IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 16, NO. 1, FEBRUARY 2008
91
MR_CHIROD v.2: Magnetic Resonance Compatible
Smart Hand Rehabilitation Device for Brain Imaging
Azadeh Khanicheh, Dionyssios Mintzopoulos, Brian Weinberg, A. Aria Tzika, and Constantinos Mavroidis
Abstract—This paper presents the design, fabrication, and
testing of a novel, one degree-of-freedom, magnetic resonance compatible smart hand interfaced rehabilitation device
(MR_CHIROD v.2), which may be used in brain magnetic resonance (MR) imaging during handgrip rehabilitation. A key
feature of the device is the use of electrorheological fluids (ERFs)
to achieve computer controlled, variable, and tunable resistive
force generation. The device consists of three major subsystems:
1) an ERF based resistive element, 2) handles, and c) two sensors,
one optical encoder and one force sensor, to measure the patient
induced motion and force. MR_CHIROD v.2 is designed to resist
up to 50% of the maximum level of gripping force of a human
hand and be controlled in real time. Our results demonstrate that
the MR environment does not interfere with the performance of
the MR_CHIROD v.2, and, reciprocally, its use does not cause
fMR image artifacts. The results are encouraging in jointly using
MR_CHIROD v.2 and brain MR imaging to study motor performance and assess rehabilitation after neurological injuries such
as stroke.
Index Terms—Electrorheological fluids, functional magnetic resonance imaging (fMRI), motor control, magnetic resonance (MR)compatible devices.
I. INTRODUCTION
UNCTIONAL magnetic resonance imaging (fMRI) has
been widely used in studying human brain mechanisms
controlling voluntary movement and reorganization of this
system in response to neurological injuries such as stroke [1].
Normalization and repeatability of commonly utilized functional activation tasks in terms of speed, kinematics, forces,
etc., are difficult to accomplish especially in impaired subjects.
Therefore, fMRI compatible devices are required in order to
study the motor performance in controllable dynamic environments which may provide important insights into human
motor control and related dysfunctions, and enable therapists
to quantify, monitor, and improve physical rehabilitation. This
motivates the development of robotic and mechatronic interfaces, which can control and measure force during movements
F
Manuscript received May 13, 2007; revised September 7, 2007. This
work was supported by National Institutes of Health (NIH) under Grant
5R21EB004665-02.
A. Khanicheh, B. Weinberg, and C. Mavroidis are with the Department
of Mechanical and Industrial Engineering, Northeastern University, Boston,
MA 02115 USA (e-mail: azadeh@coe.neu.edu, weinberg@coe.neu.edu;
mavro@coe.neu.edu).
D. Mintzopoulos and A. A. Tzika are with the Department of Surgery,
Massachusetts General Hospital and Shriners Burns Institute, Harvard Medical
School, Boston, MA 02114 USA (e-mail: dionyssi@nmr.mgh.harvard.edu;
aaria_tzika@hms.harvard.edu).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TNSRE.2007.910286
in humans and quantify the kinematics of motor task performance while performing fMRI [2]. However, the development
and use of magnetic resonance (MR)-compatible devices is a
very challenging task due to the nature of the modality which
uses high magnetic fields, fast-switching magnetic field gradients and radiofrequency pulses, as well as being very sensitive
to external noise.
Several examples of MR-compatible robotic devices have
been demonstrated for surgical applications [3]–[5]. However,
fMRI sequences are even more sensitive than anatomical MRI
sequences, as they measure heterogeneities of the magnetic
field. Heterogeneities of the magnetic field created by the
presence of a mechatronic device lead to time-varying image
distortion and signal loss and can be falsely associated with
neural activation. Additionally, the fast switching gradients
and high static magnetic field required for functional imaging
increase electromagnetic compatibility constraints. One should
take into account that, when interacting with human subjects
during functional imaging, there is a limited space within the
scanner bore (typical diameter of 60 cm) and the subject must
be comfortable with the mechatronic device while performing
the task. In addition, using the device should not create any
movement artifacts. In spite of all these difficulties and challenges, MR-compatible robotic and mechatronic interfaces
have been introduced in the past few years [5]. A nonportable,
haptic interface compatible with fMRI that uses a hydraulic
master–slave system to power the robot remotely, from the
outside of a MR room, was presented in [6] and [7]. A fMRI
compatible virtual reality system that included a data glove
equipped with tactile feedback was developed in [8]. The glove
was able to collect data from the patient’s hand motions and
transfer information through tactile feedback. However, it was
not able to apply forces and torques required in exercises of
motor rehabilitation. A robotic arm compatible with fMRI that
uses two-way, air-driven cylinders, servo valves, and linkages
has been presented to study brain regions involved in processing reach errors [9]. The force on the handle of the robot is
controlled by the inputs of the servo valves. A haptic interface
device for fMRI studies has been presented in [10]. The device
uses two coils that produce a Lorentz force induced by the large
static magnetic field of the MR scanner. Devices utilizing this
type of force actuation are very sensitive to their placement
and orientation within the MRI scanner’s magnetic field, which
significantly limits the range of motion. Also, several fMRI
compatible force sensing systems have been developed to measure forces exerted by subjects in their upper extremities for
motor function studies [11]–[13]. These systems use sensors to
quantify forces and they do not utilize any actuators to apply
forces.
