1. Introduction

Photoacoustic microscopy (PAM) [1–10], is a hybrid imaging modality, combining the merits of high optical absorption contrast with deep acoustics penetration. In PAM system, when the imaging target is illuminated by the focused short-pulsed laser beam, an acoustic impulse response is excited, which is known as PA effect. The PA waves are picked up by an ultrasonic transducer to form images of the tissue. According to the relative size of the optical focus and the acoustic focus, PAM can be further classified into OR-PAM and AR-PAM. In OR-PAM, the optical focus is smaller than the acoustic focus. It provides a maximum lateral resolution at the nanometer level at an imaging depth of about 1 mm. In contrast, AR-PAM has a larger optical focus. The imaging depth can reach several to tens of millimeters, and the lateral resolution usually ranges from a few tens to hundreds of micrometers.

In both systems, the ultrasonic transducer has an important influence on the imaging performance. The spatial resolution, imaging depth and contrast of PAM images are affected by the numerical aperture, center frequency, bandwidth, and sensitivity of the transducer. The increasing interest in obtaining higher resolution images has heightened the need for developing ultrasound transducers.

In the structure of a piezoelectric ultrasound transducer, the piezoelectric materials play the most important role, which can create electrical charge when mechanically stressed. Commonly used piezoelectric materials include piezoelectric polymers, PZT ceramics [11–13], LiNbO3 crystals [14–16] and ZnO films [17]. Among these materials, polyvinylidene fluoride [18,19] (PVDF) is a typical piezoelectric polymer. It has several merits, including the following: (1) PVDF’s high mechanical damping ability and unique permittivity make PVDF transducers exhibit wide bandwidth, leading to a higher axial resolution. (2) Its mechanical flexibility makes it possible to fabricate wearable sensors [20] and spherical focused transducers with larger numerical aperture [21,22]. (3) It has good coupling efficiency with human tissue and ultrasonic gels because of its low acoustic impedance. So, PVDF transducers always have a high receiving efficiency [23–26] even if it is not equipped with a matching layer. (4) PVDF’s transparent nature allows it to be used to make optically transparent transducers [27,28], which could enable PAM imaging systems with a simpler configuration of the optical excitation and acoustic detection. These characteristics mentioned above make PVDF transducers suitable for PAM systems.

However, when the ultrasound reaches the PVDF transducer, part of it will be converted into the electrical signal, and a non-negligible part will penetrate the PVDF film and be attenuated and lost. We can expect that the sensitivity of the transducer could be improved if the signal that penetrates the PVDF membrane can be utilized effectively.

In this study, we proposed a bionic design to enhance the sensitivity of a focused PVDF ultrasonic transducer. Our design mimics the tapetum lucidum [29,30] in the eyes of many nocturnal animals. The tapetum lucidum is in the choroid region of the eye between the lens and the retina, as shown in the Fig. 1(a). It reflects the light which were not absorbed after they passed through the photoreceptors and provides a second chance for the photoreceptors to absorb the light. It increases the amount of light absorbed by the photoreceptors, thereby improving visual sensitivity.

 

Fig. 1. A bionic design of the ultrasonic transducer with a back reflector. (a) A cross-section of the eye ball: 1 tapetum lucidum; 2 retina; 3 vitreous body; 4 choroid; 5 sclera; (b) Schematic of the transducer design and (c) Three-dimensional section view of the transducer: 1 metal housing; 2 3D printing base; 3 PVDF film; 4 Epoxy backing; 5 back-reflector; (d) Photographs of 1 the fabricated transducer and 2 the reflector.

Download Full Size | PDF

Imitating the tapetum lucidum in the eyes, we propose a new design of focused PVDF transducer by involving a back-reflector. The reflector is located behind the PVDF film and filled with epoxy resin in the middle. After the ultrasound signal passes through the PVDF film, it will be reflected and redirected by the reflector and received by the PVDF again. Then, the two signals are superimposed to increase the sensitivity of the transducer. We tested the center frequency, bandwidth, and the resolution of the transducer, and integrated it in AR-PAM and OR-PAM for biological imaging experiments with multiple resolution.

