2.1.

Imaging Parameters of Optical Ultrasound Sensors

To compare the performance of ultrasound sensors in PA imaging, we use several parameters including, frequency, sensitivity, and acceptance angle.^28,31 The working frequency of ultrasound sensors is critical for imaging quality. It contains two parameters, bandwidth and central frequency. Light-induced ultrasound waves possess a broad frequency spectrum, spanning from kHz to hundreds of MHz. Thus ultrasound sensors should ideally exhibit a wide bandwidth and high central frequency to enhance image resolution. If the impulse response of the ultrasound sensor has a Gaussian envelope, the axial resolution of PAM can be estimated by the formula, $R_a=0.88v_a/\mathrm{\Delta }f$, where $v_a$ is the speed of sound and $\mathrm{\Delta }f$ is the bandwidth of ultrasound transducer.^47,48 In addition, the axial resolution of PACT is also related to the bandwidth of ultrasound transducer.⁴⁹ The bandwidth of resonance-based optical ultrasound sensors is dependent on the two concurrent processes during the detection, the optical resonance, and ultrasonic wave propagation.⁵⁰ When the resonance mode is changed by the ultrasound wave, the resonance cavity need time to reach a steady state again. The spending time is comparable to the intracavity photon lifetime $\tau =Q/\omega$, where $Q$ is the resonator quality factor and $\omega$ is the angular frequency of the light wave. So the frequency bandwidth can be limited by the quality factor $Q$.⁵¹ In addition, the bandwidth is influenced by the interaction between the ultrasound wave and the sensor, which is dependent on the fiber material and backing material, as well as their structure.^52–54

Sensitivity is another key parameter in PA imaging. For typical resonance-based optical ultrasound sensors, the sensitivity $S$ can be expressed as the change in transmission $T$ induced by the acoustic pressure $P$, which can be expressed by $S=\frac{dT}{dP}=\frac{dT}{d\varphi }·\frac{d\varphi }{d{L}_o}·\frac{d{L}_o}{dP}$, where $\varphi$ is the round-trip phase and $L_o$ is the optical path length of the resonator. The first term $\frac{dT}{d\varphi }$ represents the maximal slope in the transmission spectral. The second term $\frac{d\varphi }{d{L}_o}$ represents the phase change caused by the modulation of the cavity length. The last term $\frac{d{L}_o}{dP}$ represents the optical path length changed by the acoustic pressure, which is related with the mechanical and optomechanical properties of the sensor. The sensitivity can also be quantified by noise equivalent pressure (NEP) and noise equivalent pressure density (NEPD). NEP is defined as the minimum detectable signal pressure that equal to the noise amplitude.^18,28 NEPD can show the sensitivity of sensors over the spectra. If the sensor sensitivity spectral density $S(f)$ and noise amplitude spectral density of the sensor $N_V(f)$ are measured, NEPD can be calculated by $N(f)=N_V(f)/S(f)$.⁵⁵

The acceptance angle is also crucial for PA imaging, which can measure the sensor’s ability to detect the ultrasound signal from different directions. Large acceptance angle is beneficial in reducing imaging artifact and improving the signal-to-noise ratio (SNR). An ultrasound sensor with a large acceptance angle can provide high imaging quality.^56–58 Due to the varying distances from the sound source to different points on the sensor, the phase of the ultrasound wave reaching each point on the sensor differs. As the ultrasound signals with different phases received by the ultrasound sensor may cancel each other out, this results in a reduction in the signal. Consequently, smaller sensors often exhibit larger acceptance angles. In addition, the shape of the sensor has significant influence on the acceptance angle. For example, a ring-shaped sensor is superior to a disk-shaped one for near-field ultrasound detection because the ring shape minimizes the phase retardation.²⁷ In addition, the acceptance angle of some sensors is tied on the detection principle, such as the fiber laser sensor. It utilizes the opposite refractive index changes in the perpendicular polarization modes. When the ultrasound wave is incident to the principal axis of the fiber laser sensor, the detected signal is the maximum. As the incident angle increases, the signal detected by the sensor gradually diminishes, reaching a minimum when the incident angle is 45 deg. This phenomenon occurs because, at this angle, the refractive index alterations in the perpendicular polarization modes are equivalent.¹⁹

2.2.

