MRI-compatible Micromanipulator - Semantic Scholar
MRI-compatible Micromanipulator Design and Implementation and MRI-compatibility Tests Yoshihiko KOSEKI, Tamio TANIKAWA, and Kiyoyuki CHINZEI Abstract— In this paper, we present a magnetic resonance imaging (MRI)-compatible micromanipulator, which can be employed to provide medical and biological scientists with the ability to concurrently manipulate and observe micron-scale objects inside an MRI gantry. The micromanipulator formed a two-finger micro hand, and it could handle a micron-scale object using a chopstick motion. For performing operations inside the MRI gantry in a manner such that the MRI is not disturbed, the system was designed to be nonmagnetic and electromagnetically compatible with the MRI. The micromanipulator was implemented with piezoelectric transducers (PZT) as actuators for micro-motion, strain gauges as sensors for closed-loop control, and a flexure parallel mechanism made of acrylic plastic. Its compatibility with a 2-Tesla MRI was preliminarily tested by checking if the MRI obtained with the micromanipulator were similar to those obtained without the micromanipulator. The tests concluded that the micromanipulator caused no distortion but small artifacts on the MRI. The signal-to-noise ratio (SNR) of the MRI significantly deteriorated mainly due to the wiring of the micromanipulator. The MRI caused noise of the order of ones of volts in the strain amplifier.
I. I NTRODUCTION Magnetic Resonance Imaging (MRI) has been developed into a powerful observation method, particularly in medical applications. In addition, it has contributed to cognitive psychological and physiological understanding due to its noninvasiveness, good soft tissue contrast, diffusion enhanced image, and biochemical shift information. In other words, a cognitive psychologist and physiologist can observe liquid flow and chemical localization inside a live tissue or animal without damaging it. MRI is considerably inferior to optical and fluoroscopic modality with regard to its resolution and acquisition time. The conventional MRI provided images with a resolution of an order of 1 × 1 × 1mm3 , which is considerably different from that of a microscope. However, many researchers have studied MR microscopy, and have attained micron-scale resolutions[1]. Currently, MR microscopy demands huge acquisition times, i.e., several hours for obtaining images of the highest quality. However, many studies have been performed to achieve fast MR microscopy, and the acquisition time is expected to further decrease. Yoshihiko KOSEKI is with the National Institute of Advanced Industrial Science and Technology (AIST), 1-2-1 Namiki, Tsukuba, Ibaraki 305-8564, Japan Tamio TANIKAWA is with the National Institute of Advanced Industrial Science and Technology (AIST), 1-1-1 Umezono, Tsukuba, Ibaraki 3058568, Japan Kiyoyuki CHINZEI is with the National Institute of Advanced Industrial Science and Technology (AIST), 1-2-1 Namiki, Tsukuba, Ibaraki 305-8564, Japan
Medical and biological scientists need to move, hold, cut, and sting microscopic biological cells and/or tissues. For such requirements, many researchers have studied mechanical and mechatronical systems for dexterous micron-scale manipulation, which has been manually difficult[2]. In this study, we propose an MRI-compatible micromanipulator that can be employed to provide medical and biological scientists with micro-scale manipulation inside the MRI gantry. The potential merits of this device are that the MRI can visualize an object’s response to mechanical changes on-site, and the micromanipulator changes the object according to the on-site information from the MRI. This can significantly save the time spent on experiments by medical and biological researchers. Many MRI-compatible manipulators, which work inside and/or nearby intraoperative MRI, have been studied to enable minimally invasive surgeries[3]. The MRI detects a weak radio wave excited by a strong radio wave in a strong and precise magnetic field. Therefore, the devices installed inside the MRI have to preserve these conditions. Our research is similar in that it is compatible with MRI, however different in that it deals with considerably smaller objects. In this paper, we first introduce our first prototype of the MRI-compatible micromanipulator. We discuss the design details required for operations inside the MRI gantry and the manipulation of micron-scale objects. Second, we perform a preliminary test to confirm its compatibility with a 2-Tesla MRI by checking if the MRI obtained with the micromanipulator are similar to those obtained without it. II. D ESIGN & I MPLEMENTATION A. Outline Fig. 1 shows a concept design of micromanipulation in MRI. Because the object is surrounded by magnets to maintain a strong magnetic field, this system provides a semi-telemanipulation from intra-MRI to extra-MRI. The system consists of MRI, a micromanipulator for precise and dexterous operations, an MRI-compatible microscope for real-time observations, a microstage for coarse and wide motion, and a master-arm for a human-machine interface. In this paper, the micromanipulator was studied in the first step of this project because it is the most important topic. In our study, the microscope is based on our MRI-compatible endoscope, which is replaced with a large-magnification optical system, and the microstage is based on a conventional MRI-compatible manipulator. These studies will be reported in the near future.
