A

atimdd©

K

1

mrcamcmtitFoet(

ei

GT

0d

Journal of Neuroscience Methods 164 (2007) 247–254

MR compatible strain gauge based force transducer

Hiske van Duinen a,b, Marijn Post a,b, Koen Vaartjes a, Hans Hoogduin b, Inge Zijdewind a,b,∗a Department of Medical Physiology, University Medical Center Groningen, University of Groningen, The Netherlands

b BCN-Neuroimaging Center, University Medical Center Groningen, University of Groningen, The Netherlands

Received 13 December 2006; received in revised form 4 April 2007; accepted 3 May 2007

bstract

In order to evaluate brain activation during motor tasks accurately one must also measure output parameters such as muscle force or musclectivity. Especially in clinical situations where the force output can be compromised by changes at different levels of the motor system, it is essentialo standardize the task or force level. We have therefore developed a magnetic resonance compatible force transducer that is capable of recording

ndex finger abduction force and to display the produced force in real-time. This transducer is based on strain-gauges techniques and designed to

easure both small and large forces accurately (range 0.7–60 N) as well as fast force fluctuations. Experiments showed that the MR environmentid not affect the force measurements or vice versa. Although, this transducer is developed for measuring index finger forces, detailed schematiciagrams are provided such that the transducer can easily be adapted for measuring forces of other muscle groups. 2007 Elsevier B.V. All rights reserved.

imraE

e2eettiii2tr

eywords: Force transducer; Strain gauges; fMRI; Force

. Introduction

Intrinsic hand muscles (including the first dorsal interosseususcle, FDI) are often used for neurophysiological studies

egarding motor function. Hand muscles have a large corti-al representation (see for review Porter and Lemon, 1993),nd they are relatively easy to activate by either transcranialagnetic stimulation or peripheral nerve stimulation. A spe-

ial advantage of the FDI is the well defined and almost uniqueechanical function (it is by far the most important muscle for

he index finger abduction force; An et al., 1983); moreover,ts muscle belly is very superficial and thus readily availableo measure muscle activity using electromyography (EMG).urthermore hand muscles are relatively easily to activate with-ut head movements. All these properties make hand muscles,xtremely suitable for studies involving measurements of cor-ical activity, such as functional magnetic resonance imagingfMRI).

For an adequate interpretation of the cortical activity it isssential to record simultaneously muscle activity. However, its challenging to record force or EMG from active muscles dur-

∗ Corresponding author at: Department of Medical Physiology, University ofroningen, A. Deusinglaan 1, 9713 AV Groningen, The Netherlands.el.: +31 50 363 2681; fax: +31 50 363 2751.

E-mail address: i.zijdewind@med.umcg.nl (I. Zijdewind).

fiwe

ugt

165-0270/$ – see front matter © 2007 Elsevier B.V. All rights reserved.oi:10.1016/j.jneumeth.2007.05.005

ng fMRI studies (see Van Duinen et al., 2005). A strong mainagnetic field is present in the scanner room and furthermore,

adio frequency (RF) and magnetic gradient fields (gradients)re induced during scanning. All may influence the force andMG recordings.

A few labs have developed MR compatible force transduc-rs (Cramer et al., 2002; Dettmers et al., 1996; Ehrsson et al.,000; Hidler et al., 2006; Kuhtz-Buschbeck et al., 2001; Liut al., 2000, 2002; Thickbroom et al., 1998). Some transduc-rs use hydraulic or air pressure systems to measure force, andhey are made of non-magnetic synthetic materials. A disadvan-age of these materials is their high compliance, often makingt impossible to measure small muscle forces or fast changesn muscle forces accurately. Studies that do use force measur-ng systems without hydraulic pressure systems (Cramer et al.,002; Ehrsson et al., 2000; Hidler et al., 2006) do not describeheir setup in much detail and it is therefore not possible toeproduce their setup. It was the aim of this study to design anMRI compatible force transducer that was capable of measur-ng small and fast force fluctuations accurately. In addition weant to describe this system in such detail that the system could

asily be adapted by other investigators.