2. Design of PVDF transducer

Figure 1(b) and 1(c) show the design details of the developed transducer. A 28µm-thickness PVDF film (3 in Fig. 1(b)) was coated with Au/Cu electrodes on both sides. The film was molded into a concave sphere with a large numerical aperture. The acoustic focal length of the transducer is 6.25 mm, the aperture diameter is 8 mm, the beam conical angle is 88.43°, which corresponds to an acoustic numerical aperture of 0.64.

The metal back-reflector (5 in Fig. 1(b)) is located on the back side of the piezoelectric film. The gap between the PVDF film and the reflector is 1 mm. The nonconductive epoxy (4 in Fig. 1(c)) was injected in the gap to keep the spherical profile of the PVDF film and maintain the transducer’s mechanical stability. The shape of the reflector’s upper surface is a smooth concave sphere with a radius of 7.25 mm, and its lower surface is flat that acts as a part of shield housing of the probe tail. There is a hole in the middle of the back-reflector to allow the laser passing through. In the manufacturing process, the centers of the back-reflector and the piezoelectric layer are placed on the same central axis, so that the concave sphere of them are focused at the same point.

The fabrication procedure of this transducer is described as follows: First, the PVDF film with electrodes was fixed on a 3D printed base (2 in Fig. 1(c)), which can be stuck together with the reflector. And the fixed parts are placed together in the pressing fixture. The pressing fixture is a home-made instrument which can press PVDF membrane into concave sphere. It contains two splints for keeping the stack (PVDF film, the back-reflector and the 3D printed base) in place and a clamp to apply force. A steel ball is used to form the concave sphere of the PVDF membrane with a radius of 6.25 mm. Next, the pressing fixture was inverted after the membrane was stretched to the expected shape. Nonconductive epoxy was injected through the small hole in the middle of the back-reflector. Then, after the epoxy resin was fully cured, the stack was taken out of the pressing fixture and trimmed into a proper size. The electrodes on both sides of PVDF were soldered with the coaxial line. Finally, the transducer was assembled into a metal shell. Electrically conductive silicone was used to seal the whole transducer. Figure 1(d) shows a photo of the fabricated transducer and its back-reflector.

It can be expected that the fabricated transducer will receive two signals when a sound source appears in its focused area. One is the direct signal from the sound source, another one is the reflective signal from the backing after the ultrasound penetrates the PVDF, as shown in Fig. 2(b). The time delay between the two signals depends on the distance between PVDF film and the back-reflector.

 

Fig. 2. Performance test by using the ultrasound pulse-echo method. (a) Schematic of the pulse-echo system; (b) Pulse-echo signal (blue line); The spectrum of signal 1(dashed green line) and signal 2(dashed-dotted orange line); (c) Direct signal 1(blue line) and superimposed signal (red dotted line); spectrum of superimposed signal (green dashed line).

Download Full Size | PDF

The received two signals can be superimposed by using the matched filtering algorithm [31]. The matched filter is one of the best linear filters, and often used in the receiver of communication, radar, and other systems. The signal received by the designed transducer in PAM system is represented as s(t). It consists of two parts with a time interval ΔT = 2 T/c, where T is the thickness of the gap between the PVDF membrane and the back-reflector, c is the speed of sound in the nonconductive epoxy. The first part s₁(t) is directly received by PVDF. The second part s₂(t) is the reflective signal from the backing. The superimposed signals can be calculated as q(t)= s(t)$\otimes$h(t). h(t) is the impulse response of the matched filter. h(t) can be estimated from the impulse response h₁(t) of the transducer without the back-reflector, that is, h(t)= h₁(t)+ h₁(t – ΔT). In contrast, the convolution s₁(t) $\otimes$ h₁(t) can be considered as the result that does not involve the reflective signal from the backing.

3. Performance test

The performance of the developed transducer was tested in experiments.

