www.elsevier.com/locate/ynimg NeuroImage 37 (2007) 761 – 768
fMRI at 7 T: Whole-brain coverage and signal advantages even infratentorially? Elke R. Gizewski, a,⁎ Armin de Greiff, b Stefan Maderwald, b Dagmar Timmann, c Michael Forsting, a and Mark E. Ladd b a
Department of Diagnostic and Interventional Radiology and Neuroradiology, University Hospital Essen, Hufelandstr. 55, D-45127 Essen, Germany Erwin L. Hahn Institute for Magnetic Resonance Imaging, University Duisburg-Essen, Germany c Department of Neurology, University Hospital Essen, Germany b
Received 14 February 2007; revised 28 May 2007; accepted 5 June 2007 Available online 14 June 2007 fMRI is one of the most likely applications to benefit from high field MRI. It profits from the higher signal-to-noise ratio (SNR) and increased BOLD contrast itself. However, this sensitivity to susceptibility brings with it problems, e.g. in-plane dephasing and signal dropouts near tissue–air boundaries. Therefore, most fMRI studies at 7 T focus on high resolution in supratentorial areas. Nine volunteers were measured at both 1.5 and 7 T using finger tapping with fMRI in a block design fashion. An EPI sequence with short TE (28 ms at 7 T) was chosen for covering the whole brain. A CP transmit/receive head coil was used for image acquisition. Statistical analyses were performed using SPM 02. The activated images were superimposed on both individual images and a standard T1-normalized brain dataset. All cerebral areas involved in finger tapping could be revealed using 7 T: SI, MI, SII, SMA, thalamus, and cerebellar areas. At 1.5 T the activation in the thalamus was only detectable in one subject using the same corrected p value for all analyses. Furthermore, the BOLD signal change was significantly higher at 7 T than at 1.5 T (factor 2 to 3). A well fitted response curve could be detected in all sensory–motor areas at 7 T in whole-brain coverage, even in areas suffering from susceptibility like the cerebellum. Supra- and infratentorial sensory–motor areas are consistently and reliably detectable using whole-brain fMRI at 7 T with good quality response functions and, as expected, higher signal compared to 1.5 T. © 2007 Elsevier Inc. All rights reserved. Keywords: fMRI; 7T; Motor
this method has the potential to be improved with increased spatial and temporal resolution. The blood oxygenation level dependent contrast (BOLD) represents signal changes in T2 or T2⁎ weighted images. These sequences are presumed to be well suited to high magnetic field strength since fMRI sequences benefit from higher SNR and higher signal in BOLD contrast images (Vaughan et al., 2001). However, their sensitivity to susceptibility also causes problems, e.g. in-plane dephasing and signal dropouts near tissue– air boundaries. To achieve greater insights into brain function, high field fMRI has been used in some studies to attain higher spatial resolutions (Duong et al., 2002; Pfeuffer et al., 2002a,b). These studies focus on high resolution images which can be acquired rapidly and with good temporal resolution. Additionally, the signal increase advantage in high field MRI has been studied (Pfeuffer et al., 2002b). Nearly all these studies therefore accepted restrictions in the field of view and number of slices for 7 T imaging and avoided areas near tissue–air boundaries. For broader application and for analyzing cognitive functions, however, a more extended coverage of the brain including multiple involved areas is needed to reveal network activation. Therefore, the aim of this study was to analyze possible benefits of high field MRI using a quadrature transmit/receive birdcage head coil and sequences covering the whole brain at 7 T in comparison to wholebrain imaging with a standard CP coil at 1.5 T. Materials and methods
Introduction
Subjects
In recent years, functional magnetic resonance imaging (fMRI) has become a widely used approach for neuroscience. However,
Seven male and two female healthy volunteers (mean age 31 years, range 24 to 49) were studied. All subjects were righthanded. No subject revealed any brain tissue abnormality on structural MRI performed at 1.5 T nor had any history of neurological or psychiatric disease. Informed written consent was obtained prior to scanning. The study was accepted by the local ethics committee.