Strain gauges are often used for force transducers that are

sed in a non-MR surrounding. Force transducers based on strainauges are very precise in measuring small and fast forces andhey are mechanically simple to use (e.g. Boonstra et al., 2005).

2 rosci

HfrsmNfoahew

2

2

ilcvcsb

Fgdthnfisfcfom

ctfT((nmaPce

wmittbcC

48 H. van Duinen et al. / Journal of Neu

owever, an environment with a strong magnetic field, radiorequency and changing gradient fields – such as a magneticesonance imaging room – may influence the strain gauge mea-urements. Moreover, transducers with strain gauges are oftenade of ferromagnetic metals which are incompatible with MRI.evertheless, we succeeded in designing an MR compatible

orce transducer with strain gauges. The transducer is capablef measuring small and relatively large forces (range: 0.7–60 N)nd fast force fluctuations accurately. Although this transduceras especially been designed for the index finger, the system canasily be adapted to measure forces of other muscle (groups) asell.

. Material and methods

.1. Force transducer

This force transducer was designed to measure isometricndex finger abduction forces accurately for both small andarge hands in an MR environment. The transducer was used inombination with an instrumentation amplifier (to convert theoltage signal into an optical signal), a power supply, an optical

able, a receiver, and a PC (with data acquisition interface) totore the data. All components will be described in more detailelow.

ig. 1. Photo of a hand holding the force transducer for measuring index fin-er abduction force. The force transducer consists of several components (foretails see text): (A) strain gauges; (B) laminate bar; (C) PVC tube; (D) connec-or between tube and bar; (E) hand grip; (F) titanium ring plus screw to adjusteight of hand grip; (G) screw to adjust angle of the bar; (H) C-shaped con-ector. The subject’s hand is taped to transducer in several ways: (I) the indexnger is taped to the connector; (J) the middle and ring fingers are taped to theubject’s hand; (K) the thumb is taped to the hand as well. The force transducerurther consists of the: L transmitter; M (cable for) power supply; N opticalable for signal transduction. To measure muscle activity simultaneously withorce, one electromyography (EMG) surface electrode is placed on the bellyf the first dorsal interosseus muscle (O) and one electrode is placed on theetacarpophalangeal joint (P).

ljtct(t

smDtPaaaspv

aNnctc

oatr

tin

ence Methods 164 (2007) 247–254

The force transducer consists of several materials (see Fig. 1,lose-up of the force transducer; the numbers in the text refer tohe numbers in the figure). Strain gauges that are compensatedor electromagnetic effects (Fig. 1A; TML® MFLA-5.350-1L;okyo Sokki Kenkyujo Co., Japan) were placed in a full bridgeWheatstone) configuration on an epoxy glass laminated barFig. 1B; Tufnol® 10G/40 20 mm diameter; RS componentsumber: 771-314; RS Components Ltd., United Kingdom). Thisaterial is strong and not easy to distort (flexural strength is

pproximately 600 MPa). Perpendicular to the bar, there is aVC tube (Fig. 1C), which is attached to the bar with a PVClamp (Fig. 1D). To hold the tube comfortably the tube isquipped with a bicycle hand grip (Fig. 1E).

The force transducer can be adjusted to hand size in threeays. First, the grip can slide up and down the PVC tube toove the laminated bar to a height horizontally parallel to the

ndex finger. If the grip is in the correct position the PVC tube isightened by a copper screw in a titanium ring (Fig. 1F). Second,he bar can be turned in relation to the PVC tube to position thear parallel to the index finger. In the correct position this baran be tightened by a brass screw (Fig. 1G). And third, a delrin-shaped connector (Fig. 1H) can slide along the epoxy glass

aminated bar to a position above the proximal interphalangealoint of the index finger. After adjusting the force transducero the subjects’ hand, the index finger is taped to the C-shapedonnector (Fig. 1I), the middle and ring fingers are taped tohe subject’s hand (Fig. 1J), and the thumb taped to the fingersFig. 1K), in order to fixate the position of the hand relative tohe force transducer.

