Functional MR imaging of visual and motor cortex stimulation at high temporal resolution using a flash technique on a standard 1.5 tesla scanner

Magnetic

Resonance

Imaging, Vol. 14. No. 5, pp. 477-483, 1996 Copyright 0 1996 Elsevier Science Inc. Printed in the USA. All rights reserved 0730-725X/96 $15.00 + .OO

PII: SO730-725X( 96) 00021-X

ELSEVIER

0 Original Contribution FUNCTIONAL MR IMAGING OF VISUAL AND MOTOR CORTEX STIMULATION AT HIGH TEMPORAL RESOLUTION USING A FLASH TECHNIQUE ON A STANDARD 1.5 TESLA SCANNER EDZARD

WIENER,

* LOTHAR R. SCHAD, * KLAUS T. BAUDENDISTEL, EDGAR MOLLER,~ AND WALTER J. LORENZ *

* MARCO

ESSIG, *

*Department of Radiology, German Cancer Research Center, Heidelberg, Germany TSiemens Medical Division, Erlangen, Germany Functional magnetic resonance imaging (fMR1) was performed on a conventional 1.5 T scanner by means of a modified FLASH-technique at temporal resolutions of 80 and 320 ms. The method’s stability was assessed by phantom measurements and by investigation of three volunteers resulting in a low amplitude (3%) periodic (4 s) signal modulation for the in vivo measurements, which was not observable in the phantom experiments. fMR1 activation studies of motor and visual cortices of four adjacent slices were carried out on 12 healthy right-handed volunteers. Stimulation was performed by a triggered single white light flash or single finger-to-thumb opposition movement, respectively. Event-related response of visual and motor activation was traced over 10.24 s with a temporal resolution of 320 ms for the four slice measurements. Brain activation maps were calculated by correlation of measured signal time courses with a time-shifted boxcar function. Activation was quantified by calculation of percentual signal change in relation to the baseline. Observed signal magnitudes were about 5-7% in visual and about 8-12% in primary motor cortex. While photic response was delayed by about 2 s, motor stimulation showed an instantaneous increase of the MR signal. MR signal responses for both stimuli had decayed completely after about 5 s. Our results show that event-related fMR1 enables mapping of brain function at sufficient spatial resolution with a temporal resolution of up to 80 ms on a conventional scanner. Functional magnetic resonance imaging; Visual cortex; Motor cortex; Fast imaging technique.

have been measured also by employing functional magnetic resonance spectroscopy ( fMRS ) . Ernst et al.’ detected an initial MR-signal undershoot 500 ms after visual stimulation followed by a signal increase after 1.5 s. While EPI techniques are capable of temporal resolutions of up to 200 ms, they require special hardware and are lacking from reduced spatial resolution and are more prone to image distortions due to offresonance effects. In contrast conventional gradientecho imaging techniques, widely used for clinical fMRI*-“’ provide a higher spatial resolution and a greater signal-to-noise ratio at the expense of increased sampling time. We have introduced event-related fMR1, allowing high temporal resolution fMR1 by modification of a conventional gradient echo sequence.‘” Temporal resolution is increased by synchro-

INTRODUCTION

Functional magnetic resonance imaging at high temporal resolution enables iaccurate tracing of stimulus induced signal changes. An increase in temporal resolution raises the prospect of deeper insight in the underlying physiological mechanisms and investigation of influences of physiological noise like respiration1 or heart beat. Studies of brain activation at high temporal resolution have been carried out using mostly EPI techniques.2-6 Several groups have addressed fast responses following visual or motor stimulation. De Yoe et a1.4have observed a two times greater response magnitude of motor cortex than of visual cortex and reported a faster signal mcrease in motor than in visual cortex. Recently, fast brain responses to stimulation

werpunkt Radiologie, Deutsches Krebsforschungszentrm, Postfach 101949, D-69009 Heidelberg, Germany.

