The feasibility of an MRI compatible
ultrasound transducer
A.H. (Anne) Schrader
BSc Report
C e
Ir. F.S. Farimani Dr.ir. J.R. Buitenweg Dr.ir. J.F. Broenink
July 2016 021RAM2016 Robotics and Mechatronics
EE-Math-CS University of Twente
P.O. Box 217
7500 AE Enschede
The Netherlands
Abstract
The goal of this thesis is to determine the feasibility of an MRI compatible ultrasound trans- ducer. Combining the capabilities of MRI and ultrasound allows us to create a superior imaging method which has a fast frame rate and a high quality image. A literature study was combined with prototype testing to determine which materials were suitable for use in an MRI compat- ible ultrasound transducer. The presence of conductive materials does not seem to be a di- rect problem for the safety of the patient and the imaging quality, while ferromagnetic metals should be avoided. Image degradation occurred when a signal was applied to the prototype.
Despite expectations set by the literature study, aluminium did not provide sufficient shield- ing, as an increase in zipper artefacts could be seen in the MRI image. This study showed that it should possible to make an MRI compatible ultrasound transducer if shielding is taken into account.
Keywords: Ultrasound imaging, MRI compatibility, Magnetic Resonance Imaging
Acknowledgements
This report is the final step in my study Biomedical Technology at the University of Twente.
Without the help of several people, this thesis would not have existed. First of all, my thanks go to Foad S. Farimani, my daily supervisor, for supporting me during the 10 weeks writing this thesis took. Also I would like to thank all the people at RaM for their support and of course my friends outside of RaM, for putting up with me at the stressful moments and standing by me.
Lastly I would like to thank Ferroperm Piezoelectrics A/S for sending me their piezoelectric elements free of charge.
Anne Schrader
Enschede, July 2016
Abbreviations
McRobot MRI compatible Robot
MURAB MRI and Ultrasound Robotic Assisted Biopsy MIS Minimally Invasive Surgery
CT Computed Tomography
RF Radio Frequent
MRI Magnetic Resonance Imaging
MHz MegaHertz
RaM Robotics and Mechatronics department
Port Port Plastics
Ciba Ciba-Araldite Products
Li Li Tungsten Co.
Si Sigri Corporation, Carbon and Graphite
Rohm Rohm and Haas
H Hysol Divison, Dexter Corp.
Contents
1 Introduction 1
1.1 Problem statement . . . . 1 1.2 Background . . . . 3
2 Overview 9
2.1 State of the Art . . . . 9 2.2 Materials . . . . 13
3 Requirements 21
4 Conceptual design 23
5 Prototyping & Testing 26
6 Results 28
7 Evaluation 30
7.1 Discussion . . . . 30 7.2 Conclusion . . . . 30 7.3 Future work . . . . 31
A Appendix 1 35
A.1 Phantom model . . . . 35
List of Figures
1.1 Several possible crystal geometries. A: Thickness-expander rectangular plate.
B: Thickness-expander circular plate disk. C: Length-expander bar. D: Width-
extensional bar or beam plate. [2] . . . . 3
1.2 Artefacts resulting from the presence of metal inside the MRI. . . . 5
1.3 Effect of paramagnetic and diamagnetic metals on a magnetic field. M denotes the metal side, while S is the surrounding material. [14] . . . . 6
1.4 Effect of paramagnetic, ferromagnetic and diamagnetic metals on a magnetic field, with µ
mthe permeability of the metal and µ
sthe permeability of the sur- rounding material. . . . 7
1.5 Zipper artefacts shown along the vertical axis. . . . 7
1.6 Schematic view of a coax cable including charges and an AC current. S is the shielding, while c is the core. . . . 8
2.1 Section view of the MARIUS transducer. [15] . . . . 9
2.2 Shielded ultrasound transducer with fiducial markers. [16] . . . . 9
2.3 Piezoelectric PZT-4 covered in Cu-Epoxy composite. [17] . . . . 10
2.4 Setup of an optical transducer with two lasers. [8] . . . . 11
2.5 Schematic view of an optical transducer. [8] . . . . 11
2.6 Transducer with two matching layers. Zp is the piezoelement, Zt is the tissue and Z1 and Z2 show the matching layers and Z is the width of both matching layers. . 16
2.7 Effect of different percentages of filling on acoustic impedance (l) and attenuation (r) at 30 Mhz. [27] . . . . 17
2.8 Attenuation of different composites at frequencies from 3 to 7 MHz in Spurr epoxy. From top to bottom: tungsten, PZT, alumina and pure Spurr epoxy. [28] . 18 5.1 The copper prototype(left) and piezo prototype(right). . . . 26
5.2 Phantom with the copper prototype inserted and wrapped in plastic foil. . . . 27
6.1 Results of the MRI-scans. A: Phantom B: Phantom + piezo C: Phantom + Copper prototype D: Phantom + transducer prototype . . . . 28
6.2 MRI-scan of phantom and piezoelectric element, deeper slice. . . . 28
6.3 Results of the MRI-scans with power on the prototype. The green square shows the location of the prototype, while the yellow square denotes roughly the inner square in the phantom. A: Copper prototype with aluminium B: Copper proto- type without aluminium. C: Copper prototype with aluminium and power. D: Copper prototype without aluminium and power. . . . 29
A.1 Drawing of the structure of the phantom used in the images. Measurements are
in mm. . . . 35
1 Introduction
In recent years image-guided interventions are used increasingly in the medical field, with one of the imaging techniques being magnetic resonance imaging (MRI). MRI is able to provide medical personnel with high resolution images of human tissues, without exposing the patient to damaging radiation. The downside of MRI is that it takes a long time to complete and the costs are much higher than those of other imaging modalities. Also, MRI is not real-time and the images are susceptible to movement artefacts.