The 29th Annual International Conference of the IEEE EMBS, ThC09.4, pp. 465-458, 2007 Linear Actuator
Micromanipulator
Lever Amplification
Magnetic Field
Needle Adjustment Mechanism Lower Needle
MRI Gantry Object
Piezoelectric Transducer
Microscope Base
Microstage
Upper Endplate
Sting
MRI Room Console Room
Cut
Hold Move
Microscopy
Fig. 1.
Upper Needle
MRI
Fig. 2.
Lower Endplate
Mechanical System of Micromanipulator
Needle Adjustment PZT(Piezoelectric Transducer) & Strain Gauges Mechanism (Acrylic Plastic & Polyacetal resin)
Operator
Concept Design of Micromanipulation in MRI
B. Mechanical System Figs. 2 and 3 show a schematic and the actual mechanical system of the micromanipulator, respectively. The micromanipulator moved two finely tapered glass needles and formed a two-finger micro-hand, as studied in [2]. The lower endplate moved both the lower needle and upper endplate, and made absolute motion corresponding to the thumb. The upper endplate moved the upper needle and made relative motion to the lower needle corresponding to the index finger. The initial relative position of the lower needle with respect to that of the upper needle was adjusted by screws. Both the needles were moved by parallel mechanisms with three degrees of freedom. Its endplate was supported by three linear actuators in parallel. When all the actuators stretched, the endplate exhibited translational motion along the longitude. When only one actuator stretched, the endplate was slightly inclined and it exhibited approximately translational motion in the transverse direction. The lever amplification mechanisms were implemented with actuators. All the joints were made of flexure hinges. The linear actuators were resin-coated multilayer piezoelectric transducers (AE0203E44H40, NEC-Tokin) because of the requirement of micron-scale linear motion and nonmagnetism. Their stroke rating was 28.0 µm at 100 V DC. The materials used for this mechanism was mainly acrylic plastic and partly polyacetal resin. The use of metals was minimized because they caused artifacts on the MRI, however they were used in electric wiring and shielding. C. Electrical System Fig. 4 shows the electrical system of the micromanipulator. The piezoelectric transducers (PZT) were driven from the region outside of MRI shield room. The driver was a noninverting amplifier (gain: 10–15) with a high-voltage op-amp (Burr-Brown, Model3582) and it could apply a maximum voltage of 150 V at 1kHz. It was connected to the PZT via an 8-m twisted and shielded cable and low-pass filters (cutoff frequency = 0.8 MHz). The input signals from the control
Needle (Glass) Flexure Parallel Mechanism (Acrylic Plastic)
Fig. 3.