9114195; ACCEPTED 212196. Address correspondence to E. Wiener, Forschungssch-

RECEIVED

477

Magnetic ResonanceImaging 0 Volume 14, Number 5, 1996

478

Fig. 1. Event-related fMR1 of photic stimulation. Four adjacent slices used for functional mapping of the visual cortex are shown. The correlation maps (cc > 0.65) overlaid on the corresponding anatomic spin-echo images indicate activation pronounced in the visual cortex. Increasing correlation-coefficients are coded in yellow and white.

nization of stimulus and data acquisition. After each stimulus, n raw data lines of n individual images are sampled. Like in ECG-gated acquisitions, data sampling is repeated until a sufficient amount of raw data is acquired. Temporal resolution is determined by the acquisition time (TR) of a single Fourier line times the number of requested slices. This article presents further investigations of high temporal resolved event-related fMR1 stimulation experiments of visual and motor cortex. These investigations include (a) stability measurements on volunteers and phantoms and (b) evaluation of postprocessing strategies for mapping of brain activation carried out at temporal resolutions of 80 ms or 320 ms.

MATERIALS

AND METHODS

Measurement technique MR images were obtained on a superconducting 1.5 Tesla whole-body scanner MAGNETOM 63/84 SP (Siemens, Erlangen, Germany) with a nonactiveshielded standard gradient system of 1 ms risetime for a maximum gradient strength of 10 mT/m using a commercially available circular polarized head coil. Event-related fMR1 was implemented by modification of a Tz*-weighted 3D-FLASH sequence.15 Sequence

parameters were TR/TE/a = 80 ms/60 ms/40”, reduced bandwidth = 40 Hz/pixel, MAT = 64 X 128 interpolated to 256 x 256, slice thickness TH = 4 mm, FOV = 200-250 mm, NEX = 1. Event-related fMR1 was performed in one or four slices simultaneously, resulting in a temporal resolution of 80 ms for the single slice or 320 ms for the four measurements. Total scan time for fMR1 was 10 min 55 s. Synchronization of sequence timing and stimulus performance is ensured by the use of a TTL-signal coming from the CCU (central control unit) prior to each acquisition. By means of a self-made “TTL-Divider-Box” the TTL-signals were counted and every 128th TTL-impulse (each 10.24 s) was used for stimulus generation. For investigation of motor activation, the trigger impulse was used to provide an acoustic signal. After each signal, the subject had to perform a single finger-to-thumb opposition movement of the left hand. Visual stimulation was carried out by external triggering of a stroboscope, providing a single white light flash (5 ys), seen by the volunteer in the dark scanner room. Event-related stimulation experiments were performed on 12 right-handed healthy volunteers (4 male, 8 female, 23-33 yr) . Stability measurements were performed on a spherical phantom and on three right-

MRI at high temporal resolution 0 E. WIENER

479

ET AL

Fig. 2. Event-related fMR1 of motor stimulation.Activation was performedby finger-to-thumboppositionmovementof the left hand. The parametricmap showsactivated regionspronouncedin the primary right motor cortex (cc > 0.65). Large activated areasposteriorthe central sulcusmay reflect large draining vesselsor sensoryinducedactivation due to a motor

handed healthy volunteers (two male, one female, 2330 yr), who were asked to rest during the experiment. A tape stripe was attached to the forehead to improve head fixation and prevent head motion during the experiment. Magnet shimming in each case resulted in a magnet field homogeneity of about 0.1-0.2 ppm. Slice selection was performed after a sagittal scan and after acquisition of a series of 20 Tr-weighted spin-echo (SE) images (TR/TE = 600 ms/ 15 ms, axial to coronary tilted by -25” for visual studies and by - 10” for motor studies, TH = #4mm). The single slice or the four adjacent slices, which showed best visualization of the visual or motor cortex, were chosen for the stimulation experiments. Informed consent from all volunteers was obtained after the nature of the experiment had been fully explained.

Data analysis After acquisition, typically 128 fMR1 images were transferred on a Workstation (VAX 4000/60, DEC, Maynard, USA) for postprocessing. Activated brain regions were computed on pixel-by-pixel basis by correlation of the experimental time course with a temporal responsefunction.‘e’ The temporal responsefunction was chosen astime-shifted box-car function, visualized

by the dotted lines in Figs. 3, 4, and 5. Accidental correlation was minimized by thresholding the correlation coefficient-maps, corresponding to a given significance level. Segmented correlation coefficientmaps were superimposed on the corresponding T ,-weighted SE images with increasing correlation coefficients (cc), color coded from red to yellow. Based on cc mapsregions of interest (ROI) were defined and the percent signal change in relation to the final baseline (average of the last 12 time points) was calculated.