The force applied to the bar will cause an unbalance in thetrain gauge bridge. The output signal is amplified by an instru-entation amplifier (300×; Analog Devices AD8230; Analogevices, USA). The amplifier is placed in a copper box (Fig. 1L)

o shield it from the RF field. The copper box is placed on theVC clamp to keep the distance to the strain gauges as shorts possible (to minimize scanner artifacts). The copper boxlso contains the transmitter system. This system consists ofvoltage-to-frequency converter, a transmitter for the optical

ignal, and a voltage stabilizer (9 V). Fig. 2A shows a sim-lified scheme of the transmitter (see Appendix A for detailedersion).

The power of the bridge and transmitter is supplied by a leadcid rechargeable battery (12 V, 1.9 Ah; Velleman® Components.V., Belgium). This battery, which is not attracted by the mag-et, is placed in a separate aluminum box and connected to theopper box with a wire (Fig. 1M). All electrical connectionso and from the copper box are decoupled with feed throughapacitors to protect the amplifier against RF effects.

A 10 m long glass fiber cable (Fig. 1N) is plugged into theptical transmitter of the box. This cable leaves the scanner roomnd enters the operator room through an RF waveguide to preventhat the radio frequencies leave the scanner room. In the operatoroom, the optical cable is connected to the receiver.

The receiver has an internal power supply which is connectedo the 230 V mains. In the receiver, the frequency-variable signals readapted to a varying voltage signal. Thereafter, the sig-al passes a low-pass filter (−3 dB at 200 Hz) and an amplifier

H. van Duinen et al. / Journal of Neur

Fig. 2. Simplified schematical representation of the transmitter (A) and thereceiver (B). The power of the transmitter is supplied by a lead acid dry bat-tery. The signal derived from the Wheatstone strain gauge bridge is amplified(AD8230, Analog Devices, USA), and converted to a frequency-modulated opti-cal signal (AD654, Analog Devices, USA). This optical signal is transmitted(HFBR1522, Agilent Technologies, USA) via an optical cable to the MR oper-ator room. The optical signal enters the receiver via the optical signal detector(HFBR2522, Agilent Technologies, USA); the signal is then re-converted tovoltage signal, linearly corresponding to the applied force (LM331N, NationalS(S

wfida(wav

mimp

mt

Fsor

as

2

sadttipow

tcts3t(aiass

4vsiec(f

emiconductor, USA), filtered (OP-07, Texas Instruments, USA), and amplifiedOP-07, Texas Instruments, USA). The analogue signal is logged to a PC viapike 2 (CED, Cambridge, UK).

ith a maximum variable gain of 6. Fig. 2B shows a simpli-ed schematic illustration of the receiver (see Appendix B foretailed diagram). The output signal of the receiver is logged inPC with Spike2 version5 for Windows, via an A/D converter

sample frequency: 500 Hz; CED, Cambridge, UK). This PCas connected to a beamer. This beamer projected force data onscreen at the back of the scanner, providing the subjects withisual feedback of the force.

The delay of the setup of transmitter and receiver was deter-ined by measuring the electrical step response; an electrical

nput was given to the transmitter, and the electrical output waseasured at the receiver (see Fig. 3A). The delay, both of the

ositive and the negative step response, was 2 ms (Fig. 3B).This total system can be used to measure forces of other

uscle groups as well. An appropriate force transducer needso be designed, and then the detailed scheme of the transmitter

ig. 3. The electrical step response of the transmitter and receiver. The electricalignal ‘A’ was the input given to the transmitter; the electrical signal ‘B’ was theutput measured by the receiver. The delays of both the positive and negativeesponse were 2 ms (from 10 till 90% of the minimal and maximal value).

awgbFKstbaw1s3

wtvTt

oscience Methods 164 (2007) 247–254 249

nd receiver, shown in Appendices A and B, can be used for theystem.

