Generating in vivo continuous ultrasound based on sub-terahertz photoacoustic effect

The photoacoustic effect is a phenomenon in which acoustic waves are generated in a sample as a result of thermal expansion involved by light irradiation.^1–4 Since acoustic waves can generally reach deep inside the sample than light, the effect is useful for biomedical and industrial applications such as non-invasive imaging of human bodies^5–7 and non-destructive inspection of subsurface structures.^8–10 To use the photoacoustic effect in practical applications, the choice of wavelength and exposure time is of fundamental importance.^11–16 Despite the importance of water in most applications, photoacoustic signal generation in water has been challenging due to the lack of water-absorptive wavelength sources. A few approaches have tackled photoacoustic signal generation in water using optical parametric oscillation around 1000¹⁷ and 1450 nm,¹⁸ Q-switched pulse generation at 1540 nm,¹⁹ and difference frequency generation at 5.1 μm.²⁰ An all-fiber hybrid optical parametric oscillator at 1930 nm, corresponding to the O–H bond vibrational absorption peak, has recently been developed.²¹ Thus, ultrasound generated in aqueous samples can be used for imaging, which can be performed even in a fully non-contact manner when combined with optical vibrometry.^18,19

Recently, it has also been demonstrated that an intense terahertz pulse generated by a free-electron laser can directly generate an acoustic pulse in water based on its high attenuation in water.²² While the use of terahertz waves offers attenuation in water comparable to that of the mid-infrared wavelengths,²³ it also provides an additional advantage in terms of transmittance through various materials. For example, the attenuation constant of terahertz waves in PLA (polylactic acid), a representative material frequently used for 3D printing, is around 2 cm⁻¹ at 0.2 THz²⁴ while it is above 40 cm⁻¹ for infrared around 1300²⁵ and 1500 nm.²⁶ Another example can be seen in terahertz body scanning based on wave delivery under clothing.²⁷ Therefore, the terahertz waves have the potential of see-through photoacoustic excitation in aqueous samples, which is relevant for practical ultrasound applications ranging from in vivo imaging to food and pharmaceutical inspections. It should also be mentioned that the use of microwaves has been investigated as a means of ultrasound excitation as well, referred to as a thermoacoustic effect.^28–33 While the microwaves can reach inside aqueous samples deeper, the relatively low attenuation constant limits the generation of high-frequency ultrasound. Since the free-electron laser generates terahertz pulses much stronger than using conventional photoconductive switching or parametric wavelength conversion³⁴ and short enough to satisfy the stress confinement condition, it is suited for acoustic pulse generation in aqueous samples.³⁵ However, the free-electron laser necessitates a large-scale facility requiring a dedicated building, thus significantly limiting the handy use of the photoacoustic effect for practical applications.

Here, we show that periodically modulated continuous-wave (CW) sub-terahertz waves can directly generate CW ultrasound in aqueous samples [Fig. 1(a)], which at resonance is detectable even in the air with a commercially available microphone. Although CW photoacoustic approaches have generally been considered in the optical regime,^36–46 its application to water is challenging due to the failure of the stress confinement condition. The sub-terahertz waves around 0.1 THz are absorbed by water with a high attenuation constant of α ∼ 70 cm⁻¹, which is comparable to that of the peak near-infrared wavelengths. Therefore, simple periodic modulation can generate CW ultrasound even if the stress confinement condition is unsatisfied. The proposed method can attain a high signal-to-noise ratio based on lock-in detection while minimizing system complexity and reducing electromagnetic and acoustic peak powers. It should be noted that the recent advancement of high-frequency electronics in the sub-terahertz regime^47–49 has allowed us to generate, modulate, and amplify signals around 0.1 THz using a compact electronics system. By leveraging such advancements and by hybridizing waveguide-based and lens-based approaches, we develop an experimental system no larger than a desktop scale with the potential for further scale-down [Figs. 1(b) and 1(c)]. To our knowledge, this is the first demonstration of an all-electronics system that generates ultrasound using the sub-terahertz irradiation frequency of around 0.1 THz. To demonstrate the great possibility opened by the proposed method, we show proof-of-concept applications of bulk modulus measurement and in vivo anatomical imaging. Ultrasound imaging is a versatile non-invasive method to inspect the inside of a human body. In recent years, its applications are expanding beyond the medical field, for example, gesture recognition from forearm muscle movements,⁵⁰ silent voice input from pharyngeal movements,⁵¹ and motor intentions from subcortical blood flow changes.⁵² While conventional ultrasound imaging requires transducers in contact with the skin for acoustic impedance matching, sub-terahertz irradiation enables ultrasound transmission in a non-contact manner. In contrast to conventional photoacoustic imaging, in which acoustic waves are generated from specific optical absorbers like blood vessels, the proposed method enables the direct generation of in vivo ultrasound under the skin.