⁎ Corresponding author. Fax: +49 201 723 5959. E-mail address:
[email protected] (E.R. Gizewski). Available online on ScienceDirect (www.sciencedirect.com). 1053-8119/$ - see front matter © 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2007.06.005
762
E.R. Gizewski et al. / NeuroImage 37 (2007) 761–768
Experimental design All subjects were measured twice: images were acquired using a 1.5 T standard scanner (Magnetom Sonata, Siemens Medical Solutions, Erlangen, Germany) and a 7 T scientific whole-body MR system (Magnetom 7 T, Siemens Medical Solutions, Germany) with a standard receive-only volume head coil for the 1.5 T system and a CP transmit/receive head coil (Invivo Diagnostic Imaging, Gainesville, Florida, USA) for the 7 T MR system. At 1.5 T a 3D FLASH sequence (TR 10 ms, TE 4.5 ms, flip angle 30°, FOV 220 mm2, matrix 512 * 512, slice thickness 1.5 mm) was acquired for individual co-registration of functional and structural images. BOLD contrast images were acquired using a gradient echo-planar technique (TR 2840 ms, TE 45 ms, flip
angle 90°, FOV 240 mm, matrix 64 * 64) with 34 transverse slices with 3 mm thickness and 0.3 mm slice gap. Three “dummy” scans were eliminated prior to data analysis. At 7 T a 3D MPRAGE sequence (TR 2500 ms, TE 2.6 ms, TI 1100 ms, flip angle 9°, FOV 256 mm2, matrix 512 * 512, slice thickness 1.0 mm) was acquired for individual co-registration of functional and structural images. BOLD contrast images were acquired using a gradient echo-planar technique (TR 2840 ms, TE 28 ms, flip angle 90°, FOV 220 mm, matrix 64 * 64) with 34 transverse slices with 3 mm thickness and 0.3 mm slice gap. Again, three “dummy” scans were eliminated prior to data analysis. Shimming at 1.5 T was performed automatically before the EPI sequence using the standard shimming algorithm available on the scanner. The shimming at 7 T was performed manually by the user
Fig. 1. (A) Statistical parametric maps of activation within all subjects performing the finger tapping task compared with rest period at 1.5 T. Task-related increase in MR signal is superimposed on three orthogonal sections of a 3D T1-weighted standard brain. Statistically corrected threshold is p b 0.05. Results show main activation in SI, SII, SMA and cerebellum. (B) Statistical parametric maps of activation within all subjects performing the finger tapping task compared with rest period at 7 T. Statistically corrected threshold is p b 0.05. Results show main activation in SI, SII, SMA, thalamus and cerebellum. (C) Statistical parametric maps of activation within all subjects performing the finger tapping task compared with rest period at 1.5 T. Statistically corrected threshold is p b 0.05. Results show main activation in SI, SII, SMA and cerebellum superimposed on individual EPI images. (D) Statistical parametric maps of activation within all subjects performing the finger tapping task compared with rest period at 7 T. Statistically corrected threshold is p b 0.05. Results show main activation in SI, SII, SMA, thalamus and cerebellum superimposed on individual EPI images.
E.R. Gizewski et al. / NeuroImage 37 (2007) 761–768
763
block of 10 scans embedded in two passive blocks were performed. The subjects were randomized for the order of measurements at 7 T and 1.5 T. During the measurements, the subjects were asked to lie relaxed inside the scanner with eyes closed and to try to perform the motor task with only finger tapping movement of the right hand.