1534-4320/$25.00 © 2008 IEEE
92
IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 16, NO. 1, FEBRUARY 2008
The objective of this study was to develop a portable, computer controlled, variable resistance, MR-compatible hand device to evaluate activation in motor cortex regions during handgrip rehabilitation. Squeezing was also of interest because of
its potential as a probe of motor function in patients with neurological disorders. The device is called MR_CHIROD (magnetic resonance compatible smart hand interfaced rehabilitation
device). A key feature of the device is the use of electro rheological fluids (ERFs) to achieve tunable, computer controlled,
resistive force generation. ERF are fluids that experience dramatic changes in rheological properties, such as viscosity or
yield stress, in the presence of an electric field. Using the electrically controlled rheological properties of ERFs, compact resistive elements capable of generating tunable and controllable
forces can be developed. Our first prototype, the MR_CHIROD
v.1 incorporated a rotary ERF brake with a pincer handle motion [14], [15]. While the device was shown to be MR-compatible and was the first system that utilized ERF in MR environment, its initial evaluation revealed that the design had to be improved to reduce friction and weight, augment its ergonomics,
and eliminate undesired eddy current effects. Our experience
with the first prototype and an increased understanding of the
MR environment led to the design and fabrication of a second,
completely different prototype called MR_CHIROD v.2 that incorporates a linear ERF damper with a linear handle motion.
This paper focuses on the design, fabrication, and MR compatibility testing of the MR_CHIROD v.2. Our study demonstrates
that there is neither an effect from the MR environment on the
performance of the MR_CHIROD v.2 (including its ERF resistive element, position and force sensors), nor significant degradation on MR images by the introduction and operation of the
MR_CHIROD v.2 in the MR scanner.
II. MR_CHIROD V.2
The MR_CHIROD v.2 consists of three major subsystems:
1) an ERF linear damper; 2) handles, and 3) two sensors, one
optical encoder and one force sensor, to measure the patient induced motion and force (Fig. 1).
The MR_CHIROD v.2 incorporates a linear handle motion
with a linear ERF damper. A linear ERF damper was chosen
over a rotary ERF brake (which was used in the MR_CHIROD
v.1) to reduce the possibility of creating eddy currents, which
are created due to the moving conductive electrode in the rotary
damper. In addition, the linear ERF damper is capable of generating larger resistive forces compared to the rotary ERF damper
of similar size. By increasing the resistive force output of the
ERF damper, we were able to eliminate the gears. Thus, we were
able to design a simpler and stronger device, with fewer moving
parts. Furthermore, the elimination of gears reduced considerably the mechanical friction of the device which means much
smaller zero voltage resistance.
Since a linear ERF damper is used in MR_CHIROD v.2, the
mechanical interface with the patient’s hand is now a linear
handle that allowed us to incorporate an adjustable handle mechanism to fit different hand sizes. Furthermore, the linear handle
motion allows direct force measurement by utilizing a linear
force sensor immediately after the handle, in contrast with what
was utilized in MR_CHIROD v.1 where the output rotary torque
Fig. 1. MR_CHIROD v.2 with linear handle motion and linear ERF damper.
was inferred by the measurement of a linear force sensor [14],
[15].
The MR_CHIROD v.2 is configured to rest next to the person,
in a desktop-like configuration, so that no weight would be felt
by the individual while exercising. Furthermore, a weight constraint was imposed on the design so that, in a future home based
training, the device could be easily transported from a person’s
home to a medical experts’ office and then to a MRI facility. The
total weight of the final prototype is only 1.1 kg which makes it
easily transportable from one place to another.