3.1 Bandwidth, central frequency and receiving sensitivity

The central frequency and bandwidth of the fabricated transducer were tested by the setup shown in Fig. 2(a). The transducer was driven by a pulser/receiver (CTS-8077PR, Guangdong Goworld Co. Ltd. China). It emitted ultrasound pulses onto a smooth glass target placed at the focal point in water. The reflected pulses were received by the transducer and amplified by a built-in amplifier of the pulser/receiver. The measured pulse-echo response was displayed and recorded by a digital oscilloscope, as shown in Fig. 2(a).

The received signals consist of two parts. One is the direct signal that is radiated from the sound source (signal 1 in Fig. 2(b)), another is the signal from the back reflector (signal 2 in Fig. 2(b)). The amplitude of the back reflection signal is about 30.9% of the direct signal. Figure 2(b) also plots the spectrums of the two signals. The central frequency of signal 1 and 2 are 13.4 MHz and 11.9 MHz, and the bandwidth at −6 dB is 94% and 104%, respectively. Therefore, the back reflection signal has similar bandwidth, center frequency and waveform to the direct signal, with a lower amplitude.

Figure 2(c) gives the signals after processing by the matched filtering. The dotted red line is the superimposed signal that contains reflective signal. While solid blue line is the direct signal without the reflective signal. The peak-to-peak value of the superimposed signal is increased by 29.30% in comparison with the direct signal, which means that the sensitivity of the transducer is improved by 29.30%. And Fig. 2(c) also shows the spectrum of the superimposed signal. The fabricated transducer has a central frequency of 13.07 MHz and a bandwidth of 96% at −6 dB.

To further confirm the increased receiving sensitivity of the transducer in PAM, a black tape experiment was conducted in PAM with the fabricated transducer. Figure 3(a) gives the schematic of the imaging setup. A Nd:YAG laser (EXPL-532-2Y, Spectra-Physics) provided laser pulses with a wavelength of 532 nm, pulse duration of about 8 ns, and repetition rate of 10 kHz. The laser beam is focused on the imaging target by an objective lens. Part of the laser is received by the photodiode (PDA10 A-EC, Thorlabs) as a trigger source. A three-dimensional linear stage was used to optimize the position of the ultrasound transducer. The focus of the ultrasonic transducer is adjusted to coincide with optical focus to maximize the sensitivity of the system. The signal received by the transducer is amplified by the amplifier (SA-230F5, NF Corp.) with a gain of 46 dB and then digitized by a data acquisition card (NI-5761, NI) at a sampling frequency of 250 MHz. Both the target and the front surface of the transducer are immersed in a water tank for acoustic coupling. The imaging target is placed in a water tank, which is fixed to a 2D motorized translational stage (M-VP-25 XA, Newport Corp.), and a two-dimensional horizontal image can be obtained by point-to-point scanning. Figure 3(b) shows a detailed enlarged image of the optical-acoustic combiner, which represents two confocal modes of OR-PAM and AR-PAM, respectively.

 

Fig. 3. Performance test of the transducer in PAM. (a) Schematic of the PAM system; (b) The focusing mode of acoustic path and optical path (left: OR-PAM, right: AR-PAM); (c) PA signal of the black tape; (d) Processed PA signal with back-reflector (red dotted line) and without back-reflector (blue line).

Download Full Size | PDF

In the experiment to test sensitivity, black tape was used as an imaging target, which was sticked to the bottom of the water tank. When the laser irradiates the black tape, the energy is absorbed and converted into ultrasound, known as PA signal. The PA signal will propagate in the water and be detected by the ultrasound transducer, as shown in Fig. 3(c). The received signals also consist of two parts and the ratio of the amplitude of signal 2 (reflected signals) to signal 1 (direct signal) is 34.30%. Figure 3(d) shows the superimposed signal achieved by the matched filtering algorithm. Its peak-to-peak value improved by 28.90%, which also demonstrates that this design can improve the imaging sensitivity by 28.90% in the PAM system.

