Recent advances in high-speed photoacoustic microscopy

Abstract

Photoacoustic (PA) microscopy (PAM) has achieved remarkable progress in biomedicine in the past decade. It is a fast-rising imaging modality with diverse applications, such as hemodynamics, oncology, metabolism, and neuroimaging. Combining optical excitation and acoustic detection, the hybrid nature of PAM provides advantages of rich contrast and deep penetration. In recent years, high-speed PAM has flourished and enabled high‐speed wide‐field imaging of functional activity. In this review, we summarize the most recent advances in high-speed PAM technologies, including high-repetition-rate multi-wavelength laser development, fast scanning techniques, and novel PA signal acquisition strategies.

1. Introduction

Photoacoustic (PA) imaging (PAI) or optoacoustic imaging has attracted increasing attention in biomedical research. It exploits a physical phenomenon called the PA effect, in which acoustic waves are generated by irradiating tissue with pulsed or amplitude-modulated optical radiations [1,2]. Combining optical illumination and ultrasonic detection, PAI harnesses both rich optical absorption contrast and deep ultrasound penetration [3,4]. The hybrid nature of PAI provides it with the key advantage of high scalability of imaging depth and spatial resolution. Besides, PAI identifies inherent sensitivity of optical absorption, as PA signal variations are based upon the unique optical absorption features of biological tissues. With these merits, PAI has been extensively applied in anatomical, functional, molecular, and histological studies.

PAI technologies can be summarized into two major categories: photoacoustic computed tomography (PACT) and photoacoustic microscopy (PAM) [5,6]. In PACT, images are reconstructed with full-field illumination and simultaneous acoustic detection at multi-spatial points using a multi-element transducer array [[7], [8], [9]]. It exhibits superior volumetric imaging with ultrasound spatial resolution beyond the optical ballistic limit, achieving a deep penetration up to a few centimeters. Distinct from algorithmically reconstructed PACT, PAM employs raster‐scanning of optical and acoustic foci and forms images directly from a set of acquired A-lines. Based on the confocal and coaxial configuration of the optical excitation and acoustic detection, PAM achieves high-resolution imaging with up to sub-micron resolution [10,11]. PAM can be further classified into acoustic-resolution PAM (AR-PAM) and optical-resolution PAM (OR-PAM). OR-PAM is advantageous over AR-PAM at high-resolution imaging, as optical foci are usually more than ten times tighter than acoustic foci. In contrast, AR-PAM that utilizes acoustic focusing is advantageous in penetration depth than OR-PAM [12,13].

Inheriting the characteristics of PAI, PAM has explored a wide range of applications in biomedical studies [[14], [15], [16], [17], [18], [19]]. In the past decades, many efforts have been devoted to improving the performance from various aspects, especially the imaging speed. With the improvement of temporal resolution, high-speed PAM holds tremendous potential for investigating fast changes in blood flow speed, oxygen saturation, and total hemoglobin concentration [[20], [21], [22], [23]]. However, in PAM systems, images with a wide field of view (FOV) are typically achieved by point-by-point raster scanning of the optical and acoustic beams. Most reported conventional PAM systems use stepper motor scanning stages, which have a relatively low scanning speed, long preventing PAM from obtaining dynamic information, such as transient drug responses and brain functions [[24], [25], [26]]. In the past decade, a variety of novel high-speed PAM technologies have been invented, making exciting progress in imaging speed improvements. In this review, we summarize the recent developments in high-speed PAM from aspects of multi-wavelength laser, scanning mechanism, and acoustic detection.

2. Advances in multi-wavelength laser

Enhancing the laser performance, e.g., laser repetition rate and multi-wavelength output, is an endeavor to improve the PAM imaging speed. In the past decade, high-repetition-rate diode-pumped lasers have been successfully implemented by manufacturers to achieve high-speed PAM [9,12,27]. However, the limited wavelength choice of commercial lasers makes PAM challenging to image with multi-contrasts simultaneously. Thus, multiple-wavelength laser technologies for fast functional PAM (fPAM) are highly needed.