The maximum grip strength that can be generated by the
, and in females
hand in males is 463.95 N
[16]. All components were de279.35 N
signed so that the device is capable of applying 200 N of force
at the human operator’s hand holding the device’s handles
(approximately 50% of a healthy hand’s gripping force). In
general, the MR_CHIROD v.2 was designed to: 1) be portable,
2) resist grasp and release, c) include a return mechanism
for hand opening and finger extension, d) provide accurate
measurement and control of the grip exercise, e) possess an
adjustable handle mechanism to fit different hand sizes, f) be
MR-compatible, g) be of compact size due to limited space
within the MR scanner bore, h) be easy to install and remove
, and i) not cause pain or injury to
in MR scanner
the patient.
A. ERF Resistive Element Design
The unique controllable variable resistance of MR_CHIROD
v.2 is achieved through an ERF linear damper that connects to
KHANICHEH et al.: MR_CHIROD V.2: MAGNETIC RESONANCE COMPATIBLE SMART HAND REHABILITATION DEVICE
93
Fig. 2. CAD drawing of linear ERF damper.
the handles in line with the direction of motion. Using the electrically controlled rheological properties of ERFs, a compact
resistive element capable of generating high resistive and controllable forces, was developed (Fig. 2). The linear ER damper
consists of a piston rod and a piston head in concentric cylinder
tubes which also act as electrodes, and it is fully filled with ER
fluid. The inner and outer cylindrical electrodes are separated
only by a thin layer of fluid and applying an electric field across
the gap alters the fluid’s properties, such as viscosity and yield
stress. By the motion of the piston, the ER fluid flows through
the gap between inner and outer cylinders and there will be a
pressure drop between upper section and lower section inside
the annular gap and the fluid will move with a certain flux or velocity. In the absence of electric fields, the ER damper produces
the damping force only caused by the fluid resistance. However,
if a certain level of electric field is supplied to ER damper, the
ER damper produces additional damping force due to the yield
stress of the ER fluid. This damping force of the ER damper can
be continuously tuned by controlling the intensity of the electric
field.
B. Handles
The handles are the haptic interface for the operator. They are
designed for a linear motion and the overall stroke of the device
can be adjusted using shaft collars for operators of varying hand
sizes. The hand grip assembly allows one degree of freedom.
The thumb grip is the stationary grip and the other four fingers
squeeze the moving handle. The handles are made of Accura 40
plastic, which is MR-compatible.
C. Sensors
Two primary sensors were employed into the device design.
The first is an optical rotary encoder (Fig. 1) to measure positions, velocities, and accelerations of the hand. The rotary encoder is modified for linear use through a rack and pinion gear
and gives a direct reading of the handles position. The sensor,
which has been included into the present design, is a Renco Low
Profile Encoder (RCML15) with a 1024 resolution. The rack and
pinion gear are made of plastic.
The second sensor is a linear force sensor for measuring the
gripping strength of the patients’ hand and for closed-loop control of the ERF resistive element. The INTERFACE force sensor
Fig. 3. MR_CHIROD v.2’s force versus position diagram during squeezing in
one stroke for different ER voltage activation.
(aluminum load cell, SML-100) links the piston rod of the ERF
linear damper to the moving handle, loading the sensor in tension or compression (Fig. 1).
III. PERFORMANCE OF MR_CHIROD V.2
The MR_CHIROD v.2 resists hand squeezing and is remotely
adjusted to different levels of resistive forces. It is operated by
closing (squeezing) its handles, using four fingers on the moving
handle and the thumb on the fixed handle of the device. The device opens with no resistance by utilizing elastic returns that
would provide the return force and can be adjusted in length
and stiffness. The position of the handles and the grip strength
are monitored and recorded at 200 Hz while the test is being
performed. By knowing the position of the handle and reaction
force of the ERF damper at small time steps, the dynamic performance can be captured and analyzed. The entire control process
is accomplished in milliseconds by a computer where information can be gathered in fine detail. All of this data is saved on
the computer so that insights can be gathered as to what is happening with the user.
Tests were performed outside of the MR scanner to verify
the capabilities of the MR_CHIROD v.2. The handgrip
strength during generation of controllable levels of force during
squeezing was obtained. By applying the voltage through the
ERF resistive element, the resistive force generated by the
ERF element is increased and more grip strength is needed to
perform the exercise with the device. Fig. 3 shows the handgrip
force during squeezing versus position for one stroke when the
ERF damper is activated with different voltages. The stroke
of the squeezing is 4.5 cm. An average force output of the
MR_CHIROD v.2 at each voltage (up to 4 kV) is plotted in
Fig. 4. The results therefore suggest that the proposed system is
capable of generating the desired force up to 200 N, depending
on the level of applied voltage to the ERF resistive element.