Fabry–Perot Sensor

The FPI is composed of two highly reflective surfaces separated by a spacing material, together forming the FP cavity. As acoustic waves reach the surface, the optical thickness of the FPI shifts, causing a minor phase alteration and modulating the optical intensity.^52,59 Zhang et al.^60,61 introduced a planar FP ultrasound sensor that formed by a thin polymer (Parylene C) film spacer sandwiched between two dielectric dichroic mirrors. The system’s ability to provide PA images was demonstrated, but the imaging speed was slow for in vivo applications. Ansari et al.^33,62 proposed a miniature forward-viewing 3D PA probe that comprises a coherent fiber bundle with an FP polymer-film ultrasound sensor at its distal end, as shown in Fig. 1. The optical fiber bundle with 50,000 cores acts as an ultrahigh-density ultrasound array, and the PA images can be obtained by sequentially scanning the input end of the bundle, which can greatly reduce the volume of the probe. The outer diameter of the probe is only 3.2 mm. However, the imaging speed of the system was limited by the pulse repetition frequency (PRF) of the excitation laser. The acquisition time of a figure can be more than 25 min. The imaging speed can be improved by increasing the excitation laser PRF and parallelizing the sensor read-out.^25,63,64 Moreover, compressed sensing techniques can further accelerate imaging speeds.^65,66

Fig. 1

(a)–(c) PAE probe with planar FP sensor head. Reprinted with permission from Ref. 62, available under a CC-BY 4.0 license.

The performance of planar FP ultrasound sensor is limited by the beam walk-off and diffraction effects around the fiber-tip. To overcome these problems, a plano-concave FP ultrasound sensor structure is proposed.⁶⁷ Guggenheim et al.⁵⁶ proposed a high $Q$-factor plano-concave microresonator that has very high sensitivity with excellent broadband acoustic frequency response and wide directivity. This microresonator ultrasound sensor has much higher $Q$-factor ($>{10}^5$) than planar FP ultrasound sensor because the plano-concave structure can precisely correct for the divergence by refocusing the light upon each round trip and preventing the beam from walking off laterally. Thus the sensitivity of the microresonator is greatly improved, and the NEP can be lower than $1.6\mathrm{mPa}/{\mathrm{Hz}}^{1/2}$. Chen et al.⁶⁸ proposed photothermally tunable FP sensor for PA mesoscopy, as shown in Fig. 2(a). The NEPD of the FP sensor is $40\mathrm{mPa}/{\mathrm{Hz}}^{1/2}$ with an acoustic detection bandwidth up to 30 MHz. The PA image of an ex vivo mouse kidney reconstructed by the FP sensor is shown in Fig. 2(b).

Fig. 2

Plano-concave FP ultrasound sensor: (a) sensor structure and (b) the imaging results of an ex vivo mouse kidney. Reprinted with permission from Ref. 68, available under a CC-BY 4.0 license.

Since fiber Bragg grating (FBG) can be used as mirrors in fiber-based interferometers, several researchers proposed FP ultrasound detector based on FBG technology. For FP interferometer that formed by FBGs, the sensing area is the region between the two FBGs, as shown in Fig. 3(a). By measuring variations of the refraction index (RI) induced by the acoustic pressure, the acoustic signal can be detected.⁷¹ Gruen et al.⁶⁹ realized an integrating line detector with a fiber-based FP interferometer formed by two FBGs. The reflectivity of an FBG is 81%, and the distance between the FBGs is 11.5 cm. The team experimented with bristle knots and ants using FP glass-fiber interferometers. Wang et al.⁷² proposed a microfiber FBG-based FP interferometric acoustic transducer and verified its performance by the imaging studies of human hairs. Ma et al.⁷⁰ proposed a Fabry–Perot ultrasound sensor that formed by a microfiber loop sandwiched by a pair of inline Bragg gratings, as shown in Fig. 3(b). Although constrained by imaging sensitivity, the needle-like focus of the microfiber serves to mitigate the degradation of both resolution and signal amplitude in out-of-focus regions.