Supercontinuum sources coupled into a highly nonlinear fiber is one of those technologies. Bondu et al. [28] reported a supercontinuum source using a carefully designed tapered photonic crystal fiber, allowing higher supercontinuum output pulse energy of 15 nJ/nm in the 500–1600 nm wavelength range. Recent research by Dasa et al. demonstrated a compact all-fiber, high pulse energy supercontinuum laser with a pulse duration of 7 ns and a repetition rate of 100 kHz. Multi-spectral photoacoustic microscopy imaging of lipids was performed, both ex vivo on adipose tissue and in vivo [29].

Another technique is based on the stimulated Raman scattering (SRS) effect, which refers to the phenomenon that laser generates scattered light with wavelengths longer than the incident light through the optical fiber. The SRS effect-based laser is ideal for multi-contrast photoacoustic imaging. Parsin et al. first reported in vivo functional imaging using a multi-wavelength fiber laser source [30]. The outputs from varying fiber lengths, input pulse energies, and PRR were explored to optimize the SRS wavelength peaks. Eventually, this laser generated 532, 545, and 580 nm with pulse energy between 300 and 500 nJ and a 40 kHz PRR by adopting a 3 m polarization-maintaining single-mode fiber. Wang’s group [31] further developed an SRS-based multi-wavelength (532/545 nm and 558 nm) pulsed laser source with a wavelength switching in 220 ns for fast functional photoacoustic imaging (Fig. 1a). A 532 nm nanosecond pulsed laser with 1 MHz PRR was employed as the pump laser. The output beam was split into two: one was coupled into a 5-m polarization-maintaining single-mode fiber in a Raman path, and the other went through a 50 m optical fiber. As shown in Fig. 1b, this technique was successfully applied to imaging oxygen saturation in trunk blood vessels and capillaries using pulse energy of ∼100 nJ/pulse. Another five-wavelength (532, 545, 558, 570, and 620/640 nm) nanosecond pulsed laser based on the SRS effect further developed by this group is illuminated in Fig. 1c [32]. Utilizing this laser source, OR-PAM was able to simultaneously image lymphatic vessel, hemodynamic, and functional information in the blood vessel in a single scan (Fig. 1d). Images of the blood and lymphatic vessels in a 2.5 × 2.5 mm² region of mouse ear were obtained with a scanning step size of 2.5 μm and an acquisition time of 250 s. Moreover, a recent research of Raman-laser-based dual-wavelength fPAM based on a novel Raman crystal provides a new explication for Raman laser development. He et al. [33] adopted potassium gadolinium tungstate [KGd (WO₄)₂, KGW] crystal as Raman crystal. Compared with other popular Raman crystal, KGW demonstrated technical superiorities such as convenient operation, high thermal damage threshold, high Raman gain coefficient, low thermal lensing, and high thermal conductivity. As Fig. 1e shows, a dual-wavelength fPAM system was developed by employing two 532 nm picosecond-pulsed lasers. The laser beam was loosely focused into a 30-mm-long KGW crystal and converted to the first Stokes line at 558 nm by SRS effect. Combined with the MEMS scanner, the fPAM system obtained sO₂ response to single-impulse stimulation in mouse brain with a 6-Hz 3-D imaging rate (Fig.1f).

Fig. 1.

PAM with fast multi-wavelength laser source.

(a) Schematic of the PAM system with a stimulated-Raman-scattering-based multi-wavelength pulsed laser source. (b) In vivo imaging of oxygen saturation in the mouse ear [31]. (c) Schematic of five-wavelength OR-PAM. (d) Images and variation of compensated oxygen saturation, blood flow speed, the blood, and lymphatic vessels. (e) Schematic of the fPAM system based on Raman laser utilizing KGW crystal as Raman material [32]. (f) Images of sO2 response to single-impulse electrical stimulation in mouse brain [33].

3. Advances in scanning mechanism

3.1. Combined optical-acoustic scanning

Combined optical-acoustic scanning is adopted by PAM systems for high sensitivity. In these systems, fast scanning of both the optical beam and acoustic beam was employed by using voice coil scanner or mirror scanner.

3.1.1. Voice coil scanner

Voice-coil scanning is an important technique employed in high-speed PAM. As a direct-drive device that uses a permanent magnet field and a coil winding (conductor), the voice coil actuator has risen in recent years as one of the most viable alternatives to leadscrew driver, with advantages of simple structure, small size, and high speed [34,35]. As shown in Fig. 2a, Wang's group constructed a fast PAM system based on an existing reflecting optical resolution OR-PAM [34], using a voice coil motor instead of the conventional linear motor as the core scanning component. This voice coil PAM improved the B-scan frame rate to 40 Hz (with a scan range of 1 mm) for the first time, realizing the real-time imaging of a single red blood cell in mouse ear vessels (Fig. 2b). Utilizing fast voice-coil linear scanner, Wang et al. developed a single-RBC photoacoustic flowoxigraphy (FOG) to enable real-time imaging [36]. Oxygen delivery from single flowing RBCs in vivo was imaged by using two synchronized lasers to periodically generate laser pulses 20 μs apart at 532 nm (an isosbestic wavelength of hemoglobin) and 560 nm (a non-isosbestic wavelength), respectively. FOG achieved a micrometer-scale spatial resolution, >100-Hz 2D imaging speed, and 20-μs oxygenation detection time. Using this system, the authors imaged the dynamic processes of single RBCs delivering oxygen to local cells and tissues in vivo at the brain (Fig. 2(c)(d)).

Fig. 2.

Fast voice⁃coil scanning PAM.

(a) Schematic of voice-coil-driven fast-scanning OR-PAM. The scanning probe is mounted on a voice-coil linear translation stage. (b) In vivo B-scan of single red blood cell flowing in a mouse ear capillary and MAP along the z-axis of the B-scan image versus time. [34] (c) Schematic of single-RBC FOG. AMP, signal amplifiers and filters; BC, acoustic–optical beam combiner; DAQ, digitizer; FC, fiber coupler; L1, pulsed laser 1; L2, pulsed laser 2; PD, photodiode; SF, single-mode fiber; ST, stepper motor scanner (y-axis); UL, ultrasound lens; UT, ultrasound transducer; VC, voice-coil scanner (x-axis); W, water. (d) Representative sO2 MAP image of the mouse brain cortex.

However, the scanning speed of voice coil scanning is fundamentally limited by the driving force from the voice coil and the mass of the scanning head.

3.1.2. Mirror scanner

Galvanometer scanner is an excellent vector scanning device with the advantages of high precision and high speed. As an implementation of an oscillating mirror, it deflects with an angle proportional to the current in the magnetic field of an electrified coil, with widespread applications in laser marking, stage lighting control, laser drilling, medical cosmetology, etc. [[37], [38], [39], [40]]. However, the integration of galvanometer scanner into PAM system is restricted by its unsatisfactory performance in imaging SNR and FOV in aqueous environment [9,41,42]. Thus, Chulhong Kim’s group developed a novel high speed and high SNR galvanometer-based OR-PAM operating in electron fluoride (3 M Novec HFE-7500) [23], a kind of non-conducting liquid acoustic coupling medium. This system allowed acoustic and optical scanning with the combination of a galvanometer and an opto-ultrasound combiner. In 2019, this group further developed a novel agent-free localization PAM system with a galvanometer scanner [18]. They placed the galvanometer scanner with the body outside water while the mirror remained inside, achieving high-speed and high-resolution imaging of mouse ears, as shown in Fig. 3a,b. This semi-water-immersible galvanometer scanner improved the temporal resolution to a B-scan rate of 500 Hz with a lateral resolution of 7.5 μm. An alternative pathway by Lee et al. was to utilize a two-axis galvanometer mirror system mounted on a custom-designed water-barricading structure (Fig. 3c) [43]. As shown in Fig. 3d, the novel waterproof galvanometer scanner-based (WGS-based) PAM system successfully obtained imaging of mouse brain in vivo with a 1 min 40 s acquisition time for a scan region of 8 × 13 mm² with 2000 × 2000 pixels.