Overall, the back reflector increases the transducer sensitivity about 30%, and almost does not change the center frequency and bandwidth.

3.2 Acoustic focus size

In the following experiment, we measured the acoustic focus size of the transducer. A small acoustic focus size can maintain a relatively high spatial resolution when the optical illumination is weakly focused. In the AR-PAM system, laser beam is not strictly focus. The optical focus is larger than the acoustic focus. In this way, the lateral resolution of the system depends on the acoustic focal diameter of the transducer, i.e., 0.71 v/(f₀×NA), where v, f₀ and NA are the sound velocity in the tissue, the center frequency and the acoustic numerical aperture of the transducer, respectively.

The diameter of the acoustic focus of the focused transducer was characterized by measuring the full width at half maximum (FWHM) of the corresponding line spread function (LSF), which is the derivative of the edge spread function (ESF). And the ESF was obtained by imaging the shape edge of a blade with the scanning step of 20µm point-by-point. Figures 4(a) and 4(b) show the maximum amplitude projection (MAP) of the blade before and after the signal superposition by the matched filtering algorithm. And the amplitude profile across the edge was fitted to compute the ESF and LSF, as shown in Fig. 4(c) and 4(d), indicating the acoustic focal size is 136µm and 140µm before and after the signal superposition. The results are close to the theoretical value 127µm obtained by formula mentioned above. In fact, the small acoustic focus also benefits from the large numerical aperture of the transducer, owing to the superior stretchability of PVDF.

 

Fig. 4. Measurement of the acoustic focal size. (a) and (b) MAP images of the sharp edge of a blade with the direct signals and the superimposed signals; (c) and (d) Profiles along the x-direction. ESF, edge spread function; LSF, line spread function; (e) The intensities of signals in the white dotted square in (a) and (b) along the acoustic axis. (f) and (g) PA signal of a single carbon fiber.

Download Full Size | PDF

In order to measure the length of the focal zone, we changed the depth of the transducer in the z direction with an interval of 0.3 mm and repeated the scanning process. We recorded the average intensity of the pixels in the white dashed box in Fig. 4(a) and 4(b) to measure the increased sensitivity of the transducer at the height near the focal area, as shown in Fig. 4(e). As shown, the sensitivity of the transducer is improved near the focal plane. And the depth of the focal zone is about 1.52 mm after superimposition, 1.65 mm before superimposition.

Besides, the axial resolution of the transducer was determined by performing an A-scan of a single carbon fiber, as shown in Fig. 4(f) and 4(g). The envelope is obtained by calculating the absolute value of the Hilbert-transformed PA signal. By measuring the FWHM of the Gaussian fitted curve, the axial resolution of the transducer is estimated to be 194.17µm without the back-reflector, 196.61µm with the back-reflector.

4. Multiple scale imaging using the PVDF transducer

Finally, we applied the transducer to AR-PAM and OR-PAM to examine its practicability.

The OR-PAM system is shown in Fig. 3(a). The pulsed laser is focused on specimens by an objective lens through the bottom of the water tank. The imaging target is a piece of fresh bamboo leaf. The spherically focused PVDF transducer is arranged above the tank and confocally aligned with the laser. A total area of 1.6 mm × 2 mm was scanned with a step size of 8µm in both the x and y directions. Figures 5(a) and 5(b) show the MAP images extracted from the direct signals and the superimposed signals. As shown, the image reconstructed by superimposed signals has greater intensity than that reconstructed by direct signals. Figure 5(c) shows a microscopic photo of the imaging area. Bamboo is a monocotyledonous plant [32] with leaves that have parallel and transverse veins. The veins divide the bamboo leaves into leaf gaps. In a leaf gap unit, it is faintly light in the middle and dark on both sides. This is because that the bulliform cells are in the middle of leaf gaps. They have higher water content and are lighter than other cells in the leaf. And the color difference is more obvious in PA images.

 

Fig. 5. OR-PAM imaging of a bamboo leaf. (a) and (b) MAP images of the direct signals and the superimposed signals; (c) A photo of the imaging area. (d) Histogram of the pixel intensity of the images.

