The elastic and viscous characteristics of biological tissues are important parameters to reflect the biomechanical properties, and we can detect changes in biomechanical properties to reflect the degree of tissue pathology^[1]. Although ultrasound elastography (USE)^[2], optical coherence elastography (OCE)^[3,4], and magnetic resonance elastography (MRE) are currently being used in clinical and related research studies^[5], most biological tissues show viscoelastic characterization, and only elastic property is not comprehensive and accurate for specifying biomechanical characteristics and pathological status^[6,7]. Photoacoustic viscoelasticity imaging (PAVEI) is a sub-degree rising imaging technique, which is a sub-degree extension of photoacoustic imaging (PAI) to probe the mechanical properties of tissues under the microscope^[8]. The technique is capable of characterizing tissue viscoelasticity information. Our previous work^[9–14] proposed a variety of biomedical research prefiguring great potential for identifying the pathological status of tissues. Examples include vulnerability assessment of atherosclerotic plaque, esophageal tumor early detection, and single-cell elasticity and viscosity mappings. As we all know, diseased tissue is accompanied by biomechanical property changes deep inside solid tissues or organs. However, PAVEI applications are limited to the observation of surfaces or luminal of organ systems, and its utility is restricted by the limited image penetration depth^[8,10,12]. Recently, we have proposed spectral interferometry depth-resolved PAVEI^[15], which breaks the dilemma of viscoelastic imaging of surfaces. Because of light scattering and absorption in biological tissue, the achievable imaging depth is approximately 500 µm below the tissue surface. However, to achieve the viscoelasticity characterization in deeper tissues, the sample arm can be made into a needle probe^[15].

Through the ultrathin needle probe, optical coherence tomography (OCT) can obtain three-dimensional structural or functional information of tissues and organs deep inside the organism in a minimally invasive way^[16–19]. The needle probe can be interfaced with the sample arm of the OCT system, thus overcoming the limitation of the imaging depth (1–3 mm) of the conventional OCT system, making it an important biomedical detection method. In 2000, Li et al. developed the first bare single-mode fiber optic probe and realized the imaging of the hamster’s leg muscle nerve bundles with an OCT system based on the probe^[19]. In 2002, Reed et al. proposed an all-fiber-type ultrathin needle probe with a “single-mode fiber + self-focusing (gradient index, GRIN) fiber” structure for low coherence interferometer research^[20]. In 2007, Mao et al. presented a GRIN fiber probe with a “single-mode fiber + coreless fiber + GRIN fiber” structure and verified that the coreless fiber can improve the working distance of the probe^[21]. By equipping the traditional lens with a graded index fiber (GIF) as the focusing element, the outer diameter of the probe can be reduced to 125 µm or less, allowing it to be placed in medical devices such as injection needles and biopsies with a diameter of 0.8 mm or 0.5 mm. Currently, the needle probe has been applied to such frontier fields as real-time monitoring of human cerebral blood vessels in neurosurgery^[16], alveolar and bronchial tube in sheep^[22], optically guided needle biopsy^[23], muscle detection in myotonic dystrophy^[17], and identification of tumor boundaries in isolated breast tissue^[24–27]. Therefore, the development of minimally invasive needle probes is conducive to pathological research, diagnosis, and treatments.

In this Letter, we present a minimally invasive needle probe (outside diameter, 0.7 mm) for deeper photoacoustic viscoelasticity (PAVE) measurements. A set of gelatin phantoms and biological tissues validated the feasibility of this needle probe. Meanwhile, experimental results from in vivo animal models provide great potential for tumor boundary detection deep inside biological tissues in medical applications.

When an intensity-modulated continuous laser is utilized to excite absorptive isotropic viscoelastic materials, light intensity is modulated in the form of $$I=\frac{1}{2}{I}_0(1+\cos \omega t),$$(1)where $I_0$ is time-averaged light intensity, and $\omega$ is the modulation frequency. The periodic changes of light intensity can result in the sinusoidal fluctuation of the temperature of the absorber, and, based on the thermoplastic expansion mechanism, the periodical heating of localized regions can cause thermal stress. The thermal stress causes the sample to generate thermal strain in the form of photoacoustic (PA) oscillations, with a dominant frequency equal to the modulated frequency. During the above process, the strain often lags behind stress; this is caused by the damping effect of viscoelastic biological tissues. According to the Rheological Kelvin–Voigt model, we know the correlation between the viscosity-elasticity ratio $\eta /E$, and the phase-delay $\delta$ can be presented as^[8]$$\delta =\arctan \frac{\eta \omega }{E},$$(2)where $E$ is Young’s modulus, $\eta$ is the coefficient of viscosity, and $\omega$ is the modulation frequency. From Eq. (2), we can obtain the viscosity–elasticity ratio of the tissues.

