2.1.

Design of the Wireless Mini-Microscope System

The overall shell and optics of our system are based on the open-source, mini-microscope system developed at UCLA,² with our specially designed new electrical circuit part. The entire microscope has a volume of $\sim 2.8{\mathrm{cm}}^3$ and a mass of 2.7 g, and it is powered by a lithium battery mounted on the back of the mouse. The internal and circuit design of the microscope is shown in Figs. 1(a) and 1(b).

Fig. 1

Wireless mini-scope imaging system. (a) Design of the wScope. (b) Custom-built video RF and control/sensor PCBs, with views of the top and bottom components. (c) Image of an assembled wireless microscope. (d) Overview of the wireless microscopy system.

The developed wScope has a blue LED with a spectral peak at $\sim 475\mathrm{nm}$ (LXML-PB01-0040). The excitation light from the LED is first collimated by a half-ball lens before passing through an excitation filter (Chroma, ET470/40x, $3.5\times 4\times 1{\mathrm{mm}}^3$), and then reflected into the objective GRIN lens by a dichroic mirror (Chroma, T495lpxr, $4\times 6\times 1{\mathrm{mm}}^3$). The emitted fluorescence light collected by the objective GRIN lens sequentially passes through the dichroic mirror and the emission filter (Chroma, ET525/50m, $4\times 4\times 1{\mathrm{mm}}^3$), and finally forms the image onto a CMOS sensor (OV7960, Omni Vision Technologies, Inc) via an achromatic lens (Edmund Optics, 5-mm diameter, 15-mm FL).

As shown in Fig. 1(b), the customized printed circuit board (PCB) contains two parts connected by a flexible connector, which is folded to form spatially separated two layers, upper layer and lower layer. The upper circuit board includes a radio frequency (RF) chip; an antenna contacts for receiving control signals and transmitting video signals; and interfaces for power LED. The main components of the lower circuit board are a CMOS image sensor and a 2.4-GHz RF transceiver with an embedded microcontroller unit. This folded two-layer design not only reduces the amount of space needed but also effectively facilitates heat dissipation. No significant behavioral difference was observed in mice whether they wore the wScope or not.

The OV7960 image sensor generates streaming pixel data at a constant frame rate of 25 Hz with analog output. The image frames are wirelessly transmitted using an RTC6705, a 5.8-GHz band FM transmitter, which are received by an analog video receiver and then viewed and saved in real time using a video recorder [Fig. 1(d)].

Our system uses nRF24LE1 chips (Nordic Semiconductor), a 2.4-GHz RF transceiver with embedded microcontroller, to realize wireless system control. Several key parameters, including the CMOS gain, the brightness of the excitation LED, and the system standby function, can be controlled remotely. Each wScope is assigned a unique ID number so that its parameters can be controlled separately during the experiment without affecting other wScopes. The controller is a customized board that has another nRF24LE1 installed, which receives commands from the computer through a serial port (Fig. S1 in the Supplementary Material). The development board sends the instructions wirelessly using a 2.4-GHz RF signal.

Besides wireless image transmission and system control, our circuit board also reserved the wired connection ports to make the mini-scope can work as a wired version. This function helps system test and performance comparison.

2.2.

System Performance Testing

The imaging resolution is measured by imaging a USAF1951 Resolution Test Target, as shown in Fig. 2(a), which demonstrates a $\sim 2\text{-}\mu \mathrm{m}$ resolution.

Fig. 2

Characterization test of the system. (a) Imaging of a USAF1951 Resolution Test Target by the wScope. $2.19\mu \mathrm{m}$ can be resolved experimentally from the cross-sectional profile along the red line in left. (b) Battery continuous working times after different standby time. (c) PSNR at different distances. (d) Histogram of per frame PSNR and SSIM of the wireless conditions, as compared to the wired condition (at 2 m).

We measured continuous working time of batteries with different capacities (Table S1 in the Supplementary Material). Additionally, to verify the effectiveness of the standby function in conservation of battery life, we also measured the working time after half an hour and one hour of standby, respectively [Fig. 2(b)]. Balanced with battery life and weight, we chose a lithium battery with a capacity of 100 mAh, a mass of 1.7 g, and a size of $4\mathrm{mm}\times 10\times 25{\mathrm{mm}}^3$ for the in vivo experiment.

To evaluate the wireless image transmission quality of our wScope, we employed two parameters, namely, the peak signal-to-noise ratio (PSNR) and structural similarity (SSIM), to quantify the similarity between images by wireless and wired mini-scopes.^12,13 To calculate the PSNR, we first calculate the mean squared error (MSE). For two m × n images, $x$ (the wired image) and $y$ (the wireless image), the MSE is defined as

The PSNR is calculated from the MSE with the following formula:

MAX is the maximum value of the color scale that is applied to each point in the image. If each sampling point is 8 bits, then MAX is 255. The smaller the MSE is, the larger the PSNR, and the better the image quality. Then, the SSIM between $x$ and $y$ can be calculated as follows:

We evaluated the image quality as a function of the distance between wScope and the wireless receiver, as shown in Fig. 2(c). Taking the image at 0 distance as the reference image, we took images at 0.5-m intervals from 0 to 3 m, and the calculated PSNR values were all above 30. Thus, our system can guarantee image quality at a distance of 3 m.