We next investigate the temperature dependence of the sound speed for the generated acoustic waves. Figure 3(a) shows the frequency response of the acoustic pressure acquired at different water temperatures. The circles and error bars show the averages and standard errors of ten measurements at each temperature, respectively, while the solid lines are the result of Lorentzian fitting. We observe that the resonance frequency increases as the temperature increases in accordance with the sound speed acceleration. We notice that the peak height also varies with the resonance frequency shift. This is probably attributed to the effect of the standing waves [Fig. 2(c)]; the apparent signal level measured at a fixed microphone position changes when the resonance frequency changes. By substituting the experimental resonance frequencies into Eq. (1), we estimate the sound speed for each temperature. Figure 3(b) shows a comparison of the experimental values (circles) with Greenspan’s empirical values⁵³ (solid line). The vertical error bars are derived by the 3 dB widths of the spectral peaks in Fig. 3(a), while the horizontal error bars are derived by the standard errors of the sample temperatures. The agreement not only validates the modeling based on Eq. (1) but also demonstrates a potential of non-contact measurement applications as discussed in the section “Bulk modulus measurement of gelatin gels.”

This section presents a proof of concept of non-contact bulk modulus measurement based on the sub-terahertz photoacoustic effect. As a sample under test, we prepare gelatin gels used frequently as a model of human tissues.^60–62 For sample preparation, we dissolve granular gelatin (A-U α, Jellice Co., Ltd.) in water with different concentrations of 3%, 6%, 9%, 12%, and 15% at 50 °C and cooling it in a refrigerator at 4 °C for 10 h. We then leave the samples at room temperature until their temperature reaches 24 °C. Similar to the section “Non-contact generation and detection of ultrasound in water,” we use the microphone to detect acoustic pressures generated from the gel samples with different gelatin concentrations formed in the same container, i.e., $Lx,Ly,Lz=22.2,22.2,37.5$ mm. The results are shown in Fig. 4(a). The circles and the error bars are the averages and standard errors of ten measurements at each concentration, respectively, while the solid lines are the result of Lorentzian fitting. We observe that the higher gelatin concentration leads to a higher resonance frequency, i.e., the faster sound speed. Figure 4(b) shows the comparison of the sound speeds derived from our experiment (red circles) and literature^58,59 (blue triangles and gray diamonds) as a function of the gelatin concentration. The error bars are calculated from the 3 dB width of the spectra in Fig. 4(a). The increasing trend of the sound speed to the gelatin concentration is common for all the cases, confirming the consistency between the experimental result and the literature. Considering the high water content, the bulk modulus of the gel, κ, can be estimated as κ = ρc², where ρ is the density⁶³ as summarized in Fig. 4(c). Thus, the proposed method enables a fully non-contact estimation of the bulk modulus of a sample. The ability to inspect the mechanical property of samples such as food and pharmaceutical products without touching them will be important for hygienic and non-destructive inspection.

In this section, we show a proof of concept of in vivo anatomical imaging using ultrasound, which is generated directly under the human skin irradiated with sub-terahertz waves. We use this ultrasound to line scan a hand by translating the sub-terahertz irradiation spot. In this study, we use a hydrophone in contact with the skin as a receiver instead of the microphone for the sake of better impedance matching, resulting in a higher signal-to-noise ratio. As a result, the imaging system is not entirely non-contact. However, as no mechanical contact is required for the transmitter side, transmission imaging can be easily performed using a simple configuration with only one contact point for the receiver side. This stands in contrast to conventional transmission imaging, which requires two contact points to sandwich the object.