First Prototype of Micromanipulator
PC were optically transmitted and the system was electrically isolated from an environment not compatible with the MRI. For precise positioning, the nonlinear behavior of the PZT had to be compensated by close-loop control. The strain gauge was bonded to the PZT in order to measure its extension. The dummy gauge was also bonded orthogonally to cancel the thermal effect. The strain amplifier was set close to the gauges to prevent noise from long transmission paths, however, it was set at a distance of 0.4 m from the micromanipulator to prevent artifacts from the amplifier box on the MRI. Similar to the driver, the signals were transmitted to the control PC through a twisted and shielded cable, low-pass filter, and optical isolation. III. E XPERIMENTS & R ESULTS The compatibility with the MRI was tested by comparing the similarity between MRI obtained with the micromanipulator and those obtained without the micromanipulator. Bruker BioSpin’s Biospec 20/30 with AVANCE was used for the MRI (right-hand side image in Fig. 5). It is manufactured for scientific use and provides a 2.0-Tesla magnetic field. As shown in the lower-left side of Fig. 5, the MRI phantom, an anthropogenic signal source, comprised water in an acrylic plastic container with test patterns. The phantom was scanned with three major protocols, named FLASH, FSE ETL16, and MSME respectively. Their major parameters are shown in Table I. The micromanipulator was introduced step-by-step. (1) Only the phantom was installed at the center of the MRI
466
The 29th Annual International Conference of the IEEE EMBS, ThC09.4, pp. 465-458, 2007 Strain Gauge PZT(Piezoelectric Transducer)
Optical Piezo Isolation Amp.
From PC
To PC
Strain Gauge (Sensor for Closed-loop Control) 0.4m
Fig. 4.
Electrical System of Micromanipulator
Fig. 5. MRI setup. The arrangement of the phantom, micromanipulator, and receiving coil inside the MRI gantry is shown in the upper-left image. the phantom, anthropogenic signal source is shown in the lower-left image. The image on the right-hand side shows the MRI’s magnet, Bruker BioSpin’s Biospec 20/30 TABLE I M AJOR SCAN PARAMETERS OF MRI
Sequence Field of View Matrix Resolution Echoes Average Slice Thickness TR TE Flip Angle Bandwidth Scan Time
FLASH Gradient Echo
5.0 mm 28.3 ms 4.3 ms 30 deg 104167 Hz 7s
π-type Low Pass Filter Cut off = 0.8 MHz 8m
Strain Gauge Difference Bridge Amp. Amp. Box(Aluminum H60xW90xD125mm)
FSE ETL16 MSME Fast Spin Echo Spin Echo 100 mm 256 0.391 mm/pixel 1 1 1.0 mm 5000 ms 2500 ms 14 ms 25 ms 180 deg 90 deg 50000 Hz 1 m 20 s 10 m 40 s
gantry and it was scanned as a reference image. Subsequently, the phantom was never moved. (2) The micromanipulator without wires was installed beside the phantom, as shown in the upper-left side of Fig. 5. Strictly speaking, the micromanipulator and the wires until the amplifier box were installed however the rest wires or amplifier box were not installed. The needle adjustment mechanism was approximately 2 cm away from the surface of the phantom. The glass needle was not attached in this experiment. (3) All the wires were connected to the micromanipulator, but
the piezo amplifier and strain amplifier were not powered. (4) The piezo amplifier and strain amplifier were powered, but the feedback was not activated. (5) The feedback was activated. (6) The reference image was scanned again to confirm that the phantom was not moved. To test the noise from the MRI to the strain amplifier, its output was measured by an oscilloscope, Iwatsu’s DS-8812P, while the MRI did not scan and while the MRI scanned with the three protocols mentioned above. Fig. 6 shows the raw images of the phantom scanned with FSE ETL16 for the five conditions, i.e., from (1) to (5). In addition, it shows the difference in the images as compared to the reference image. In the case of “(1) Phantom (no manipulator),” the difference between the two reference images is shown. Small dots are found in the raw images