also using the standard shimming algorithm, but with multiple repetitions and close verification of the result before starting the EPI sequence. At both field strengths a global optimum was determined with no per slice shimming. At both field strengths the phase corrections parameters were calculated slice by slice using three non-phase-encoded navigator echoes before the EPI readout (Heid, 1997). Each subject underwent two functional sessions each using finger tapping as the active condition in a block design and alternated with resting periods every 28 s. Each run was divided into four epochs. Additionally, runs with only one active condition
Data analysis All images were acquired using an EPI sequence with prospective motion correction. For data analysis, SPM 02 software
Table 1 Signal changes for each subject and activated area (corrected p b 0.001) Subject
Region
Signal value 1.5 T
Voxel 1.5 T
Signal value 7 T
Voxel 7 T
Tal 1.5 T
Tal 7 T
1
SI/MI SII SMA Thalamus Cerebellum SI/MI SII SMA Thalamus Cerebellum SI/MI SII SMA Thalamus Cerebellum SI/MI SII SMA Thalamus Cerebellum SI/MI SII SMA Thalamus Cerebellum SI/MI SII SMA Thalamus Cerebellum SI/MI SII SMA Thalamus Cerebellum SI/MI SII SMA Thalamus Cerebellum SI/MI SII SMA Thalamus Cerebellum
3.6 1.2 1.1 – 2.4 3.0 0.9 1.1 0.7 2.9 1.4 0.5 0.6 – 0.7 2.4 1.3 1.4 – 2.2 1.4 0.7 0.9 – 1.3 4.5 0.8 1.2 – 2.4 3.6 1.9 3.1 – 3.2 4.9 1.5 – – 2.7 2.5 – 1.2 – 1.4
983 5 17 – 377 2748 13 174 4 1549 1026 27 15 – 70 1089 16 3 – 297 98 4 7 – 279 258 12 2 – 115 1292 98 180 – 538 646 5 – – 548 79 – 4 – 37
6.5 2.7 3.5 2.2 5.0 10.5 1.5 4.2 1.9 8.0 9.8 2.7 5.4 1.2 4.3 6.6 3.0 3.2 1.9 5.0 5.8 1.2 2.5 1.6 4.5 18.0 2.5 5.9 2.5 6.0 10.0 4.5 4.4 2.7 9.0 11.4 2.8 2.8 1.9 6.4 5.0 2.4 2.0 0.9 3.8
1298 24 58 25 557 3812 40 195 10 3437 6636 79 851 5 1293 2306 29 236 10 318 219 22 25 10 318 4046 21 82 5 1430 4253 187 1372 49 5081 1519 58 17 6 990 651 40 9 10 210
−36, − 28, 70 −52, − 20, 20 −04, − 06, 60
− 44, − 30, 62 − 52, − 20, 20 − 02, − 10, 62 − 14, − 20, 0 14, − 50, − 28 − 36, − 30, 62 − 40, − 24, 19 − 04, − 04, 52 − 14, − 24, 0 22, −52, − 28 − 38, − 24, 70 − 62, − 14, 28 − 02, − 02, 58 − 24, − 10, 06 18, − 48, − 22 − 44, − 24, 58 − 52, − 24, 22 − 08, − 04, 56 − 16, − 24, 04 14, − 52, − 18 − 44, − 18, 60 − 50, − 28, 20 − 10, − 02, 54 − 22, − 20, − 02 18, − 46, − 26 − 40, − 26, 64 − 50, − 26, 18 − 02, − 04, 54 − 14, − 20, 06 18, − 50, − 28 − 42, − 24, 66 − 58, − 18, 18 − 06, − 10, 50 − 16, − 22,08 24, −54, − 28 − 46, − 18, 54 − 60, − 18, 08 − 10, − 08, 52 − 14, − 22, 0 18, − 50, − 28 − 38, − 22, 54 − 54, − 24, 22 − 06, 02, 48 − 22, − 26, 06 16, − 48, − 22
2
3
4
5
6
7
8
9
16, − 56, − 24 −36, − 26, 70 −48, − 24, 28 −02, − 10, 62 −14, − 20, 08 16, − 52, − 18 −44, − 24, 56 −60, − 16, 28 −04, − 02, 58 26, − 52, −24 −34, − 18, 68 −54, − 20, 20 −04, − 02, 56 18, − 46, − 24 −38, − 32, 64 −48, − 30, 20 −08, −06, 52 20, − 52, −24 −44, − 26, 60 −58, − 22, 28 00, − 04, 58 18, − 54, − 26 −42, − 20, 64 −58, − 20, 22 −06, − 08, 52 22, − 56, −24 −46, − 12, 56 −52, − 18, 12 20, − 54, −20 −40, − 32, 64 −02, − 02 − 46 18, − 50, − 24
t-values and signal changes for each activated region and each subject are presented from the standard block design and the single active block analysis. Only statistically reliable areas (corrected p b 0.001 (FWE)) are included. Talairach (Tal) coordinates of signal maximum and voxel extent for each area are given in x, y and z direction.
764
E.R. Gizewski et al. / NeuroImage 37 (2007) 761–768
(Welcome Department of Cognitive Neurology, London, UK) was used. Prior to statistical analysis images were realigned utilizing the sinc interpolation and normalized to the standard stereotactic space corresponding to the template from the Montreal Neurological Institute (http://www.mrc-cbu.cam.ac.uk/Imaging/ mnispace.html). Bilinear interpolation was applied for normalization. The slightly different image contrast of the acquired images did not lead to any problems in the normalization. The typical scanner drift due to temperature differences and shim changes was low. The images were smoothed with an isotropic Gaussian kernel of 7 mm. A voxel-by-voxel comparison according to the general linear model was used to calculate differences in activation between the active and resting condition. The model consisted of a box-car function convolved with the hemodynamic response function (hrf) (Friston et al., 2000) and the corresponding temporal derivative. High pass filtering with cut-off of 128 s and low pass filtering with the hrf were applied. Significant signal changes for each contrast were assessed by means of t-statistics on a voxel-by-voxel basis (Friston et al., 1995). The resulting set of voxel values for each contrast constituted a statistical parametric mapping (SPM) of the t-statistic. The threshold was set to p b 0.05 (FWEcorrected (Family Wise
Error)) for single subject analysis. The effect of interest was defined for each subject with a contrast vector producing a contrast image (con-image) containing the contrast of the parameter estimated at each voxel (Penny and Holmes, 2003). For group analysis, single-subject con-images of the images acquired with a 64 matrix were entered into a random effects model. The conimages were feed in a general linear model that implements a onesample t-test. The threshold was set to p b 0.05 (FDRcorrected (false discovery rate)) for group analysis. Activated images were superimposed on standardized T1-weighted images as well as single-subject EPI and T1-weighted images. For all activated areas, time–response curves were calculated in SPM 02. Additionally, the average fractional signal change between on and off condition was calculated for all activated areas at 1.5 T and 7 T at the local maximum for each region. These data were then averaged across subjects. The activated areas were defined according to the Talairach coordinate system and visually checked for individual single-subject analysis (Talairach and Tournoux, 1988). To reveal the extension of activation for each scanner the convalues were analyzed in relation to the activated voxels over the t-value up to the maximum value at 7 T in a logarithmic scale.
Fig. 2. (A) Plot of fitted response function at the main cluster in the cerebellar sensory–motor areas at 1.5 T (representative subject). (B) Plot of fitted response function at the main cluster in the one subject revealing a thalamic activation at 1.5 T. (C) Plot of fitted response function at the main cluster in the cerebellar sensory–motor areas at 7 T (representative subject). (D) Plot of fitted response function at the main thalamic cluster at 7 T (representative subject).
E.R. Gizewski et al. / NeuroImage 37 (2007) 761–768
765
Results During the finger tapping task, the known task-related cortical activations were observed on both MR systems. Results at 1.5 T All cerebral areas involved in finger tapping could be revealed using the standard EPI sequence at 1.5 T. Consistent activation with acceptable t-values were found in all subjects in the cortical areas (SI, MI) and the contralateral cerebellar areas (medial cerebellum within lobules IV, V and VIII predominantly on the right) known to be involved in sensory–motor tasks (Fig. 1A). Activation in the SII and SMA was found in all but one subject. Activation in the thalamus could only be revealed in one subject using corrected p values b 0.05. This p value and FWE correction were used for all single-subject analyses in this study. Table 1 summarizes all individual signal changes for each area at 1.5 T and 7 T. Even at 1.5 T some susceptibility artefacts were present in the basal brain structures as show in Fig. 1C, where the activation maps are superimposed on individual EPI images of a representative subject. A well-fitted response curve could be detected in all sensory– motor areas at 1.5 T using whole-brain coverage. Response functions in the cerebellum were acceptable in all subjects. For the thalamic area, the response function was only calculated in subjects revealing an activation in this area at corrected p b 0.05. Even in this restricted group of subjects the response curve did not reveal an excellent fit; especially within the first two active phases the response revealed a significant deviation from the assumed hrf function. Fig. 2A shows time–response curves of one representative subject for the cerebellar activation and Fig. 2B for the thalamic area. Analysis of single activated block design revealed activation in SI, MI, SII, SMA and the cerebellum in all subjects. The thalamic activation was only revealed in one subject. Relative signal change was between 0.5 and 1.4. Results at 7 T Activation was revealed in all sensory–motor areas at 7 T: SI, MI, SII, SMA, thalamus and contralateral cerebellar areas (medial cerebellum within lobules IV, V and VIII predominantly on the right) involved in sensorimotor processing (Fig. 1B). Furthermore, the signal change was significantly higher at 7 T than at 1.5 T (factor 2 to 5; see Table 1). The signal changes of all subjects are presented in Fig. 3, where the value of signal change in SI and cerebellum is plotted against field strength. The results of the convalues analyzed in relation to the activated voxels over the t-value up to the maximum value at 7 T in a logarithmic scale revealed a higher signal over the different of t-values at 7 T images. The coordinates for the activated areas are given as the standard calculation of local maximum in the activated cluster including SI/ MI and for the further areas in a similar manner. Therefore, the coordinates show a distribution in 1.5 T and similar in 7 T images (see Table 1). The differences of the automatically generated center of activation for each area were calculated for each subject from Table 1 and are shown in Fig. 4. At 7 T susceptibility artefacts were present especially in the basal brain structures as shown in Fig. 1D, where the activation
Fig. 3. The signal changes of all subjects are presented. The value of signal change in MI/SI (first diagram) and cerebellum (second diagram) is plotted for both field strengths.
maps are superimposed on individual EPI images of a representative subject. A well-fitted response curve could be detected in all sensory– motor areas at 7 T using whole-brain coverage, even in areas suffering from susceptibility such as the cerebellum. In contrast to the results at 1.5 T, thalamic activation was found in all subjects and revealed an excellent response function. Fig. 2C shows the time–response curve for the cerebellar activation of one representative subject and Fig. 2D for the thalamic area. Analysis of single activated block design revealed activation in SI, MI, SII, SMA and the cerebellum in all subjects. Relative signal change in SI was between 3.0 and 11.2. Therefore, the single block analyses revealed similar or even higher response strength than multi-block measurements at 1.5 T. The results for SI activation in a representative subject during single block at 1.5 T and 7 T are given in Fig. 5. Discussion Our results reveal a consistent and reliable activation in supraand infratentorial sensory–motor areas using fMRI at 7 T. Supratentorial areas involved in sensory motor tasks are the primary and secondary sensory motor areas (S I, SII, MI and MII) and the SMA (Ball et al., 1999; Del Gratta et al., 2002; Freund, 2002). Somatotopic representation in the cerebellum with two homunculi, one located upside down in the anterior lobe and a second one in the posterior lobe, has first been described by Adrian and Snider (Adrian, 1943; Snider and Stowell, 1944). A similar
766
E.R. Gizewski et al. / NeuroImage 37 (2007) 761–768
Fig. 4. The differences of the maximum of the activation at 1.5 T and at 7 T in each subject and each activated area are shown.
somatotopic representation has been found in humans using fMRI (Grodd et al., 2001; Rijntjes et al., 1999). The main activated clusters in lobules IV/V and VIII correspond to the hand areas of the two homunculi in the anterior and posterior cerebellar lobes. Furthermore, it is known that activity during sensorimotor tasks is more pronounced ipsilaterally which correspond with our results. Vermal activity in lobules V and VII during the sensorimotor tasks, as shown in our study, has also been described in the literature
(Nitschke et al., 1998). The results indicate that fMRI can be robustly performed at 7 T covering the whole brain using a t/r CP head coil. As expected, higher signal compared to 1.5 T was revealed in the analyses of the 7 T images. In our study, a two to threefold increase in relative signal change was detected using an EPI sequence with a TE of 28 ms. The optimum TE for 7 T was reported to be around 25 ms in focused fMRI on the occipital cortex (Yacoub et al., 2001). In accordance, the shortest TE
Fig. 5. Signal changes in SI during single activated block design with 1 active condition at 1.5 T (upper row) and 7 T (lower row).