.2. Tests

To determine the linearity of the force transducer, we recon-tructed a response curve by plotting the measured voltagegainst the force applied to the transducer. We held the trans-ucer upside down when hanging calibrated weights (1–60 N) athe bar. This induced a force in the same direction as the volun-ary index finger abduction force. The calibration was performedn both ascending and descending directions. This procedure waserformed both in and out of the scanner room. The repeatabilityf the force measuring system was analyzed by measuring twoeights 10 times successively.Scanning was performed on a 3 T Philips MRI scanner (Best,

he Netherlands) using the standard eight channel SENSE headoil as receiver and the body coil as RF transmitter. We usedhe following pulse sequence parameters: fast field echo (FFE)ingle shot echo planar images (EPI); 39 slices; slice thickness.5 mm; no gap; field of view 224 mm; scanning matrix 64 × 64;ransverse slice orientation; repetition time (TR) = 2 s; echo timeTE) = 30 ms; minimal temporal slice timing (1957 ms); flipngle 90◦. To test whether the force transducer system wouldnfluence the echo planar images, we scanned both a subject andphantom with and without the presence of the force measuring

ystem (three runs per scanning condition, only four runs of theubject without force transducer).

Furthermore, nine right-handed subjects (mean age:1.1 ± 10.7 years; 7 females) were asked to produce maximaloluntary index finger abductions during scanning (using theame pulse sequence parameters as above), after signing annformed consent. The experiment was approved by the medicalthical committee of the University Medical Center Groningen,onform the standards set out in the Declaration of Helsinki2000). To measure the maximal voluntary contraction (MVC)orce, the force transducer was adjusted to the subjects hand sizes described above. When the force transducer was adjusted,e taped the proximal interphalangeal joint of the index fin-er to the connector at the laminated bar to maintain contactetween the finger and the transducer during relaxation (Fig. 1I).urthermore, the thumb, ring and middle fingers (Fig. 1J and) were also taped to the subjects own hand to prevent the

ubject from repositioning his hand in relation to the forceransducer. The contractions lasted 8 s (four scans), followedy 52 s rest, and were repeated three times. Furthermore, wecquired T1 weighted anatomical images of the entire brainsith the following pulse sequence parameters: 160 slices ofmm in transverse slice orientation, field of view: 256 mm,

canning matrix: 256 × 256, TE: 4.6 ms, TR: 25 ms, flip angle:0◦.

To evaluate the MR compatibility of the force transducer itas checked whether the transducer emitted any RF radiation in

he frequency range of 127.78 ± 0.44 MHz using a protocol pro-ided by the scanner manufacturer (Philips, Best, Netherlands).he measurement to detect ‘spurious’ frequencies was repeated

wice: once with an empty scanner bore (‘baseline’), once with

2 roscience Methods 164 (2007) 247–254

act

fwctaiF

2

tpfwsestf

DeIdmtvdsdTwm((

Airimwup

fdessrs

Fit

3

3

raic

d0tnt2t

out2

aettrmaximal amplitude was reached within 5 ms and within 15 msthe evoked oscillations remained below 5% of the maximalamplitude, which shows that the transducer can measure fast

50 H. van Duinen et al. / Journal of Neu

fully operational transducer on the scanner bed at a locationlose to the usual hand position (‘force transducer’). In all cases,he body coil was used as receiver coil.

Electrically, the rise time of the transmitter and receiver wasast (only a delay of 2 ms); however, we also wanted to knowhether the complete setup, including the transducer itself,

ould measure rapid force fluctuations. This was tested outsidehe MR room by hitting the laminated bar of the transducer withhammer. The dampening of the recorded oscillations gives an

ndication of the frequency response of the force transducer (seeig. 6).

