1. Introduction

Two-photon microscopy (2PM) leverages near-infrared wavelength excitation light, characterized by reduced scattering and absorption, to effectively explore deep brain regions within biological tissues. Because two-photon excitation efficiency is quadratically proportional to the photon density, the excitation light is usually focused on a single spot and the focal plane is raster-scanned via galvanometer-based scanning mirrors [1,2]. Unlike two-photon imaging in a single focal plane, two-photon volumetric imaging requires a greater number of light pulses to excite numerous voxels across a large area. Thus, to achieve volumetric imaging with a high voxel rate, techniques capable of enabling axial scanning while increasing the number of excitation pulses and/or decreasing effective voxels are required. Integrating such techniques into volumetric imaging is essential for elucidating neuronal functions in living animals [3].

The past 20 years have witnessed the development of various methods aimed at enhancing scanning efficiency or voxel rate through a standard single-beam 2PM expansion. Temporal multiplexing of excitation pulses is an effective volumetric imaging technique [4–7]. This method splits an excitation pulse into 4–80 sub-pulses with different optical paths, allowing each pulse to be focused on a separate position with several nanoseconds delay. However, the implementation of such highly optimized optical and electronic systems is complex, which may limit their usability to optics experts. Another approach involves point spread function (PSF) engineering [8–10]. This method utilizes an axially elongated focus (e.g., a Bessel beam), enabling a z-projection image acquisition of 3D volume by single lateral scanning [11–13]. This approach suffers from fluorescence signal overlap along the z-axis, and several solutions have recently been suggested [11,14,15]. The third method entails optimizing the scanning trajectory based on 2D spiral trajectories with continuous axial scanning using a piezo-z-scanner [16]. However, concerns arise regarding difficulties in correcting unavoidable motion artifacts in in vivo brain imaging [17], as the relationship between the spatial position and temporal order of the scanned voxels is less consistent compared with raster-scanning.

Multibeam 2PM with a confocal spinning-disk scanner and a 2D detector demonstrated higher scanning efficiency than standard single-beam 2PM in the late 1990s [18]. In a spinning-disk scanner, microlenses are arranged in a spiral pattern with a constant pitch on the disk, and its rapid rotation enables lateral scanning. The excitation beam is split into multiple foci through a single microlens-array disk, and this mechanics may contribute to the simplicity of an efficient 3D scanning system for volumetric imaging. A recent study utilized the spinning-disk system for in vivo one-photon Ca^2 + imaging in the neocortex in a 1-mm² field of view (FOV) with an 8 K CMOS camera [19]. Integrating a microlens array and line scan into in vivo two-photon imaging using a galvanometer-based scanning mirror yielded a 1-kHz frame rate [20]. Multibeam 2PM with a confocal spinning-disk scanner has recently been optimized by modifying the pinhole size and pitch [21] and introducing a high-peak power excitation laser light source [22,23]. Despite utilizing the multibeam approach in bioimaging and several technical improvements, its application to in vivo two-photon volumetric mouse brain imaging remains largely unreported. Hence, the multibeam approach harbors untapped potential for further development.

In this study, we implemented axial scanning into a spinning-disk 2PM based on a remote focusing approach using an electrically tunable lens (ETL) [24–26]. By incorporating both axial and lateral scanning mechanisms, we developed a multibeam continuous axial scanning 2PM (MCAS-2PM) for volumetric imaging. Compared to the previously devised volumetric imaging system as extensions of standard single-beam 2PM, MCAS-2PM offers the benefits of a simpler optical and control system, an easily adjustable axial resolution, and stable four dimensional (4D) reconstruction from planar images acquired in a living mouse brain.