E.R. Gizewski et al. / NeuroImage 37 (2007) 761–768
available for whole-brain imaging with EPI sequences was used. An optimal TE at 1.5 T is given with 40 ms (Triantafyllou et al., 2005); for whole brain covering we used 45 ms in this study. Some former studies revealed a signal increase up to fivefold. These studies employed different imaging parameters as they were focused on increase of resolution and small field of view (Pfeuffer et al., 2002b). Therefore, the increase in BOLD signal was more extensive than in our study, which emphasized whole-brain fMRI at high field strength. The matrix in our study was chosen to be 64 × 64 in order to cover the whole brain. The focus was set on good response functions even in areas suffering from e.g. in-plane dephasing and signal dropouts near tissue–air boundaries. The sensitivity was somewhat constrained by the SNR characteristics of the CP head coil used, and the voxel size in our study was about nine times larger than in the study of Pfeuffer et al. It has been shown that a reduction in voxel size leads to an improvement in time series SNR through a decrease in physiological noise (Triantafyllou et al., 2005). The relatively small BOLD changes in our study might be explained by this effect. We decided to use the larger voxels to ensure whole-brain coverage and to ensure good comparability to the 1.5 T images. Previous studies have reported the benefit of fMRI at high field strength because of an increase in SNR and BOLD signal. It is likely that many future studies will not stick to exceptional resolution of one area but be targeted at analyzing complex networks. Especially cognitive functions will require more slices and coverage of extended brain areas. Furthermore, some interesting structures like the hippocampal region can, like the cerebellum, suffer from signal dropouts near tissue–air boundaries. The excellent response functions and signal change elevations shown in this study using a well-established, simple sensory–motor paradigm indicate that even in such problematic brain regions high field fMRI is possible. The response function was chosen as a standard human hrf function as it was assumed that the field strength should not have an important influence on the hrf (Friston et al., 2000). The adjusted response curves revealed no significant deviations in the cortical and cerebellar activations. Only the thalamic activation revealed some deviations from the curve within the first two active conditions at 1.5 T. This finding may be due to the irregular and lower signal change in this region at 1.5 T and the lower SNR in comparison to 7 T. An influence of partial volume effects due to the voxel size is not likely as the voxel size was equal to the 7 T measurements. The distribution of the activated maxima was variable comparing 1.5 T and 7 T measurements in each subject and area. Most areas revealed a small and acceptable distribution of about 5 mm except SI/MI with a maximal variation of 15.7 mm. SI/MI, however, covers a large area and separate analysis of SI and MI, e.g. using a smaller p value, would reveal more consistent results in individual subjects. Nevertheless, there are increased susceptibility artefacts compared to 1.5 T. Much more improvement will be reached using more advanced head coils than the t/r CP coil. Multichannel coils allow application of parallel acquisition techniques (Mirrashed et al., 2004). Multiple channels will provide further increases in SNR and signal strength, and coupled with parallel imaging will reduce artefacts e.g. due to susceptibility differences near tissue–air boundaries as is known from experience at 1.5 T. However, a disadvantage at high field could be a restriction of the number of slices and inhomogeneous resolution over the brain (Wiggins et al., 2005). Therefore, the coils and sequences have to be chosen depending on the paradigms to be applied. Again,
767
parallel imaging can be useful for reducing the RF load on the tissue and enabling more slices. Further studies should also address cognitive functions involving more challenging brain areas susceptible to large inhomogeneities, as for example the hippocampal region is an important structure located near tissue– air boundaries, and reliable detection of activation might be hampered at high field. The problem that physiological noise dominates the SNR of the fMRI time-course at low spatial resolutions at high field strengths was not considered to be a prominent issue in this study since we used a 64 matrix and a slice thickness of 3 mm. Otherwise, the physiological noise can limit some benefits of high field acquisition since increases in image SNR produce only small increases in time-course SNR (Triantafyllou et al., 2006). The signal increase using the finger tapping paradigm with the chosen EPI sequence was two- to fivefold higher than at 1.5 T. The relatively wide range of relative signal changes compared to 1.5 T may be by the difficulty of achieving a uniform magnetic field shim and RF excitation field at 7 T. The fMRI experiments at 7 T are therefore more dependent on field inhomogeneities and these have to be taken into account during image analysis. But, even at the lowest level, the signal change increase led to a similar signal when analyzing one active block at 7 T compared to standard block designs with four active phases at 1.5 T. Therefore, the results indicate that single block designs can be applied in wholebrain imaging at 7 T. Besides fundamental experimental interests, e.g. for cognitive studies, clinical indications can be imagined. Pre-surgical fMRI in patients with brain tumors could benefit from either higher resolution or faster imaging. Even patients impaired with respect to motor function are for the most part able to perform a short finger movement for a single block examination. In conclusion, supra- and infratentorial sensory–motor areas are consistently and reliably detectable using fMRI at 7 T covering the whole brain, with good response functions and, as expected, higher signal compared to 1.5 T, enabling a wide field of applications e.g. in the analysis of activated networks or for clinical indications using sensory–motor imaging. Acknowledgments We would like to thank Karsten Wicklow of Siemens Medical Solutions and Charlie Saylor of Invivo Diagnostic Imaging for their help in optimizing the performance of the CP coil on the 7 T system.