RESULTS Twelve fMR1 examinations (six visual and six motor stimulations) were performed and compared with three nonactivation experiments and phantom measurements.Two examinations (one visual, one motor) had to be excluded becauseof insufficient cooperation of the volunteers (head motion). Each stimulation experiment covered four contiguous slices, resulting in a total cross section of 16 mm. Figures 1 and 2 show Tr-weighted images overlaid with segmentedcc maps (cc > 0.65, significance level ((Y = 0.0002) for visual and motor stimulation, respectively. Significant activation (pixels with high cc val-

Magnetic ResonanceImaging 0 Volume 14, Number 5, 1996

480

10 10 5

5

E 0

% 5

0

-5 -10

1 0

, 2

,

,

,

,I

4 6 delay [s]

8

10

Fig. 3. Signal time course of a ROI in the visual cortex. The dotted line visualizes the theoretical response (time-shifted boxcar function) used for data postprocessing shown in Fig. 1. First significant increase of the MR signal is detected after 2 s. Rise-time from stimulus onset to maximum is about 3 s and baseline is reached again in about 5 s after stimulus onset. Signal enhancement is about 7% in relation to baseline obtained from the last 12 measurement points.

ues, coded in yellow), occurred predominantly in areas occupied by the gray matter of primary right motor cortex and the calcarine fissure. Large activated areas posterior to the central sulcus of primary right motor cortex may reflect large draining vessels or sensory induced activation due to a motor task. Less pronounced correlation, coded in red, can be seen in the sulci. Figure 3 is representing an exemplary signal time course in response to photic stimulation of a ROI placed in activated visual cortex. The theoretical response function is represented by the dashed line. Response to visual stimulation started with a delay of about 2 s, maximal response is reached in about 3 s. Response has decayed completely after 5 s. The measured percent signal changes were about 5 - 7%. Figure 4 shows an exemplary signal time course of a ROI located in primary right motor cortex for left hand stimulation. Motor cortex response can be observed instantaneously after stimulus onset. A plateau is reached in about 2 s and baseline is reached again after about 5 s. The motor response magnitudes were about 8-12%. As demonstrated by Fig. 5, a large signal enhancement of about 20-40% can be observed in the sulci of motor and visual cortex. Figure 6 shows the results of stability measurements. Signal time courses at temporal resolutions of 320 ms and 80 ms were obtained from two volunteers and of a phantom measurement at 80 ms temporal resolution. Signal time courses obtained from ROIs located in the visual cortex show a periodic signal

0

2

4 6 delay [s]

8

10

Fig. 4. Time course of the fMR1 signal obtained from a ROI located in the primary motor cortex. The dotted line represents the theoretical response function (time-shifted boxcar function) used for data postprocessing shown in Fig. 2. MR signal increases instantaneously after the stimulus. Rise-time from stimulus to the plateau is about 2 s and baseline is reached again in about 5 s after stimulus onset. Signal enhancement is about 9% in relation to the baseline obtained from the last 12 measurement points.

waving of about 4 s and an amplitude modulation of about 3%, which can be seen in the images as artifactual band moving in phase-encoding direction, also observable in activation studies. In contrast, phantom experiments show no waving and signal discontinuities below 2% as demonstrated by Fig. 6. Figure 7 represents the averaged signal time courses of the five visual and five motor stimulation experiments 50 1

I

I

I

I

I

40

-10 I I 4 6 delay [s] Fig. 5. Time course of the MR signal of a ROI located in the sulci of primary motor cortex with a temporal resolution of 320 ms. In this case a signal increase up to 40% can be observed.