To test the quality of the wireless transmission, we use the wScope to image the fluorescence micrometer and sent and analyzed video signals via both wired and wireless transmission to evaluate the quality of the images obtained in real time. Our wireless miniscope reserved the port for wired connection to be function like a traditional wired version, which is very useful for system debug and test. We also experimentally confirmed that the wired connect won’t affect simultaneously wireless image transmission. Therefore, when we compare the imaging quality between wireless and wired modes, instead of doing one after another, we connected the cable with the miniscope and acquire images from wired and wireless modes simultaneously. The PSNR value of the image received after wireless transmission was $\sim 28$ to 30 dB, and the SSIM value was $\sim 0.7$ to 0.8 [Fig. 2(d)], which demonstrates that our wireless transmission can guarantee reliable image quality.^12,15,16

3.1.

Animal Surgery

All animal experiments were conducted at Capital Medical University Laboratory Animal Center, an Association for Assessment and Accreditation of Laboratory Animal Care-approved animal facility. All animal experiments were undertaken in accordance with the Guide for the Care and Use of Laboratory Animals (eighth edition), and all experimental protocols were approved by the Institutional Animal Care and Use Committee of Capital Medical University. Wild-type C57BL/6 mice (male, aged 8–12 weeks) were purchased from Vital River Laboratories (Beijing, China) and maintained on a 12/12-h reversed dark-light cycle. All experiments were performed during the light cycle. Before surgery, the mice were anesthetized with tribromoethanol ($240\mathrm{mg}/\mathrm{kg}$, Sigma), and a craniotomy was performed over the primary visual cortex of the right hemisphere (coordinates with respect to bregma: anteroposterior, $-2.8\mathrm{mm}$; mediolateral, 2.5 mm; and dorsoventral, 0.2 mm). After the craniotomy, a 1.8-mm diameter GRIN lens (Edmund Optics) was placed on the surface of the cortex without damaging the tissue. The gap between the GRIN lens and the skull was covered with Kwik-Sil (World Precision Instruments) to protect the cortex. A metal baseplate was mounted on the skull using cyanoacrylate adhesive and dental acrylic. After surgery, the animals were allowed to recover for 1 week, during which they were intraperitoneally injected with ceftriaxone sodium ($200\mathrm{mg}/\mathrm{kg}$) and dexamethasone ($5\mathrm{mg}/\mathrm{kg}$) every day to prevent inflammation and edema. One week after the surgery, the mice mounted with a 3D printed dummy microscope on their heads, for rousting the locomotor behavior.¹⁷ After training, the mice showed robust locomotor behavior (Fig. S2 in the Supplementary Material).

3.2.

Experimental Area

To evaluate the performance of the wScope system for in vivo brain imaging, we built an arena and placed a camera above it to monitor the behavior of the animals. The size of this experimental area is $50\times 50\mathrm{cm}$. Two sides of the arena were made of transparent acrylic plates, allowing the animals to see visual stimuli presented on screens outside the transparent walls [Fig. 3(a)].

Fig. 3

In vivo CBF imaging in freely moving mice. (a) Illustration of mice fitted with the wScope system in an open arena. (b) Example image of brain vessel imaging using the wireless microscope system. (c) Changes in the diameter of the vessel during visual stimulation. (d) Merged image showing the vessel and the diameter change. (e) Example trace (c1) and line scan heatmap (c2) of vessel diameter change during stimulus presentation. (f) The absolute diameter changes for vessel dilation and contraction during visual stimulation. (g) The delays of the initial phase, the peak of the dilation phase and the peak of the contraction phase of vessels with respect to the onset of visual stimulation.

3.3.

Cerebral Vessel Imaging in Awake Mice

Before imaging, the animals were anesthetized with isoflurane. Fluorescein isothiocyanate-dextran (2% w/v, MW 70,000, 50 ml, Sigma) was administered into the caudal vein to label the blood plasma. A wScope was mounted on the mouse’s head, and the focal plane was adjusted manually to show a clear image of the vessels. One piece of self-gripping fastener material was taped to the battery, and the complementary piece was sutured to the back of the mouse. Subsequently, the wScope body and battery were connected with a soft thin wire. The gain and LED power were adjusted to the proper values (vessel is clearly visible, and no over exposure), and the device was powered down immediately to conserve battery life. All these adjustments were made by remote control. Then, the anesthetized animal was transferred to the experimental arena. The wScope was wirelessly turned on 10 min after waking. The behavior of the mouse was observed by the experimenters, and visual stimuli were presented for 2 s when the animal’s head was turned toward one of the screens placed behind two transparent walls of the arena.