To irradiate the human skin with sub-terahertz waves, we follow the Radio Radiation Protection Guidelines (RRPG), which specifies the maximum permissible exposure (MPE) of 10 mW/cm² (6 min average) for partial body absorption for frequencies from 30 to 300 GHz under a controlled environment.^64, Figure 5(a) shows the experimental setup to line scan a hand. The sub-terahertz output from the power amplifier is focused on the dorsal side of the hand with a 1/e beamwidth of 3.6 mm (radius) at the focal length of 25 mm (supplementary material, Fig. S2). To improve the lateral resolution, we use a double lens system composed of two plano-convex lenses, which forms a sub-terahertz focus with a 1/e beamwidth of 3.6 mm (radius) at the focal length of 25 mm (supplementary material, Fig. S2). We then laterally translate the hand fixed on a jig attached to an automated liner stage at a step of 1.5 mm while the transmitter and receiver (hydrophone) are fixed. During translation, the palm smoothly slides on the hydrophone, which is in contact with ultrasound gel. The hydrophone is connected to the lock-in amplifier. At each measurement point, we sweep the modulation frequency from 20 to 200 kHz at an interval of 9 kHz. At each frequency, we emit and pause the sub-terahertz waves for 0.335 s each. When the irradiation spot is translated with a step of 1.5 mm for a line scan on the skin, each point experiences five times irradiation processes due to the spatial overlap as estimated by the ratio of the spot diameter and the translation step. Under these conditions, we set the duty ratio to be 7.2% resulting in the average irradiation power density of 9.1 mW/cm² [supplementary material, Eq. (S6)]. The data acquisition time for each subject is about 20 min. The experimental protocol has been approved by the Research Ethics Committee of the University of Tokyo 22-22. Four subjects including two males and females in their 20s participated in the study. Figure 5(b) shows the hands of the four subjects. Figure 5(c) shows time-domain representations of the scan results, in which the amplitude of the inverse Fourier transform of the frequency-domain acoustic signals is represented. The horizontal and vertical axes indicate the position and propagation time, respectively. The images reveal the distribution of muscles and phalanges, which facilitate and impede ultrasound propagation due to the contrast of the lower and higher acoustic impedances, respectively. The ultrasound first appears at around t₀ ≃ 16 µs, which is consistent with the propagation time through the thickness of the hand. The temporal profiles exhibit a cyclic pattern occurring at ∼11 µs intervals across all subjects. We attribute this to the weak resonant response of the hydrophone used in this study, which occurs around 90 and 190 kHz.⁶⁵ Since the time-domain signal thus includes two major oscillatory signals at those frequencies, its absolute value will involve the apparent difference frequency of 100 kHz, corresponding to the cyclic pattern of 10 µs. Such an artifact could be suppressed by implementing a filter on the receiver side or pre-emphasis on the transmitter side. The temporal resolution of the frequency-domain acoustic imaging is governed by the bandwidth, similar to swept-source optical coherence tomography.⁶⁶ Although we have used frequencies below 200 kHz in this study due to the detectable range of the hydrophone and the lock-in amplifier, the generation of higher acoustic frequencies up to the MHz range will be possible simply by increasing the modulation frequency without hardware change (supplementary material, Fig. S5). Meanwhile, the lateral resolution of our photoacoustic imaging is governed by the greater of the two parameters, the acoustic wavelength and the irradiation spot size. Considering the sound speed in the water, the acoustic wavelength at f MHz is 1.5/f mm, where f is 0.2 in this study. In this case, the irradiation wavelength of 3 mm (0.1 THz) is smaller and the lateral resolution does not improve even if the irradiation frequency is increased. For higher acoustic frequencies when f > 0.5, using sub-terahertz waves with higher frequencies can improve lateral resolution although generating high-power waves at higher frequencies becomes more challenging. Yet, considering the high degree of freedom in amplitude and phase modulation of sub-terahertz waves using electronics-based systems, it will be possible to apply super-resolution techniques,⁶⁷ such as time-reversal⁶⁸ and structured acoustic illumination,⁶⁹ to improve the lateral resolution. In this study, the focal length is fixed, and variations in the height of the hand surface profile cause the spot size to change, resulting in image blur. To address this issue, variable focusing of sub-terahertz irradiation will help improve lateral resolution. Although challenging, there is also a possibility of making the receiver non-contact based on optical interferometry, which has been a hot topic in the field of photoacoustics.⁷⁰ When combined with our method, such a detection scheme will enable fully-noncontact in vivo ultrasound anatomical imaging.

In conclusion, we have demonstrated the generation of CW ultrasound in aqueous samples, including in vivo human bodies, using modulated CW sub-terahertz waves. The high absorption coefficient of water to sub-terahertz waves allows efficient generation of ultrasound in water without any additional absorber. Although we have used frequencies below 200 kHz in this study due to the limitation of our acoustic receiving system, the generation of higher frequencies up to the MHz range will be possible simply by increasing the modulation frequency. After quantitatively characterizing the fundamental properties of the proposed method, we have shown proofs-of-concept sensing applications including non-contact bulk modulus measurement and in vivo anatomical imaging. The former is based on the through-wall delivery of sub-terahertz waves to a sample in a dielectric container. We showed that its acoustic resonance is detectable even in the air with a microphone. The fully non-contact characterization is advantageous for, for example, the hygienic and non-destructive inspection of food and pharmaceutical products. The latter is based on ultrasound transmission imaging with a simple experimental configuration that requires only one contact point for the receiver. We have visualized the distribution of muscles and phalanges in the hands of the subjects. While we have adopted a simple mechanical line scan in this study, spatiotemporal modulation of the sub-terahertz irradiation pattern will open greater possibilities. For example, it is conceivable that the wavefront of in vivo ultrasound can be synthesized based on the phased array principle when irradiation timing among multiple points is controlled. This will contribute not only to improving the current experimental setup in terms of scanning speed and surface profile compensation but also to synthesizing stronger in vivo ultrasound without increasing the irradiation power density by distributing the exposure load to a wider area of the skin.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

See the supplementary material for experimental details such as the sub-terahertz beam power and the water temperature measurement.

The authors thank Professor Masahiko Inami with the University of Tokyo for fruitful discussions. This work was supported by JST PRESTO (Grant Nos. JPMJPR18J9 and JSPS KAKENHI 21K18307).

The authors have no conflicts to disclose.

N.I. and Y.M. conceived the experiments, N.I. conducted the experiments, N.I. and Y.M. analyzed the results and wrote the manuscript.

Natsumi Ichikawa: Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Software (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Yasuaki Monnai: Conceptualization (lead); Data curation (equal); Formal analysis (equal); Funding acquisition (lead); Investigation (equal); Methodology (equal); Supervision (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal).