Journal of Medical Physics
: 2009  |  Volume : 34  |  Issue : 1  |  Page : 37--42

Ultrafast bold fMRI using single-shot spin-echo echo planar imaging

Said Boujraf1, Paul Summers2, Faouzi Belahsen3, Klaas Prussmann4, Spyros Kollias2,  
1 Department of Biophysics and Clinical MRI Methods, Faculty of Medicine and Pharmacy, University of Fez, Fez, Morocco; Institute of Neuroradiology, University Hospital of Zurich, Zurich, Switzerland
2 Institute of Neuroradiology, University Hospital of Zurich, Zurich, Switzerland
3 Neurology Department, University Hospital of Fez, Fez, Morocco
4 Institute of Biomedical Engineering, ETH, Zurich, Switzerland

Correspondence Address:
Said Boujraf
Department of Biophysics and Clinical MRI Methods, Faculty of Medicine and Pharmacy, University of Fez, BP. 1893; Km 2.200, Sidi Hrazem Road, Fez 30000; Morocco


The choice of imaging parameters for functional MRI can have an impact on the accuracy of functional localization by affecting the image quality and the degree of blood oxygenation-dependent (BOLD) contrast achieved. By improving sampling efficiency, parallel acquisition techniques such as sensitivity encoding (SENSE) have been used to shorten readout trains in single-shot (SS) echo planar imaging (EPI). This has been applied to susceptibility artifact reduction and improving spatial resolution. SENSE together with single-shot spin-echo (SS-SE) imaging may also reduce off-resonance artifacts. The goal of this work was to investigate the BOLD response of a SENSE-adapted SE-EPI on a three Tesla scanner. Whole-brain fMRI studies of seven healthy right hand-dominant volunteers were carried out in a three Tesla scanner. fMRI was performed using an SS-SE EPI sequence with SENSE. The data was processed using statistical parametric mapping. Both, group and individual subject data analyses were performed. Individual average percentage and maximal percentage signal changes attributed to the BOLD effect in M1 were calculated for all the subjects as a function of echo time. Corresponding activation maps and the sizes of the activated clusters were also calculated. Our results show that susceptibility artifacts were reduced with the use of SENSE; and the acquired BOLD images were free of the typical quadrature artifacts of SS-EPI. Such measures are crucial at high field strengths. SS SE-EPI with SENSE offers further benefits in this regard and is more specific for oxygenation changes in the microvasculature bed. Functional brain activity can be investigated with the help of single-shot spin echo EPI using SENSE at high magnetic fields.

How to cite this article:
Boujraf S, Summers P, Belahsen F, Prussmann K, Kollias S. Ultrafast bold fMRI using single-shot spin-echo echo planar imaging.J Med Phys 2009;34:37-42

How to cite this URL:
Boujraf S, Summers P, Belahsen F, Prussmann K, Kollias S. Ultrafast bold fMRI using single-shot spin-echo echo planar imaging. J Med Phys [serial online] 2009 [cited 2022 Nov 28 ];34:37-42
Available from:

Full Text


Functional magnetic resonance imaging (fMRI) based on a blood oxygen-level dependent (BOLD) contrast mechanism, has become a powerful tool for neuroscientists to investigate the functional organization of the human brain. [1],[2],[3],[4] Gradient echo planar imaging (GE-EPI) at a resolution of 64 64 is the most commonly used technique for this purpose. This approach has proven to be robust in the investigation of human brain function at 1.5 Tesla, as it is sensitive to the BOLD effect without being prone to subject motion artifacts and tissue pulsations. [5],[6],[7]

However, these techniques require excellent gradient performance, leaving the images subject to blurring and warping, caused by imperfect gradient performance, T * 2 , and off-resonance effects. Increasing the static field strength B0 only enhances the sensitivity of GE-EPI scans to blurring and wrapping caused by short T *2 and off-resonance effects. Although this may be partially accommodated for by reducing the echo times, doing so, will place an additional demand on the gradient performance. [5],[6],[7]

SE-EPI is less prone to these artifacts. [8],[9] Recently, SE-EPI experiments were suggested on both 1.5 Tesla and 3 Tesla systems to obtain spin-echo (SE) fMRI data from both the brain and spinal cord. In contrast to what is usually expected of the BOLD effect, the SE fMRI data do not show any signal changes that approach zero as the echo time (TE) approaches zero. [8],[9]

A second approach has been introduced to reduce both the blurring and warping seen in SS-EPI, which is the combination of EPI with recently introduced parallel imaging techniques. [10],[11],[12],[13],[14],[15],[16],[17] These techniques use the spatial differences in sensitivity profiles of the individual channels in detector arrays [11],[17] to reconstruct MR images from reduced field-of-view (FOV) data. For a given resolution, this under-sampling strategy could be used to reduce image artifacts by shortening the data acquisition window, [11],[17] or to reduce gradient switching rates. [13]

Parallel imaging techniques have shown great promise in applications such as cardiac imaging, [14],[15] angiography [16] and diffusion-weighted imaging. [12] However, the benefit for single-shot fMRI has not yet been completely demonstrated.

The preliminary demonstrations of SENSE fMRI showed that a twofold increase in scan speed can be achieved at a relatively small reduction in signal stability. [17] Moreover, the selection of the MRI method may however, have an impact on localization accuracy by affecting image quality and the degree of BOLD contrast achieved. [18] In order to reduce artifacts and improve spatial resolution in fMRI, [17] we have examined a combination of SENSE with single-shot spin-echo EPI imaging. This combination should offer reduced off-resonance artifacts and should improve the anatomical localization and detectability of the functional signal.

The goal of this work was to investigate the BOLD response of a SENSE-adapted single-shot spin-echo EPI using a motor task on a 3 Tesla scanner and optimize the sequence parameters for detection of activation-induced signals.

We scrutinized the ability of this approach to localize the motor-sensory activation using a simple hand-clenching task.

 Materials and Methods


Sensitivity encoding (SENSE) is a parallel imaging method that uses the spatial inhomogeneity in the sensitivity of receiver coil arrays to reduce the number of phase-encoding steps used in conventional imaging approach. This reduction is critical in decreasing the scan time and improving the spatial and temporal resolutions. To perform the unfolding of aliased single-coil images in image space, the SENSE approach requires a set of calibration images to define coil sensitivities throughout the desired FOV. [11] The number of coil elements involved in the measurement constitutes the upper limit for the reduction of the number of phase-encoding steps acquired. [11]

The signal-to-noise ratio (SNR) in the SENSE imaging approach is defined by the following relationship: [11],[12],[13],[14],[15],[16],[17],[18]


where SNR SENSE is the SNR when using the SENSE imaging approach, SNR Full is the SNR when using the conventional imaging approach, g is the local geometry factor, and R denotes the factor by which the number of samples is reduced with respect to the conventional imaging approach.

Notably, both R and g are always greater or equal to the unit [11],[12],[13] with the effect that the SNR SENSE in SENSE imaging is less than that of the equivalent image acquired without SENSE; the mathematical expressions of g and R are given in the reference. [11]

The combination of SENSE with single-shot EPI methods can be exploited to increase the matrix size while the total length of the readout period is kept constant, or to shorten the acquisition window by reducing the echo train length.

To achieve good sensitivity of the BOLD signal in fMRI measurement, a relatively long echo time is used (TE≥55ms at 1.5 Tesla and TE≥40 at 3 Tesla) [8] at higher static field strengths, however, a shorter T *2 relaxation times are limiting factor. SENSE can help in shortening the scan time when decreasing the echo times while moving to higher field strengths.

Alternatively, a general increase in spatial resolution can be achieved in areas with extremely short T *2 , [19] and typical EPI artifacts such as image distortions can be clearly reduced by the use of faster k-space traversal. [20] Recent studies have shown the possibility of combining the SENSE imaging approach with EPI techniques when operating at a static field of 3 Tesla to avoid the typical distortion occurring in typical EPI images. [21],[22],[23] Achieving images free of distortions and artifacts with high resolution is not enough in fMRI studies in ultra fast modes, because the most important issue is to achieve images with enough sensitivity to the BOLD signal induced in the brain by its involvement in a given function. Thus, the reproducibility of the localization of the activation map corresponding to a given brain activity is critical within that respect for whatever imaging technique used. Recent studies using SENSE in fMRI studies have demonstrated this possibility at low spatial resolution with lower SENSE reduction factors. [13],[14],[15]

We have used single shot spin echo EPI combined with SENSE at high resolution in our study.