2. Materials and methods

2.1 Optical configuration of MCAS-2PM

The MCAS-2PM system was built on an upright microscope (BX51WI, Olympus) configured with a 25× 1.05 NA water-immersion objective lens (XLPLN25XWMP, Olympus) and a Yb-based laser light source (1,042-nm wavelength, 4-W average power, 10-MHz repetition rate, 350-fs pulse width, femtoTrain, Spectra-Physics, Inc.). A confocal spinning-disk (CSUMPΦ100, Yokogawa Electric Corp.) was employed as a multibeam lateral scanner that splits an excitation beam into multiple foci through the microlens-array disk [21,22]. To ensure sufficient laser power for two-photon excitation, an excitation beam was narrowed and split into 40–60 beamlets in the effective FOV [21]. The diameter of the excitation beam, corresponding to the FOV, was unchanged through all experiments. Axial scanning in MCAS-2PM was implemented by adjusting the excitation beam's divergence angle using an ETL (EL-16-40-TC-VIS-5D, Optotune Switzerland AG) [24], prior to introducing the beam into the objective lens. A convergence beam brings the focal spot closer to the objective lens than the nominal working distance. Inversely, a divergence beam takes the focal spot far away from the objective lens (Fig. 1(a)). For axial scanning with a constant z pitch (axial range of xy-images), a sawtooth waveform was adopted to drive the ETL (Figs. 1(b), (c), and S1). For every volumetric imaging session, the proposed system tweaked the sawtooth drive-waveform for the specific axial range based on the nonlinear relationship between axial shifts and ETL current (Fig. S2(a)). This relationship was measured using a fluorescent bead phantom with a refractive index of 1.33. Fluorescence signals were captured as monochromatic images using an sCMOS camera (2,304 × 2,304 pixels, ORCA Fusion BT, Hamamatsu Photonics K.K.; Fig. 1(d)).

 

Fig. 1. Schematic of MCAS-2PM. (a) Axial scanning based on different focal lengths of the incident beam controlled by an ETL. Convergence, parallel, and divergence incident beams, produced by the respective ETL current (I⁽⁺⁾, I₀, I^(-)), focus at z₀− Δz, and z₀, z₀ + Δz, respectively. (b) 3D reconstruction of example xy-images captured during axial scanning. Colored areas represent xy-images corresponding to their respective depths. z_range indicates the total axial range of the field of view. z_pitch indicates the axial range for each xy-image. (c) Relationship between exposure time and z_pitch during axial scanning with a sawtooth drive waveform. Colored areas indicate the example xy-images shown in (b). (d) Optical and electronic layout of MCAS-2PM. HWP: Half-wave plate; GLP: Glan-laser polarizer; DM: Dichroic mirror; MLA: Microlens array; Ex.: Excitation light; Fluor.: Fluorescence; DAQ: Data acquisition board.

Download Full Size | PDF

2.2 4D image reconstruction

4D (xyz-t) images were reconstructed from xy-images with respective axial information, continuously acquired in volumetric imaging. To determine the axial range of each xy-image, the driving voltages and corresponding timings recorded by the data acquisition boards containing the information on tick counts, ETL currents, and camera triggers (i.e., elapsed time, axial positions, and frame numbers, respectively) were integrated with each other. Using this information, the corresponding area of 3D space for each frame was updated frame-by-frame basis, and the entire set of 3D information at the time was output as a part of the 4D image at a specified interval. The xy-images acquired while the focal plane returned to the initial z-position (approximately 10 ms) were excluded from the image reconstruction.

The FOV of each xy-image changed during axial scanning due to ETL-induced alteration in the focal length of the objective lens, as previously reported [24]. Therefore, in the 4D image reconstruction, the FOV changes were corrected by resizing the xy-images to match the FOV and the pixel size of the xy-image captured at an ETL current of 0 mA, using 2D bilinear interporation. In this study, a scaling factor was derived from the measured data of a 1-μm bead phantom with a 25× 1.05 NA objective lens (Fig. S2). Briefly, the xy-images were resized at a rate of approximately −1% per 10 μm axial shifts (see Supplement 1 for details).

The image reconstruction code was written by Python3 and powered by the Tifffile package [27] to create 4D image files in the OME-TIFF format. The 3D and projection views were created using an image analysis software (NIS-Elements AR, Nikon; ImageJ/Fiji [28,29]).