MRI at high temporalresolution 0 I

10

I

481

E. WIENER ET AL.

I

I

I

volunteer

5 E 0

2 r al

-5

-10

I

I

2

4

I

I

I

6

6

10

10

5 E 5 E

0

-5

-10 10

5 E 2 5

0

-5

-10

0

time

[s]

Fig. 6. Stability measurements of event-related fTvIR1. Signal time courses of a ROI located in visual cortex obtained from two volunteer studies at temporal resolutions of 320 ms and 80 ms in comparison with a phantom experiment. Please note a periodic 4 s signal waving with an amplitude modulation of about 3% in the volunteer studies which is absent in the phantom measurement. with significant increased MR-signal responses. The averaged response magnitudes are about 6% after photic

and about 10% after motor stimulation. Motor cortex response started instantaneously after stimulus onset, whereas visual cortex response started with a delay of about 2 s. The baseline is reached again after about 5 s. The error bars represent one standard deviation.

DISCUSSION Several studies on fMR1 of the visual and motor cortex at high temporal resolution

have been reported

during the last few years. Most of them were performed using EPI techniques. Use of event-related fMR1 raises the prospect of high temporal and increased spatial

Magnetic ResonanceImaging0 Volume 14, Number 5, 1996

482 ,

I

1

I

I

visual

10 t

I

I

I

I

4

[ST

II

8

10

I

delay

10

-

5

-

0

-

t

,

I

I

I

I

I

I

2

4

E -G W ?I

W

* -

-5

’ 0

[sy

8

I

II 10

delay

Fig. 7. Averaged signal time courses (2 SD) of the five visual and five motor stimulation experiments with signifi-

cant increased MR signal responses.

resolution by application of conventional imaging sequences. Event-related fMR1 relies critically on both, exact stimulus repetition, and timing of stimulus presentation and data acquisition. While exact timing of stimulus presentation is ensured by the sequence-controlled stimulus generation, consecutive stimulus repetition may lead to adaptation effects, causing small changes in stimulus response, thus resulting in physiological variation of the experimental conditions. The presence of periodic noise during the volunteer experiments may reflect either periodic physiological changes or instability of the MR scanner. We speculate that the observed (4 s) waving signal reflects most likely respiration or pulsation (brain, CSF), because variations of NRI signal induced by heart beat and respiration1 are known and the periodic signal fluctuations were absent in our phantom studies. The presence of this physiological noise at the activation experi-

ments also limits the detectable extent of stimulation effects, but ROI signal time courses of activated brain regions indicate signal changes exceeding physiological noise in most of the cases. For volunteers’ convenience a small matrix size was used to minimize total scan time, but further improvement of the in-plane resolution at the expense of total scanning time would be possible. Tracking of MRsignal response after motor and visual stimulation for 10.24 s proved sufficient for visualization of stimulus induced signal changes, which had decayed completely 5-6 s after stimulus onset. As a temporal resolution of 320 ms proved sufficient for tracing brain responses, stimulation experiments were carried out simultaneously on four slices to ensure a maximal coverage of activated cortical areas. The high temporal resolved stimulation experiments demonstrate a significant signal increase in cortical layers of gray matter in the primary motor and visual cortices after stimulus-induced brain activity. In agreement with another observation,4 a delayed response of visual compared with motor stimulation and a more pronounced signal change for motor activation could be observed. Previous findings of the delayed response of about 2 s after visual stimulation were also observed using fMRS7 and EPI techniques.4 The difference in response patterns may be due to different stimulus designs. While the flash light used for visual stimulation is a short, precise stimulus, motor stimulation included attention to the acoustic signal and the performance of relative complex task. Due to the low periodic signal modulation small signal changes like the reported initial decrease’ of the MRsignal were not detectable. The extremely strong signal enhancements between 20-40s observed in the sulci of motor and visual cortex may most likely result from large vessels. Event-related fMRI combines the advantages of conventional sequences (good signal to noise ratio and sufficient spatial resolution) with high temporal resolution recommended for fMRI. The method was able to reproduce previous results of visual and motor activation studies obtained by fMRS or fMR1 using EPI techniques. Its application may encourage investigation of brain response to stimulation at high temporal resolution for clinical studies on standard equipments and may lead to improved modeling of the physiological mechanisms underlying the stimulus induced signal changes. REFERENCES 1. Nell, DC.; Schneider, W. Respiration artifacts in functional brain imaging: Sources of signal variation and compensation strategies. In: Book of Abstracts. Second

MRI at high temporal resolution 0

2. 3.

4.

5.

6.

7. 8.

9.