After the presentation of stimuli, the vessels in V1 quickly began to dilate ($0.45\mathrm{s}\pm 0.11$) and reached their peak dilation amplitude ($19.46\%\pm 4.21$) in $1.66\mathrm{s}\pm 0.10$, followed by contraction in $5.69\mathrm{s}\pm 0.36$ to reach their peak contraction amplitude ($7.70\%\pm 1.57$), after which they gradually returned to baseline [Figs. 3(b)–3(d)]. The vascular kinetics were correlated with visual stimulation, and the amplitude of the vessel dilations was significantly higher than that of the contractions. Our data are well consistent with reported works.^18–20 These results demonstrate that our wireless microscope system can reliably record the changes in cerebral vessels in freely moving mice, enabling brain hemodynamic imaging studies for a variety of topics, such as cerebrovascular disorders and neurovascular kinetics.

The wScope has similar imaging performance to existing wired microscopes, while its wireless hardware allows the experimenter to study the behavior of multiple freely moving mice at once; it can even be used while the animals are in enclosed space (Fig. S3 in the Supplementary Material). Using wScopes, we successfully imaged 4 mice simultaneously in the same arena (Figs. S4 and S5 in the Supplementary Material). We further compared these data with that from previously published literature on cerebral hemodynamic changes in response to stimuli; our data were comparable to those in the literature, demonstrating the reliability of our system.

3.4.

In Vivo Recording the neuronal Activation Study Simultaneously

To demonstrate the in vivo real-time imaging performance for recording the neuron activities, we first imaged neurons in mice cortex. A virus (AAV2/9.syn.GCaMP6m.WPRE.SV40, diluted to a titer of $4\times {10}^{12}\mathrm{vg}/\mathrm{ml}$) was drawn into a prepared glass pipette (tip size: 0.3 to 0.5 mm) through negative pressure. To create a seal against the surface of the brain, the glass pipette was slowly lowered again by $\sim 450\mu \mathrm{m}$ to press on the pia without breaking through it. Viral infusion was started after 2 min and proceeded at $60\mathrm{nl}/\min$ for 10 min.²¹ The craniotomy was covered with a small piece of glass coverslip after infusion. The animals recovered for 7 days after the craniotomy and were intraperitoneally administered ceftriaxone sodium ($200\mathrm{mg}/\mathrm{kg}$) every day to prevent inflammation. These results show that our system has good sensitivity and resolution to study GCaMP fluorescence signals in mouse (Fig. 4).

Fig. 4

Example of GCaMP fluorescence signals in mouse (Video 1, MP4, 7.91 MB [URL: https://doi.org/10.1117/1.JBO.28.2.026503.s1]).

We next presented the performance of the system in simultaneously recording neuronal activity from another two free-moving mice. The location of V1 was determined using stereotactic coordinates. Four weeks after the injection of rAAV2/9-hsyn-GCamp6m in V1, the mice were trained using dummy microscopes for one week. During the experiments, the mice were allowed to freely explore and engage in social interaction in the arena. The arena and visual stimuli were the same as those in Fig. 3(a). The population neuronal activity in V1 was analyzed using extended constrained nonnegative matrix factorization (CNMF-E) and ImageCN.^21–23 The spatial components were extracted from the time-series data [Fig. 5(a) and 5(b)]. The temporal components of neuronal activities were also extracted, and calcium signals were distinguishable from the background noise [Fig. 5(c)]. The neural population in the V1 areas of mice demonstrated strong activity beginning $\sim 0.1\mathrm{s}$ after the onset of stimulus presentation [Fig. 5(d)]. To evaluate the performance of the wireless system in capturing the two recordings independently and simultaneously and confirm the absence of interference between them, we used the absolute value of the Pearson correlation coefficient, a statistic for measuring the linear relationship between each pair of neurons, to quantify the inter- and intragroup correlations of the neuron populations. The population correlation between the two mice was low during the dark phase, the signal received by two mice were independent, indicating that no interference was detected during simultaneous recording with two microscopes [Fig. 5(e)]. However, during visual stimulus presentation, the neuron populations tended to fire simultaneously. Thus, the correlation of activity between pairs of neurons in both mice was higher in this phase than during the dark phase, when these populations were quiescent [Fig. 5(f)]. The behavior of the animals was recorded with a camera; the corresponding movement trajectories of the two mice are shown in Fig. 5(g). Overall, we recorded the calcium transients of V1 neurons in two mice at the same area simultaneously, and no signal interference was observed between different channels of microscopes, indicating that our system provides a powerful tool for recording the activity of neural populations in multiple freely moving mice with excellent spatial and temporal resolution.

Fig. 5

Primary visual cortex imaging in two freely moving mice. (a) Average projection image from a 5-min recording session. (b) Extracted contours of neurons using CNMF-E, colored randomly. (c) Representative transient calcium traces extracted from the time-series data. (d) Heatmap of neuronal calcium transients during visual stimulation. The red dashed lines indicate the stimulus phase. (e) Matrix of the Pearson correlation coefficients between the calcium transients of pairs of neurons from the two mice during the dark phase. (f) Matrix of the Pearson correlation coefficients between the calcium transients of different neurons from the two mice during the dark phase (left) and the stimulus phase (right). The black line separates the neurons of two mice from each other; the red dashed lines group highly correlated neural populations. (g) Locomotion trajectories of two mice fitted with the wireless microscope system within the arena (right, Scale bar: 5 cm).