Fig. 3

FBGs-based FP ultrasound sensor system. (a) FP interferometer formed by two FBGs. Reprinted with permission from Ref. 69. (b) Schematic of the transparent microfiber ultrasound sensor for PA imaging. Reprinted with permission from Ref. 70, available under a CC-BY 4.0 license.

2.3.

Fiber Laser Sensor

A fiber laser is a type of laser where the active gain medium is an optical fiber infused with rare-earth elements such as erbium, ytterbium, and neodymium. When equipped with two reflective mirrors, a fiber laser configuration is established. A specific type of short-cavity fiber laser, known as either a distributed feedback or distributed Bragg reflector laser, is adept at detecting ultraweak signals, including strains and acoustic waves. Furthermore, the application of the wavelength-division multiplexing technique allows these fiber laser sensors to be integrated into a sensor array.⁷³

Liang et al.¹⁹ introduced a fiber laser sensor to detect high-frequency ultrasound waves and the ultrasonic sensing system is shown in Fig. 4. This sensor amplifies the acoustic response based on the frequency change of the signal light to measure ultrasound, specifically by gauging the acoustically induced optical phase change. The NEPD of this system is below $1.5\mathrm{mPa}/{\mathrm{Hz}}^{1/2}$ within a measured frequency range of 5 to 25 MHz. The optical phase detection uses the beating signal between two different-polarized laser beams, offering resilience against thermal drift and vibrational disturbances.

Fig. 4

Fiber laser ultrasonic sensing system. Reprinted with permission from Ref. 19, available under a CC-BY 4.0 license.

Guan et al.^35,74,75 proposed a sensor with a sensitive primitive consisting of a fiber laser. This laser is constructed by inscribing two wavelength-matched FBGs into an Er-Yb co-doped fiber, as depicted in Fig. 4. These gratings exhibit a strong reflection at $\sim 1550\mathrm{nm}$, which facilitates the generation of laser light. Notably, this wavelength corresponds to the peak luminescence efficiency of the fiber gain ion, enabling the laser to achieve a significant gain. The gratings have a length ranging from 2 to 6 mm, and their spacing can vary between 0.5 and 10 mm. Emitting laser light from cavities shorter than 2 mm is challenging due to the restricted doping concentration. The sensor leverages the inherent birefringence of the optical fiber. The fiber’s natural processing makes it weakly birefringent, leading to the generation of two laser beams with distinct frequencies on the $x$ and $y$ polarization axes. The two orthogonal modes produce a beat signal in a detectable GHz frequency range. When ultrasonic waves induce vibrations in the optical fiber, the refractive index changes in the $x$- and $y$-polarized modes are equal but opposite. This phenomenon is utilized in the optical heterodyne detection method. Both modes respond similar to low-frequency disturbances, such as thermal and mechanical vibrations, which can be minimized in the beat signal.

In the fiber laser ultrasonic sensing system, an erbium-doped fiber amplifier is used to boost the light power, allowing the photodetector to function in the shot-noise-limited region, further improving the SNR. Finally, the photodetector transforms the optical signal into a radio frequency signal.

Ultrasonic sensing systems primarily experience two types of noise: phase noise and intensity noise.⁷⁶ In fiber laser PA imaging systems, the predominant sources of phase noise include the inherent noise from the fiber laser (a sensitive element), spontaneous radiation noise from the fiber amplifier, thermal and scattering noise from the photodetector, and noise from the data acquisition (DAQ) system. The total system noise is primarily attributed to the fiber laser, optical amplifier, and DAQ system. At a frequency of 3 MHz, when the noise from the DAQ system is approximately $-140\mathrm{dBc}/\mathrm{Hz}$, the system’s noise power density stands at about $-130\mathrm{dBc}/\mathrm{Hz}$. The sensing system shown in Fig. 4 remains unaffected by intensity fluctuations and the coupling efficiency between intensity and phase noise is minimal, ranging from 1% to 3%. Consequently, the acoustic sensitivity is solely limited by the phase noise.

Liang et al.^19,77 and Zhou et al.⁷⁸ applied the fiber laser to the PAM and PAE. The PAE system consists of a PA probe, a dual-wavelength laser source, a sensor interrogation unit, mechanical scanners, and DAQ and control modules [Fig. 5(a)]. In addition, Bai et al.⁷⁹ applied the sensor to the PACT [Fig. 5(b)]. The imaging depth can be tuned by bending the fiber laser sensor into different curvatures using the customized holder.