.3. Analyses

Offline, we used Spike 2 version 5 for Windows to analyzehe force data. For the calibration, we determined the mean out-ut per weight, and we measured the standard deviation of theorce during baseline with and without scanning. Furthermore,e determined the maximal voluntary contraction force of nine

ubjects. The highest peak of the three contractions was consid-red the MVC force. Also, the dampening effects of the hammertroke were analyzed, both the time until the oscillation ampli-ude was below 5% of the maximal amplitude and the oscillationrequency.

We used SPM2 (http://www.fil.ion.ucl.ac.uk/spm, Wellcomeepartment of Imaging Neuroscience; Friston et al., 1995) to

valuate the effect of the force transducer on the EPI images.mage series with and without the presence of the force trans-ucer were realigned to the first image of all series to reduceotion artifacts. Furthermore, we applied a regression technique

o the realigned data using the movement parameters as modelectors to remove residual motion effects. The pooled standardeviations (i.e., the averaged standard deviation of the voxel timeeries) were calculated on the residual data. The pooled standardeviations were also determined for the phantom measurements.hereafter, we calculated the difference between the EPI slicesith and without the presence of the force transducer equip-ent, and also the difference between the EPI slices without

before) and without (after) the presence of the force transducerFig. 7).

SPM2 was also used to analyze the functional MRI results.t subject level, the functional images were realigned to the first

mage (see above). Furthermore, the functional images were co-egistered with the individual anatomical image. Together thesemages were normalized to a T1 template. Thereafter, the maxi-

al voluntary contractions were modeled and an contrast imageas calculated. The contrast images of the nine subjects weresed in a one sample t test (second-level analysis; uncorrected< 0.001; minimal cluster size: 10 voxels).

As mentioned above, the measurements to detect spuriousrequencies were repeated twice (‘baseline’ and ‘force trans-ucer’). A mean noise spectrum (in time) was determined forach frequency in the range of 127.78 ± 0.44 MHz. The noise

pectrum of the ‘force transducer’ was divided by the noisepectrum of the baseline. When the MR does not detect spu-ious frequencies in the frequency range of interest, the quotienthould be around 1.

Fa

ig. 4. Calibration of the force transducer, the actual recordings were madenside the scanner room during scanning. The input weights (range: 1–60 N) vs.he output (V).

. Results

.1. Force

Fig. 4 shows the calibration of the force transducer. Linearegression analysis showed a good fit between the used weightsnd the voltage changes of the transducer (R2 = 0.9994), and thentercept of the fitted line was close to zero (0.017 N, with a 95%onfidence interval from −0.023 to +0.057).

At rest, both during scanning and non-scanning the stan-ard deviations of the force baselines were small (0.065 and.062 N, respectively; F(1,44) = −0.920, n.s.). This demonstrateshat the changing magnetic fields inside the scanner room doot influence the force recording. The repeatability of the forceransducer was estimated by measuring two weights (4.9 and7.2 N) 10 times. All 10 measurements varied less than 5% fromhe average value (range: −3.5 to 4.1%).

Fig. 5 shows an example of a maximal voluntary contractionf the index finger in abduction direction. The highest peak wassed to determine the MVC. The mean maximal voluntary con-raction force (the mean of the peak values) of all subjects was5.8 ± 8.6 N.

Fig. 6 shows the result of hitting the force transducer withhammer. The amplitude of the signal was 3.68 V, which

quals 37.38 N (this falls within the range of the MVCs ofhe nine pilot subjects). The frequency response of the sys-em was approximately 250 Hz, which is high compared to theelatively slow force signals. During the hammer stroke the

ig. 5. An example of a maximal voluntary contraction of the right index fingerbductor recorded during MR-scanning.

H. van Duinen et al. / Journal of Neur

Fig. 6. Response as a result of a hammer stroke on the force transducer. Thehorizontal line at 0 V represents the baseline. The horizontal line at 3.68 V rep-resents the maximal amplitude of the hammer stroke. The lines at 0.18 and−lo

fb

3

o(dmtp

aBEbthimwwsFdriE

mmrta

3

wt

Ftac

0.18 represent +5 and −5% of maximal amplitude, respectively. The verticalines represent the start of the response (at 3.570 s) and the end at which thescillations pass the 5% of maximal amplitude for the last time (3.579 s).

orce changes accurately and that the signal quickly returns toaseline.