2.3 Relationship between imaging speed and axial resolution

In the MCAS scheme, the relational expression of the key parameters for volumetric imaging can be written as follows [Eq. (1)]:

2.4 PSF evaluation and volumetric imaging with fluorescent bead samples

For PSF measurements, fluorescent orange beads with a diameter of 0.2 µm (FluoSpheres, 0.2 µm, orange (540/560), Life Technologies Corp.) embedded in 1% agarose were observed using a 25× 1.05 NA water-immersion objective lens (XLPLN25XWMP, Olympus) with a 1,042-nm excitation beam. The lateral pixel size was 0.104 μm. The axial step size of the z-stack images was 0.2 µm controlled by a mechanical focus drive motor. To calculate the full width at half maximum (FWHM) of the PSF, the fluorescence bead intensity profiles were fitted by the Gaussian function for the x-, y-, and z-axes using a custom code of ImageJ/Fiji. For PSF measurement at various ETL currents, the focal plane was consistently set near the coverslip (∼10 µm) to minimize imaging depth-induced aberrations.

To demonstrate MCAS-2PM-based volumetric imaging, fluorescent Nile red beads with a diameter of 1.0 µm (FluoSpheres carboxylate-modified, 1.0 µm, Nile red (535/575), Life Technologies Corp.) embedded in 1% agarose were used.

2.5 Cranial window surgery

Male Thy1-EYFP-H transgenic mice [30] (18-week-old) expressing enhanced yellow fluorescent protein (EYFP) in cortical neurons [31] were used to examine the penetration depth of the proposed imaging system in vivo. Male wild-type C57BL/6J mice (5–9-week-old) were used for in vivo volumetric Ca²⁺ imaging. All mice were housed under a 12 h/12 h light/dark cycle. The experimental procedures were approved by the Institutional Animal Care and Use Committee of the National Institutes of Natural Sciences. The facility used for the care and management of the laboratory animals was approved by the Institutional Animal Care and Use Committee of the National Institute for Physiological Sciences (Approval number: 23A062).

For in vivo mouse brain imaging, the opaque skull overlying the observation area was replaced with a glass coverslip. First, the mice were anesthetized with 0.5%–1.5% isoflurane, and their bodies were warmed by a disposable heat pad during all experiments. Local anesthesia with 2% xylocaine was applied to the surgical field, and their skin was incised to expose their skulls. A custom-made stainless steel head chamber was attached to the skull centered on the right parietal bone, and glycerol (Taiyo Pharma Co., Ltd.) was intraperitoneally administered (15 μL/g) to decrease the intracranial pressure and loosen the dura mater. About 15 min after the administration, the skull overlying the parietal lobe was partially removed for an ∼4.2 mm diameter circle using a dental drill. The exposed brain was sealed using a Φ4.2-mm glass coverslip (0.17-mm thickness; Matsunami Glass Ind., Ltd.), cyanoacrylate glue (Aron Alpha, Toagosei Co., Ltd.), and an ultraviolet curable resin (Luxa Flow Star, Yoshida Dental Trade Distribution Co., Ltd.).

For multicell bolus loading of the Ca²⁺ indicator, after skull removal, the dura mater was also removed for smooth insertion of a glass pipette, and the brain was left unsealed. After injecting the Ca²⁺ indicator into the cortex, the exposed brain was sealed following the described procedure.