Fig. 5

Fiber laser PA system: (a) PAE system, reprinted with permission from Ref. 19, available under a CC-BY 4.0 license. (b) PACT system, reprinted with permission from Ref. 79 with permission.

2.4.

$\pi$-Phase-Shifted FBG

Another FBG-based sensor, $\pi$-phase-shift FBG ($\pi$-FBG) is an alternative optical sensor that used in PA imaging. Bragg grating is a transparent structure with a periodic variation of refractive index. $\pi$-FBG contains a phase jump of $\pi$ at the center of the FBG, forming a region analogous to the cavity of FPI [Fig. 6(a)].^80–82 This phase jump leads to a narrow spectral notch at the center of the reflection bandwidth of the grating, allowing highly sensitive ultrasonic detection.^83–85

Fig. 6

Schematic and reflection spectrum of a $\pi$-FBG. Λ is the grating pitch. Reprinted with permission from Ref. 80, available under a CC-BY 4.0 license.

The imaging systems based on $\pi$-FBG have been successfully applied in PA imaging. Rosenthal et al.⁸⁶ proposed an intravascular PA catheter with a diameter of 1 mm that consisted of a fiber with a $\pi$-FBG written close to its tip and an additional illuminating fiber. The NEP of the $\pi$-FBG over 16 MHz bandwidth was found to be 100 Pa. The narrow transmission spectrum of $\pi$-FBG was utilized to reduce the amplified spontaneous emission noise and improve the sensitivity. To translate the ultrasound signal into intensity shifts, the fiber-based Mach– MZI was used for active demodulation. A healthy stented artery ex vivo was imaged by the catheter and the results showed that the system had great stability even strong vibrations were applied to the catheter. Wissmeyer et al.⁸⁷ presented an all-optical PA microscope and its biological imaging results. The adopted $\pi$-FBG had a narrow resonance width of 8 pm at $-3\mathrm{dB}$ and two distinct frequency bands at $-6\mathrm{dB}$, ranging from 7 to 27 MHz and from 62 to 77 MHz, which contributed to obtain the high-resolution PA images. Shnaiderman et al.⁸⁸ described a miniaturized PA sensor with $\pi$-FBG embedded in an acoustic cavity, as shown in Fig. 7(a). Though the $Q$ factor of the sensor is moderate, the sensitivity could be compensated by acoustic cavity signal amplification, achieving the NEP of 88 Pa. A mouse ear imaged by the sensor in vivo is shown in Fig. 7(b).

Fig. 7

$\pi$-FBG PA microscope: (a) schematic of the $\pi$-FBG-based sensor and (b) the imaging results of a mouse ear in vivo. Reprinted with permission from Ref. 88, available under a CC-BY 4.0 license.

To further reduce the sensor dimensions and achieve a higher center frequency, fiber-optic waveguide has been applied to fabricate the $\pi$-FBG sensor. Rosenthal et al.⁸⁹ had demonstrated a miniaturized wideband ultrasound sensor based on $\pi$-phase shifted waveguide Bragg grating ($\pi$-WBG) that embedded in a silicon-on-insulator (SOI) photonic platform. Shnaiderman et al.³⁶ developed a point-like silicon waveguide–etalon detector (SWED) with a sensing area of only $220\mathrm{nm}\times 500\mathrm{nm}$ using SOI technology. The point-like SWED reached an ultrawide bandwidth of 230 MHz at $-6\mathrm{dB}$ and could provide super-resolution detection and imaging performance. In addition, the small size of SWED provides a potential method to build very dense ultrasound arrays on a silicon chip. Hazan et al.³⁷ proposed a miniaturized silicon-photonics acoustic detector (SPADE) with NEPs down to $2.2\mathrm{mPa}{\mathrm{Hz}}^{-1/2}$ and a bandwidth above 200 MHz that capable of tomographic imaging. The $\pi$-WBG in SOI that coated with the elastomer polydimethylsiloxane, which can enhance the sensitivity and reduce the parasitic effect of surface acoustic waves. The imaging performance of SPADE was tested by both dark knot ex vivo and mouse ear in vivo, as shown in Fig. 8(b).