.2. Brain activity

Fig. 7 shows the mean of the time series of EPI slice 22 ofne subject (upper panel) obtained without (before; left: A), withmiddle: B), and without (after; right: C) the force system. The

ata were realigned to the very first scanned volume to reduceotion artifacts. The signal intensity of the slices was scaled with

heir mean signal, so that they were comparable. The lower grayanel below shows the differences between the mean EPIs with

td1r

ig. 7. The upper panel shows the mean time series of EPI slice 22 without (left: A; bransducer; the lower panel shows the differences between the mean EPI slices with after the presence of the force transducer (both without the presence; A-B). The signaomparison of the slices.

oscience Methods 164 (2007) 247–254 251

nd without the presence of the force transducer (left: before ‘A-’; middle: after ‘B-C’), and the difference between the meanPIs before and after the presence of the force transducer (right:oth without the presence of the force transducer ‘A-C’). If theransducer had had an effect somewhere on the EPIs, it wouldave been visible in this image. Only small differences are vis-ble at the borders of the brain; however, these differences are

ore likely to be movement artifacts, even though the imagesere realigned. The pooled standard deviations of the subjectith and without the presence of the force transducer were not

ignificantly different (13.15 and 11.27, respectively; ANOVA:(1,5) = 3.32, p = 0.13). This was also the case for the pooled stan-ard deviations of the phantom (6.68 ± 0.56 and 6.35 ± 0.35,espectively; ANOVA: F(1,4) = 0.474, p = 0.529). These resultsndicate that the force measuring equipment did not affect thePIs.

Fig. 8 shows the group result of brain activation during theaximal voluntary contractions projected on an individual nor-alized anatomical image. During right contractions with the

ight hand, activation was observed in the left sensorimotor cor-ex, the supplementary motor area, the bilateral premotor areas,nd the right cerebellum.

.3. MR compatibility

The force measuring equipment did not affect the EPIs, asas confirmed by the test for spurious frequencies. Fig. 9 shows

he results of this test. The test revealed no differences between

he empty scanner bore and the bore containing the force trans-ucer at the usual hand location in the frequency range of27.78 ± 0.44 MHz, as dividing ‘force transducer’ by ‘baseline’esulted in values around 1.

efore), with (middle: B), and without (right: C; after) the presence of the forcend without the presence of the force transducer (A-B and B-C), and before andl intensity of the slices was scaled to the mean signal intensity to allow a better

252 H. van Duinen et al. / Journal of Neuroscience Methods 164 (2007) 247–254

Fig. 8. Brain activation during maximal force production. Group brain activity (n = 9; one sample t test, uncorrected p < 0.001) during maximal force production,projected on the normalized anatomical image of a single subject. The numbers in the upper left corner indicate the Talaraich Z coordinate in MNI space. The numbersin the brain refer to specific activated areas—1: left sensorimotor cortex, extending inpremotor cortex; 5: right cerebellum, lobule VI; 6: right cerebellum, lobule VIII. Thand/or anatomical image acquisition.

Fig. 9. Noise spectrum of the fMRI scanner with the force transducer on thescanner bed divided by the noise spectrum of an empty scanner bore. Note thelack of significant peaks.

4

emron

dtwwi

rFZg

the left lateral premotor cortex (4); 2: supplementary motor area; 3: right lateralis figure shows that the force measuring system does not affect the functional

. Conclusions

Accurate measurements of force and force fluctuations arextremely important for understanding the relationship betweenuscle output and fMRI data. Especially, studies concerning the

elation between the timing of muscle activation and the activityf various brain areas are of large interest both from a basiceurophysiological and pathological view point.