for the conditions (2) to (5). This suggestes that the existence of the micromanipulator caused the artifacts. In the difference images for the conditions (2) to (5), no shift or distortion was found. In the raw images for the conditions (3) to (5), the signal from water became weaker and the noise became stronger in comparison to (1) and (2). The images scanned with other protocols showed similar tendencies. Fig. 7 shows the signal-to-noise ratio (SNR) of each image of each protocol in order to know how the introduction of the micromanipulator affects the image quality of the MRI. The signal is defined as the average of the pixel values within the lower-right circle, and the noise is defined as the standard deviation of that area. In Fig. 7, the SNR of the reference images for the conditions (1) and (6) are shown in advance. Fig. 7 suggests that SNR varies for the same condition. However, the existence of wires and not the micromanipulator, significantly debased the SNR. The changes in the SNR caused by power-on or feedback were considerably smaller than those caused by wiring. Fig. 8 shows the output of the strain amplifier measured by the oscilloscope while the MRI did not scan (left) and while the MRI scanned with FLASH (Gradient Echo). The vertical and horizontal scales for the absence of MRI scanning are 50 mV and 0.1 ms, respectively. Even while the MRI did not scan, the strain amplifier outputted noise of the order of tens of millivolts. The vertical and horizontal scales of the FLASH scanning are 500 mV and 1 ms, respectively. While the MRI scanned, the strain amplifier outputted significant noise of the order of ones of volts. The other protocols
467
The 29th Annual International Conference of the IEEE EMBS, ThC09.4, pp. 465-458, 2007
12.00
Difference from the Reference Image
FLASH Signal-to-Noise Ratio
Raw Image
(1) Phantom (No Manipulator)
10.00
FSE_ETL16 MSME
8.00 6.00 4.00 2.00
(5) Phantom+Manipulator (Feedback ON)
(4) Phantom+Manipulator (Power ON)
(3) Phantom+Manipulator (Power OFF)
(2) Phantom+Manipulator (No Wire)
(2) Phantom + Manipulator (No Wire)
(6) Phantom (No Manipulator)
(1) Phantom (No Manipulator)
0.00
Fig. 7. Signal to Noise Ratio of MRI vs. introduction of micromanipulator
(3) Phantom + Manipulator (Power OFF)
Fig. 8. Noise on Strain amplifier, Left: During the absence of MRI scanning (Vert. Scale = 50 mV, Horiz. Scale = 0.1 ms). Right: During FLASH scanning (Gradient Echo Imaging) (Vert. Scale = 500 mV, Horiz. Scale = 1 ms).
(4) Phantom + Manipulator (Power ON)
(5) Phantom + Manipulator (Feedback ON)
Fig. 6. MRI (FSE ETL16) of phantom vs. the introduction of micromanipulator and its differential image as compared to the reference image
showed a similar tendency. Basically, the micromanipulator does not have to move during the MRI scanning. To avoid non-intentional motion, the micromanipulator should be locked during the MRI scanning, i.e. the PZT amplifier should output constant voltages by open-loop. IV. C ONCLUSIONS AND F UTURE W ORKS
artifacts and significant debasement in the signal-to-noise ratio of the MRI. The debasement in the signal-to-noise ratio was mainly caused by the wiring of the micromanipulator. The results also suggest that the MRI caused noise of the order of ones of volts in the strain amplifier. In practice, this would not be a problem because the micromanipulator does not have to move during the MRI scanning. Improvements are necessary with regard to the wiring of the micromanipulator. The length, shielding, and grounding have to be readjusted. An optical strain gauge (e.g., fiber Bragg grating) provides a definitive solution against noise from the MRI scanning. Lock the micromanipulator during the MRI scanning would be a practical solution. V. ACKNOWLEDGMENTS This research was partially funded by Grants-in-Aid for Young Scientists (#B17700419) from the Japan Society for the Promotion of Science. We wish to thank Mr. Homma and Mr. Numano for their contributions to this research study.
In this paper, we introduced our MRI-compatible micromanipulator, to provide medical and biological scientists with the ability to concurrently manipulate and observe micronscale objects inside the MRI gantry. First, we discussed the design details required for the operation inside the MRI gantry and the manipulation of a micron-scale object. Second, we performed a preliminary test to confirm its compatibility with a 2.0-Tesla MRI. The results suggest that it caused no shift or distortion, however it caused small 468
R EFERENCES [1] L. Ciobanu, et al., ”Magnetic resonance imaging of biological cells,” Progress in Nuclear magnetic Resonance Spectroscopy, Vol. 42, 2003, pp. 69-93. [2] T. Tanikawa, T. Arai, N. Koyachi, ”Development of Small-sized 3 DOF Finger Module in Micro Hand for Micro Manipulation,” Proc. IEEE/RSJ IROS 1999, pp. 876-881, 1999. [3] K. Chinzei, R. Kikinis, F.A. Jolesz, ”MR Compatibility of Mechatronic Devices: Design Criteria,” Proc. MICCA 1999, pp. 1020-1031.
I. I NTRODUCTION Magnetic Resonance Imaging (MRI) has been developed into a powerful observation method, particularly in medical applications. In addition, it has contributed to cognitive psychological and physiological understanding due to its noninvasiveness, good soft tissue contrast, diffusion enhanced image, and biochemical shift information. In other words, a cognitive psychologist and physiologist can observe liquid flow and chemical localization inside a live tissue or animal without damaging it. MRI is considerably inferior to optical and fluoroscopic modality with regard to its resolution and acquisition time. The conventional MRI provided images with a resolution of an order of 1 × 1 × 1mm3 , which is considerably different from that of a microscope. However, many researchers have studied MR microscopy, and have attained micron-scale resolutions[1]. Currently, MR microscopy demands huge acquisition times, i.e., several hours for obtaining images of the highest quality. However, many studies have been performed to achieve fast MR microscopy, and the acquisition time is expected to further decrease. Yoshihiko KOSEKI is with the National Institute of Advanced Industrial Science and Technology (AIST), 1-2-1 Namiki, Tsukuba, Ibaraki 305-8564, Japan Tamio TANIKAWA is with the National Institute of Advanced Industrial Science and Technology (AIST), 1-1-1 Umezono, Tsukuba, Ibaraki 3058568, Japan Kiyoyuki CHINZEI is with the National Institute of Advanced Industrial Science and Technology (AIST), 1-2-1 Namiki, Tsukuba, Ibaraki 305-8564, Japan
Medical and biological scientists need to move, hold, cut, and sting microscopic biological cells and/or tissues. For such requirements, many researchers have studied mechanical and mechatronical systems for dexterous micron-scale manipulation, which has been manually difficult[2]. In this study, we propose an MRI-compatible micromanipulator that can be employed to provide medical and biological scientists with micro-scale manipulation inside the MRI gantry. The potential merits of this device are that the MRI can visualize an object’s response to mechanical changes on-site, and the micromanipulator changes the object according to the on-site information from the MRI. This can significantly save the time spent on experiments by medical and biological researchers. Many MRI-compatible manipulators, which work inside and/or nearby intraoperative MRI, have been studied to enable minimally invasive surgeries[3]. The MRI detects a weak radio wave excited by a strong radio wave in a strong and precise magnetic field. Therefore, the devices installed inside the MRI have to preserve these conditions. Our research is similar in that it is compatible with MRI, however different in that it deals with considerably smaller objects. In this paper, we first introduce our first prototype of the MRI-compatible micromanipulator. We discuss the design details required for operations inside the MRI gantry and the manipulation of micron-scale objects. Second, we perform a preliminary test to confirm its compatibility with a 2-Tesla MRI by checking if the MRI obtained with the micromanipulator are similar to those obtained without it. II. D ESIGN & I MPLEMENTATION A. Outline Fig. 1 shows a concept design of micromanipulation in MRI. Because the object is surrounded by magnets to maintain a strong magnetic field, this system provides a semi-telemanipulation from intra-MRI to extra-MRI. The system consists of MRI, a micromanipulator for precise and dexterous operations, an MRI-compatible microscope for real-time observations, a microstage for coarse and wide motion, and a master-arm for a human-machine interface. In this paper, the micromanipulator was studied in the first step of this project because it is the most important topic. In our study, the microscope is based on our MRI-compatible endoscope, which is replaced with a large-magnification optical system, and the microstage is based on a conventional MRI-compatible manipulator. These studies will be reported in the near future.