Fig. 8

SPADE PA microtomography: (a) illustration of the silicon-photonics layer structure and (b) the imaging results of a mouse ear in vivo. Reprinted with permission from Ref. 37, available under a CC-BY 4.0 license.

2.5.

Whispering Gallery Mode

In optical WGM, total internal reflection can confine light waves within a closed circular microcavity. The resonance is achieved when the optical path length matches an integer multiple of the laser’s wavelength.^90,91 This section delves into ultrasound transducers featuring various resonant microcavity shapes, such as microrings, microspheres, and microbubbles.

The microring resonator (MRR) is composed of a bus waveguide and a circular waveguide, depicted in Fig. 9(a). A laser is introduced from one end of the bus and is evanescently coupled to the ring waveguide through a low-dielectric gap separating the bus from the circular waveguide. When the circumference of the circular waveguide corresponds to integer multiples of the laser’s wavelength, resonance is achieved, resulting in distinct dips in the transmission spectrum, as illustrated in Fig. 9(b).^92,93 Ultrasound waves can deform the MRR, altering the effective RI of the guided mode due to the elasto-optic effect. This causes a shift in the resonance wavelength, allowing the detection of ultrasound waves by monitoring the modulated output intensity.

Fig. 9

Microring resonator sensor: (a) geometry and (b) typical transmission spectrum.

The MRR’s submicron thickness induces an acoustomechanical resonance within the gigahertz range, enabling a uniform frequency response from DC up to several hundred megahertz.⁹⁴ This ultrasound sensor stands out due to its high $Q$ factor and pronounced resonance from multibeam interference, offering both exceptional sensitivity and a broad bandwidth. Zhang et al.⁹⁵ introduced a polystyrene (PS) microring sensor. This sensor is characterized by a ring and bus structure with dimensions of $60\mu \mathrm{m}$ in diameter and $1.4\mu \mathrm{m}$ in height. The sensor’s resonance bandwidth is 6 pm with $Q$ factor of $1.3\times {10}^5$. Furthermore, it has a bandwidth of 350 MHz and a NEP of 105 Pa within this range.

Encasing MRR sensors in an acoustic impedance-matched protective layer enhances their reliability and stability, making them more suitable for in vivo applications. Rong et al.⁹⁴ introduced an MRR sensor with an $80\text{-}\mu \mathrm{m}$ diameter, crafted using nanoimprint lithography. This sensor has a $Q$ factor of $4.6\times {10}^4$ and a NEP of 81 Pa, complemented by its $\sim 23\mathrm{MHz}$ detection bandwidth and a 90 deg acceptance angle. Building on this, Rong et al. developed a 3D-PACT system centered around this MRR sensor, as shown in Fig. 10(a). This system employs a low RI polymer known for its biocompatibility. The sample undergoes scanning on a motorized three-axis stage, with a narrowband tunable laser serving as the detection light source. This system can provide lateral and axial resolutions of $\sim 114$ and $\sim 57\mu \mathrm{m}$. The system was validated by imaging human hair, leaf veins, isolated mouse brains as well as in vivo mouse ears and tadpoles. The results showed that the system can obtain high SNR and high-contrast 3D PA images. The PA images of the mouse ear that reconstructed by this system are shown in Fig. 10(b).

Fig. 10

MRR 3D-PACT: (a) schematic of the imaging system and (b) the imaging results. Reprinted with permission from Ref. 94, available under a CC-BY 4.0 license.

Due to the ability of digital optical frequency comb (DOFC) to generate ultranarrow and tunable combs, it can be used to locate the resonance frequency of an array of microring sensors in parallel with high precision, enabling a one-off measurement of the transmission spectrum using only a single photoreceiver, which simplifies the use of PACT systems with arrays of microrings. Pan et al.⁹⁶ proposed a PACT system based on an array of 15 microring sensors, as shown in Fig. 11. The chalcogenide-based MRR sensors have high $Q$ factors ranging from $5\times {10}^5$ to $7\times {10}^5$, while each element has a bandwidth of 175 MHz at $-6\mathrm{dB}$, a NEP of $2.2\mathrm{mPa}{\mathrm{Hz}}^{-1/2}$, and an acceptance angle of $\pm 30\mathrm{deg}$. These sensors were tuned to slightly different resonant frequencies, and by DOFC, Pan et al. implemented a PACT system and scanned leaf veins, zebrafish at different growth stages.

