Dual-wavelength, nanosecond, miniature Raman laser enables efficient photoacoustic differentiation of water and lipid

Photoacoustic microscopy (PAM) has broad application prospects in biology and medicine as it combines the advantages of optical excitation and ultrasonic detection.^1–4 To generate PA waves, a pulsed laser beam of the corresponding wavelength is required to illuminate the target tissue. As an important branch of PAM, functional PAM requires excitation sources with multiple wavelengths.^5,6 Lipid and water are important components of living organisms, composing 60%–80% of the human body, and are inextricably linked to cellular functions.⁷ The concentration and distribution of lipid is an important biomarker for studying diseases such as atherosclerosis and cancer,^8–10 while studying the content of water can help us understand the biological properties of cells, tissues, and organs^11,12 and identify diseased tissues with respect to healthy ones.⁷ Photoacoustic identification of lipid can be achieved by exploiting the first and second overtone resonances of the C–H bond, which are at ∼1.7 and ∼1.2 μm, respectively.^13,14 The advantages of ∼1.2 μm for imaging lipid over ∼1.7 μm have been explored:^15–18 higher lateral resolution and penetration depth due to weaker water absorption, stronger contrast, and standard optical elements without removing water vapor for beam guidance. Therefore, the 1.2-μm laser is an ideal excitation source for imaging lipid in the tissue. On the other hand, to distinguish water and lipid in tissues, the other wavelength is required, where the excited PA intensity in water should be significantly greater than that in lipid. Fortunately, the easily achievable 1064-nm laser meets the requirements for PA imaging of water.^7,19 Thus, we propose to use the excitation difference of these two wavelengths to identify water and lipid in PAM. At present, most PA imaging studies only map a single content, whether water^20–22 or lipid,^15,23 and it is difficult to map water and lipid in biological tissues simultaneously because of the lack of suitable multi-wavelength excitation sources.

Wavelengths at which lipid and water have high absorption coefficients are typically generated by expensive optical parametric oscillator (OPO) lasers,^{13,20–22,24} supercontinuum lasers,^25,26 or stimulated Raman scattering (SRS) fiber amplifiers.^15,27 In contrast, a passively Q-switched (PQS) solid-state laser (SSL) is cheap, while it has the characteristics of a compact structure and high laser pulse energy (PE).²⁸ PQS-SSL uses a saturable absorber for Q-switching without active modulators, which is beneficial for reducing the laser cavity length. Due to the short length of the laser cavity, nanosecond pulse width (PW, typically <10 ns) can be easily achieved. In addition, the pulse repetition rate (PRR) of the output laser is tunable by adjusting the pump power. Furthermore, combining the PQS-SSL with the Raman crystal can acquire multiple wavelengths at the same time by using the principle of Raman frequency shift.^29–31 Raman crystals are ideal frequency-shift media for multiple biomedical imaging modalities, including functional PAM,³² multiphoton microscopy,³³ and SRS microscopy.³⁴ Thus, PQS-SSL provides a new light source option for PA imaging.

In this paper, we combine the Nd:YAG/Cr⁴⁺:YAG PQS-SSL and a YVO₄ Raman crystal to propose a compact Raman PQS-SSL that produces a 1064-nm fundamental laser and an 1176-nm Raman laser simultaneously. These two wavelengths can efficiently distinguish water and lipid in PAM. The 1064-nm laser (1.4 µJ) and 1176-nm laser (1.7 µJ) are simultaneously obtained with the same PRR of 34 kHz. The PW of the output lasers is less than 2 ns with a compact laser cavity of ∼5.5 mm. Implementing a PAM system with our designed optical source, the distribution of water and lipid in beef slices and the mesenteric fat are mapped. The presented laser structure introduces a new excitation source with optional wavelengths for PA microscopic differentiation of lipid and water.