A nonlinear proportional integral derivative (PID) force controller was developed and implemented for the MR_CHIROD
v.2. This controller was based on our previous torque controllers
for ERF resistive elements, as described in [17] and [18]. Fig. 5
shows the measured handgrip force for different desired input
94
IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 16, NO. 1, FEBRUARY 2008
Fig. 4. MR_CHIROD v.2’s force versus voltage diagram.
Fig. 5. Handgrip force measurement using MR_CHIROD v.2 for desired input
forces.
forces in a repetitive open and close operation of the device.
The detailed description of the device closed loop force control
is beyond the scope of this paper.
Fig. 6. MR_CHIROD v.2 being used inside the scanner during fMRI testing.
The EPI used was the same gradient-echo EPI protocol using
parallel-imaging (GRAPPA) currently used for an fMRI experi,
ment. Acquisition parameters were:
, voxel size
3.0 mm,
GRAPPA
128 128 acquisition matrix /200 mm 200 mm FOV, 48
slices (5% skip), 90 flip angle. Maximum gradient is approximately 40 mT/m, while the operational EPI readout gradient is
approximately 25 mT/m, allowing nominal 128-step acquisition
matrix with 1.562 kHz/px bandwidth.
The power supply (Trek 610C) that supplied a voltage to the
MR_CHIROD v.2 and the computer that reads the sensor data
and control the device were located outside of the MRI room. To
minimize electromagnetic interface (EMI), the wires were properly shielded and cables of appropriate size and impedance were
used. The cables were shielded by a braided copper meshing and
passed through the penetration panel into the shielded MR room.
The low amperage current required to activate the ERF ensures
that the electromagnetic interference is kept to a minimum, both
in the cables and the ERF components. The low amperage current also results in a low power consumption in the cables and
ERF components which avoids increased temperature in the test
device.
IV. MR COMPATIBILITY TESTS OF MR_CHIROD V.2
The following experiments were performed to demonstrate
the compatibility of the MR_CHIROD v.2 on a 3-T Siemens
Trio whole-body MRI equipped with 12-channel Siemens TIM
head coil, of the Athinoula A. Martinos Center for Biomedical
Imaging located at the Massachusetts General Hospital, Boston.
Tests were performed to demonstrate that the strong magnetic
field of the scanner and sensitive functional imaging sequences
does not affect the performance of the MR_CHIROD v.2 in the
MR environment. The performance of the encoder and force
sensor was evaluated individually, and then the complete hand
device was tested (Fig. 6). In addition, thorough testing was
performed to study possible degradation of the MR images
during the operation of the MR_CHIROD v.2 during functional
imaging.
Images were collected using gradient-echo echo-planar
imaging (GE-EPI), commonly used in functional imaging, as it
is the sequence to be used with an operating MR_CHIROD v.2.
A. Performance Evaluation of the Encoder in MR Environment
The optical rotary encoder was modified for linear position
measurement through rack and pinion gear. The device was
placed about 80–90 cm from the isocenter of the magnet, where
the hand of the subject lying in the supine position will be located, to evaluate its performance in the MR environment. Then
the handles were opened and closed and the encoder’s direct
reading of the handles’ positions in real time were recorded in
the computer. Fig. 7 shows the plot of the position versus time
when the device was outside the MR environment and when it
was in the MR scanner, during EPI scans. The results represent
that the optical encoder works with no problem in the MR environment. It needs to be noted that in the graphs of Fig. 7 the
frequency of opening and closing of the handles is slightly different as the two graphs correspond to two different tests performed by a human in two different instances and hence it is
virtually impossible to repeat the exact same frequency.
KHANICHEH et al.: MR_CHIROD V.2: MAGNETIC RESONANCE COMPATIBLE SMART HAND REHABILITATION DEVICE
Fig. 7. Optical encoder MR compatibility, outside the MR scanner (left), inside
the scanner (right).
Fig. 8. Force sensor MR compatibility.
Fig. 9. Effect of MR_CHIROD v.2 on phantom images: (a)–(e). (a) Control.
(b) MR_CHIROD v.2 with ERF not activated. (c) MR_CHIROD v.2 with ERF
activated at 1.5 kV. (d) MR_CHIROD v.2 with ERF not activated and squeezing
the device with 45 N. (e) MR_CHIROD v.2 with ERF activated at 0–2 kV and
squeezing the device with 90 N. (a)–(d) Subtraction of the control (a) from
(b)–(e).