In this thesis the main focus will be on interventions in the form of an image-guided biopsy, as it is essential to know if the needle is in the correct position during a biopsy. Using MRI to determine the position of the needle takes a long time as the patient has to be taken out of the MRI scanner repeatedly to adjust the needle and later check the position of the needle again inside the MRI. On the other hand, using another imaging modality like ultrasound has its downsides too. While ultrasound is real-time and cheap to perform, the resulting images are of poor quality. If there are similar tissues next to each other it becomes increasingly difficult to determine the correct location of the needle.
In general combining ultrasound and MRI increases the susceptibility of imaging, as each modality can differentiate between different tissues and materials. The combination also al- lows real-time images to be taken with the resolution of an MRI image. For RaM the usefulness lies with the Murab and McRobot project.
Murab aims to create a robot that can perform biopsies, by means of mechanical imaging and image fusion outside of the MRI. This does not only speed up the process, but will also be able to detect more lesions as ultrasound and MRI both have the ability to detect different lesions. By making a transducer which can be present during an MRI scan, we can link the MRI image to the ultrasound image. This MRI compatible transducer enables merging of the separate images later on.
McRobot on the other hand aims to create a fully MRI compatible robot. Ultrasound will be used to guide the robot to the correct positions inside the body while the MRI is switched on. Because of this, the patients do not have to be removed from the MRI at any point during the biopsy. To achieve this, it has to be shown that it is feasible to create an MRI compatible ultrasound transducer within the department at a reasonable cost.
1.1 Problem statement
While MRI is used more often, the MRI is both time consuming and costly. A breast biopsy with MRI takes around 1 hour, but can take even longer if the needle placement has to be rechecked.
This makes MRI waiting times very long, up to a 100 days for routine scans.
In this thesis the feasibility of an MRI compatible ultrasound transducer will be determined, which could enable MRI and ultrasound image fusion to take place. For Murab the aim is that the full biopsy can be done outside the MRI after one scan, as the ultrasound and MRI image will be fused which creates a high-quality real-time image of the biopsy area. In com- bination with the McRobot the full biopsy can take place inside the MRI. Because of this the patient does not have to be moved in and out the MRI and as a result the biopsy time is reduced.
Currently ultrasound transducers cannot be used within an MRI because they contain metals,
cannot be tracked inside the MRI and do not contain shielding to block generated radio fre- quent(RF) waves. Ferromagnetic metals can be dangerous for the patient and cause damage to the machine. Non-ferromagnetic metals are a danger because of inductive heating, eddy currents and deformation of the magnetic field, possibly burning the patient and decreasing image quality of both the MRI and the ultrasound image.
During this study the following questions will be answered in depth:
1. Is it possible to have an MRI compatible ultrasound?
2. Why are current ultrasound transducers not compatible with MRI?
3. What has been done before in regard to MRI compatible ultrasound?
4. How is an ultrasound transducer made?
5. What materials can be used inside the MRI?
1.2 Background
To answer the questions posed above, it is important to know how MRI and ultrasound works.
This will be covered in the following sections.
Ultrasound imaging
Ultrasound is one of the most used medical imaging technologies [1]. Ultrasound waves are generated by transducers, which are reflected by tissue and again received by the transducer.
The elements that are used nowadays to generate and receive these ultrasound waves are piezo- electric elements. When the material is excitated electrically it expands or contracts, which in turn produces ultrasound waves. While receiving ultrasound, the piezoelectric material is de- formed slightly, which creates an electrical pulse. This pulse can be detected by machines and later transformed into an image. The ratio of this conversion from the electrical domain to the mechanical domain is dependent on the coupling coefficient, which is in turn dependent on the inherent material properties and geometry [2]. The different crystal geometries that can be used for piezoelectric elements can be seen in figure 1.1.
Figure 1.1: Several possible crystal geometries. A: Thickness-expander rectangular plate. B: Thickness- expander circular plate disk. C: Length-expander bar. D: Width-extensional bar or beam plate. [2]
In the past thickness expander elements, which have good conversion in the thickness direc- tion, were mostly use in single element transducers. Now with the existence of array transduc- ers, width and thickness expander transducers are used more often.
The wavelength of a wave determines the resolution and imaging depth. Higher frequencies create a higher resolution, but they cannot penetrate as deep into the body [3]. Lower frequen- cies can penetrate deeper into the body. Because of this, broadband transducers, which are transducers that span several frequencies, are needed to correctly image several layers of the body. The downside of a broadband transducer is that the broader the signal, the lower the sensitivity of the transducer [4].
The frequency of an ultrasound wave with wavelength λ and speed v is given by equation 1.1 [5].
F r equenc y = v
λ (1.1)
The wavelength and amount of periods per pulse (ppp) can be used to calculate the spatial pulse length (SPL) as seen in equation 1.2.
SP L = ppp • λ (1.2)
The axial resolution of a transducer is equal to half the SPL and a shorter SPL creates a broader
bandwidth, which is the range of frequencies that a transducer outputs. This in turn tells us
that a broader bandwidth gives a smaller axial resolution. The bandwidth is often given as a
percentage of the central frequency, so a 5 MHz transducer with a bandwidth of 50% has signal ranging from 2.5 MHz to 7.5 MHz. Bandwidth is important for imaging transducers, as without bandwidth only one layer of tissue can be imaged.
Approximate imaging frequencies in medical imaging for different areas can be seen in table 1.1.
Table 1.1: Approximate imaging frequencies used in medical ultrasound. [6]
Frequency Imaging area
2.5 MHz Deep abdomen, obstetric and gynaecological imaging 3.5 MHz General abdomen, obstetric and gynaecological imaging 5.0 MHz Vascular, breast, pelvic imaging
7.5 MHz Breast, thyroid
10.0 MHz Breast, thyroid, superficial veins, superficial masses, musculoskeletal imaging 15.0 MHz Superficial structures, musculoskeletal imaging
To generate the ultrasound waves needed for imaging, two principles can be used. The most common option is the use of piezoelectric elements, for which the generated wavelength is approximately equal to twice the thickness of the element [7]. Higher frequencies thus mean thinner piezoelectric elements, which in turn limits the highest frequency possible as materials become increasingly fragile. Most piezoelectric medical transducers operate between a range of 1-15 MHz.
Another option for ultrasound generation is an optical transducer. Optical transducers can be made by using the thermoelastic effect to generate ultrasound. The downside of this is that most optical transducers work in the frequency range of 20-50 MHz [8] and cannot be focused on a central frequency correctly. As an optical transducer does not contain electrical parts, it requires less shielding than a piezoelectric transducer.
Magnetic Resonance Imaging
The second imaging modality that will be covered is MRI. In MRI, the spin of the nuclei of hy- drogen atoms is used to image different tissues bases on their hydrogen content. A full expla- nation is beyond the scope of this thesis, but in short the scanner sends out an radio frequent (RF) pulse, which causes the hydrogen atoms to move slightly and thus create a change in the magnetic field of the which can be detected [9].
There are several scans that can be used, but in this thesis only the Spin Echo T1 will be used for testing. In a spin echo a 90 degree RF excitation pulse is followed by a 180 degree RF refocus- ing pulse. When a scan is T1 weighted, as the Spin Echo T1 is, fat appears bright, while water appears darker [10]. While taking an scan there are several parameters that affect the image quality [10].
• Repetition time (TR) is defined as the time between two consecutive scans. TR influ- ences T1 contrast strongly.
• Echo time (TE) is the time between the RF excitation pulse and the pulse echo.
• Slice thickness is the depth of each slice and thus determines the details that can be seen in the MR image.
• Number of acquisitions shows the amount of times that an RF pulse is applied to the same voxel. More acquisitions increase the signal to noise ratio.
• Field of View (FOV) determines the dimensions of the imaging area.
For each scanning method different settings are required. Besides the options discussed above, there is also the acquisition matrix, which is the total number of independent data samples in the image and depends on the settings used in the scan. When a scan is taken, it has to be made sure that any object present inside the while testing is MRI compatible.
MRI compatibility
Before MRI compatibility can fully be taken into account, a definition has to be known. In 2005 ASTM International [11] published new definitions for MRI compatibility, which included the terms MRI-safe, MRI-unsafe and MRI-conditional.
• MRI-Safe: it poses no known hazards in de MRI environment.