PA differentiation of water and lipid requires a suitable dual-wavelength excitation source, and the choice of the wavelengths is crucial. The basic concept is to find two wavelengths that excite lipid and water with a distinguishable PA intensity difference. Usually, the optical absorption spectrum of the matter is chosen as the basis for wavelength selection, as the PAM is an absorption-dependent imaging technique. There is no doubt that lipid has an absorption peak near 1.2 μm, where its excited PA intensity is significantly large. Based on this, a wavelength near 1.2 μm can be selected for imaging lipid, so the 1176-nm wavelength in our Raman laser can be used. Now, the question is to check whether the 1064-nm wavelength can be utilized for water. However, we found that the optical absorption spectra of water and lipid are not consistent in different reports.^7,14,35–39 For example, the water is over the lipid in the spectrum in Refs. 14, 35, and 36, while it is opposite in Refs. 37–39. This may be related to the measurement method, conditions, or matter contents.⁷ On the other hand, the excited PA spectrum is not always consistent with the optical absorption spectrum of the matter; usually, they follow a similar trend, but the small variations are not exactly the same.^16,40,41 Therefore, the distinction between water and lipid in PAM cannot simply be based on the absorption spectra of the two substances. In fact, the PA excitation efficiency is not only dependent on the optical absorption but also on the Grüneisen parameter of the substances, which is unique for different tissues.¹³ 

To confirm that the 1064- and 1176-nm lasers can be used for PA differentiation of water and lipid, we measured the intensity of the PA signals at 1064 and 1176 nm for water and lipid substances in PAM. As shown in Fig. 1(a), the PA intensity for beef lipid, butter, mineral water, and distilled water at 1064 and 1176 nm was measured ten times. The mean values and standard deviations of ten sets of data were calculated. The water used here can be regarded as pure samples, and the butter and beef lipid both contain ∼80% fat, where the light absorption of the C–H bond is dominant. Obviously, in Fig. 1(a), the PA intensity of beef lipid and butter at 1176 nm is far stronger than that at 1064 nm, while the PA intensity of mineral and distilled water at 1176 nm is much weaker than that at 1064 nm. Figure 1(b) shows the PA amplitudes of beef lipid and mineral water at 1064 and 1176 nm, respectively. The peak-to-peak value of the PA amplitude was calculated as the PA intensity. Figure 1(c) shows the ratio of the PA intensity at 1064 nm to that at 1176 nm. According to the formula, the ratios for beef lipid and butter are 0.07 and 0.12, respectively, and then, the ratios for mineral water and distilled water are both ∼11. Therefore, the excited PA intensity difference between water and lipid at 1064 and 1176 nm is large enough to be used as an indicator.

Another advantage of using these two wavelengths is that the lasers of 1064 and 1176 nm can be easily generated simultaneously in the Raman SSL. The 1064-nm fundamental laser can be converted to an 1176-nm Raman laser through the 890 cm⁻¹ Raman shift of the YVO₄ crystal. The ratio of the fundamental laser converted to the Raman laser can be adjusted to realize the simultaneous output of 1064- and 1176-nm lasers instead of achieving the maximum Raman conversion, broadband generation,⁴² or higher-order Raman shift.⁴³ Therefore, we select 1064- and 1176-nm dual-wavelength PQS-SSL as a single excitation source to efficiently distinguish water and lipid in PAM.

Figure 2 shows the experimental setup of the dual-wavelength PA imaging system and the energy level transition diagram of the laser. For the laser source, the pump source was an 808-nm fiber-coupled laser diode (core diameter: 105 μm, numerical aperture: 0.22). Two 11-mm spherical lenses were used to collimate and focus the pump beam. The size of the pump beam focus was ∼105 μm. A Nd:YAG crystal (3 mm, 1 at. % Nd³⁺ ions) serving as the gain medium and a Cr⁴⁺:YAG crystal (0.5 mm, 80% initial transmission) acting as the saturable absorber are bonded as a composite crystal. An a-cut YVO₄ crystal with a thickness of 2 mm was used as the Raman gain medium. The surface of the composite crystal near spherical lenses was coated with antireflection at 808 nm and high reflection at 1064 nm to act as the rear cavity mirror (M1). The entrance of the OC (M2) was coated with 98% reflection at 1064 nm to act as the front cavity mirror (M2). The overall cavity length is ∼5.5 mm. The increase in thickness of Nd:YAG, Cr⁴⁺:YAG, and YVO₄ crystals generally contributed to a higher gain, deeper modulation depth, and higher frequency-shift efficiency, respectively, in a certain range.

The energy level transition of the dual-wavelength laser is as follows: First, the Nd³⁺ ions absorb 808-nm pump laser energy, transiting the level from the ground state E₀ to the excited state E₃, then transferring from E₃ to the upper laser level E₂ through the non-radiation transition, where the population inversion is formed. After that, it transits from E₂ to the lower laser level E₁, generating the 1064-nm laser. Finally, E₁ goes back to the ground state E₀ after the other non-radiation transition. After having the fundamental laser, a part of the generated 1064-nm laser is converted to the first-order Stokes laser by the 890 cm⁻¹ Raman shift of the YVO₄ crystal, corresponding to a wavelength of 1176 nm. In this way, the dual-wavelength laser of 1064 and 1176 nm can be easily and efficiently produced in this compact laser cavity.