The 29th Annual International Conference of the IEEE EMBS, ThC09.4, pp. 465-458, 2007 Linear Actuator
Micromanipulator
Lever Amplification
Magnetic Field
Needle Adjustment Mechanism Lower Needle
MRI Gantry Object
Piezoelectric Transducer
Microscope Base
Microstage
Upper Endplate
Sting
MRI Room Console Room
Cut
Hold Move
Microscopy
Fig. 1.
Upper Needle
MRI
Fig. 2.
Lower Endplate
Mechanical System of Micromanipulator
Needle Adjustment PZT(Piezoelectric Transducer) & Strain Gauges Mechanism (Acrylic Plastic & Polyacetal resin)
Operator
Concept Design of Micromanipulation in MRI
B. Mechanical System Figs. 2 and 3 show a schematic and the actual mechanical system of the micromanipulator, respectively. The micromanipulator moved two finely tapered glass needles and formed a two-finger micro-hand, as studied in [2]. The lower endplate moved both the lower needle and upper endplate, and made absolute motion corresponding to the thumb. The upper endplate moved the upper needle and made relative motion to the lower needle corresponding to the index finger. The initial relative position of the lower needle with respect to that of the upper needle was adjusted by screws. Both the needles were moved by parallel mechanisms with three degrees of freedom. Its endplate was supported by three linear actuators in parallel. When all the actuators stretched, the endplate exhibited translational motion along the longitude. When only one actuator stretched, the endplate was slightly inclined and it exhibited approximately translational motion in the transverse direction. The lever amplification mechanisms were implemented with actuators. All the joints were made of flexure hinges. The linear actuators were resin-coated multilayer piezoelectric transducers (AE0203E44H40, NEC-Tokin) because of the requirement of micron-scale linear motion and nonmagnetism. Their stroke rating was 28.0 µm at 100 V DC. The materials used for this mechanism was mainly acrylic plastic and partly polyacetal resin. The use of metals was minimized because they caused artifacts on the MRI, however they were used in electric wiring and shielding. C. Electrical System Fig. 4 shows the electrical system of the micromanipulator. The piezoelectric transducers (PZT) were driven from the region outside of MRI shield room. The driver was a noninverting amplifier (gain: 10–15) with a high-voltage op-amp (Burr-Brown, Model3582) and it could apply a maximum voltage of 150 V at 1kHz. It was connected to the PZT via an 8-m twisted and shielded cable and low-pass filters (cutoff frequency = 0.8 MHz). The input signals from the control
Needle (Glass) Flexure Parallel Mechanism (Acrylic Plastic)
Fig. 3.
First Prototype of Micromanipulator
PC were optically transmitted and the system was electrically isolated from an environment not compatible with the MRI. For precise positioning, the nonlinear behavior of the PZT had to be compensated by close-loop control. The strain gauge was bonded to the PZT in order to measure its extension. The dummy gauge was also bonded orthogonally to cancel the thermal effect. The strain amplifier was set close to the gauges to prevent noise from long transmission paths, however, it was set at a distance of 0.4 m from the micromanipulator to prevent artifacts from the amplifier box on the MRI. Similar to the driver, the signals were transmitted to the control PC through a twisted and shielded cable, low-pass filter, and optical isolation. III. E XPERIMENTS & R ESULTS The compatibility with the MRI was tested by comparing the similarity between MRI obtained with the micromanipulator and those obtained without the micromanipulator. Bruker BioSpin’s Biospec 20/30 with AVANCE was used for the MRI (right-hand side image in Fig. 5). It is manufactured for scientific use and provides a 2.0-Tesla magnetic field. As shown in the lower-left side of Fig. 5, the MRI phantom, an anthropogenic signal source, comprised water in an acrylic plastic container with test patterns. The phantom was scanned with three major protocols, named FLASH, FSE ETL16, and MSME respectively. Their major parameters are shown in Table I. The micromanipulator was introduced step-by-step. (1) Only the phantom was installed at the center of the MRI
466
The 29th Annual International Conference of the IEEE EMBS, ThC09.4, pp. 465-458, 2007 Strain Gauge PZT(Piezoelectric Transducer)
Optical Piezo Isolation Amp.