2.6 Dye loading and in vivo Ca²⁺ imaging

For in vivo Ca²⁺ imaging, the multicell bolus loading approach [32,33] was employed to introduce synthetic Ca²⁺-sensitive acetoxymethyl (AM) dyes into the cortex. A red Ca²⁺ indicator (Cal-590 AM), which was effectively excited at 1,050 nm [34], was used owing to its compatibility with the 1,042 nm excitation laser light source employed in this study. Cal-590 AM dye (50 μg, AAT Bioquest, Inc.) was dissolved in DMSO + 20% Pluronic F-127 (i.e., 5-mL DMSO + 1-g Pluronic F-127). After dissolving the dye to 1.5 mM in a solution containing 150-mM NaCl, 2.5-mM KCl, and 10-mM HEPES at pH 7.4, dextran-conjugated Alexa Fluor 488 (25 μM, 10,000 MW; Thermo Fisher Scientific, Inc.) was added for fluorescence guidance during bolus loading. A glass pipette was made from borosilicate glass capillaries (GD-1.5, Narishige) using a micropipette puller (P-1000, Sutter Instrument Co.). The pipette tip diameter was adjusted to approximately Φ20–30 μm under the bright field microscope (SMZ25, Nikon). The glass pipette was backfilled with an ∼10-μL dye solution.

Dye loading was performed via standard single-beam scanning 2PM (A1R-MP, Nikon) equipped with 16× 0.8 NA water-immersion objective lens (N16XLWD-PF, Nikon) and a Ti:Sa laser (MaiTai eHP, Spectra-Physics, Inc.) tuned to a 920-nm wavelength. The glass pipette was angled at ∼30° and slowly inserted into the cortex of anesthetized mice at 150–300-µm depth from the surface using a 3-axis manipulator. To eject the dye from the pipette, 2–3 psi air pressure was applied for ∼5 min using a custom tool (made from a syringe and F-clamp) with a pressure monitor (PM015D, World Precision Instruments, Inc.). About 30–45 min after the injection, spontaneous Ca²⁺ activity was briefly confirmed using the same microscope setup at 1,000–1,040-nm excitation wavelength. An in vivo volumetric Ca²⁺ imaging in awake mice with MCAS-2PM was started 1 h after the dye loading, and it lasted up to 6 h.

To avoid photodamage to the mouse brain, the excitation laser power was evaluated. We observed that a laser power of 190 mW or higher under the objective lens induced photodamage, characterized by persistently abnormally high calcium concentrations in cell bodies around the center area of the shallowest focal plane (approximately 130 µm from the surface). Consequently, the excitation laser power was set to 160 mW under the objective lens, which did not induce photodamage during in vivo Ca²⁺ imaging.

2.7 Data analysis

To evaluate the penetration depth in in vivo imaging, signal-to-background ratio was calculated as the ratio of the peak intensity to the background intensity. The background intensity was defined as the mean value of a 20 × 20-pixel area within a 50 pixels radius from the peak intensity position.

The reconstructed in vivo volumetric Ca²⁺ imaging data were processed using the CaImAn software package [35] to extract Ca²⁺ transients in neuronal populations. Motion correction of a 3D volume was performed with NoRMCorre, a piecewise rigid motion correction algorithm [36]. A constrained nonnegative matrix factorization algorithm was used to extract spatial footprints and temporal fluorescence traces of neurons [37]. Background and neuropil signals were subtracted in the CaImAn pipeline using a three-component background model. The normalized changes in fluorescence intensity relative to the baseline, ΔF/F, were calculated after detrending the fluorescence traces with a 30-second time window. Finally, the candidate regions of interest (ROIs) were manually verified based on their shape and size.

3. Results

3.1 PSF measurement

To evaluate the spatial resolution of the volumetric imaging system and potential changes in PSF while axial scanning with ETL, PSFs were measured using 0.2-μm orange beads embedded in 1% agarose. Measurements were conducted using 25× 1.05 NA water-immersion objective lens at various axial shifts ranging from +200 to −200 μm (Figs. 2(a)–(c)). This range corresponded to an ETL current range from approximately −211 to 145 mA (Fig. S2). In the ideal condition, characterized by parallel incidence of the excitation beam into the objective pupil via an ETL, the theoretical FWHM values of the PSFs for x-, y-, and z-axes were 0.345 µm, 0.422 µm, and 1.29 µm, respectively. The results showed that FWHMs increased according to the axial shift from the initial position of z = 0 (Figs. 2(a)–(c)), as previously reported [24]. The FWHM values for the y-axis exceeded those for the x-axis, along the polarized direction of excitation light [22]. Overall, the PSFs were smaller than the size of neuronal somata (10–15 μm), enabling volumetric imaging with single-cell resolution within 400 μm along z-axis.