Then, the laser source was combined with a transmission-mode PAM system^28,44,45 to enable multi-contrast PA imaging. First, the output laser from the source passed an 808-nm long-pass filter (LPF) to filter out the residual pump light. A 1-GHz photodiode (PD) detected a small portion of the laser beam reflected by the LPF. The PD converted the received optical signals into electrical signals that served as the trigger of the oscilloscope (OSC, 1 GHz, MDO3104, Tektronix). Then a 1064-nm bandpass filter (BPF) and a 1100-nm LPF were, respectively, placed to separate the 1064- and 1176-nm lasers to conduct the imaging. It is needed to mention that the two wavelengths can be easily mapped in a dual-pulse sequence for acquiring the information simultaneously in a single scan by adding a delay line to either wavelength.⁴⁶ The size of the laser beam was then expanded by a telescope consisting of L1 and L2. By regulating the focus lengths of L1 and L2, or the distance between them, the beam size was matched to the back-aperture size of the objective lenses (0.1 and 0.4 NA). The laser beam was focused by the objective lens and illuminated on the sample, which was placed in a water sink. The water sink with the sample was controlled by an x–y motorized translation stage for raster scanning. A spot-focusing ultrasound transducer (UT, Shantou Institute of Ultrasonic Instruments, 5P20F25, focal length: 25 mm, central frequency: 4.7 MHz, −6 dB bandwidth: 50.1%) collected the PA waves from the samples and converted them to electrical signals. The electrical signals were amplified by a custom-made radio frequency amplifier (RFA) with 46 dB gain and then filtered by an 18-MHz electronic low-pass filter (ELPF) before being captured by the OSC. Ultimately, a personal computer (PC) was applied to synchronize the sample scan, data acquisition, and data processing.

The performance of the laser source was examined before conducting the PA imaging. The laser spectra were measured by an optical spectra analyzer (Anritsu, MS9740A, resolution: 0.03 nm). As shown in Fig. 3(a), the PQS-SSL oscillated around 1064 nm at first. As the incident pump power (P_in) increased, the 1176-nm Raman laser began to oscillate. The 1176-nm Raman laser is Stokes radiation from the fundamental laser at 1064 nm through the 890 cm⁻¹ Raman shift line. Thus, the 1064- and 1176-nm dual-wavelength lasers oscillated. The tunable power intensity ratio between 1064- and 1176-nm lasers was achieved by regulating the P_in. It can be seen from Fig. 3(b) that the fundamental laser threshold is 0.6 W and the Raman laser is excited when P_in > 0.8 W. The average output power of the total laser increases linearly with P_in. The maximum total output power is achieved at P_in = 2.8 W, which is 103 mW, and no power saturation occurs in this stage. Among them, the fundamental laser power is 46 mW, and the Raman laser power is 57 mW. The Gaussian-shaped beam profiles and excellent beam quality factors (M²) of the dual-wavelength lasers are shown in Figs. 3(b) and 3(c), respectively. The measured M² of 1064- and 1176-nm lasers are 1.12 and 1.67, respectively, which are close to the theoretical fundamental mode. Figures 3(e) and 3(h) show the properties of the dual-wavelength laser pulses. The PRR both increases with the pump power, while the PW (1064 nm, 1.5–1.8 ns; 1176 nm, 1.0–1.3 ns) and PE (1064 nm, 1.4–2.5 μJ; 1176 nm, 1.5–1.9 μJ) maintain a relatively constant level. The PW and PE are mainly determined by the cavity length and modulation depth of the SA, respectively; thus, the two parameters are nearly steady in a certain PQS laser. This feature greatly benefits PA microscopic imaging: the pulse properties are capable of being maintained during the PRR adjustment, resulting in constant imaging performance under different scanning rates. The typical pulse profiles of the dual-wavelength lasers at the maximum P_in = 2.8 W are shown in Figs. 3(f) and 3(i), where no satellite pulses exist. As illustrated in Figs. 3(g) and 3(j), the PRRs of the dual-wavelength lasers are tunable by regulating the pump power. The scalable PRR of the laser source was examined at PRR = 12, 16, 21, and 25 kHz. The corresponding pulse trains demonstrate flat laser outputs under different PRRs, indicating a stable operation of the dual-wavelength PQS-SSL. It is needed to mention that the average output power and PRR of the laser can be further improved by increasing the pump power, as no power saturation occurs in Fig. 3(b).

