dashed boxes mark the slight discrepancies between them. (c) The correlation of the MAP images over the entire course of the experiment. (d) The MAP images of the entire imaging domain and four arbitrarily selected areas within the original FOV. See Visualization 2.

Download Full Size | PDF

Besides, we applied different initial bias voltages to four independent actuators of the MEMS mirror to achieve region-of-interest imaging within the original FOV. This scanning mechanism is beneficial for focusing on areas of interest for different experimental models. Figure 3(d) represents both MAP and four sub-images of select areas in a mouse brain. We reduced the number of A-lines and B-scans to 200 for the sub-image with a suppressed FOV of ${1.2}\;{\rm mm} \times {1.2}\;{\rm mm}$. The time cost is thus reduced to about 0.8 s, resulting in an increased volume rate of 1.25 volume/s.

To illustrate the detachable feature of the probe, we also performed long-term monitoring experiments on a C57 mouse brain for 7 days. We first fixed the mounting base on the mouse skull by using tissue adhesive and dental powder. Each day, the probe is attached onto the mounting base and immobilized through three magnets. We maintained the skull moisture with another homemade holder and filled the holder with medical ultrasonic gel after daily experiments. Figure 4 presents the imaging results in the 1st, 3rd, 5th, and 7th days. All images clearly show an intact vasculature with large and small vessels and capillaries. However, there are still some slight changes in vascular morphology. Figure 4(b) shows the relative vascular ratio of the area marked by the white dashed box. The statistical result shows that the number of blood vessels slightly decreases over time due to the long-term exposure of the skull. Although we filled the ultrasonic gel on the skull after daily experiment, it became dry in next day, and some marginal small vessels disappeared. Besides, the removal of the mouse scalp might induce inflammation of the tissue and physiological reactions of the mouse. Hence, the diameters of vessels such as V1 and V2 marked in day 1 show a significant increase, and the imaging of the sagittal sinus was varied in the next few days [Fig. 4(c)]. The quantitative parameters in Figs. 4(b) and 4(c) were normalized, and the day 1 data served as the baseline.

 

Fig. 4. (a) Long-term freely moving imaging results of a C57 mouse brain. (b) The relative vascular ratio of the area marked by the white dashed box. (c) The relative diameter changes of V1 and V2 marked by orange lines.

Download Full Size | PDF

In this Letter, we presented a miniaturized ORPAM probe and demonstrated that it is suitable for investigating brain hemodynamics in freely moving mice. The probe contains the optical fiber, two lenses, one prism, one MEMS scanner, and one miniaturized ultrasonic transducer. The outer size of this probe is ${12}\;{\rm mm} \times {6}\;{\rm mm} \times {20}\;{\rm mm}$, and it weighs ${\sim}{1.8}\;{\rm g}$, which is less than 10% of an adult mouse. Both the size and weight of this probe are suitable for adult mice to wear. The lateral resolution of this probe is measured to be 2.8 µm, which is sufficient to resolve most capillaries in the cerebral cortex. The maximum FOV of this probe is estimated to be ${\sim}{3}\;{\rm mm} \times {3}\;{\rm mm}$, which is large enough to cover several key brain regions in reaction and motion control. Besides, a random and adjustable area scanning mechanism allows for observing the interested area for specific animal model with the requirement of a fast imaging speed. The imaging quality and stability of this probe are evaluated by short- and long-term in vivo experiments. In short-term experiments, the correlation variation of the MAP images acquired is always under 10%, indicating a high stability of this probe. As for long-term experiments, the detachable design of this probe also makes it convenient. A C57 mouse is imaged every day within a week. The performance of this probe in long-term experiments is not as good as its performance in short-term experiments. We observed slight displacements and lateral shifts that might be caused by the relative movements of the brain to the skull. From these results, we may conclude that our probe will be a promising tool for investigating mice brain activities in freely moving conditions.