Ultrasonics xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Ultrasonics journal homepage: www.elsevier.com/locate/ultras
Acoustic impedance microscopy for biological tissue characterization Kazuto Kobayashi a,⇑, Sachiko Yoshida b, Yoshifumi Saijo c, Naohiro Hozumi d a
Honda Electronics Co., Ltd., 20 Koyamazuka, Oiwa-cho, Toyohashi 441-3193, Japan Toyohashi University of Technology, 1-1 Tempaku, Toyohashi 441-8580, Japan c Tohoku University, 6-6 Aoba, Aramaki, Aoba-ku, Sendai 980-8579, Japan d Aichi Institute of Technology, 1247 Yachigusa, Yakusa, Toyota 470-0392, Japan b
a r t i c l e
i n f o
Article history: Received 4 April 2011 Received in revised form 2 April 2014 Accepted 3 April 2014 Available online xxxx Keywords: Tissue characterization Acoustic impedance Acoustic microscope
a b s t r a c t A new method for two-dimensional acoustic impedance imaging for biological tissue characterization with micro-scale resolution was proposed. A biological tissue was placed on a plastic substrate with a thickness of 0.5 mm. A focused acoustic pulse with a wide frequency band was irradiated from the ‘‘rear side’’ of the substrate. In order to generate the acoustic wave, an electric pulse with two nanoseconds in width was applied to a PVDF-TrFE type transducer. The component of echo intensity at an appropriate frequency was extracted from the signal received at the same transducer, by performing a time–frequency domain analysis. The spectrum intensity was interpreted into local acoustic impedance of the target tissue. The acoustic impedance of the substrate was carefully assessed prior to the measurement, since it strongly affects the echo intensity. In addition, a calibration was performed using a reference material of which acoustic impedance was known. The reference material was attached on the same substrate at different position in the field of view. An acoustic impedance microscopy with 200 200 pixels, its typical field of view being 2 2 mm, was obtained by scanning the transducer. The development of parallel fiber in cerebella cultures was clearly observed as the contrast in acoustic impedance, without staining the specimen. The technique is believed to be a powerful tool for biological tissue characterization, as no staining nor slicing is required. Ó 2014 Elsevier B.V. All rights reserved.
1. Introduction In most of optical observation of biological tissue, the specimen is sliced into several micrometers in thickness, and fixed on a glass substrate. The microscopy is obtained by transmitted light through the specimen. As it is normally not easy to get a good contrast by local difference in refraction and/or transmission spectrum, the specimen is usually stained before being observed. It can be classified as a kind of chemical imaging, since only a portion that has a specific chemical property can be stained by selecting an appropriate staining material. However the staining has some disadvantages. It normally takes from several hours to several days to finish the process. Furthermore, the tissue, after being stained, often completely loses its biological functions; i.e., the observation with staining process is chemically destructive. On the other hand, acoustic imaging can be performed without staining process; i.e., it is chemically non-destructive. The observation can be finished in a very short time, as it does not need the staining process. The idea of ultrasonic microscopy for biological ⇑ Corresponding author. Tel.: +81 532 41 2511. E-mail address: [email protected] (K. Kobayashi).
tissue is based on this advantage, and it is considered to become a powerful tool for tissue characterization that can image elastic parameters. Most of ultrasonic microscopes are scanning type, in which the response to a focused acoustic signal is successively acquired as the beam is mechanically scanned [1,2]. The authors previously proposed a pulse driven ultrasonic sound speed microscopy that can obtain sound speed image in a short time [3,4]. Although a small roughness of the specimen was approved in this type of microscope, slicing the specimen into several micrometers was still required for the observation. However it is often required that the observation can be performed without slicing process, as slicing may damage some functions of the tissue. Based on the above background, the authors newly propose the acoustic impedance microscopy that can image the local distribution of cross sectional acoustic impedance of tissue. As acoustic impedance is given as a product of sound speed and density, it would have a good correlation with sound speed, when the variance in density was not significant. In this paper, the methodology for micro-scale imaging of cross sectional acoustic impedance, and its application to the observation of cerebellar tissue of a rat will be described.
http://dx.doi.org/10.1016/j.ultras.2014.04.007 0041-624X/Ó 2014 Elsevier B.V. All rights reserved.
Please cite this article in press as: K. Kobayashi et al., Acoustic impedance microscopy for biological tissue characterization, Ultrasonics (2014), http:// dx.doi.org/10.1016/j.ultras.2014.04.007
2
K. Kobayashi et al. / Ultrasonics xxx (2014) xxx–xxx
2. Preparation of specimens The cerebellum tissue of a rat was employed as the specimen to be observed. Rats were dissected and removed their whole brains. Some of the isolated cerebellum were sliced at 200-micrometers thick using a rotor slicer (Dohan EM, Kyoto, Japan). The slices were incubated in oxygenated phosphate buffer solution (PBS) on ice for one hour. They were chemically fixed with 4% formaldehyde fixative, for 20 min. For optical observation, some slices were subjected to immunohistochemical staining against calbindin D-28k. The other specimens, the intact ones, were cut at an appropriate cross section. Both intact and fixed slices were rinsed and observed in same PBS. All experimental procedures were approved by the committees for the use of animals in Toyohashi University of Technology, and all animal care followed the Standards Relation to the Care and Management of Experimental Animals (Notification No. 6, March 27, 1980 of the Prime Minister’s Office in Japan). The cross section of each specimen was in contact with the substrate. The substrate was a flat plastic plate made of polymethylmetacrylate (PMMA), its thickness being 0.5 mm. A reference material, of which acoustic impedance was known, was also placed on the same substrate. In many cases, the target tissue was observed together with the reference, by including it in the same field of view. In some cases, the surface of the substrate was treated beforehand with an atmospheric plasma for three seconds by a plasma surface treatment equipment (Keyence ST-7000 Plasma Surface Treater), in order to upgrade its hydrophilic property. In this report, a silicone rubber, distilled water or agar was employed as a reference material, choosing one of them depending on the convenience of the measurement. In case of using silicone rubber, the observation was performed after having waited for more than 24 h since the rubber had been hardened, in order to retain the stability of the material.