• MRI-Conditional: it poses no known hazards in the MRI environment with specific con- ditions of use.
• MRI-Unsafe: it poses known hazards in all MRI environment.
As can be seen an MRI-Safe or MRI-conditional material does guarentee safety of the patient, but not quarantee good image quality. Besides ASTM other companies have tried to define MRI compatibility. GE guidelines [12] define four zones of MRI compatibility as follows:
Zone 1 If it may remain in the imaging volume and in contact with the patient throughout MRI scanning.
Zone 2 If it can remain in the imaging volume and in contact with the patient throughout the procedure and scanning, but is not located in the imaging field.
Zone 3 If it is used within the imaging volume, but will be removed during scanning or when not in use.
Zone 4 If it is suitable for use in the magnet room during procedure when kept at least one meter from the isocenter or beyond the 20 mTesla line.
The aim is to design a transducer that fits in zone 1 according to GE guidelines and a MRI-Safe or MRI-Conditional transducer according to ASTM definition. Besides these formal guidelines, image quality also has to be sufficient for any solution to work.
Image quality is dependent on the presence of image artefacts in the final image. As MRI makes use of a magnetic field to put the hydrogen atoms in their starting position and then sends out RF pulses to change the position of the hydrogen atoms, it is important for the magnetic field to be the same in all places. When there is a change in the magnetic field of the MRI, an artefact can be seen at the place of the inhomogeneity. This can be seen in figure 1.2 and is called a local inhomogeneity of the magnetic field [10].
Figure 1.2: Artefacts resulting from the presence of metal inside the MRI.
These changes in the magnetic field due to metal pieces can be explained by Maxwell’s equa- tions, which can be seen in equations 1.3 and 1.4 [13].
B
⊥m= B
s⊥(1.3)
1
µ
m• B
m||= 1
µ
s• B
s||(1.4)
Here B is the magnetic field strength, with B
⊥and B
||signifying the strength in different di- rections as can be seen in figure 1.3. µ is the magnetic permeability, which shows how the material affects the surrounding magnetic field. Ferromagnetism is basically the same as para- magnetism, in the sense that it pulls in the magnetic field, but for ferromagnetism the effect is much stronger, which is shown by the fact that a force acts upon ferromagnetic materials inside a magnetic field. Comparing the magnetic permeability relative to the magnetic permeability of the vacuum of both materials, that of paramagnetic materials is only slightly higher than 1, while ferromagnetic materials can have a magnetic permeability up to the hundreds.
Figure 1.3: Effect of paramagnetic and diamagnetic metals on a magnetic field. M denotes the metal side, while S is the surrounding material. [14]
From the above equations we obtain the following equation [13], 1
µ
m• B
||mB
⊥m= 1
µ
s• B
s||B
⊥s(1.5)
With this equation it can be concluded that when µ
m≈ µ
sthe magnetic field is not affected
much, as B
m||/B
m⊥≈ B
s||/B
s⊥. But if µ
m>> µ
s, then B
m||/B
⊥m>> B
||s/B
s⊥, which means that α
tis
almost 90
◦. This results in B being perpendicular to the surface, which can be seen in figure
1.4 as ferromagnetism. Paramagnetism has the same effect as ferromagnetism, except much
weaker. Diamagnetism has the opposite affect, thus pushing the magnetic field away and this
happens when µ
m<< µ
s.
Figure 1.4: Effect of paramagnetic, ferromagnetic and diamagnetic metals on a magnetic field, with µ
mthe permeability of the metal and µ
sthe permeability of the surrounding material.
Currently there is no method to fully eliminate this type of artefacts, but the effect can be de- creased. A shorter TE, larger acquisition matrix and smaller FOV may decrease the size of the artefact [10]. Another artefact that can show up in the MRI image is the zipper artefact, which looks like lines of alternating bright and dark pixels running through the image as can be seen in figure 1.5 They are caused by RF noise from outside and can thus be reduced by decreasing the amount of RF noise [9].
Figure 1.5: Zipper artefacts shown along the vertical axis.
Reducing RF noise can be done with shielding, for which a good example is a coax cable. Fig-
ure 1.6 shows a schematic view, with I being the current, which changes over time and L the
circumference of the cable.
Figure 1.6: Schematic view of a coax cable including charges and an AC current. S is the shielding, while c is the core.
The shield will generate a charge opposite to the charge of the core, so when the charges are exactly the same in magnitude the electric fields cancel each other. To avoid an electric charge taking place on the outside of the shielding, it should be grounded, allowing the charge to be dispersed.
For the magnetic field we know that;
I
C
B • dl = µ
0Z
S
I d A (1.6)
For a wire with circular symmetry this means that equation 1.7 is true.
I
C