From PC
To PC
Strain Gauge (Sensor for Closed-loop Control) 0.4m
Fig. 4.
Electrical System of Micromanipulator
Fig. 5. MRI setup. The arrangement of the phantom, micromanipulator, and receiving coil inside the MRI gantry is shown in the upper-left image. the phantom, anthropogenic signal source is shown in the lower-left image. The image on the right-hand side shows the MRI’s magnet, Bruker BioSpin’s Biospec 20/30 TABLE I M AJOR SCAN PARAMETERS OF MRI
Sequence Field of View Matrix Resolution Echoes Average Slice Thickness TR TE Flip Angle Bandwidth Scan Time
FLASH Gradient Echo
5.0 mm 28.3 ms 4.3 ms 30 deg 104167 Hz 7s
π-type Low Pass Filter Cut off = 0.8 MHz 8m
Strain Gauge Difference Bridge Amp. Amp. Box(Aluminum H60xW90xD125mm)
FSE ETL16 MSME Fast Spin Echo Spin Echo 100 mm 256 0.391 mm/pixel 1 1 1.0 mm 5000 ms 2500 ms 14 ms 25 ms 180 deg 90 deg 50000 Hz 1 m 20 s 10 m 40 s
gantry and it was scanned as a reference image. Subsequently, the phantom was never moved. (2) The micromanipulator without wires was installed beside the phantom, as shown in the upper-left side of Fig. 5. Strictly speaking, the micromanipulator and the wires until the amplifier box were installed however the rest wires or amplifier box were not installed. The needle adjustment mechanism was approximately 2 cm away from the surface of the phantom. The glass needle was not attached in this experiment. (3) All the wires were connected to the micromanipulator, but
the piezo amplifier and strain amplifier were not powered. (4) The piezo amplifier and strain amplifier were powered, but the feedback was not activated. (5) The feedback was activated. (6) The reference image was scanned again to confirm that the phantom was not moved. To test the noise from the MRI to the strain amplifier, its output was measured by an oscilloscope, Iwatsu’s DS-8812P, while the MRI did not scan and while the MRI scanned with the three protocols mentioned above. Fig. 6 shows the raw images of the phantom scanned with FSE ETL16 for the five conditions, i.e., from (1) to (5). In addition, it shows the difference in the images as compared to the reference image. In the case of “(1) Phantom (no manipulator),” the difference between the two reference images is shown. Small dots are found in the raw images for the conditions (2) to (5). This suggestes that the existence of the micromanipulator caused the artifacts. In the difference images for the conditions (2) to (5), no shift or distortion was found. In the raw images for the conditions (3) to (5), the signal from water became weaker and the noise became stronger in comparison to (1) and (2). The images scanned with other protocols showed similar tendencies. Fig. 7 shows the signal-to-noise ratio (SNR) of each image of each protocol in order to know how the introduction of the micromanipulator affects the image quality of the MRI. The signal is defined as the average of the pixel values within the lower-right circle, and the noise is defined as the standard deviation of that area. In Fig. 7, the SNR of the reference images for the conditions (1) and (6) are shown in advance. Fig. 7 suggests that SNR varies for the same condition. However, the existence of wires and not the micromanipulator, significantly debased the SNR. The changes in the SNR caused by power-on or feedback were considerably smaller than those caused by wiring. Fig. 8 shows the output of the strain amplifier measured by the oscilloscope while the MRI did not scan (left) and while the MRI scanned with FLASH (Gradient Echo). The vertical and horizontal scales for the absence of MRI scanning are 50 mV and 0.1 ms, respectively. Even while the MRI did not scan, the strain amplifier outputted noise of the order of tens of millivolts. The vertical and horizontal scales of the FLASH scanning are 500 mV and 1 ms, respectively. While the MRI scanned, the strain amplifier outputted significant noise of the order of ones of volts. The other protocols