3. Experimental setup Fig. 1 illustrates the outline of the acoustic impedance microscope. Distilled water was used for the coupling medium between the substrate and transducer. A sharp electric pulse of about 40 V in peak voltage and 2 ns in width was generated by the pulse generator (AVTEC, AVP-AV-HV3-C). The maximum repetition rate of the pulse was as high as 10 kHz. The transducer was PVDF-TrFE type. It was 1.5 mm in aperture diameter, and 3.0 mm in focal length. An acoustic wave with a wide frequency component was generated by applying the voltage pulse. The acoustic wave, being focused on the interface between the substrate and tissue, was transmitted and received by the same transducer.
Reference
The reflection was detected and digitized by the oscilloscope (Tektronix, TDS-7145B). The waveform was acquired with 2.5 G samplings/s in sampling rate, up to 1 GHz in frequency range and 8 bits in resolution. The input was terminated with 50 X. Considering the focal distance and the sectional area of the transducer, the diameter of the focal spot was estimated as about 26 lm at 80 MHz. The distance between the nearest two points was typically set at 10 lm, in order to retain a sufficient lateral resolution. Two-dimensional profile of acoustic impedance was obtained by mechanically scanning the transducer using the stage driver (Sigma Koki, MARK-202), keeping the focal point on the rear surface of the substrate. A typical field of view of 2 mm 2 mm was covered with 200 200 pixels. A part of the system is overviewed in Fig. 2. The transducer embedded on the coupling circuit box was mechanically scanned by the X–Y stage. The substrate was placed on the other stage, so that the acoustic wave can be transmitted from its bottom side. It took typically 2–3 min for one observation. In order to save the time for data transfer from the oscilloscope to the computer, the waveforms through each X-scan were once stored in the oscilloscope using its fast-frame mode before being transferred through the LAN interface. In order to reduce random noise, three times of responses at the same point were averaged.
4. Results 4.1. Waveforms Fig. 3 shows the reflected acoustic signals. In this particular case, a water droplet was used as the reference. A part of cerebellum tissue was used as a target. The signal from the target tissue is very similar to that from the reference, suggesting the acoustic impedance of the tissue is close to that of water (1.5 106 Ns/ m3). Fig. 4 shows the intensity spectrum of the target signal normalized by the reference signal, and cross power spectrum of the target and reference signals. The intensity spectrum is almost flat from 15 to 100 MHz, the intensity being a little smaller than 1.0. It indicates that the impedance of the target is somehow different from the reference. The calibration method for the acoustic impedance will be described in the following section. The cross-power spectrum of the target and reference signals shows that a wideband acoustic signal had been successfully generated.
Target
Substrate 2 ns Water
Delay Line 17 m
Transduce
500 pF
Pulse generator Oscilloscope LAN / GPIB Trigger
Stage Driver
Fig. 1. Experimental setup.
100 Ω
r
Mechanical Scan Fig. 2. Overview of a part of the experiment system.
Please cite this article in press as: K. Kobayashi et al., Acoustic impedance microscopy for biological tissue characterization, Ultrasonics (2014), http:// dx.doi.org/10.1016/j.ultras.2014.04.007
3
K. Kobayashi et al. / Ultrasonics xxx (2014) xxx–xxx
Starget ¼
Z target Z sub S0 ; Z target þ Z sub
ð1Þ
where S0 is the transmitted signal, Ztarget and Zsub are the acoustic impedances of the target and substrate, respectively. On the other hand, the reference signal can be described as
Sref ¼
Z ref Z sub S0 ; Z ref þ Z sub
ð2Þ
where Zref is the acoustic impedance of the reference material. We can measure Starget and Sref, however, S0 cannot be directly measured. The acoustic impedance of the target is subsequently calculated as a solution of the simultaneous equations for Ztarget and S0, as
Z target ¼ Fig. 3. Waveforms of reflected acoustic signals.
1 1þ
Starget S0 Starget S0
Z sub ¼
Z
Z
1
Starget Sref
Zsub þZref
1þ
Starget Sref
Zsub þZref
sub
Z
ref
Z
sub
Z sub ;
ð3Þ
ref
assuming that S0 is constant throughout the observation process. In case of using water as the reference, its acoustic impedance was assumed to be 1.5 106 Ns/m3. On the other hand, in case of using silicon rubber, the acoustic impedance of itself was calibrated, by using water as the standard reference material. In this report, 0.985 106 Ns/m3 was used. The acoustic impedance of
Acoustic Impedance (106 Ns/m3)
1.62 1.60 1.58 1.56 1.54 1.52 1.50 1.48
0
1
2
3
4
5
NaCl Content (wt%) Fig. 4. Analysis for signals in the frequency domain. The normalized intensity represents the intensity of the target signal by the reference signal. The cross-power is calculated from the target and reference signals.
Fig. 6. Acoustic impedance of salt water as a function of NaCl constant.
Immature
Z sub
So Reference Z ref
Sref So
Object
Maturation
Target ML Substrate (PMMA )
Mature Ztarget
PL
Starget
Fig. 5. Illustration for calibration of the acoustic impedance.
IGL
4.2. Calibration
WM Coronal
Fig. 5 illustrates the calibration of acoustic impedance. The target signal is compared with the reference signal. Hereafter, the signal component at an arbitrary frequency will be symbolized by S. Considering the reflection coefficient, the target signal Starget can be described as
Sagittal Fig. 7. Illustration for the development of cerabellar cortex.
Please cite this article in press as: K. Kobayashi et al., Acoustic impedance microscopy for biological tissue characterization, Ultrasonics (2014), http:// dx.doi.org/10.1016/j.ultras.2014.04.007
4
K. Kobayashi et al. / Ultrasonics xxx (2014) xxx–xxx
agar was calibrated in the same manner short time before the observation. It was calculated to be 1.65 106 Ns/m3. As the sound speed of the substrate at about 80 MHz at 25 °C and its density at 25 °C were 2.78 km/s and 1.16 mg/mm3, respectively, the acoustic impedance of the substrate was calculated to be 3.22 106 Ns/m3. The accuracy was ensured by observing saline droplets with different salinities. The acoustic impedance measured by this system was compared with that estimated as the product of sound speed and density. As shown in Fig. 6, these values agree well. 4.3. Observation of cerebellar cortex of a rat Fig. 7 illustrates the development of cerebellar cortex [5,6]. Parallel fibers in molecular layer are axons of granule cells and play an important role in cerebella neuronal connections. Migrating granule cells elongate them horizontally and form a lot of excitatory synapses to dendrites of Purkinje cells. These are major neuronal circuits of cerebellum so that parallel fibers are expected to construct rich molecular layer with development. But it was hard to
evaluate a degree of parallel fiber development with over molecular layer. We have little sufficient histochemical tools to visualize the developing parallel fibers. Although electron microscopy shows fine structure, it is limited to a very local image. As the field of view of the proposed acoustic technique is as wide as that of optical microscope, it is expected to be a suitable replacement that can observe without any histochemical tools. The granule cells migrate away from the ventricular zone, over the top of the developing Purkinje cells to form a secondary zone of neurogenesis, called the external granule layer. After the birth the cells in this layer continue to actively proliferate, generating an enormous number of granule cell progeny at postnatal 7–10 days. Soon after their generation, after their final mitotic division, the granule cells change from a very round cell to take on a more horizontally oriented shape as they begin to extend axons tangential to the cortical surface, called parallel fibers. Next, the cell body of granule cell migrate to deep into the cerebellum, so the cell assumes a T shape. The cell body eventually migrates past the Purkinje cell layer and then begins to sprout dendrites in the granule cell layer until postnatal 20 day.
1 mm
(a) P1
200 μm
Inner layer
Inner layer EGL
EGL
PL
(b) P7 WM IGL
EGL
PL
PL
EGL
(c) P20
ML PL IGL ML PL WM
IGL
Fig. 8. Two-dimensional profiles of acoustic impedance (106 Ns/m3) of cerabellar cortex (left) and optical microcopies (right). Specimen: rat, sagittal cross section, chemically fixed. Frequency range: 60–100 MHz. Substrate: not treated with plasma.
Please cite this article in press as: K. Kobayashi et al., Acoustic impedance microscopy for biological tissue characterization, Ultrasonics (2014), http:// dx.doi.org/10.1016/j.ultras.2014.04.007
5
K. Kobayashi et al. / Ultrasonics xxx (2014) xxx–xxx
Purkinje cells, on the other hand, have arranged on the Purkinje cell layer at birth, whereas their dendrites are short and immature. They form synapses to parallel fibers of the granule cells and construct cerebellar neuronal network. The network layers containing parallel fibers and Purkinje cell dendrites are called the molecular layers. Parallel fibers are thin and unmyelinated neuronal fibers. Because parallel fibers run transverse between left and right hemisphere, sagittal sections of cerebellar cortex show the cross view of the parallel fibers and coronal sections show the side view. Fig. 8 shows the observed images of cerebellar cortex of a rat at immature (P1; postnatal 1 day), transient (P7), and mature (P20) stages. All the specimens in Fig. 7 had been chemically fixed. In the immature cerebellar cortex (P1), the external granular layer (EGL), the outer layer of the cortex, showed higher impedance compared to the inner layer. The area indicated by the rectangle in the acoustic image is morphologically corresponding to the immunohistochemical observation, although the scale is not completely corresponded because the tissue was somehow subjected to compression during the acoustic observation. At this stage, as myelin is not yet generated, the existence of white matter (WM) is not clearly observed. In the transient stage, four different layers; the WM, internal granular layer (IGL), Purkinje layer (PL) and EGL become to be comprehensive. The EGL and IGL showed higher impedance than the PL and WM. Morphological correspondence between acoustic and immunohistological observation is however not clear in these images. In the mature stage, the EGL, which is composed of small neuronal cell bodies, has developed into the molecular layer (ML), which is composed of elongated axon (neurite), called parallel fibers. The four layers, WM, IGL, PL and ML are more clearly observed in acoustic image. The correspondence with immunohistological observation is also clearly seen. Fig. 9 shows the observed images without chemical fixation. A cerebellum of mature rat (P36) was employed as a specimen. The sagittal cross section was placed on the substrate. The same structure as Fig. 7(c) is clearly seen. The enlarged image was taken by scanning the transducer with 2 lm pitch. Although the image is not very clear, it suggests that the observation of each Purkinje cell, its scale being as small as 20 lm, will be quite feasible if the focal diameter is reduced by using a little higher frequency range.
5. Discussion Considering the precision of the calibration, the reference material should be stable in both physical and chemical properties, and should strongly adhere to the substrate. It is recommended that the acoustic impedance of the reference is close to that of the target. Furthermore, as for the substrate, most of available materials have higher acoustic impedance than biological tissues. In such cases, the phase of the transmitted signal is reversed at the interface. The acoustic impedance of the substrate should be sufficiently high compared to that of the target, in order to retain a strong reflection. However extremely high acoustic impedance of the substrate may increase the reflection coefficient at the interface between the coupling medium and substrate, and reduce the intensity of transmitted signal to the target. This would obviously reduce the S/N ratio. Therefore, in order to obtain a good S/N ratio, the materials should be carefully selected considering their accordance. As the transducer was designed for usage with water as the coupling medium, the existence of the plastic plate between the transducer and focal point may bring an aberration. This will be significant if the thickness of the substrate is very thick, and the convergence angle is very large. The good agreement would derive
5 mm
Tissue on substrate 2 mm
ML PL IGL
200 μm
ML
PL
IGL
Fig. 9. Two-dimensional profiles of acoustic impedance (106 Ns/m3) of cerabellar cortex. Specimen: rat, postnatal 36 days, sagittal cross section, not chemically fixed. Frequency range: 60–100 MHz. Substrate: treated with plasma.
from the fact that the angle of focusing of the transducer was as small as 14° . In case that the angle of focusing is larger, we may have to consider the existence of aberration due to oblique incidence. In such a case, analysis similar to Fourier optics by decomposing the beam into plane waves will be required. Nevertheless, a quantitative analysis is needed in order to precisely assess the acoustic impedance, especially when a thick substrate is employed. As the reflected signal includes wide range of frequency component, the image can be reconstructed using different frequency ranges. Fig. 10 shows the image reconstructed from the same dataset as Fig. 8(c). These corresponding images are reconstructed by extracting the frequency range of 15–100 MHz, and 60–100 MHz, respectively. It can be comprehended that limiting the low frequency range results in the clearer image. It is therefore recommended to extract higher frequency range from the signal, however, it should be also noted that extreme limitation of low frequency component would lead to the reduction in S/N ratio. It occasionally took place that the image looked differently depending on the compatibility between the tissue and plastic substrate. Fig. 11 shows an example that the inner structure was not clearly observed. This kind of image was often obtained with mature rats. It is considered that the cerebellar parallel fibers in a mature rat is more developed than an immature, leading to a
Please cite this article in press as: K. Kobayashi et al., Acoustic impedance microscopy for biological tissue characterization, Ultrasonics (2014), http:// dx.doi.org/10.1016/j.ultras.2014.04.007
6
K. Kobayashi et al. / Ultrasonics xxx (2014) xxx–xxx
1.5
1 mm
1.0 0.5 0.0 0
ML
20
40 60 Frequency (MHz)
80
100
PL IGL
WM Fig. 10. Two-dimensional profiles of acoustic impedance (106 Ns/m3) of cerabellar cortex. Specimen: rat, postnatal 36 days, sagittal cross section, not chemically fixed. Frequency range: 15–100 MHz. Substrate: not treated with plasma.
2 mm
Fig. 12. Analysis for signals in the frequency domain. The normalized intensity represents the intensity of the target signal with water layer between the substrate by the reference signal.
density and expectedly low sound speed. Further pharmacological investigation is required. Until now, we are convinced the contact between the tissue and slide glass is well retained. However, there is two kinds of possibility of delamination. The first is that the air layer is formed in between the substrate and tissue. In such case, perfect optical reflection can be observed with a relatively small incident angle. We noticed no reflection formed when the contact is good. The second case is that a thin water layer is formed between the substrate and tissue. It is difficult to check with an optical microscope, because the optical refraction index of the tissue is close to water. We noticed multiple acoustic reflections in the water layer take place in such case, as the acoustic impedance of the tissue is a bit larger than water. We can find bad contact by observing distortion of the reflected waveform. As shown in Fig. 12 the similarity of waveforms from the tissue and reference would not be retained. As shown in Fig. 4, the normalized spectrum is very flat throughout wide frequency range suggests that the waveform from the target is similar to that from the reference. We have proved the contact is well retained for the experiment with above proves. 6. Summary
Fig. 11. Two-dimensional profiles of acoustic impedance (106 Ns/m3) of cerabellar cortex. Specimen: rat, postnatal 36 days, sagittal cross section, not chemically fixed. Frequency range: 60–100 MHz. Substrate: not treated with plasma.
bad compatibility between the tissue and substrate, probably because of difference in degree of cytoskeltal development [7]. This may be overcome by upgrading the hydrophilic property of the substrate. In this investigation, as an attempt, the surface of the substrate was treated with an atmospheric plasma in air. In fact Fig. 8, in which the internal structure is clearly seen, was observed using the same kind of specimen as that shown in Fig. 11. The image of Fig. 8 was obtained with the surface treatment, whereas that of Fig. 10 was without the treatment. The contact angle of water was improved from 55° to 39° after the treatment. Polar groups induced on the surface of the substrate would have brought a good compatibility with the tissue. The compatibility may also be upgraded by employing different kind of plastic materials, as well as performing some other surface treatments such as silane coupling treatment. As the WM is rich with fat, its acoustic impedance would be lower than the IGL. The ML is composed of axon, which have a lot of actin fibers with high elasticity [7]. It would lead to a high acoustic impedance. The reason why the Purkinje layer has low impedance can be considered as Purkinje cell has relatively small nuclei and mostly filled with cytoplasm that has relatively low
(1) A new method for two-dimensional acoustic impedance imaging for biological tissue characterization with microscale resolution was proposed. (2) Calibration was performed using a reference material of which acoustic impedance was known. Quantitative imaging of acoustic impedance was made possible. (3) Acoustic impedance microscopy with 200 200 pixels, its typical field of view being 2 2 mm, was obtained by scanning the transducer. (4) The development of cerebella cultures of a rat was clearly observed as the contrast in acoustic impedance, without staining the specimen. (5) The technique is believed to be a powerful tool for biological tissue characterization, as no staining nor slicing is required.
Acknowledgements The authors would like to express their sincere thanks to C.-K. Lee, T. Morishima and E. Fukushi of Toyohashi University of Technology for their assistance with the experiment. This study was financially supported by Grants-in-Aid for Scientific Research (Scientific Research (B)15360217, (B) 15300178), and Japan Society for the Promotion of Science and Health and Labor Sciences Research Grants from the Ministry of Health, Labor and Welfare for the Research on Advanced Medical Technology (H17-Nano-001).
Please cite this article in press as: K. Kobayashi et al., Acoustic impedance microscopy for biological tissue characterization, Ultrasonics (2014), http:// dx.doi.org/10.1016/j.ultras.2014.04.007
K. Kobayashi et al. / Ultrasonics xxx (2014) xxx–xxx
References [1] Y. Saijo, M. Tanaka, H. Okawai, H. Sasaki, S. Nitta, F. Dunn, Ultrasonic tissue characterization of infracted myocadium by scanning acoustic microscopy, Ultrasound Med. Biol. 23–1 (1997) 77. [2] H. Okawai, K. Kobayashi, S. Nitta, An approach to acoustic properties of biological tissues using acoustic micrographs of attenuation constant and sound speed, J. Ultrasound Med. 20 (2001) 891. [3] N. Hozumi, R. Yamashita, C. -K Lee, M. Nagao, K. Kobayashi, Y. Saijo, M. Tanaka, N. Tanaka, S Ohtsuki, Ultrasonic sound speed microscope for biological tissue characterization driven by nanosecond pulse, Acoustic Sci. Technol. 20 (2003) 386.
7
[4] N. Hozumi, R. Yamashita, C-K. Lee, M. Nagao, K. Kobayashi, Y. Saijo, M. Tanaka, N. Tanaka, S. Ohtsuki, Time – frequency analysis for pulse driven ultrasonic microscopy for biological tissue characterization, Ultrasonics 42 (2003) 717. [5] D.H. Sanes, T.A. Reh, W.A. Harris, Development of the Nervous System, Academic Press, 2000. [6] E.R. Kandel, J.H. Schwarts, T.M. Jessell, Principles of Neural Science, McGrawHill, 2000. [7] H.W. Wu, T. Kuhn, V.T. Moy, Mechanical properties of L929 cells measured by acoustic force microscopy: effects of anticytoskeletal drugs and membrane crosslinking, Scanning 20 (1998) 389.