Depth-resolved PAVEI used to be performed by a spectral domain low coherence interferometry (SDLCI) device^[15], as illustrated in Fig. 1(a). When the excitation light is used to radiate the sample, PA oscillations modify the local optical path length $\mathrm{\Delta }\mathrm{OPL}$ and the optical phase difference $\mathrm{\Delta }\mathrm{\Phi }$ of the detection light^[15]. The relationship between the depth-resolved mechanical phase delay $\delta (x,z)$ and optical phase difference $\mathrm{\Delta }\mathrm{\Phi }(x,z,t)$ can be expressed as^[15]$$\mathrm{\Delta }\mathrm{\Phi }(x,z,t)=a·\cos [\omega t+\delta (x,z)]+{\phi }_n(t),$$(3)where $(x,z)$ indicates the lateral and depth coordinates of the scatter position within sample. $a=4\pi z_0{\epsilon }_A/{\lambda }_0$ is the optical phase oscillations’ amplitude, where ${\lambda }_0$ is the central wavelength of the light source, ${\epsilon }_A$ is the amplitude of the complex strain, $z_0$ is the minimum depth resolution, ${\phi }_n(t)$ is optical phase noise. Therefore, once we obtain $\mathrm{\Delta }\mathrm{\Phi }(x,z,t)$, the mechanical phase delay in the depth direction is available. According to the quadrature demodulation method, ${\phi }_n(t)$ is suppressed, and $\delta (x,z)$ can be extracted. The extraction method of the depth-resolved mechanical phase delay is shown in Fig. 1(b), where the black curve is the reference curve, and the red dotted curve is the optical phase difference oscillation curve.

Figure 2(a) illustrates a schematic diagram of the needle probe, which includes a needle (0.7 mm outer diameter) with a side output window, a customized prism (45° reflection angle), a GRIN lens (0.5 mm diameter and 1.5 mm focal distance), a single-mode fiber, and a capillary tube. The GRIN lens was used to focus the output laser and collect the returned detection light. The prism was used to reflect the laser to excite the tissue. The prism and the GRIN lens were bonded with an ultraviolet (UV) adhesive that is interfaced with a single-mode fiber, and the fiber was inserted into a capillary tube to ensure maximum collection of scattering light. Then, the probe is encased within a 22 G needle. An image of the assembled needle probe is shown in Fig. 2(b). After the above process, the needle probe was installed on the motorized pull-back stage assembly and connected to the sample arm of the SDLCI system. Figure 2(c) shows the schematic layout of the SDLCI system. A continuous pump laser at 532 nm is being modulated by a chopper (MC2000B, Thorlabs), which is coupled to a wavelength division multiplexer (WDM, WD202A2-FC, Thorlabs) by a multimode fiber. In the SDLCI system, the light source is a super luminescent diode (SLD1325, Thorlabs) with a central wavelength of 1325 nm and a full width at half-maximum (FWHM) bandwidth of 94 nm. The measured axial resolution is 11.7 µm in air. After passing through an isolator, the light was coupled to a 50/50 fiber coupler and was separated into a reference arm and sample arms. Simultaneously, the excitation beam is coupled to the sample arm via a WDM and then coupled to a needle probe. The interference light signal is detected by a homemade spectrometer, which includes a collimator, a spectro-grating (1145 lines/mm), a doublet lens, and a line scan CCD (GL2048R, Sensors Unlimited). In order to enable SDLCI detection and PA excitation at the same time, the output signal of the chopper is used for the trigger signal of the data acquisition card (DAQ, PCIe-1433, National Instruments). For maximum suppression of the jitter of the external environment, the entire system is placed on an air-floating optical platform. A phase-sensitive method was adopted to obtain the phase sensitivity of the system. We measured three thousand A-lines using a static mirror as a sample to obtain the phase deviation. The phase sensitivity of the SDLCI system was measured as 9.1 mrad; this is equivalent to a displacement sensitivity of 949 pm, as shown in Fig. 2(d).