B. Performance Evaluation of the Force Sensor in MR
Environment
To evaluate the performance of the force sensor in the MR
environment, known weights were placed on it while the sensor
was placed about 80–90 cm from the isocenter of the magnet,
where the hand of the subject lying in the supine position will
be located. The electrical output of the sensor in mV/V was read
by a computer, filtered, and then converted to the corresponding
force applied on the sensor using the output-load calibration
equation provided by the manufacturer. A series of tests were
performed for different weights, while performing EPI scans.
Fig. 8 shows the sensor force measurement for a set of known
weights. The results of Fig. 9 show that the force sensor has the
same accuracy inside and outside the MR environment.
95
C. Performance Evaluation of the ERF Damper in MR
Environment
To evaluate the effects of the strong magnetic field and EPI
scans on the ERF in the MR environment, the forces generated
from the ERF linear damper at different voltages were acquired
while the device was placed about 80–90 cm from the isocenter
of the magnet. Fig. 4 shows the force generated at different voltages, using the MR_CHIROD v.2 outside MR environment and
inside MR scanner, during functional imaging. The results show
that the ER linear damper performs without any problem in the
MR environment.
D. Effects of the Operation of MR_CHIROD v.2 on MR Images
To ensure that using the MR_CHIROD v.2 had no degradation in the MR images and to visualize the possible artifacts
caused by the introduction of the device in the magnetic field,
a series of phantom and human tests were conducted on the assembled MR_CHIROD v.2 using the GRAPPA EPI sequence
used for human imaging. Phantom and human control images
were acquired in the absence of MR_CHIROD v.2.
1) Phantom Tests: The Compatibility of the MR_CHIROD
v.2 was first examined by computing the variation of signal to
noise ratio (SNR) in a phantom, placed within the MR scanner.
The imaging object was a cylindrical phantom filled with a
solution of 1.24 g NiSO4 6 H2O/2.62 NaCl per 1000 g H O.
The control image was acquired without any device. Then,
the MR_CHIROD v.2 was attached to the scanner table in
the approximate position, where the hand of the subject lying
in the supine position would be located while squeezing the
handles and performing the rehabilitation exercises (about
80-90 cm from the isocenter of the magnet or about 500 mT
line). Phantom EPI images were acquired first simply in the
presence of MR_CHIROD v.2 with ERF not activated and ERF
activated at 1.5 kV [Fig. 9(b)–(c)], and second with a person
near the magnet operating the device and squeezing the handles
with the device generating maximum 45 N desired resistive
force (ERF is not activated) and 90 N desired resistive force
(ERF is activated between 0 and 2 kV), [Fig. 9(d)–(e)]. The
phantom was not moved during the entire tests.
Signal-to-noise ratio (SNR) was calculated in the image domain: the signal was calculated from the mean of a region of interest (ROI) encircling the center of the image and the width of
the noise was computed by choosing an ROI near the image edge
outside the object [19]. Image SNR was then computed from:
, where is the mean signal intensity in the cen, is the width of the histogram of the intensities
tral ROI and
of the background ROI, or the empirical width of the noise profile. For the noise calculation, an ROI was drawn near the edge
of the volume, and the intensity profile of that ROI was fitted.
The MRI background is known to follow a Rician distribution
[20]; our images, however, are the result of a GRAPPA EPI acquisition. The GRAPPA reconstruction is comprised of mathematical operations that may affect the distribution of background intensities in the image domain; empirically it was found
that histograms of background intensities (“image noise”) are
fitted better with Gaussian distribution than with a Rayleigh
distribution (Fig. 10) [20]. The noise profiles in all images reported below are overlapping each other, i.e., are drawn from
96
IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 16, NO. 1, FEBRUARY 2008
Fig. 10. Noise profile histogram (from image background), Gaussian fit ( =
39, = 9), histograms are normalized to area 1.
TABLE I
SIGNAL STABILITY AND SNR FOR PHANTOM IMAGES
ACQUIRED WITH GRAPPA EPI
Fig. 11. Operation of the MR_CHIROD v.2 does not affect EPI human images. (a) Without the MR_CHIROD v.2 in the scanner room (control). (b) Subject squeezing handles of MR_CHIROD v.2 continuously with constant 60 N
resistive force (ERF activated, maximum voltage approximately 2 kV). (c) Difference of (b) minus (a); (d) the histogram of (c), calculated for an ROI encompassing brain matter N (0; = 59). Images (a), (b) are in arbitrary intensity
units, with maximum intensity approximately 2000, or approximately 33 . All
volumes were coregistered to the control volume prior to subtraction in order to
minimize the effects of motion between scans. (e) Image noise histogram profiles: (– – –), control image without RF excitation, volunteer in scanner but
MR_CHIROD v.2 not in the scanner room; (– –), background noise of EPI
image (a) with volunteer in scanner but MR_CHIROD v.2 not in the scanner
room; (– –), noise of (b), volunteer in scanner squeezing at constant resistive
force 60 N All histograms of (e) are fitted with a Rician distribution using nonlinear least squares. The noise histograms corresponding to images (a) and (b)
show no statistical difference (P Value = 0:64, F-Test of norms).
0
one and the same probability distribution. For the signal calculation, an ROI was drawn around the center of images. ROIs
were drawn from the same plane near the “sweet spot” of the
coil at the magnet isocenter. The same coordinates (pixel locations) were used for all ROIs. The phantom signal remained
remarkably stable despite the introduction and operation of the
MR_CHIROD v.2 and activation of the ERF. Signal means of
the central ROI deviated less than 1% of the overall mean for
all five acquisitions. Signal stability and SNR values upon the
introduction and operation of MR_CHIROD v.2 into the MR
scanner are reported in Table I. In all cases, simple paired t-tests
) failed
comparing condition to control (two-tailed,
to reach significance, with P-values ranging from 0.06 to 0.6.
(Note, the specific numbers on the table correspond to a 3-D
signal ROI spanning 10 slices, but averages across fewer or more
slices gave similar results). Results show that in all cases, the
loss of SNR observed was not significant regardless of the introduction and operation of MR_CHIROD v.2 and activation of
the ER fluid.
2) Human Tests: The successful phantom tests were completed by tests conducted with a human subject performing
handgrip exercises with MR_CHIROD v.2. Human images
were acquired with a volunteer lying in the normal supine
position in the scanner. Padding was used to immobilize the
volunteer’s head inside the phased-array head-coil. Extra foam
padding was used at the elbow, such that the arm is padded
against the magnet bore wall on one side and against the volunteer’s own body on the other side to minimize elbow flexion
and reflexive movement of the upper body and the head as
a person squeezes the handles of the MR_CHIROD v.2. EPI
human images were acquired for 1) control (no MR_CHIROD
v.2 in the scanner), 2) with the MR_CHIROD v.2 attached to
the scanner table in the position, the volunteer is squeezing the
handles and performing handgrip exercise with 60 N constant
resistive force (ERF activated between 0 and 2 kV). Fourteen
EPI volumes were collected, the first four volumes serving
to allow the magnetization to reach equilibrium and being
discarded in the analysis. A typical control EPI image and a
similarly acquired EPI image where the MR_CHIROD v.2 was
operated with 60 N are presented in Fig. 11(a) and (b). Their
difference and the histogram of their difference are shown in
Fig. 11(c) and (d). Background noise distributions [Fig. 11(e)]
, F-Test). Mean values
show no statistical difference (
and standard errors of SNR estimated from signal ROIs near
and
the center of the EPI images were
from Fig. 11(a) and (b), respectively, and were not statistically
; two-tailed t-test). The volumes were
different (
coregistered to each other using FSL tools before subtraction
in order to minimize the effects of motion between scans. The
KHANICHEH et al.: MR_CHIROD V.2: MAGNETIC RESONANCE COMPATIBLE SMART HAND REHABILITATION DEVICE
results represents that the introduction of MR_CHIROD v.2
did not affect fMR image quality.
V. CONCLUSION
A 1DOF fMRI compatible force feedback device for hand
rehabilitation that generates computer controlled resistive forces
and measures the position, velocity, acceleration, and gripping
force of the hand has been presented. These properties were
achieved by utilizing ERFs, which can produce large resistive
forces upon activation with an electric field; an optical encoder;
and a force sensor.
Tests with the prototype showed that it was able to provide
200 N resistive forces for activation voltages up to 4 kV,
which is approximately 50% of the maximum level of gripping
force that a human hand can apply. The series of experiments
described here demonstrate that the scanning environment did
not interfere with the ability to generate controllable resistive
forces with ER fluid damper, and accurately record handle
motion and handgrip strength with utilizing the optical encoder
and force sensor. Furthermore, tests were performed both with
a phantom and a human subject and our results demonstrated
that the MR_CHIROD v.2 had no degradation effect in the MR
images and its use does not cause fMR image artifacts.