Fig. 11

MRR array PACT: (a) schematic of the imaging system and (b) the imaging result of a 3-month-old adult zebrafish. Reprinted with permission from Ref. 96, available under a CC-BY 4.0 license.

The microsphere resonator sensor consists of a taper fiber and a microsphere cavity, as shown in Fig. 12.⁹⁷ Ultrasound deforms the microsphere cavity and causes changes in the RI of surrounding medium and spheres at the same time, which affects the coupling mode.⁹⁸ Sun et al.³⁸ proposed an method to fabricate microsphere sensors and developed two types of encapsulated microsphere resonators with different cavity materials. The silica microsphere sensor has a $Q$ factor of $\sim {10}^6$ and 160 Pa NEP at 20 MHz, whereas the PS microsphere sensor has a $Q$ factor of $\sim {10}^5$ and 100 Pa NEP at 20 MHz. Sun et al. applied the microsphere sensor to PAM with a lateral resolution of $\sim 5\mu \mathrm{m}$ and successfully imaged hairs and leaf veins in 3D.

Fig. 12

Sensing system of microsphere resonator sensor. Reprinted with permission from Ref. 97, available under a CC-BY 4.0 license.

Microbubble resonators (MBRs) are manufactured using hollow capillary tubes, which have greater deformation compared to microrings and microspheres. It can have a $Q$ factor of ${10}^7$, and the schematic of the sensing system with MBR sensor is shown in Fig. 13.⁹⁹ Tu et al.¹⁰⁰ proposed an packaged optical MBRs sensor with a broad bandwidth (10 Hz to 100 kHz). This size of the sensor is $140\mu \mathrm{m}$ in diameter with a wall thickness of $5\mu \mathrm{m}$. It has a $Q$ factor of $5.2\times {10}^5$ and a NEP of $2.2\mathrm{mPa}/{\mathrm{Hz}}^{1/2}$. Tu et al. applied the sensor to underwater acoustic wave detection. They recorded responses in two orthogonal directions, spanning angular ranges of 75.6 deg and 105.5 deg (at $-6\mathrm{dB}$). MBRs have been successfully applied to PA sensing and have the potential to be used for PA imaging.

Fig. 13

Schematic of the experimental setup for pressure wave detection using MBR sensor. Reprinted with permission from Ref. 99, available under a CC-BY 4.0 license.

2.6.

Summary of Resonance-Based Ultrasound Sensors

The primary characteristics of resonance-based ultrasound sensors are detailed in Table 1. These ultrasound sensors with optical resonance are small but have high sensitivity and broad bandwidth. FP ultrasound sensors are commonly used for ultrasound detection. They can achieve high $Q$ factor, have low NEP, and large acceptance angle, empowering them to capture PA images with superior resolution and contrast. The FP ultrasound sensor can be fabricated at the tip of optical bundle and acting as a high-density ultrasound array to realize 3D forward-viewing PA imaging. However, FP ultrasound sensors require locking the probe laser wavelength to the resonance frequency of the FP cavity. As for fiber laser sensor, drawing on the birefringence principle, it adopts heterodyne phase detection for detecting ultrasound waves, which does not require an additional laser to scan and is insensitive to perturbations, such as temperature and optical intensity change. As a result, the fiber laser sensor can be sensitive and stable. But the fiber laser sensor has directivity, it should be rotated to the most sensitive angle before measuring. $\pi$-FBG sensors are increasingly used in PA imaging due to their broad bandwidth and high sensitivity. Based on SOI technology, $\pi$-FBG can have compact configuration for high-density arrays. However, akin to FP sensors, $\pi$-FBG sensors also need use additional laser for detection. WGM sensors can achieve the highest $Q$ factors and have ultrabroad bandwidth and wide angular response. The small size and transparent WGM sensors can be conveniently integrated into OR-PAM with high-NA objective lens that has a limited working distance.⁵⁰ Yet current fabrication techniques for WGM remain challenging.

Table 1

Performance of optical ultrasound sensors.