Download Full Size | PDF

We conducted statistical analysis by calculating the intensity distribution of the image pixels, as shown in Fig. 5(d). Quantitatively, the average intensities of Fig. 5(a) and 5(b) are 0.29 and 0.22, which demonstrate that the sensitivity of the transducer with reflector is improved by 31.8%. And the standard deviations of the sample intensity are approximately 0.17 and 0.13, respectively, which shows a 30.8% improvement of the standard deviation. Thus, these results demonstrate that the designed transducer can provide images with a higher sensitivity and contrast, the structure of the bamboo leaves can be seen more clearly.

Then, we apply the transducer to an AR-PAM to image a leaf skeleton. In AR-PAM, the laser is no longer strictly focused, and the size of the optical focus is larger than that of the acoustic focus, as shown in Fig. 3(b).

The tip of the leaf skeleton was cut and evenly colored with black ink, laid flat on the bottom of the water tank, as shown in Fig. 6(c). To ensure high lateral resolution, the sample was placed in the focal plane of the transducer, and the laser beam was strictly collimated with the acoustic axis of the transducer. A total area of 15mm × 15 mm was scanned with a step size of 40µm. Figure 6(a) and 6(b) are the MAP images of the direct signals and the superimposed signals, respectively. As can be seen, the intensities of the images have been improved after superimposing the reflected signal. In order to evaluate the effect of the back reflector on the image intensity and resolution, we compared the thickness of the leaf veins in two images. The profiles along the white dashed line in Fig. 6(a) and 6(b) were read out, as shown in Fig. 6(d). By Gaussian fitting, the peak amplitude of the profile is approximately 0.677 in (a), which is 1.26 times higher than that of 0.536 in (b). It shows that the image intensity is improved. And the FWHM are calculated to be 150µm in (a) and 144µm in (b), which means that the lateral resolution is barely affected. This experiment also confirms that the designed transducer with back-reflector can increase the image intensity and contrast in PAM. Moreover, benefiting from the flexible nature of PVDF, the transducer has good acoustic focusing and provides high lateral resolution for an AR-PAM.

 

Fig. 6. AR-PAM imaging of the tip of a leaf skeleton. (a) and (b) MAP images extracted from the direct signals and the superimposed signals; (c) A photo of the imaging area; (d) Profile along the white dashed lines in (a) and (b).

Download Full Size | PDF

5. Conclusions

In summary, we design and manufacture a bionic PVDF transducer suitable for multiscale-resolution imaging in PAM system. This transducer involves a back reflector, which models after the tapetum lucidum in the eyes of nocturnal animals. When ultrasound reaches the transducer, a part of ultrasound will be converted into electric signals. The left part of ultrasound penetrates the PVDF layer, which is usually attenuated and wasted in common designs. In our design, the penetration signal will be reflected by the back-reflector and reach the PVDF again, and be converted into an additional electric signal eventually.

The bionic structure provides a second chance for the PVDF to convert the ultrasound to electric signals. Then, the two signals can be superposed by using the matched filtering. This design increases the amount of ultrasound absorbed by the PVDF, thereby improving the detection sensitivity. The experimental measurements show that the fabricated transducer has a center frequency of 13.07 MHz, and a bandwidth of 96% at −6 dB. More importantly, the bionic design improves the detection sensitivity of the transducer about 30%. Finally, we apply the fabricated transducer to OR-PAM and AR-PAM, respectively. Then, we image the structure of the bamboo leaf by using the OR-PAM and the tip of a leaf vein by using the AR-PAM. The imaging results demonstrate that the transducer can provide better imaging intensity and contrast. Moreover, benefitting from the flexible nature of the PVDF, the transducer has a large acoustic numerical aperture (NA) of 0.64 and good focusing performance, which provides a lateral resolution of 140µm for the AR-PAM. Therefore, the proposed transducer design will be useful to enhance the performance of multiscale-resolution PAM.