We have demonstrated that it is possible to use a force trans-ucer with strain gauges in an MR scanner environment. Despitehe magnetic field (both static and changing), the transduceras able to measure a large range of forces accurately. In otherords, the functioning of the transducer was not affected by MR

maging. Moreover, the transducer did not affect the MR images.The voluntary contractions of the index finger abduction

esulted in similar data as described earlier (Enoka et al., 1989;uglevand et al., 1993, 1995; Rutherford and Jones, 1988;ijdewind et al., 1998). Generally, one of the advantages of strainauges is that they can measure fast force fluctuations. Even

Neur

wtbtp

csafeatEstmmsr1

bbp

A

ttNl

A(

tb

U(mo

Rawp

A

dnttalU

iTo

R

A

B

C

D

H. van Duinen et al. / Journal of

ith the MR adjustments, this advantage still holds, since theransducer was able to measure fast force fluctuations (causedy a hammer stroke) accurately. This suggests that the func-ional properties of our force transducer were compatible to theroperties of other non-MR compatible transducers.

The brain activation pattern during the maximal voluntaryontraction showed activation of the traditional motor areas,imilar to results described by other studies (e.g. Dettmers etl., 1996; Van Duinen et al., 2007). In other papers, we describeunctional data recorded using these force transducers morextensively (Post et al., 2007; Van Duinen et al., 2007). Thenatomical image of a single subject demonstrated no artifactshat could be caused by the force transducer. Furthermore, thePI slices were not affected by the force transducer, althoughome differences were visible along the borders of the brain inhe lower panel of Fig. 7. These were probably due to residual

ovement artifacts. Both the human and phantom measure-ents showed no effects of the force transducer on the pooled

tandard deviations of the EPI time series. Furthermore, no spu-ious frequencies were detected in the RF frequency range of27.78 ± 0.44 MHz (Fig. 8).

In summary, our data showed the possibility of measuring aroad range of index finger forces and fast force changes duringrain imaging, which will be helpful in studies on human motorerformance.

cknowledgements

The authors are grateful to Anita Kuiper for her help withhe fMRI scanning and At Hoff for his help on testing the forceransducers. This work was supported in the framework of theWO Cognition Program with financial aid from the Nether-

ands Organization for Scientific Research (NWO).

ppendix A. The detailed schematic of the transmitterincluding the Wheatstone strain gauge bridge)

The power of the transmitter is supplied by a lead acid dry bat-ery (J1). The signal derived from the Wheatstone strain gaugeridge (J5, J6) is amplified (U1; AD8230, Analog Devices,

E

E

oscience Methods 164 (2007) 247–254 253

SA), and converted to a frequency-modulated optical signalU2; AD654, Analog Devices, USA). This optical signal is trans-itted (U6; HFBR1522, Agilent Technologies, USA) via an

ptical cable to the MR operator room.All resistors are made of SMD thick film 0805, except R6 and

7 which are made of 250 mW TH metal film. All capacitorsre made of SMD ceramic multilayer 0805, except C5 and C6hich are made of tantalum, and C1 which is made of metallizedolyester TH.

ppendix B. The detailed schematic of the receiver

The optical signal enters the receiver via the optical signaletector (U1; HFBR2522, Agilent Technologies, USA); the sig-al is then re-converted to voltage signal, linearly correspondingo the applied force (U2; LM331N, National Semiconduc-or, USA), filtered (U3; OP07, Texas Instruments, USA), andmplified (U4; OP07, Texas Instruments, USA). The ana-ogue signal is logged to a PC via Spike 2 (CED, Cambridge,K).All resistors are made of 250 mW TH metal film. The capac-

tors C1, C3, C9, C10, and C11 are made of ceramic multilayerH; C6–C8 are made of tantalum TH; C2, C4, and C5 are madef metallized polyester TH.

eferences

n KN, Ueba Y, Chao EY, Cooney WP, Linscheid RL. Tendon excursion andmoment arm of index finger muscles. J Biomech 1983;16:419–25.