Annual Meeting of the Society of Magnetic Resonance. San Francisco: SMRM; 1994: p. 647. Mansfield, P. Multi-planar image formation using NMR spin echoes. J. Phys. C lO(L55):349-352; 1977. Kwong, K.K.; Belliveau, J.W.; Chesler, D.A.; Goldberg, I.E.; Weisskoff, R.M.; Poncelet, B.P.; Kennedy, D.N.; Hoppel, B.E.; Cohen, MS.; Turner, R.; Cheng, H.M.; Brady, T.J.; Rosen, 13.R. Dynamic magnetic resonance imaging of human brain activity during primary sensory stimulation. Proc. Natl. Acad. Sci. USA 895675-5679; 1992. DeYoe, E.A.; Neitz, J.; Bandettini, P.A.; Wong, EC.; Hyde, J.S. Time course of event-related MR signal enhancement in visual and motor cortex. In: Book of Abstracts: Eleventh Annual Meeting of the Society of Magnetic Resonance in Medicine. Berlin: SMRM; 1992: p. 1824. Bandettini, P.A.; Wong, EC.; Hinks, R.S.; Tikofsky, R.S.; Hyde, J.S. Time course EPI of human brain function during task activation. Magn. Reson. Med. 25:390397; 1992. Turner, R.; Jezzard, P.; Wen, H.; Kwong, K.K.; Le Bihan, D.; Zeffiro, T.; Balaban, R.S. Functional Mapping of the human visual cortex at 4 and 1.5 Tesla using deoxygenation contrast EPI. Magn. Reson. Med. 29:277-279; 1993. Ernst. T.; Hennig, J. Observation of a fast response in functional MR. Magn. Reson. Med. 32: 146- 149; 1994. Haase, A.; Frahm, J.; Matthaei, D.; Hlnicke, W.; Merboldt, K.D. FLASH imaging. Rapid NMR imaging using low flip-angle pulse:, Magn. Reson. Med. 14:68-78; 1990; J. Magn. Reson. 67:258-266; 1986. Ogawa, S.; Lee, TM.; Nayak, A.S.; Glynn, P. Oxygen-

10.

Il.

12.

13.

14.

15.

16.

E. WIENER

ETAL.

483

ation-sensitive contrast in magnetic resonance image of rodent brain at high magnetic fields. Magn. Reson. Med. 14:68-78; 1990. Ogawa, S.; Tank, D.W.; Menon. R., Ellermann, J.M.; Kim, S.G.; Merkle, H.; Ugurbil, K. Intrinsic signal changes accompanying sensory stimulation: Functional brain mapping with magnetic resonance imaging. Proc. Natl. Acad. Sci. USA 89:5951-5955; 1992. Frahm, J.; Merboldt, K.D.; H%nicke, W. Functional MRI of human brain activation at high spatial resolution. Magn. Reson. Med. 29( 1):139-144; 1993. Schad, L.R.; Trost, U.; Knopp, M.V.; Mtiller, E.; Lorenz, W.J. Motor cortex stimulation measured by magnetic resonance imaging on a standard 1.5 Tesla clinical scanner. Magn. Reson. Imaging 11(4) :461-464; 1993. Constable, R.T.; McCarthy, G.; Allison, T.; Anderson, A.W.; Gore, J. Functional brain imaging at 1.5T using conventional gradient echo MR imaging techniques. Magn. Reson. Imaging 11(4):451-459; 1993. Schad, L.R.; Wenz, F.; Knopp, M.V.; Baudendistel, K.; Mtlller, E.; Lorenz, W.J. Functional 2D and 3D magnetic resonance imaging of motor cortex stimulation at high spatial resolution using standard 1.5 Tesla imager. Magn. Reson. Imaging 12( 1):9-15; 1994. Schad, L.R.; Wiener, E.; Baudendistel, K.T.; Mtiller, E.; Lorenz, W.J. Event-related functional MR imaging of visual cortex stimulation at high temporal resolution using a standard 1.5 Tesla Imager. Magn. Reson. Imaging (in press). Bandettini, P.A.; Jesmanowicz, A.; Wong, EC.; Hyde, J.S. Processing strategies for time-course data sets in functional MRI of the human brain. Magn. Reson. Med. 30:161-173: 1993.