To verify the feasibility of needle depth PAVE measurement, a set of gelatin phantoms was prepared. The schematic diagram of the gelatin phantoms is illustrated in Fig. 3(a). The gelatin phantoms were made with a mixture of Indian red ink, intralipid solution, distilled water, and gelatin powder. Indian red ink (0.2%) was applied to supply optical absorption, and intralipid solution (0.3%) was used to increase light scattering to improve the signal-to-noise ratio (SNR). The gelatin powder concentrations of the three gelatin phantoms were 45% (${\mathrm{G}}_1$), 33% (${\mathrm{G}}_2$), and 14% (${\mathrm{G}}_3$). These phantoms have the same optical absorption coefficient but different viscoelasticity, which were stacked together to form two boundaries. After completing the above process, the experiment was carried out. The needle probe was inserted in the phantoms along the black dotted line over a distance of 15 mm, with a speed of 0.05 mm/s and a discrete step of 0.2 mm. Three-cycle phase oscillation curves for all depths were acquired by intensity-modulated continuous laser-induced PA oscillations at 1000 A lines at each excitation position (M-mode) with the SDLCI system. Figure 3(b) illustrates the optical phase oscillation curve and the extraction process of the mechanical phase delay at the positions marked with three black arrows in Fig. 3(a). The phase delay of the three optical phase oscillation curves is calculated by the method of quadrature demodulation with the reference curve^[15]. The mechanical phase delays at the three black arrows are 7°, 19°, and 27°, respectively. Figure 3(c) exhibits the mechanical phase delay distribution along the needle trajectory. The average phase delays are 7°, 18°, and 27°. It is obvious that there is a significant variation tendency of increased phase delay, illustrating that the three gelatin phantoms have different viscoelasticity. To confirm the reliability of the results, we measured the internal fraction angle of the phantoms using a rheometer (MARS III). The results are shown in Fig. 3(d), which indicates that the mechanical phase delay and internal fraction angle have the same variation tendency. These prove that the needle probe is an effective technique for depth PAVE measurement.

To illustrate the ability of the needle probe to detect the phase delay of different biological tissues and to distinguish tissue interfaces by different biomechanical properties, we validated it with porcine muscle and myocardium. PA excitation and SDLCI detection were performed during the slow insertion of the needle probe into the tissue. Figure 4(a) presents the distribution with mechanical phase delays along the needle probe trajectory. The average phase delay for muscle is 64°, which is higher than the 44° for myocardium. As we expected, it is clear that there is significant phase delay variation tendency between muscle (1–5 mm) and myocardium (5–10 mm), which indicates that muscle has greater viscoelasticity than myocardium. Further, the accuracy of the results was verified by a rheometry (MARS III).

Figure 4(b) shows the statistical results of the mechanical phase delay and the internal fraction angle. As can be seen, both have the same variation tendency, indicating that the needle probe has the ability to distinguish mechanical contrast and determine biological tissue interfaces.

Next, to further illustrate the ability of the needle probe to detect tumor interfaces in vivo, we used a nude mouse model of hepatocellular carcinoma (HCC). A mouse model of HCC was established by injecting a Hepg-2 tumor cell suspension (0.1 mL, $2\times {10}^7\mathrm{cells}/\mathrm{mL}$) into the normal tissue of the mouse leg. The mouse was deeply anesthetized with a single dose of pentobarbital and secured on a platform. Then, the needle probe was inserted at 0.05 mm/s under ultrasound guidance and advanced toward the tumor border. Figure 5(a) shows an ultrasound image of the tumor and surrounding normal tissues captured during needle insertion, with the probe running from right to left about 1 cm below and parallel to about 1.5 mm of the mouse’s leg surface. The tumor mass presents as a hypoechoic region in the image, which indicates that the contrast between the tumor and the surrounding normal tissue is high, and the tumor interfaces can be clearly identified. The blood in the tissue provides light absorption, and PA oscillations are detected by the SDLCI system. When the needle passes through the tumor interfaces, the mechanical phase delay decreases significantly, as shown in Fig. 5(a). This suggests that the tumor may be stiffer than the muscle. The average phase delays of muscle and tumor are 49° and 28°, respectively. To verify the accuracy of the results, we measured the internal fraction angle of the tissues using a rheometer (MARS III), which agrees well with the mechanical phase delay in Fig. 5(b). According to the variation tendency of the mechanical phase delay between the tumor and surrounding normal tissue, we can clearly distinguish the tumor interfaces in deeper tissues.

In summary, we have developed and demonstrated a PAVE measurement needle probe that allows the characterization of the viscoelasticity at a depth of 1 cm within the sample. The needle probe combines the techniques of PAVEI and SDLCI. The effectiveness was verified by the detection of gelatin phantoms and in vitro biological tissues. The results show that the needle probe has the ability to characterize the mechanical properties of deep tissues. In addition, the experiment results of the in vivo animal model show that it has great potential for tumor boundary detection. The interference signal is unstable during experiments, mainly due to low SNR and partly due to jitter in the experimental environment. However, the needle probe comprising miniaturized focusing optics consisting of coreless fiber and GRIN fiber may improve SNR and reduce the probe diameter^[22]. Sensitivity can be improved by laying the reference and sample arms into a common path configuration, thus reducing the phase turbulence caused by environmental variables. In this way, detection speed and SNR can be maximized.

In conclusion, we present for the first time, to the best of our knowledge, a needle probe that measures PAVE at a depth of 1 cm below the sample surface. This needle probe opens up a new approach to clinical potential differentiated tumor boundaries through the different viscoelasticity of biological tissues. We believe that this technique can be used to detect tumor boundaries in deeper tissues, such as gliomas or breast tumors intraoperatively.