oonstra TW, Clairbois HE, Daffertshofer A, Verbunt J, van Dijk BW, Beek PJ.MEG-compatible force sensor. J Neurosci Meth 2005;144:193–6.

ramer SC, Weiskoff RM, Schaechter JD, Nelles G, Foley M, Finklestein SP,et al. Motor cortex activation is related to force of squeezing. Hum BrainMapp 2002;16:197–205.

ettmers C, Connelly A, Stephan KM, Turner R, Friston KJ, Frackowiak RSJ, etal. Quantitative comparison of functional magnetic resonance imaging withpositron emission tomography using a force-related paradigm. Neuroimage1996;4:201–9.

hrsson HH, Fagergren A, Jonsson T, Westling G, Johansson RS, ForssbergH. Cortical activity in precision- versus power-grip tasks: an fMRI study. JNeurophysiol 2000;83:528–36.

noka RM, Robinson GA, Kossev AR. Task and fatigue effects on low-thresholdmotor units in human hand muscle. J Neurophysiol 1989;62:1344–59.

2 rosci

F

F

F

H

K

L

L

P

P

R

T

V

Vfatigue on human brain activity, an fMRI study. Neuroimage 2007;35:

54 H. van Duinen et al. / Journal of Neu

riston KJ, Holmes AP, Worsley KJ, Poline JB, Frith C, Frackowiak RSJ. Sta-tistical parametric maps in functional imaging: a general linear approach.Hum Brain Mapp 1995;2:189–210.

uglevand AJ, Zackowski KM, Huey KA, Enoka RM. Impairment of neuro-muscular propagation during human fatiguing contractions at submaximalforces. J Physiol 1993;460:549–72.

uglevand AJ, Bilodeau M, Enoka RM. Short-term immobilization has a mini-mal effect on the strength and fatigability of a human hand muscle. J ApplPhysiol 1995;78:847–55.

idler J, Hodics T, Xu B, Dobkin B, Cohen LG. MR compatible force sensingsystem for real-time monitoring of wrist moments during fMRI testing. JNeurosci Meth 2006;155:300–7.

uhtz-Buschbeck JP, Ehrsson HH, Forssberg H. Human brain activity in thecontrol of fine static precision grip forces: an fMRI study. Eur J Neurosci2001;14:382–90.

iu JZ, Dai TH, Elster TH, Sahgal V, Brown RW, Yue GH. Simultaneous mea-

surement of human joint force, surface electromyograms, and functionalMRI-measured brain activation. J Neurosci Meth 2000;101:49–57.

iu JZ, Zhang L, Yao B, Yue GH. Accessory hardware for neuromuscular mea-surements during functional MRI experiments. MAGMA 2002;13:164–71.

Z

ence Methods 164 (2007) 247–254

orter R, Lemon RN. Corticospinal function and voluntary movement. Oxford,UK: Oxford University Press; 1993.

ost M, van Duinen H, Steens A, Renken R, Kuipers B, Maurits N, et al. Reducedcortical activity during maximal bilateral contractions of the index finger.Neuroimage 2007;35:16–27.

utherford OM, Jones DA. Contractile properties and fatiguability of the humanadductor pollicis and first dorsal interosseus: a comparison of the effects oftwo chronic stimulation patterns. J Neurol Sci 1988;85:319–31.

hickbroom GW, Phillips BA, Morris I, Byrnes ML, Mastaglia FL. Isometricforce-related activity in sensorimotor cortex measured with functional MRI.Exp Brain Res 1998;121:59–64.

an Duinen H, Zijdewind I, Hoogduin H, Maurits NM. Surface EMG measure-ments during fMRI at 3 T: accurate EMG recordings after artifact correction.Neuroimage 2005;27:240–6.

an Duinen H, Renken R, Maurits NM, Zijdewind I. Effects of motor

1438–49.ijdewind I, Zwarts MJ, Kernell D. Influence of a voluntary fatigue test on

the contralateral homologous muscle in humans? Neurosci Lett 1998;253:41–4.