Fig. 3.

Galvanometer scanner.

(a) Configuration of the agent-free localization PAM system. (b) Depth-encoded PA images of mouse brain [18]. (c) Schematic of the WGS-based PAM system. (d) Maximum amplitude projection (MAP) images and depth-encoded PA images of mouse brain with intact skull [43].

Water immersible micro-electro-mechanical system (MEMS) scanner was also proposed to enhance imaging speed. Yao et al. [44] developed a fast PAM system with high sensitivity and wide FOV by using a water‐immersible MEMS scanning mirror, as shown in Fig. 4a,b. By simultaneously steering the excitation laser beam and the PA waves, the PAM can maintain confocal alignment, ensuring high sensitivity over a wide FOV. This system achieved a B-scan speed of 400 frames /s within a range of 3 mm, which is about 400 times faster than the second-generation mechanical scanning system [12] and about 20 times faster than the fast voice coil scanning system [34]. On the strength of this MEMS scanner, Yao et al. published another research work in functional imaging of both resting and stimulated states in the mouse brain (Fig. 4c) [22]. The lateral spatial resolution of structural imaging was improved from 15 μm to 3 μm. High-speed imaging of the oxygen saturation of hemoglobin (sO₂) illuminated in Fig. 4d was performed by using a single-wavelength pulse-width-based method, with a 3D volumetric rate of 1 Hz over a 3 × 2 mm² field of view. Kim et al. developed a 2-axis water-proofing MEMS scanner made of flexible PDMS. The scanner was able to cover a 9 × 4 mm² scan range [45]. The fabrication of miniaturized MEMS scanner provided the potential for handheld PAM systems. In 2016, Lin et al. presented a compact handheld OR-PAM based on a 2-axis water-immersible MEMS scanner [46]. Handheld OR-PAM was demonstrated on mouse ear and human skin with a scanning range of 2.5 × 2.0 mm², and B-scan imaging rate of 220 Hz. Recently, Baik et al. overcame the limitations of slow imaging speed and/or small FOV of conventional AR-PAM systems [47]. By incorporating a 1-axis water-proofing MEMS scanner with two linear stepper motor stages, the MEMS-AR-PAM system demonstrated a super wide-field scanning over a region of 36 × 80 mm² within 224 s (Fig. 4e). In vivo image of subcutaneous microvascular networks in human fingers, palm, and forearm was acquired by mosaicking multiple volumetric images, as shown in Fig. 4f.

Fig. 4.

Water‐immersible MEMS scanning.

(a) Schematic of MEMS-OR-PAM with a 9 × 9 mm² MEMS mirror plate. (b) Images of blood flow dynamics of the vasculature in a mouse ear [44]. (c) Schematic of the PAM system for functional imaging. (d) Fractional photoacoustic amplitude changes (shown in yellow) in response to left hindlimb stimulation (LHS) and right hindlimb stimulation (RHS) [22]. (e) A system configuration of the MEMS-AR-PAM system. (f) In-vivo super wide-field PA imaging of fingers, palm, and forearm in a human volunteer without skin signals [47].

As another scheme, the polygon-mirror scanner is well suited for high-speed imaging, thus Yao’s group developed a wide-field high-speed OR-PAM system based on a water-immersible polygon-mirror scanner [20]. As shown in Fig. 5a, this system scanned the sample using a hexagon mirror actuated by a water-immersible brushed micro-DC motor. Made of BK-7 glass, the hexagon mirror with six facets coated with protective aluminum was capable of wideband optical and acoustic reflection. The B-scan rate can reach up to 900 Hz over a consistent scanning range of 12 mm, which has never been achieved by previous PAM systems. Drug responses in the skin were performed by injecting epinephrine into the mouse’s leg and skin vasoconstriction imaging was obtained within an area of ∼10 × 10 mm² at a laser pulse repetition rate (PRR) of 600 kHz and a B-scan rate of 420 Hz (Fig. 5b). And on this basis, Chen et al. [48] implemented sO₂ imaging by adopting a 1-MHz dual-wavelength pulsed laser system (Fig. 5c). They achieved a 1-MHz sO₂ A-line rate over a large FOV of 12 × 12 mm² in imaging microvasculature in the mouse ear (Fig. 5d).

Fig. 5.

Polygon-mirror scanner.

(a) Setup of the wide-field high-speed OR-PAM with a polygon-mirror scanner. (b) Drug responses in a mouse ear at different time points after the epinephrine injection [20]. (c) A schematic of polygon-scanning OR-PAM for sO₂ imaging. BPF, bandpass filter; DM, dichroic mirror; HWP, half-wave plate; NDF, neutral density filter; PBS, polarization beam splitter; CL, correction lens; UT, ultrasonic transducer; AL, acoustic lens; PS, polygon scanner. (d) Snapshots of microvasculature and imaging of sO2 at 1-MHz A-line rate in mouse ear [48].

3.2. Pure optical scanning

Pure optical scanning employs stationary PA signal detection, and thus avoids the time penalties associated with mechanical scanning of the sample/transducer. Imaging speed is limited only by the speed of PA signal generation.

3.2.1. Fiber-optic sensor-based stationary acoustic detection

In another imaging configuration, the acoustic detection was kept stationary during the laser beam scanning. In such approach, the FOV is limited by the acoustic detection area. Thus, an ultrasound sensor with both large detection area and high sensitivity is needed. Planoconcave fiber-optic sensor can provide high detection sensitivity and broad bandwidth. Its characteristic of large acoustic acceptance angle (±90 deg) provides a near omnidirectional response, enabling PA signals generated over a large lateral FOV to be recorded without translating the sensor [49,50]. Recently, Allen et al. reported an ultrafast laser-scanning PAM system that has a range-ambiguity limited signal acquisition speed at 2 million A-line/s for the first time [51]. As Fig. 6a shows, this system utilized a fast galvanometer-based scanner to raster scan the laser beam over the sample surface, and used a static fiber-optic ultrasound detector to record the PA signals. Based on the large acoustic acceptance angle (±90 deg) of planoconcave fiber-optic sensor, large FOVs can be imaged without any mechanical scanning of the sample or the receiver, and thereby avoiding the time penalties associated with mechanical scanning. Photoacoustic image shown in Fig. 6b was acquired in vivo from an area of 10 × 10 mm² of mouse ear microvasculature. This image was composed of 16 million A-lines and was acquired in just 8 s when operating the fiber laser at a PRF of 2 MHz.

Fig. 6.

Sationary acoustic detection.

(a) Schematic of the ultrafast laser-scanning PAM system. (b) Image obtained in vivo of a mouse ear at 2 million A-lines/s within a 10 mm × 10 mm area [51]. (c) Schematic of FOUS-based fast scanning PAM system. (d) Image obtained in vivo of a mouse ear with a frame rate of 0.2 Hz over a 2.2 mm × 2.2 mm area [52].

Another presentation of this category utilized an unfocused side-looking fiber optic ultrasound (FOUS) sensor. [52]. Combining stationary acoustic detection with side-looking FOUS sensor and scanning the laser beam with a 2D galvo mirror (Fig. 6(c)), hemodynamics imaging within a 2 × 2 mm² area at 2 Hz frame rate was demonstrated in the mouse ear (Fig. 6(d)). The high B-scan rate of 400 Hz enabled imaging physiological dynamics in both trunk vessels and capillaries. Imaging speed can be further accelerated by increasing the PRF of the laser.

3.2.2. Encoded optical excitation

Yang et al. proposed a motionless volumetric spatially invariant resolution photoacoustic microscopy (SIR-PAM) (Fig. 7a), which further develops 2D single-pixel Fourier-spectrum acquisition imaging to overcome limitation of depth resolving ability [14]. SIR-PAM uses a series of propagation-invariant sinusoidal fringes (PISFs), which are generated by a digital micromirror device, to produce in-focus fringes at any depth and induces PA signals inside object. Detected by a single-element ultrasonic transducer, signals from each depth were extracted to obtain the Fourier spectrum and subsequently reconstruct the image. As Fig. 7b shows, non-invasive 3D imaging of whole zebrafish larvae in vivo by SIR-PAM was clearly resolved with a spatially invariant resolution over a large depth range. The resolution of SIR-PAM system remained around 1.89 μm (≤ 8% degradation) within a resolution-invariant axial range (RIAR) of 1800 μm, whereas the conventional PAM could achieve a lateral resolution of 2.86 μm within a 55 μm depth of focus. Therefore, SIR-PAM images were much superior in resolution at all depths compared with conventional PAM system. Limited by the pulse repetition rate (1 kHz) of the laser, it took about 21 s to obtain the volumetric image of zebrafish. Yet, for potential video-rate volumetric imaging, the speed can be further improved by increasing the laser pulse repetition rate and incorporating compressed sensing.

Fig. 7.

Encoded optical scanning.

(a) Principle of spatially invariant resolution photoacoustic microscopy (SIR-PAM). (b) Depth-encoded whole-body images of a zebrafish larva obtained by SIR-PAM (top) and conventional PAM (bottom) [14].

4. Advances in acoustic detection

4.1. Parallel acoustic detection

For the typical design of single-element detector-based scanning PAM systems, the imaging speed was limited since the photoacoustic signals were obtained by a single ultrasound transducer through point-by-point scanning. Song et al. [53] first invented a multifocal OR-PAM to improve the imaging speed through parallel acoustic detection (Fig. 8a,b). In this OR‐PAM system, 20 diffraction‐limited focal spots were optically stimulated via a microlens array, and the excited PA signals were then detected by a linear 48-element ultrasound array without mechanical scanning. The imaging speed of this multiple foci scanning technique significantly increased compared with that of the single focus scanning method. A volumetric image was eventually obtained with the size of 1000 × 500 × 200 voxels at ∼10 μm lateral resolution within 4 min. Nonetheless, this OR-PAM system was designed as a transmission mode, limiting it to imaging of thin samples. Thus Li et al. [54] developed a reflection-mode multifocal OR-PAM by placing a microlens array directly underneath the ultrasonic array transducer, as shown in Fig. 5c. The reflection-mode OR-PAM was capable of acquiring volumetric images in a mouse ear and brain with the size of 6 × 5 × 2.5 mm³ within 2.5 min (Fig. 8d). While multifocal OR-PAM has shown its feasibility, the potential of microlens and transducer arrays has not been fully realized. For that reason, Xia et al. [55] presented a multifocal photoacoustic-computed microscopy using a 2D microlens array and a 512-element full-ring ultrasonic transducer array (Fig. 5e,f). They transmitted 1800 optical foci within the focal plane of the transducer and raster scanned with a 25 μm step size. A 64-channel data-acquisition system was adopted to digitize photoacoustic signals with a 40 MHz sampling rate. As a result, they acquired a cross-sectional PAM image over a 10 × 10 mm² field of view within 36 s. This system was further applied in the research of ultraviolet photoacoustic microscopy. Besides, as shown in Fig. 8g,h, Imai, et al. [56] developed a high-throughput multifocal UV-PAM with a 256-channel data acquisition system. They achieved a lateral resolution of ∼1.6 μm and produced images 40 times faster than the conventional UV-PAM scanning on a point-by-point basis.

Fig. 8.

Parallel acoustic detection.

(a) Schematic of the multifocal OR-PAM system with slit (width = 50 μm) and the aperture (width = 5 mm). (b) Photoacoustic image of a mouse ear microvasculature [53]. (c) Schematic of the whole reflection-mode multifocal OR-PAM system. (d) Trans-cranial PA images of a nude mouse brain with the skin removed [54]. (e) Schematic of the multifocal photoacoustic-computed microscopy with a 512-element full-ring ultrasonic transducer array. (f) PA images of an intact uterus from a pregnant mouse [55]. (g) Schematic of multifocal ultraviolet laser photoacoustic microscopy system. (h) PA MAP image of a fixed mouse brain slice [56].

4.2. Encoded acoustic detection

In addition to parallel acoustic detection, a recent breakthrough of encoded acoustic detection also demonstrated the feasibility of combining high speed, high frame rate, wide field of view, and low cost into a single system [57]. This technique, to a great extent, breaks through the limitation of complex and expensive array-based detection. Li et al. [57] developed an ergodic relay (PATER) for a low-cost high-throughput AR-PAM system based on an acoustic ergodic relay (ER), which was coupled with a single-element ultrasonic transducer (Fig. 9a,b). Without scanning that normally applied in conventional PAM systems, PATER captured a snapshot wide-field image with only a single laser shot. The parallelly detected PA signals were decoded mathematically, based on the encoding spatial information with temporal signatures through the calibrated impulse responses. Through this system, they applied functional imaging of mouse brain hemodynamic responses and high-speed imaging of blood pulse wave propagation in vivo. Super-resolution imaging of vascular patterns was further demonstrated via tracking and localizing flowing melanoma cells at a high frame rate of 2 kHz. Another hybrid method of encoded acoustic detection by Li et al. was implemented in multifocal OR-PAM through an ergodic relay (MFOR-PAMER) [19] (Fig. 9c,d). In the MFOR-PAMER system, the acoustically defined image resolution (AR) was improved to optically defined image resolution (OR) by utilizing a 2D microlens array for optical illumination with 400 optical foci. An acoustic ER simultaneously detected parallelly generated PA signals with a single-element ultrasonic transducer. This system improved the lateral resolution to 13 μm and enabled at least 400-folds scanning speed increase compared with conventional OR-PAM at the same imaging resolution. Imaging resolution and speed were expected to be further improved by optimizing the ultrasonic transducer detector and microlens' performance.

Fig. 9.

Encoded acoustic detection.

(a) Schematic of the PATER system. (b) Image of mouse brain hemoglobin responses to front-paw stimulations through an intact skull [57]. (c) Schematic of the MFOR-PAMER system. (d) Wide-field images and zoomed-in views of AR-PAMER image and MFOR-PAMER image [19].

5. Conclusions

The modern incarnation of PAM for biological imaging has sustained a great momentum of development, due to the advances in modern laser technologies, inverse reconstruction, sensitive ultrasonic detectors, and effective signal processing. The imaging performance of PAM has been markedly improved in aspects of all key imaging parameters, including spatial resolution, penetration depth, detection sensitivity, and imaging speed. Therefore, PAM has developed as a powerful tool for anatomical, functional, and molecular imaging, with great potential for clinical applications.

In this concise review, we have summarized recent efforts in developing high-speed PAM, including the laser source, scanning mechanism, and acoustic detection, listed in Table 1. Nevertheless, most of the reported PAM systems are much more versatile in imaging within a small region less than 12 × 12 mm², limited by the tradeoff of scanning range, imaging resolution, and scanning speed. Hence, a novel scanning technique for high-speed PAM with high resolution and ultra-wide FOV is highly desired. Parallel and encoded acoustic detection provide innovative solutions for high-speed PAM, yet may suffer from the bulky system volume and complex imaging algorithm. Moreover, the capability of high-speed PAM can be further enhanced by integrating with high repetition rate pulsed laser sources with more wavelength output.

Table 1.

Summary of advances in high-speed PAM.

Methods	Category			PRR of laser	Scanning Speed	Imaging range	Lateral resolution	Ref
Laser performance	Supercontinuum source			100 kHz	3.2 s/volumetric scan	7 mm × 7 mm × 7 mm	11.46 μm	[29]
	Raman laser (multi-wavelength)			1 MHz (5 wavelength)	6 Hz/volumetric scan; 1.5 Hz/volumetric scan	0.6 × 1.8 mm2; 2.2 × 2.0 mm2	2.7 μm	[33]
Scanning mechanism	Combined optical-acoustic scanning	Voice coil scanner		4 kHz	20 Hz/B-scan 40 Hz/B-scan	9 mm 1 mm	3.4 μm	[34]
		Mirror scanner	Galvanometer scanner	800 kHz	500 Hz/ B-scan rate	∼2.4 mm	7.5 μm	[18]
			Water immersible MEMS scanner	100 kHz+500kHz	400 Hz/B-scan	3 × 2 mm2	3 μm	[22]
			Polygon-mirror scanner	600 kHz	900 Hz/B-scan	12 mm	10 μm	[20]
	Pure optical scanning	Fiber-optic sensor-based stationary acoustic detection		2 MHz	2 million A line/s	10 × 10 mm2	7 μm	[51]
		Encode optical scanning		1 kHz	21 s/volumetric scan	1.2 mm × 0.9 mm × 1.8 mm	1.89 μm	[14]
Acoustic detection mechanism	Parallel acoustic detection			1.35 kHz	2.5 min/volumetric scan	6 × 5 mm2	16 μm	[54]
	Encoded acoustic detection			2 kHz	10 s/volumetric scan	10 × 10 mm2	13 μm	[19]

For point-scanning based imaging, different area in the biological tissue undergoes laser exposure only once and the energy of each laser pulse is regulated by the maximum permissible exposure set by American National Standards Institute [58]. For some systems with stationary detection, laser energy may be increased within the safety limit, in order to maintain good sensitivity over the entire FOV. For methods involving repeated exposure, such as the encoded optical excitation method, heat accumulation also needs to be assessed. Therefore, for safety and stability, laser energy needs to be considered in the high-speed PAM system development.

In summary, with a series of long-standing challenges to overcome, high-speed PAM will draw more attention from various research communities and find broader impacts in fundamental research and clinical practice.

Declaration of Competing Interest

The authors declare that there are no conflicts of interest related to this article.

Biographies

Kaiyue Wang received her B.S. degree from Northeastern University in 2017 and her M.S. degree from Zhejiang University in 2020, both in Biomedical Engineering. She is currently an engineer at Zhejiang Lab, Hangzhou, China. Her research interests include photoacoustic microscopy, acoustic neuromodulation, and functional brain imaging.

Chiye Li is a research scientist at Zhejiang Lab. He received his B.S. from University of Science and Technology of China in 2011 and his Ph.D. in Biomedical Engineering from Washington University in St. Louis in 2016. His research interests include photoacoustic imaging and biomedical optics.

Ruimin Chen received his B.S. degree from University of Electronics Science and Technology of China, Chengdu, China and his M.S. degree from University of Southern California, Los Angeles, CA, in 2006 and 2008, respectively, both in Biomedical Engineering. He is currently a research scientist at Zhejiang Lab, Hangzhou, China. His research interests include the design, modeling, and fabrication of high-frequency ultrasonic transducers and arrays for medical imaging applications, piezoelectric material characterization, and photoacoustic imaging.

Junhui Shi received his BSc in chemical physics from the University of Science and Technology of China. Then, he continued to study chemistry and received his PhD at Princeton University. He was working on theoretical chemical dynamics and experimental nuclear magnetic resonance spectroscopy. Afterward, he worked on photoacoustic imaging in the Department of Biomedical Engineering of Washington University in St. Louis, Missouri, and then in the Department of Medical Engineering of Caltech in Pasadena, California. Currently, he is a senior researcher at Zhejiang Lab, Hangzhou, China. His research interests include photoacoustic microscopy, photoacoustic tomography, nuclear magnetic imaging, and their applications in biomedical studies.