Please cite this article in press as: K. Kobayashi et al., Acoustic impedance microscopy for biological tissue characterization, Ultrasonics (2014), http:// dx.doi.org/10.1016/j.ultras.2014.04.007
Contents lists available at ScienceDirect
Ultrasonics journal homepage: www.elsevier.com/locate/ultras
Acoustic impedance microscopy for biological tissue characterization Kazuto Kobayashi a,⇑, Sachiko Yoshida b, Yoshifumi Saijo c, Naohiro Hozumi d a
Honda Electronics Co., Ltd., 20 Koyamazuka, Oiwa-cho, Toyohashi 441-3193, Japan Toyohashi University of Technology, 1-1 Tempaku, Toyohashi 441-8580, Japan c Tohoku University, 6-6 Aoba, Aramaki, Aoba-ku, Sendai 980-8579, Japan d Aichi Institute of Technology, 1247 Yachigusa, Yakusa, Toyota 470-0392, Japan b
a r t i c l e
i n f o
Article history: Received 4 April 2011 Received in revised form 2 April 2014 Accepted 3 April 2014 Available online xxxx Keywords: Tissue characterization Acoustic impedance Acoustic microscope
a b s t r a c t A new method for two-dimensional acoustic impedance imaging for biological tissue characterization with micro-scale resolution was proposed. A biological tissue was placed on a plastic substrate with a thickness of 0.5 mm. A focused acoustic pulse with a wide frequency band was irradiated from the ‘‘rear side’’ of the substrate. In order to generate the acoustic wave, an electric pulse with two nanoseconds in width was applied to a PVDF-TrFE type transducer. The component of echo intensity at an appropriate frequency was extracted from the signal received at the same transducer, by performing a time–frequency domain analysis. The spectrum intensity was interpreted into local acoustic impedance of the target tissue. The acoustic impedance of the substrate was carefully assessed prior to the measurement, since it strongly affects the echo intensity. In addition, a calibration was performed using a reference material of which acoustic impedance was known. The reference material was attached on the same substrate at different position in the field of view. An acoustic impedance microscopy with 200 200 pixels, its typical field of view being 2 2 mm, was obtained by scanning the transducer. The development of parallel fiber in cerebella cultures was clearly observed as the contrast in acoustic impedance, without staining the specimen. The technique is believed to be a powerful tool for biological tissue characterization, as no staining nor slicing is required. Ó 2014 Elsevier B.V. All rights reserved.
1. Introduction In most of optical observation of biological tissue, the specimen is sliced into several micrometers in thickness, and fixed on a glass substrate. The microscopy is obtained by transmitted light through the specimen. As it is normally not easy to get a good contrast by local difference in refraction and/or transmission spectrum, the specimen is usually stained before being observed. It can be classified as a kind of chemical imaging, since only a portion that has a specific chemical property can be stained by selecting an appropriate staining material. However the staining has some disadvantages. It normally takes from several hours to several days to finish the process. Furthermore, the tissue, after being stained, often completely loses its biological functions; i.e., the observation with staining process is chemically destructive. On the other hand, acoustic imaging can be performed without staining process; i.e., it is chemically non-destructive. The observation can be finished in a very short time, as it does not need the staining process. The idea of ultrasonic microscopy for biological ⇑ Corresponding author. Tel.: +81 532 41 2511. E-mail address: [email protected] (K. Kobayashi).
tissue is based on this advantage, and it is considered to become a powerful tool for tissue characterization that can image elastic parameters. Most of ultrasonic microscopes are scanning type, in which the response to a focused acoustic signal is successively acquired as the beam is mechanically scanned [1,2]. The authors previously proposed a pulse driven ultrasonic sound speed microscopy that can obtain sound speed image in a short time [3,4]. Although a small roughness of the specimen was approved in this type of microscope, slicing the specimen into several micrometers was still required for the observation. However it is often required that the observation can be performed without slicing process, as slicing may damage some functions of the tissue. Based on the above background, the authors newly propose the acoustic impedance microscopy that can image the local distribution of cross sectional acoustic impedance of tissue. As acoustic impedance is given as a product of sound speed and density, it would have a good correlation with sound speed, when the variance in density was not significant. In this paper, the methodology for micro-scale imaging of cross sectional acoustic impedance, and its application to the observation of cerebellar tissue of a rat will be described.
http://dx.doi.org/10.1016/j.ultras.2014.04.007 0041-624X/Ó 2014 Elsevier B.V. All rights reserved.
Please cite this article in press as: K. Kobayashi et al., Acoustic impedance microscopy for biological tissue characterization, Ultrasonics (2014), http:// dx.doi.org/10.1016/j.ultras.2014.04.007
2
K. Kobayashi et al. / Ultrasonics xxx (2014) xxx–xxx
2. Preparation of specimens The cerebellum tissue of a rat was employed as the specimen to be observed. Rats were dissected and removed their whole brains. Some of the isolated cerebellum were sliced at 200-micrometers thick using a rotor slicer (Dohan EM, Kyoto, Japan). The slices were incubated in oxygenated phosphate buffer solution (PBS) on ice for one hour. They were chemically fixed with 4% formaldehyde fixative, for 20 min. For optical observation, some slices were subjected to immunohistochemical staining against calbindin D-28k. The other specimens, the intact ones, were cut at an appropriate cross section. Both intact and fixed slices were rinsed and observed in same PBS. All experimental procedures were approved by the committees for the use of animals in Toyohashi University of Technology, and all animal care followed the Standards Relation to the Care and Management of Experimental Animals (Notification No. 6, March 27, 1980 of the Prime Minister’s Office in Japan). The cross section of each specimen was in contact with the substrate. The substrate was a flat plastic plate made of polymethylmetacrylate (PMMA), its thickness being 0.5 mm. A reference material, of which acoustic impedance was known, was also placed on the same substrate. In many cases, the target tissue was observed together with the reference, by including it in the same field of view. In some cases, the surface of the substrate was treated beforehand with an atmospheric plasma for three seconds by a plasma surface treatment equipment (Keyence ST-7000 Plasma Surface Treater), in order to upgrade its hydrophilic property. In this report, a silicone rubber, distilled water or agar was employed as a reference material, choosing one of them depending on the convenience of the measurement. In case of using silicone rubber, the observation was performed after having waited for more than 24 h since the rubber had been hardened, in order to retain the stability of the material.