References Adrian, E.D., 1943. Afferent areas in the cerebellum connected with the limbs. Brain 66, 289–315. Ball, T., Schreiber, A., Feige, B., et al., 1999. The role of higher-order motor areas in voluntary movement as revealed by high-resolution EEG and fMRI. NeuroImage 10, 682–694. Del Gratta, C., Della Penna, S., Ferretti, A., et al., 2002. Topographic organization of the human primary and secondary somatosensory cortices: comparison of fMRI and MEG findings. NeuroImage 17, 1373–1383. Duong, T.Q., Yacoub, E., Adriany, G., et al., 2002. High-resolution, spinecho BOLD, and CBF fMRI at 4 and 7 T. Magn. Reson. Med. 48, 589–593. Freund, H.J., 2002. fMRI studies of the sensory and motor areas involved in movement. Adv. Exp. Med. Biol. 508, 389–395.
768
E.R. Gizewski et al. / NeuroImage 37 (2007) 761–768
Friston, K.J., Holmes, A.P., Poline, J.B., et al., 1995. Analysis of fMRI timeseries revisited. NeuroImage 2, 45–53. Friston, K.J., Mechelli, A., Turner, R., et al., 2000. Nonlinear responses in fMRI: the Balloon model, Volterra kernels, and other hemodynamics. NeuroImage 12, 466–477. Grodd, W., Hulsmann, E., Lotze, M., et al., 2001. Sensorimotor mapping of the human cerebellum: fMRI evidence of somatotopic organization. Hum. Brain Mapp. 13, 55–73. Heid, O., 1997. Robust EPI phase correction. Paper presented at: ISMRM (Vancouver). Mirrashed, F., Sharp, J.C., Cheung, I., et al., 2004. High-resolution imaging at 3 T and 7 T with multiring local volume coils. Magma 16, 167–173. Nitschke, M.F., Hahn, C., Melchert, U.H., et al., 1998. Activation of the cerebellum by sensory finger stimulation and by finger opposition movements. A functional magnetic resonance imaging study. J. Neuroimaging 8, 127–131. Penny, W.D., Holmes, A.J., 2003. Random-effects analysis, In: Frackowiak, R.S.J., Friston, K.J., Frith, C., Dolan, R., Friston, K.J., Price, C.J., Zeki, S., Ashburner, J., Penny, W.D. (Eds.), Human Brain Function, 2nd edition. Academic Press. Pfeuffer, J., Adriany, G., Shmuel, A., et al., 2002a. Perfusion-based highresolution functional imaging in the human brain at 7 Tesla. Magn. Reson. Med. 47, 903–911.
Pfeuffer, J., van de Moortele, P.F., Yacoub, E., et al., 2002b. Zoomed functional imaging in the human brain at 7 Tesla with simultaneous high spatial and high temporal resolution. NeuroImage 17, 272–286. Rijntjes, M., Buechel, C., Kiebel, S., et al., 1999. Multiple somatotopic representations in the human cerebellum. NeuroReport 10, 3653–3658. Snider, R.S., Stowell, A., 1944. Receiving areas in the cerebellum connected with the limbs. J. Neurophysiol. 7, 331–357. Talairach, J., Tournoux, P., 1988. Co-Planar Stereotactic Atlas of the Human Brain, Vol. 1. Thieme, New York. Triantafyllou, C., Hoge, R.D., Krueger, G., et al., 2005. Comparison of physiological noise at 1.5 T, 3 T and 7 T and optimization of fMRI acquisition parameters. NeuroImage 26, 243–250. Triantafyllou, C., Hoge, R.D., Wald, L.L., 2006. Effect of spatial smoothing on physiological noise in high-resolution fMRI. NeuroImage 32, 551–557. Vaughan, J.T., Garwood, M., Collins, C.M., et al., 2001. 7 T vs. 4 T: RF power, homogeneity, and signal-to-noise comparison in head images. Magn. Reson. Med. 46, 24–30. Wiggins, G.C., Potthast, A., Triantafyllou, C., et al., 2005. Eight-channel phased array coil and detunable TEM volume coil for 7 T brain imaging. Magn. Reson. Med. 54, 235–240. Yacoub, E., Shmuel, A., Pfeuffer, J., et al., 2001. Imaging brain function in humans at 7 Tesla. Magn. Reson. Med. 45, 588–594.