 

Fig. 2. Characteristics of optical property and volumetric imaging in bead phantoms via MCAS-2PM. (a–c) PSF measurements along the x-, y-, and z-axes under various axial shifts by ETL using 0.2-µm orange beads. The positive axial shifts indicate deeper penetration into the sample. All FWHM data were represented as mean ± SEM (n = 11). (d, e) 3D view of the reconstructed bead phantom of 1-µm Nile red beads in 180 × 180 × 400 µm³ area with volume rates (f_z) of 1 and 3 Hz for (d) and (e), respectively. (f, g) xz-slice images of the volumes shown in (d) and (e), respectively. (h, i) xz-images cropped from (f) and (g), respectively. The distances between dotted gray lines indicate z-pitches, the range of quantized fluorescence intensities in the different volume rates.

Download Full Size | PDF

3.2 Volumetric imaging of bead phantom via MCAS-2PM

To demonstrate the capability of MCAS-2PM, volumetric imaging was performed in a 1-μm Nile red bead phantom. A 180 × 180 × 400 μm³ area was imaged at a volume rate of 1 or 3 Hz and a frame rate of 80 Hz with an effective pixel count of ∼1,700 × 1,700 pixels/frame. The excitation laser light, with a pulse repetition rate of 10 MHz, was split into ∼60 foci. Based on these parameters, the voxel rate and the number of excitation pulses were estimated as ∼230 M voxels/sec and ∼600 M pulses/sec, respectively. The excitation laser light intensity was 330 mW under the objective lens. Figures 2(d)–(i) shows the 3D and y-projection views with volume rates of 1 and 3 Hz. In the 3D views, most of the beads were clearly resolved through the FOV (Figs. 2(d), (e)). Comparing the volume rates of 1 and 3 Hz, the z_pitch was increased from 5 to 15-µm (Figs. 2(f)–(i)), which resulted from the faster movement of the focal plane within a constant exposure time in accordance with (Eq. (1)). Additionally, although the axial intensity distribution of the fluorescent beads in the 3-Hz setup expanded more than that in the 1-Hz setup, the same beads were still visualized. These results indicated that sawtooth-driven continuous axial scanning provided uniformly sampled volumetric images with negligible spatial gaps along the z-axis.

3.3 In vivo deep imaging in mouse brain via multibeam 2PM

To examine the penetration depth of multibeam 2PM equipped with ETL as an axial scanner, the neocortex of an anesthetized Thy1-EYFP-H mouse was observed through the cranial window. The z-stack image was captured in a 170 × 170 × 500-μm³ area (Fig. 3(a)). Since the evaluated range of ETL (±200 µm; Figs. 2(a)–(c)) was insufficient to image the entire axial range of 500 µm, the ETL was partially used for axial scanning from a depth of 200 to 500 μm, alternative to a standard focus drive motor. The ETL current was set to 0 mA at a depth of 300 µm. To enhance the two-photon excitation efficiency in deeper regions, the excitation laser light intensity was gradually increased from 160 to 380 mW under the objective lens, and the exposure time was adjusted from 1.0 to 1.5 sec/frame. As a result, apical dendrites and pyramidal neurons in the cortical layer 5 [31] were successfully visualized over a depth of 450 μm in the center area of 100 μm in diameter (Figs. 3(b)–(d)). As the observation depth increased, the FOV was gradually decreased. Signal-to-background ratios of the xy-images at the depth of 100 μm, 200 μm, and 460 μm were 20.1, 6.28, and 2.00, respectively (Figs. 3(b)–(d)). The z-projection images showed slight convergence towards the center due to FOV correction (Figs. 3(c), (d)). Overall, a multibeam 2PM with axial scanning via ETL enabled to visualize within the cortical layer 5 in a living mouse brain with subcellular resolution.