The laser beam was sent to an optical-resolution PAM system. Then, the lateral and axial resolutions of the PAM system at 1064 and 1176 nm lasers were separately measured by scanning the edge of a carbon fiber with a scanning step of 0.2 μm. The raw data of the PA intensity were obtained by calculating the peak-to-peak intensity of the PA signal, which was then fitted with the edge spread function (ESF). The line spread function (LSF) could be obtained by taking the first-order derivative of the ESF. The full width at half-maximum (FWHM) of the LSF was calculated as the lateral resolution of the PAM system. Through the above data processing steps, the measured lateral resolution for the 1064- and 1176-nm lasers was 4.2 and 4.4 μm, respectively, as shown in Figs. 4(a) and 4(c). The theoretical lateral resolutions (λ/2NA, where λ is the laser wavelength) at 1064 and 1176 nm were calculated to be 1.3 and 1.5 μm, respectively, with a NA of 0.4. The deviation of the measured resolution from the theoretically calculated results was mainly attributed to the short working distance of the objective lens, which caused the sample to be slightly out of focus and the carbon fiber to have a very sharp edge. The axial resolution was measured by extracting the envelope of a single PA signal with Hilbert transformation, as shown in Figs. 4(b) and 4(d). The measured axial resolutions at 1064 and 1176 nm were 318 and 322 μm, respectively, which were very close. In a PAM system, the axial resolution can be calculated by 0.88v/B, where v is the speed of sound and B is the center response frequency (CRF) of the UT. The theoretical axial resolution was 292 μm with v = 1.56 μm/ns and B = 4.7 MHz, which was independent of the wavelength of the excitation light. The main limitation of the axial resolution of the PA imaging system is the low CRF of the UT. Thus, better axial resolution can be effectively achieved by applying a high-frequency UT.

To demonstrate the differentiation of lipid and water in PAM, the beef slice [Fig. 5(a)] is selected for imaging, as it contains rich and separated lipid and water substances. The sample was placed at the bottom of the sink and immersed in water to reduce the attenuation of the ultrasonic signal propagation. The lasers at the two wavelengths were sequentially illuminated on the sample using different optical filters. Figure 5(b) shows photos of the imaged area of the samples and their PA images at 1064 and 1176 nm. The imaging regions of pure beef lipid and beef slice were 1.2 × 1.1 and 1.2 × 1.2 cm², respectively. Multi-contrast imaging results can be achieved by merging the two PA images obtained by the dual-wavelength lasers. It can be seen that the PA images of the dual-wavelength lasers are complementary, that is, the 1176-nm laser images the lipid (green) only and the 1064-nm laser excites the water (red) only. As shown in the first row of Fig. 5(b), by demonstrating PA imaging of pure beef lipid (in the center) and water (surrounding the lipid), it can be found that the two substances can be easily distinguished using 1064- and 1176-nm lasers. As illustrated in the second and third rows of Fig. 5(b), more complete details of the PA images of beef slices were achieved by increasing the NA of the objective lens because the lateral resolution of the PAM system was related to the focus size of the scanning beam. The PA images at 1064 and 1176 nm were, respectively, obtained under illumination energies of 1.4 and 1.9 μJ with a laser PRR of 25 kHz. The scanning step was set to be 100 μm. Figure 5(c) shows the normalized intensity profiles of pure beef lipid and water at 1064 and 1176 nm, as indicated by the dashed line r1. The boundaries of lipid and water along the line direction were clear and complementary. Similarly, PA images of beef slice at 1064 and 1176 nm, indicated by the dashed line r2, were also evaluated, as shown in Fig. 5(d). The water contained in the muscle of the beef slices can be PA excited by the 1064-nm laser. The muscle contains a relatively low water content and also has a certain thickness, leading to weak PA signals being received. To enhance the PA signals, high-NA objectives were utilized for focusing the beam into the sample. The SNRs of the PA images are compared under different NAs, as illustrated in Fig. 5(e). Here, we define the SNR as 10 × log(I_max/I_b), where I_max and I_b are the signal and background intensities, respectively. As indicated by the dashed box r3, the SNR of the PA image under NA = 0.4 is 9.7 dB, which is higher than that of 8.0 dB under NA = 0.1. Thus, a tightly focused beam spot can be applied for PA imaging to enhance the SNR when the absorption coefficient of biological tissues is small.^47–49 The imaging performance of the beef slice illustrates that the dual-wavelength laser with 1064 and 1176 nm is capable of efficiently identifying lipids and water in PAM.