We envision that the MR_CHIROD v.2 will be useful in fMRI
motor studies to achieve a more detailed understanding of the
relationship between movement kinematics, the force exerted
and features of brain activation in the healthy and injured subjects. The device could also permit evaluating the efficacy of
therapeutic interventions, for example physical rehabilitation in
stroke patients, using a combination of fMRI and online dynamic and kinematic data.
97
[9] J. Diedrichsen, Y. Hashambhoy, T. Rane, and R. Shadmehr, “Neural
correlates of reach errors,” J. Neurosci., vol. 25, pp. 9919–9931, Oct.
2005.
[10] R. Riener, T. Villgrattner, R. Kleiser, T. Nef, and S. Kollias, “FMRICompatible electromagnetic haptic interface,” in 2005 27th Annu. Int.
Conf. IEEE Eng. Med. Biol. Soc., China, 2005, pp. 7024–7027.
[11] J. Z. Liu, T. H. Dai, T. H. Elster, V. Sahgal, R. W. Brown, and G. H.
Yue, “Simultaneous measurement of human joint force, surface electromyograms, and functional mri-measured brain activation,” J. Neurosci. methods, vol. 101, pp. 49–57, Aug. 2000.
[12] J. Hidler, T. Hodics, B. Xu, B. Dobkin, and L. G. Cohen, “MR compatible force sensing system for real-time monitoring of wrist moments
during fmri testing,” J. Neurosci. Methods, vol. 155, pp. 300–307, Sep.
2006.
[13] S. C. Cramer, R. M. Weisskoff, J. D. Schaechter, G. Nelles, M. Foley, S.
P. Finklestein, and B. R. Rosen, “Motor cortex activation is related to
force of squeezing,” Human Brain Mapp., vol. 16, pp. 197–205, Aug.
2002.
[14] A. Khanicheh, A. Muto, C. Triantafyllou, L. Weinberg, L. Astrakas, A.
Tzika, and C. Mavroidis, “MR compatible ERF driven hand rehabilitation device,” in 2005 IEEE 9th Int. Conf. Rehabil. Robot., Chicago,
IL, United States, 2005, pp. 7–7.
[15] A. Khanicheh, A. Muto, C. Triantafyllou, B. Weinberg, L. Astrakas,
A. Tzika, and C. Mavroidis, “FMRI-compatible rehabilitation hand device,” J. Neuroeng. Rehabil., vol. 3, pp. 24–124, 2006.
[16] V. Mathiowetz, N. Kashman, G. Volland, K. Weber, M. Dowe, and S.
Rogers, “Grip and pinch strength: Normative data for adults,” Arch.
Phys. Med. Rehabil., vol. 66, pp. 69–74, Feb. 1985.
[17] M. A. Vitrani, J. Nikitczuk, G. Morel, C. Mavroidis, and B. Weinberg, “Torque control of electrorheological fluidic resistive actuators
for haptic vehicular instrument controls,” Trans. ASME J. Dynamic Systems, Meas. Control, vol. 128, pp. 216–226, 2006.
[18] J. Nikitczuk, B. Weinberg, and C. Mavroidis, “Control of electro-rheological fluid based resistive torque elements for use in active rehabilitation devices,” Smart Mater. Struct., vol. 16, pp. 418–428, 2007.
[19] M. J. Firbank, A. Coulthard, R. M. Harrison, and E. D. Williams, “A
comparison of two methods for measuring the signal to noise ratio on
MR images,” Phys. Med. Biol., vol. 44, pp. N261–N264, Dec. 1999.
[20] H. Gudbjartsson and S. Patz, “The rician distribution of noisy MRI
data,” Magn Reson Med, vol. 34, pp. 910–914, Dec. 1995.
REFERENCES
[1] J. B. Rowe and R. S. Frackowiak, “The impact of brain imaging technology on our understanding of motor function and dysfunction,” Curr.
Opin. Neurobiol., vol. 9, pp. 728–734, Dec. 1999.
[2] N. Hogan and H. I. Krebs, “System and method for medical imaging
utilizing a robotic device, and robotic device for use in medical
imaging,” U.S. Patent 5 794 621, 1998.
[3] K. Masamune, E. Kobayashi, Y. Masutani, M. Suzuki, T. Dohi, H.
Iseki, and K. Takakura , “Development of an mri-compatible needle
insertion manipulator for stereotactic neurosurgery,” J. Img. Guided
Surg., vol. 1, pp. 242–248, 1995.