 

Fig. 3. In vivo deep imaging in Thy1-EYFP-H mouse brain via multibeam 2PM. (a) 3D representation of the z-stack image observed in the mouse brain neocortex in a 170 × 170 × 500-μm³ area using a 25× 1.05 NA water-immersion objective lens. (b–d) 3D views of the sub-regions from (a) at depth ranges of 50–150 μm (b), 150–250 μm (c), and 450–500 μm (d). The pseudo-color indicates the depth of the fluorescence signals. White arrowheads in (d) represent pyramidal neurons.

Download Full Size | PDF

3.4 In vivo volumetric Ca²⁺ imaging via MCAS-2PM

To investigate MCAS-2PM performance in volumetric Ca²⁺ imaging, spontaneous neuronal activity was observed in the living mouse brain using a red Ca²⁺ indicator, Cal-590 AM. For bolus loading of the Ca²⁺ indicator, a pipette containing a mixture of Cal-590 AM and dextran-conjugated Alexa Fluor 488 (10,000 MW) was carefully injected into the brain at a depth of 150–300 µm from the surface under fluorescence guidance using standard 2PM. Volumetric Ca²⁺ imaging was performed at a depth of 130–200 µm in a 100 × 100 × 70-µm³ area with a 4.8-µm z_pitch at 1.5 volume/sec. Because excitation pulses with a 1,042-nm wavelength rarely excite Alexa Fluor 488, almost all fluorescence signals came from Cal-590 and were detected with a single channel. The excitation laser light intensity was 160 mW under the objective lens and obvious photodamage was not induced during Ca²⁺ imaging. Figure 4(a) shows the maximum intensity projection (MIP) image along the time-axis, which visualizes the structures of several neuronal somata in volume. The fluorescence intensity variations in neurons were indicated as paired xy-images at two time points following the specific cell bodies at distinct depths (Figs. 4(b)–(d)). Figure 4(f) shows the spontaneous Ca²⁺ transients recorded from neuronal somata corresponding to the ROIs (Fig. 4(e)). The Ca²⁺ activity in neurons was observed in the region of 100 µm in diameter. These results demonstrate that MCAS-2PM can be applied to Ca²⁺ imaging up to a 200-μm depth from the surface of the living mouse brain.

 

Fig. 4. In vivo volumetric Ca²⁺ imaging of spontaneous activity with Cal-590 AM in mouse brain via MCAS-2PM. (a) Maximum intensity projection (MIP) image along the time-axis. The pseudo-color represents the depth of the fluorescence signals. (b–d) Paired xy-slice images showing an increase in fluorescence intensity in neurons at distinct depths (134 µm (b), 172 µm (c), and 188 µm (d)). Yellow arrowheads represent neuronal somata. (e) Projection view of the 3D ROIs. The pseudo-color of the ROIs represents the depth of their centroids. The ROIs are numbered sequentially from the shallowest to the deepest. (f) ΔF/F traces extracted from neurons corresponding to the numbered ROIs shown in (e). Offset: 0.075 × ΔF/F.

Download Full Size | PDF

4. Discussion

This study introduces an alternative optical design (MCAS-2PM) for in vivo two-photon volumetric imaging. By combining spinning-disk-based lateral scanning with continuous axial scanning enabled through an ETL, we visualized a 3D volume of 1-μm bead phantom across a z_range of ∼400 μm with a z_pitch of ∼5 μm (Fig. 2). Continuous axial scanning facilitates 3D scanning with no settling time during focal plane shifts, outperforming stepwise multiplane scanning, which costs a few tens of milliseconds for the damped oscillation at each focal plane. However, continuous axial scanning may introduce distortion to xy-images, especially in single-beam 2PM, due to different z-position at the early and later scanned areas within the same xy-image. In contrast, integrating multibeam scanning and 2D detection enabled the fast and quasi-uniform sampling of the focal plane every ∼3 ms under the maximum rotation speed (10,000 rpm) of the spinning-disk [22,23], thus reducing xy-image distortion. Consequently, 3D volumes were stably reconstructed as a z-projection stack with negligible axial spatial gaps (Figs. 2 and 4), which may help reveal the densely distributed neurons in a specific brain region, thereby revealing individual differences in neuronal representations in a functionally equivalent circuit across animals [38].