467
The 29th Annual International Conference of the IEEE EMBS, ThC09.4, pp. 465-458, 2007
12.00
Difference from the Reference Image
FLASH Signal-to-Noise Ratio
Raw Image
(1) Phantom (No Manipulator)
10.00
FSE_ETL16 MSME
8.00 6.00 4.00 2.00
(5) Phantom+Manipulator (Feedback ON)
(4) Phantom+Manipulator (Power ON)
(3) Phantom+Manipulator (Power OFF)
(2) Phantom+Manipulator (No Wire)
(2) Phantom + Manipulator (No Wire)
(6) Phantom (No Manipulator)
(1) Phantom (No Manipulator)
0.00
Fig. 7. Signal to Noise Ratio of MRI vs. introduction of micromanipulator
(3) Phantom + Manipulator (Power OFF)
Fig. 8. Noise on Strain amplifier, Left: During the absence of MRI scanning (Vert. Scale = 50 mV, Horiz. Scale = 0.1 ms). Right: During FLASH scanning (Gradient Echo Imaging) (Vert. Scale = 500 mV, Horiz. Scale = 1 ms).
(4) Phantom + Manipulator (Power ON)
(5) Phantom + Manipulator (Feedback ON)
Fig. 6. MRI (FSE ETL16) of phantom vs. the introduction of micromanipulator and its differential image as compared to the reference image
showed a similar tendency. Basically, the micromanipulator does not have to move during the MRI scanning. To avoid non-intentional motion, the micromanipulator should be locked during the MRI scanning, i.e. the PZT amplifier should output constant voltages by open-loop. IV. C ONCLUSIONS AND F UTURE W ORKS
artifacts and significant debasement in the signal-to-noise ratio of the MRI. The debasement in the signal-to-noise ratio was mainly caused by the wiring of the micromanipulator. The results also suggest that the MRI caused noise of the order of ones of volts in the strain amplifier. In practice, this would not be a problem because the micromanipulator does not have to move during the MRI scanning. Improvements are necessary with regard to the wiring of the micromanipulator. The length, shielding, and grounding have to be readjusted. An optical strain gauge (e.g., fiber Bragg grating) provides a definitive solution against noise from the MRI scanning. Lock the micromanipulator during the MRI scanning would be a practical solution. V. ACKNOWLEDGMENTS This research was partially funded by Grants-in-Aid for Young Scientists (#B17700419) from the Japan Society for the Promotion of Science. We wish to thank Mr. Homma and Mr. Numano for their contributions to this research study.
In this paper, we introduced our MRI-compatible micromanipulator, to provide medical and biological scientists with the ability to concurrently manipulate and observe micronscale objects inside the MRI gantry. First, we discussed the design details required for the operation inside the MRI gantry and the manipulation of a micron-scale object. Second, we performed a preliminary test to confirm its compatibility with a 2.0-Tesla MRI. The results suggest that it caused no shift or distortion, however it caused small 468
R EFERENCES [1] L. Ciobanu, et al., ”Magnetic resonance imaging of biological cells,” Progress in Nuclear magnetic Resonance Spectroscopy, Vol. 42, 2003, pp. 69-93. [2] T. Tanikawa, T. Arai, N. Koyachi, ”Development of Small-sized 3 DOF Finger Module in Micro Hand for Micro Manipulation,” Proc. IEEE/RSJ IROS 1999, pp. 876-881, 1999. [3] K. Chinzei, R. Kikinis, F.A. Jolesz, ”MR Compatibility of Mechatronic Devices: Design Criteria,” Proc. MICCA 1999, pp. 1020-1031.