3. Experimental setup Fig. 1 illustrates the outline of the acoustic impedance microscope. Distilled water was used for the coupling medium between the substrate and transducer. A sharp electric pulse of about 40 V in peak voltage and 2 ns in width was generated by the pulse generator (AVTEC, AVP-AV-HV3-C). The maximum repetition rate of the pulse was as high as 10 kHz. The transducer was PVDF-TrFE type. It was 1.5 mm in aperture diameter, and 3.0 mm in focal length. An acoustic wave with a wide frequency component was generated by applying the voltage pulse. The acoustic wave, being focused on the interface between the substrate and tissue, was transmitted and received by the same transducer.
Reference
The reflection was detected and digitized by the oscilloscope (Tektronix, TDS-7145B). The waveform was acquired with 2.5 G samplings/s in sampling rate, up to 1 GHz in frequency range and 8 bits in resolution. The input was terminated with 50 X. Considering the focal distance and the sectional area of the transducer, the diameter of the focal spot was estimated as about 26 lm at 80 MHz. The distance between the nearest two points was typically set at 10 lm, in order to retain a sufficient lateral resolution. Two-dimensional profile of acoustic impedance was obtained by mechanically scanning the transducer using the stage driver (Sigma Koki, MARK-202), keeping the focal point on the rear surface of the substrate. A typical field of view of 2 mm 2 mm was covered with 200 200 pixels. A part of the system is overviewed in Fig. 2. The transducer embedded on the coupling circuit box was mechanically scanned by the X–Y stage. The substrate was placed on the other stage, so that the acoustic wave can be transmitted from its bottom side. It took typically 2–3 min for one observation. In order to save the time for data transfer from the oscilloscope to the computer, the waveforms through each X-scan were once stored in the oscilloscope using its fast-frame mode before being transferred through the LAN interface. In order to reduce random noise, three times of responses at the same point were averaged.
4. Results 4.1. Waveforms Fig. 3 shows the reflected acoustic signals. In this particular case, a water droplet was used as the reference. A part of cerebellum tissue was used as a target. The signal from the target tissue is very similar to that from the reference, suggesting the acoustic impedance of the tissue is close to that of water (1.5 106 Ns/ m3). Fig. 4 shows the intensity spectrum of the target signal normalized by the reference signal, and cross power spectrum of the target and reference signals. The intensity spectrum is almost flat from 15 to 100 MHz, the intensity being a little smaller than 1.0. It indicates that the impedance of the target is somehow different from the reference. The calibration method for the acoustic impedance will be described in the following section. The cross-power spectrum of the target and reference signals shows that a wideband acoustic signal had been successfully generated.
Target
Substrate 2 ns Water
Delay Line 17 m
Transduce
500 pF
Pulse generator Oscilloscope LAN / GPIB Trigger
Stage Driver
Fig. 1. Experimental setup.
100 Ω
r
Mechanical Scan Fig. 2. Overview of a part of the experiment system.
Please cite this article in press as: K. Kobayashi et al., Acoustic impedance microscopy for biological tissue characterization, Ultrasonics (2014), http:// dx.doi.org/10.1016/j.ultras.2014.04.007
3
K. Kobayashi et al. / Ultrasonics xxx (2014) xxx–xxx
Starget ¼
Z target Z sub S0 ; Z target þ Z sub
ð1Þ
where S0 is the transmitted signal, Ztarget and Zsub are the acoustic impedances of the target and substrate, respectively. On the other hand, the reference signal can be described as
Sref ¼
Z ref Z sub S0 ; Z ref þ Z sub
ð2Þ
where Zref is the acoustic impedance of the reference material. We can measure Starget and Sref, however, S0 cannot be directly measured. The acoustic impedance of the target is subsequently calculated as a solution of the simultaneous equations for Ztarget and S0, as
Z target ¼ Fig. 3. Waveforms of reflected acoustic signals.
1 1þ
Starget S0 Starget S0
Z sub ¼
Z
Z
1
Starget Sref
Zsub þZref
1þ
Starget Sref
Zsub þZref
sub
Z
ref
Z
sub
Z sub ;
ð3Þ
ref
assuming that S0 is constant throughout the observation process. In case of using water as the reference, its acoustic impedance was assumed to be 1.5 106 Ns/m3. On the other hand, in case of using silicon rubber, the acoustic impedance of itself was calibrated, by using water as the standard reference material. In this report, 0.985 106 Ns/m3 was used. The acoustic impedance of
Acoustic Impedance (106 Ns/m3)
1.62 1.60 1.58 1.56 1.54 1.52 1.50 1.48
0
1
2
3
4
5
NaCl Content (wt%) Fig. 4. Analysis for signals in the frequency domain. The normalized intensity represents the intensity of the target signal by the reference signal. The cross-power is calculated from the target and reference signals.
Fig. 6. Acoustic impedance of salt water as a function of NaCl constant.
Immature
Z sub
So Reference Z ref
Sref So
Object
Maturation
Target ML Substrate (PMMA )
Mature Ztarget
PL
Starget
Fig. 5. Illustration for calibration of the acoustic impedance.
IGL
4.2. Calibration
WM Coronal
Fig. 5 illustrates the calibration of acoustic impedance. The target signal is compared with the reference signal. Hereafter, the signal component at an arbitrary frequency will be symbolized by S. Considering the reflection coefficient, the target signal Starget can be described as
Sagittal Fig. 7. Illustration for the development of cerabellar cortex.
Please cite this article in press as: K. Kobayashi et al., Acoustic impedance microscopy for biological tissue characterization, Ultrasonics (2014), http:// dx.doi.org/10.1016/j.ultras.2014.04.007
4
K. Kobayashi et al. / Ultrasonics xxx (2014) xxx–xxx
agar was calibrated in the same manner short time before the observation. It was calculated to be 1.65 106 Ns/m3. As the sound speed of the substrate at about 80 MHz at 25 °C and its density at 25 °C were 2.78 km/s and 1.16 mg/mm3, respectively, the acoustic impedance of the substrate was calculated to be 3.22 106 Ns/m3. The accuracy was ensured by observing saline droplets with different salinities. The acoustic impedance measured by this system was compared with that estimated as the product of sound speed and density. As shown in Fig. 6, these values agree well. 4.3. Observation of cerebellar cortex of a rat Fig. 7 illustrates the development of cerebellar cortex [5,6]. Parallel fibers in molecular layer are axons of granule cells and play an important role in cerebella neuronal connections. Migrating granule cells elongate them horizontally and form a lot of excitatory synapses to dendrites of Purkinje cells. These are major neuronal circuits of cerebellum so that parallel fibers are expected to construct rich molecular layer with development. But it was hard to
evaluate a degree of parallel fiber development with over molecular layer. We have little sufficient histochemical tools to visualize the developing parallel fibers. Although electron microscopy shows fine structure, it is limited to a very local image. As the field of view of the proposed acoustic technique is as wide as that of optical microscope, it is expected to be a suitable replacement that can observe without any histochemical tools. The granule cells migrate away from the ventricular zone, over the top of the developing Purkinje cells to form a secondary zone of neurogenesis, called the external granule layer. After the birth the cells in this layer continue to actively proliferate, generating an enormous number of granule cell progeny at postnatal 7–10 days. Soon after their generation, after their final mitotic division, the granule cells change from a very round cell to take on a more horizontally oriented shape as they begin to extend axons tangential to the cortical surface, called parallel fibers. Next, the cell body of granule cell migrate to deep into the cerebellum, so the cell assumes a T shape. The cell body eventually migrates past the Purkinje cell layer and then begins to sprout dendrites in the granule cell layer until postnatal 20 day.