While axial scanning for volumetric imaging can be implemented using a standard piezo z-scanner, a properly calibrated ETL offers distinct advantages compared to piezo z-scanners, such as elimination of mechanical vibrations transmitted to the sample and easy commercial availability. In the use of an ETL with a sawtooth drive waveform at higher frequencies (≥10 Hz), oscillation during the reset movement has been reported [39], indicating that the part of the ETL drive waveform needs to be optimized for faster volumetric imaging. Additionally, the difference in the refractive index between the specimen and the calibration sample (e.g., fluorescent beads phantom), which is used to establish the conversion between axial shifts and the ETL current, can possibly lead to nonnegligible errors in the axial position, particularly when the axial range becomes extended.

The developed imaging system demonstrated a penetration depth of over 450 μm in the in vivo observation of cortical structures (Fig. 3), which surpasses previously reported depths achieved with the multibeam approach in living mouse brains: under 200 μm for one-photon systems [19,40] and 300 μm for two-photon system [20]. Moreover, we demonstrated volumetric Ca²⁺ imaging at a volume rate of 1.5 Hz with up to a depth of 200 μm, based on the video rate acquisition of xy-images while axial scanning (Fig. 4). Ca²⁺ transients were successfully observed in multiple neurons located at various z-positions. Using a higher frame rate can yield improvements in imaging parameters (volume rate, z_range, and z_pitch) and further enhance MCAS-2PM imaging performance, according to the relationship shown in Eq. (1). Recently, several effective techniques to enable a higher frame rate by improving fluorescence intensity and signal-to-noise ratio (SNR) have been demonstrated: axially confined two-photon excitation by temporally focused excitation pulses [4,41], localization of Ca²⁺ probes to neuronal soma for reducing the fluorescence emitted from axons and dendrites overlapped with the soma [42,43], and uniform illumination of excitation laser light across the entire FOV by converting Gaussian beam into flat-top beam [20,44]. In addition, self-supervised deep-learning denoising approaches have been reported to infer two-photon fluorescence from shot-noise limited signals [45–47]. By incorporating technologies that improve fluorescence intensity and SNR with less invasiveness into MCAS-2PM, improved volumetric imaging performance is anticipated.

The developed multibeam scanning 2PM-based volumetric imaging system has a simpler optical and electronic layout than standard 2PM-based systems using galvanometer-based scanning mirrors [5,6,11,12]. The simplicity of the multibeam approach results from the single spinning-disk scanner performing two indispensable roles: 1) splitting the excitation beam into multiple foci for increased voxel rate and 2) lateral scanning at the focal plane. Thus, unlike standard 2PM-based systems, our approach typically does not require additional optical modules for high-speed 3D scanning, such as multiplexing excitation pulses and/or PSF engineering. Additionally, in this study, volume images were reconstructed by sequentially arranging xy-images along the z-axis, and the z-pitches of xy-images followed the relationship between the imaging parameters (Fig. 2). Therefore, MCAS-2PM offers a simple yet flexible volumetric imaging platform, readily reconfigurable for imaging parameters through the control computer to suit various imaging purposes.

In summary, the proposed MCAS-2PM is a simple and practical volumetric imaging approach that effectively integrates multibeam lateral and continuous axial scanning. Using the constructed system, neuronal activity was successfully visualized in a living mouse brain. The feature of a simple microscope design may be paramount for volumetric imaging to be widespread in the future. Furthermore, we anticipate that the compact 3D scanner and control system, utilizing a spinning-disk, liquid lens, and 2D imager, could find applications in the medical field, especially for quick 3D visualization of pathological sections.