To demonstrate the clinical application of the lipid and water differentiation by the dual-wavelength laser, the mesenteric fat of rats was selected as the sample for PA imaging, as shown in Fig. 6. All experiments with these samples were approved and performed in accordance with the institutional guidelines of Xiamen University. Six male Sprague–Dawley (SD) rats (300 ± 20 g in body weight) were purchased from the Laboratory Animal Center at Xiamen University. Once the rats had been sacrificed, mesenteric fat of about 0.5 × 0.5 × 0.5 cm³ was harvested rapidly and soaked in 10% formaldehyde. The scanning step was set to be 20 μm, and the imaging region was 2.2 × 2.2 mm². Here, the system resolution was not fully utilized in order to scan the sample faster for a wider region. As shown in the first row of Fig. 6, the intestinal crypt can be clearly visualized using the PA imaging system at 1176 nm. In the second row of Fig. 6, the blood vessel can be illustrated using the 1064-nm laser, which is also complementary to the lipid (green) image. The imaging results of the mesenteric fat of rats indicate our dual-wavelength laser source is promising for lipid and water differentiation in clinical applications.

We propose to use a compact dual-wavelength PQS-SSL as the excitation source to directly achieve PA differentiation of water and lipid simultaneously. The novelty of this work comes from two parts: the single compact dual-wavelength laser and the differentiation of water and lipid. The main contribution of our work is to use the excitation difference under 1064- and 1176-nm lasers for mapping water and lipid in PAM, respectively, rather than exploring the highest absorption peak of each substance. Furthermore, 1064- and 1176-nm lasers can be easily generated in a single laser source, and these two wavelengths can also promisingly excite efficient PA signals of water and lipid, respectively. Meanwhile, the miniature structure (cavity size: ∼10 × 10 × 5.5 mm³) of the laser source is not only promising for portable applications but also benefits the PA-desired nanosecond (<2 ns) laser pulse establishment. The dual-wavelength PQS-SSL can provide a 1.4-μJ 1064-nm laser and a 1.7-μJ 1176-nm laser with a maximum PRR of 34 kHz. Our technique is confirmed by efficient PA imaging of water and lipid in beef slices and clinical application of the mesenteric fat in rats. This dual-wavelength, nanosecond, miniature Raman laser provides a novel imaging source for PAM to track changes in water and lipid distribution in biological tissues and study the growth processes of organisms.

See the supplementary material for supplemental Figs. S1–S5.

We acknowledge the National Natural Science Foundation of China Youth Fund (Grant No. 62305274), Natural Science Foundation of Xiamen City, China (Grant No. 3502Z202371001), President’s Foundation of Xiamen University (Grant No. 20720240025), Xiang An Biomedicine Laboratory (Grant No. 2023XAKJ0101031), National Natural Science Foundation of China (Grant Nos. 61275143 and 61475130), Program for New Century Excellent Talents in University (Grant No. NCET-09-0669), and Natural Science Foundation of Fujian Province of China (Grant No. 2021J01052).

The authors declare no conflicts of interest. All experiments with these samples were approved and performed in accordance with the institutional guidelines of Xiamen University.

H.W., L.Z., and H.Y. conducted the experiments. H.W. and H.H. drafted the manuscript. H.W. and Q.Z. provided bio-samples and supported the analysis. X.D. supervised the coding. S.B. supported the laser establishment. H.H. and J.D. conceived and supervised the project. All authors read and edited the manuscript.

Hanjie Wang: Data curation (equal); Writing – original draft (equal). Lin Zhao: Data curation (equal). Huiyue You: Data curation (equal); Software (equal). Huiling Wu: Resources (equal). Qingliang Zhao: Funding acquisition (equal); Supervision (equal). Xin Dong: Software (equal). Shengchuang Bai: Supervision (equal). Hongsen He: Conceptualization (equal); Funding acquisition (equal); Investigation (equal); Supervision (equal). Jun Dong: Conceptualization (equal); Funding acquisition (equal); Supervision (equal).

The data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.