[4] A. Krieger et al., “Design of a novel MRI compatible manipulator for
image guided prostate intervention,” IEEE Trans. Biomed. Eng., vol.
52, no. 2, pp. 306–313, Feb. 2005.
[5] N. V. Tsekos, A. Khanicheh, E. Christoforou, and C. Mavroidis,
“Magnetic resonance-compatible robotic and mechatronics systems
for image-guided interventions and rehabilitation: A review study,”
Annu. Rev. Biomed. Eng., vol. 9, pp. 351–387, 2007.
[6] R. Gassert, R. Moser, E. Burdet, and H. Bleuler, “MRI/fMRI-compatible robotic system with force feedback for interaction with human motion,” IEEE/ASME Trans. Mechatronics, vol. 11, pp. 216–224, 2006.
[7] R. Gassert, L. Dovat, O. Lambercy, Y. Ruffieux, D. Chapuis, G.
Ganesh, E. Burdet, and H. Bleuler, “A 2-DOF fmri compatible haptic
interface to investigate the neural control of arm movements,” in Proc.
IEEE Intl Conf Robotics Autom., Orlando, FL, 2006, pp. 3825–13825.
[8] J. Ku, R. Mraz, N. Baker, K. K. Zakzanis, J. H. Lee, I. Y. Kim, S. I.
Kim, and S. J. Graham, “A data glove with tactile feedback for FMRI of
virtual reality experiments,” Cyberpsychol. Behav., vol. 6, pp. 497–508,
Oct. 2003.
Azadeh Khanicheh received the B.S. degree in
mechanical engineering from Sharif University of
Technology, Tehran, Iran, in 2000, the M.S. degree
and the Ph.D. degree in mechanical engineering
from Northeastern University, Boston, MA, in
2003 and 2007, respectively. Her doctoral research
investigated the use of magnetic resonance compatible robotic and mechatronic systems in functional
imaging.
She is now a postdoctoral fellow at the Robotics
and Mechatronics Laboratory at Northeastern University. Her areas of research include rehabilitation robotics, force feedback devices, compact actuators, and MR imaging.
Dionyssios Mintzopoulos received the Ph.D. degree
in physics (dynamical chaos theory) from New York
University (NYU), in 2004, and then moved to MR
physics as a fellow to the Department of Neuroscience at NYU.
He is now a fellow at Massachusetts General
Hospital, NMR Surgical Laboratory. He works in
fMRI of stroke rehabilitation and MR imaging.
Dr. Mintzopoulos received a Best Paper Award
(Honorable Mention) at the 2007 meeting of the
International Society for Magnetic Resonance in
Medicine.
98
IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 16, NO. 1, FEBRUARY 2008
Brian Weinberg received the B.S. degree in mechanical engineering from Rutgers University, New
Brunswick, NJ, and received the M.S. degree in mechanical engineering from Northeastern University,
Boston, MA, where he is currently working toward
the Ph.D. degree.
He is a Senior Research Engineer in the Robotics
and Mechatronics Laboratory at Northeastern University. His areas of research include compact actuators and rehabilitation devices.
A. Aria Tzika received the B.Sc. degree with highest
honors from the University of Patras, Greece, and
the Ph.D. degree from the University of California,
Berkeley. Her graduate work was immediately followed by postdoctoral research at the Department of
Radiology, University of California, San Francisco.
She came to Harvard Medical School and the
Department of Radiology, Boston Children’s
Hospital as an Assistant Professor of Radiology
in 1993, and was appointed in the Department of
Surgery Harvard Medical School in 2001 as the
Director of the NMR Surgical Laboratory.
Dr. Tzika was elected as a Radiological Society of North America Scholar in
1995 and the same year she received the first Derek-Harwood Nash Award.
Constantinos Mavroidis received the Diploma in
mechanical engineering from the National Technical University of Athens, Greece in 1988, and
the M.S. and Ph.D. degrees in mechanical engineering/robotics from the University of Paris VI,
France, in 1989 and 1993, respectively.
He is a Professor of Mechanical and Industrial Engineering at Northeastern University, Boston, MA,
since July 1, 2006. He was an Associate Professor
(with tenure) in the same department from January
1, 2004 to June 30, 2006. He was a Tenured Associate Professor at the Department of Mechanical and Aerospace Engineering at
Rutgers University (2001–2004), an Assistant Professor at the same department
(1996–2001), and Visiting Associate Professor of Surgery at Harvard University Medical School (2001–2003). He has authored and co-authored more than
120 journal and conference papers and book contributions.
Dr. Mavroidis is a Fellow of the ASME.