1 mm
(a) P1
200 μm
Inner layer
Inner layer EGL
EGL
PL
(b) P7 WM IGL
EGL
PL
PL
EGL
(c) P20
ML PL IGL ML PL WM
IGL
Fig. 8. Two-dimensional profiles of acoustic impedance (106 Ns/m3) of cerabellar cortex (left) and optical microcopies (right). Specimen: rat, sagittal cross section, chemically fixed. Frequency range: 60–100 MHz. Substrate: not treated with plasma.
Please cite this article in press as: K. Kobayashi et al., Acoustic impedance microscopy for biological tissue characterization, Ultrasonics (2014), http:// dx.doi.org/10.1016/j.ultras.2014.04.007
5
K. Kobayashi et al. / Ultrasonics xxx (2014) xxx–xxx
Purkinje cells, on the other hand, have arranged on the Purkinje cell layer at birth, whereas their dendrites are short and immature. They form synapses to parallel fibers of the granule cells and construct cerebellar neuronal network. The network layers containing parallel fibers and Purkinje cell dendrites are called the molecular layers. Parallel fibers are thin and unmyelinated neuronal fibers. Because parallel fibers run transverse between left and right hemisphere, sagittal sections of cerebellar cortex show the cross view of the parallel fibers and coronal sections show the side view. Fig. 8 shows the observed images of cerebellar cortex of a rat at immature (P1; postnatal 1 day), transient (P7), and mature (P20) stages. All the specimens in Fig. 7 had been chemically fixed. In the immature cerebellar cortex (P1), the external granular layer (EGL), the outer layer of the cortex, showed higher impedance compared to the inner layer. The area indicated by the rectangle in the acoustic image is morphologically corresponding to the immunohistochemical observation, although the scale is not completely corresponded because the tissue was somehow subjected to compression during the acoustic observation. At this stage, as myelin is not yet generated, the existence of white matter (WM) is not clearly observed. In the transient stage, four different layers; the WM, internal granular layer (IGL), Purkinje layer (PL) and EGL become to be comprehensive. The EGL and IGL showed higher impedance than the PL and WM. Morphological correspondence between acoustic and immunohistological observation is however not clear in these images. In the mature stage, the EGL, which is composed of small neuronal cell bodies, has developed into the molecular layer (ML), which is composed of elongated axon (neurite), called parallel fibers. The four layers, WM, IGL, PL and ML are more clearly observed in acoustic image. The correspondence with immunohistological observation is also clearly seen. Fig. 9 shows the observed images without chemical fixation. A cerebellum of mature rat (P36) was employed as a specimen. The sagittal cross section was placed on the substrate. The same structure as Fig. 7(c) is clearly seen. The enlarged image was taken by scanning the transducer with 2 lm pitch. Although the image is not very clear, it suggests that the observation of each Purkinje cell, its scale being as small as 20 lm, will be quite feasible if the focal diameter is reduced by using a little higher frequency range.
5. Discussion Considering the precision of the calibration, the reference material should be stable in both physical and chemical properties, and should strongly adhere to the substrate. It is recommended that the acoustic impedance of the reference is close to that of the target. Furthermore, as for the substrate, most of available materials have higher acoustic impedance than biological tissues. In such cases, the phase of the transmitted signal is reversed at the interface. The acoustic impedance of the substrate should be sufficiently high compared to that of the target, in order to retain a strong reflection. However extremely high acoustic impedance of the substrate may increase the reflection coefficient at the interface between the coupling medium and substrate, and reduce the intensity of transmitted signal to the target. This would obviously reduce the S/N ratio. Therefore, in order to obtain a good S/N ratio, the materials should be carefully selected considering their accordance. As the transducer was designed for usage with water as the coupling medium, the existence of the plastic plate between the transducer and focal point may bring an aberration. This will be significant if the thickness of the substrate is very thick, and the convergence angle is very large. The good agreement would derive
5 mm
Tissue on substrate 2 mm
ML PL IGL
200 μm
ML
PL
IGL
Fig. 9. Two-dimensional profiles of acoustic impedance (106 Ns/m3) of cerabellar cortex. Specimen: rat, postnatal 36 days, sagittal cross section, not chemically fixed. Frequency range: 60–100 MHz. Substrate: treated with plasma.
from the fact that the angle of focusing of the transducer was as small as 14° . In case that the angle of focusing is larger, we may have to consider the existence of aberration due to oblique incidence. In such a case, analysis similar to Fourier optics by decomposing the beam into plane waves will be required. Nevertheless, a quantitative analysis is needed in order to precisely assess the acoustic impedance, especially when a thick substrate is employed. As the reflected signal includes wide range of frequency component, the image can be reconstructed using different frequency ranges. Fig. 10 shows the image reconstructed from the same dataset as Fig. 8(c). These corresponding images are reconstructed by extracting the frequency range of 15–100 MHz, and 60–100 MHz, respectively. It can be comprehended that limiting the low frequency range results in the clearer image. It is therefore recommended to extract higher frequency range from the signal, however, it should be also noted that extreme limitation of low frequency component would lead to the reduction in S/N ratio. It occasionally took place that the image looked differently depending on the compatibility between the tissue and plastic substrate. Fig. 11 shows an example that the inner structure was not clearly observed. This kind of image was often obtained with mature rats. It is considered that the cerebellar parallel fibers in a mature rat is more developed than an immature, leading to a
Please cite this article in press as: K. Kobayashi et al., Acoustic impedance microscopy for biological tissue characterization, Ultrasonics (2014), http:// dx.doi.org/10.1016/j.ultras.2014.04.007
6
K. Kobayashi et al. / Ultrasonics xxx (2014) xxx–xxx
1.5
1 mm
1.0 0.5 0.0 0
ML
20
40 60 Frequency (MHz)
80
100
PL IGL
WM Fig. 10. Two-dimensional profiles of acoustic impedance (106 Ns/m3) of cerabellar cortex. Specimen: rat, postnatal 36 days, sagittal cross section, not chemically fixed. Frequency range: 15–100 MHz. Substrate: not treated with plasma.
2 mm
Fig. 12. Analysis for signals in the frequency domain. The normalized intensity represents the intensity of the target signal with water layer between the substrate by the reference signal.
density and expectedly low sound speed. Further pharmacological investigation is required. Until now, we are convinced the contact between the tissue and slide glass is well retained. However, there is two kinds of possibility of delamination. The first is that the air layer is formed in between the substrate and tissue. In such case, perfect optical reflection can be observed with a relatively small incident angle. We noticed no reflection formed when the contact is good. The second case is that a thin water layer is formed between the substrate and tissue. It is difficult to check with an optical microscope, because the optical refraction index of the tissue is close to water. We noticed multiple acoustic reflections in the water layer take place in such case, as the acoustic impedance of the tissue is a bit larger than water. We can find bad contact by observing distortion of the reflected waveform. As shown in Fig. 12 the similarity of waveforms from the tissue and reference would not be retained. As shown in Fig. 4, the normalized spectrum is very flat throughout wide frequency range suggests that the waveform from the target is similar to that from the reference. We have proved the contact is well retained for the experiment with above proves. 6. Summary
Fig. 11. Two-dimensional profiles of acoustic impedance (106 Ns/m3) of cerabellar cortex. Specimen: rat, postnatal 36 days, sagittal cross section, not chemically fixed. Frequency range: 60–100 MHz. Substrate: not treated with plasma.
bad compatibility between the tissue and substrate, probably because of difference in degree of cytoskeltal development [7]. This may be overcome by upgrading the hydrophilic property of the substrate. In this investigation, as an attempt, the surface of the substrate was treated with an atmospheric plasma in air. In fact Fig. 8, in which the internal structure is clearly seen, was observed using the same kind of specimen as that shown in Fig. 11. The image of Fig. 8 was obtained with the surface treatment, whereas that of Fig. 10 was without the treatment. The contact angle of water was improved from 55° to 39° after the treatment. Polar groups induced on the surface of the substrate would have brought a good compatibility with the tissue. The compatibility may also be upgraded by employing different kind of plastic materials, as well as performing some other surface treatments such as silane coupling treatment. As the WM is rich with fat, its acoustic impedance would be lower than the IGL. The ML is composed of axon, which have a lot of actin fibers with high elasticity [7]. It would lead to a high acoustic impedance. The reason why the Purkinje layer has low impedance can be considered as Purkinje cell has relatively small nuclei and mostly filled with cytoplasm that has relatively low
(1) A new method for two-dimensional acoustic impedance imaging for biological tissue characterization with microscale resolution was proposed. (2) Calibration was performed using a reference material of which acoustic impedance was known. Quantitative imaging of acoustic impedance was made possible. (3) Acoustic impedance microscopy with 200 200 pixels, its typical field of view being 2 2 mm, was obtained by scanning the transducer. (4) The development of cerebella cultures of a rat was clearly observed as the contrast in acoustic impedance, without staining the specimen. (5) The technique is believed to be a powerful tool for biological tissue characterization, as no staining nor slicing is required.
Acknowledgements The authors would like to express their sincere thanks to C.-K. Lee, T. Morishima and E. Fukushi of Toyohashi University of Technology for their assistance with the experiment. This study was financially supported by Grants-in-Aid for Scientific Research (Scientific Research (B)15360217, (B) 15300178), and Japan Society for the Promotion of Science and Health and Labor Sciences Research Grants from the Ministry of Health, Labor and Welfare for the Research on Advanced Medical Technology (H17-Nano-001).
Please cite this article in press as: K. Kobayashi et al., Acoustic impedance microscopy for biological tissue characterization, Ultrasonics (2014), http:// dx.doi.org/10.1016/j.ultras.2014.04.007
K. Kobayashi et al. / Ultrasonics xxx (2014) xxx–xxx
References [1] Y. Saijo, M. Tanaka, H. Okawai, H. Sasaki, S. Nitta, F. Dunn, Ultrasonic tissue characterization of infracted myocadium by scanning acoustic microscopy, Ultrasound Med. Biol. 23–1 (1997) 77. [2] H. Okawai, K. Kobayashi, S. Nitta, An approach to acoustic properties of biological tissues using acoustic micrographs of attenuation constant and sound speed, J. Ultrasound Med. 20 (2001) 891. [3] N. Hozumi, R. Yamashita, C. -K Lee, M. Nagao, K. Kobayashi, Y. Saijo, M. Tanaka, N. Tanaka, S Ohtsuki, Ultrasonic sound speed microscope for biological tissue characterization driven by nanosecond pulse, Acoustic Sci. Technol. 20 (2003) 386.
7
[4] N. Hozumi, R. Yamashita, C-K. Lee, M. Nagao, K. Kobayashi, Y. Saijo, M. Tanaka, N. Tanaka, S. Ohtsuki, Time – frequency analysis for pulse driven ultrasonic microscopy for biological tissue characterization, Ultrasonics 42 (2003) 717. [5] D.H. Sanes, T.A. Reh, W.A. Harris, Development of the Nervous System, Academic Press, 2000. [6] E.R. Kandel, J.H. Schwarts, T.M. Jessell, Principles of Neural Science, McGrawHill, 2000. [7] H.W. Wu, T. Kuhn, V.T. Moy, Mechanical properties of L929 cells measured by acoustic force microscopy: effects of anticytoskeletal drugs and membrane crosslinking, Scanning 20 (1998) 389.
Please cite this article in press as: K. Kobayashi et al., Acoustic impedance microscopy for biological tissue characterization, Ultrasonics (2014), http:// dx.doi.org/10.1016/j.ultras.2014.04.007