MRI hardware

All the experiments were carried out in a 3 Tesla scanner (Gyroscan Intera, Philips Medical Systems, Best, Holland) using the body coil for rf-excitation and an eight-element head coil (MRI Devices Corporation, Waukesha WI, USA) connected to six independent channels. The maximum gradient strength is 30mT/m with a maximum slew rate of 150mT/ms.

fMRI protocol

Seven healthy right-handed volunteers were scanned (four males and three females, average age 33 6 years) using a paradigm designed to produce activation in the motor and sensory regions of the brain.

The subjects performed a self-paced (~1.5Hz), simple motor task with the dominant hand for 30 seconds, alternating with rest in a block design for three minutes, resulting in the collection of 72 time points. The functional experiment was performed twice for each echo time. The volunteers were instructed and trained before the scan session, and were reminded of the instructions immediately prior to each fMRI scan.

Imaging protocol

Following a three-plane localizer scan, a full-FOV reference scan was performed with the body coil and each array element. [11]

Due to its large FOV, which comprised the FOV of all subsequent SENSE imaging protocols and appropriate interpolations, a 3D spoiled gradient echo sequence was used as the reference scan. The measurement parameters were as follows: matrix size 64 64 80 slices, FOV 270mm, slice thickness 3.75mm, TE 1.53ms, TR 8ms, flip angle 7, and an acquisition time of 2 min, 36s. A reference scan was acquired once for each subject. The generation of sensitivity maps and SENSE reconstruction was performed at the scanner console using standard routines (Philips Medical Systems).

The functional experiments were conducted using single-shot SE-EPI imaging protocols with a SENSE reduction factor of R=2.75, and a high spatial resolution (128 128) acquisition. The other scan parameters were defined as follows: TR=2400ms, FOV=225mm, flip angle of 90, and 22 slices, to facilitate whole brain coverage; the slice thickness was 4mm with an interslice gap of 0.4mm. Scan duration was 3 min when 72 volumes were acquired. Likewise, the repetition time (TR) was kept constant at 2400ms while the echo time was varied from 30 to 60ms at increments of 5ms. Using the same paradigm has thus, yielded seven fMRI measurements with echo times of 30, 35, 40, 45, 50, 55, and 60ms. The bandwidth was varied for each echo time to the optimal value yielding the best signal-to-noise ratio. Phase encoding was chosen in the anterior-posterior direction because the artifact behavior is often more favorable in this direction. However, phase encoding in the left-right direction would allow for a rectangular FOV and matrix, and hence, a further reduction in measurement times. Fat suppression was achieved using spectral presaturation with inversion recovery (SPIR).

A 3-D high-resolution T1-weighted image was acquired for anatomical reference, using a multi-shot turbo gradient echo with a turbo factor of 32, a 256 256 matrix, FOV of 290mm, slice thickness of 3mm, TE=2.30ms, TR=20ms, and a flip angle of 20.

Data postprocessing

Native fMRI data was collected in DICOM, and converted to the IMG format for postprocessing.

All data underwent identical analysis with Statistical Parametric Mapping 99 (SPM 99; Wellcome Department of Cognitive Neurology, London, UK) [24] (see also

The echo-planar images were realigned using a rigid body transformation to the first volume of the time series for each subject. After this, data were spatially smoothed with a Gaussian filter (FWHM 5 5 5mm), and spatially normalized. The box-car designed the task was convolved with a hemodynamic response function was used.

T statistics were calculated for each voxel, and P *2 weighting has higher sensitivity in detecting brain activations. [26],[27],[28]

Comparing activation maps for the scans corresponding to the three echo times [Figure 1a], larger activation areas were identified in the images obtained at a higher echo time (TE=55) due to their longer effective TE, whereas more specific activations were detected in the BOLD images obtained at lower echo times.

Time courses of the BOLD signals in the activated areas from the same subject and the same echo times are shown in [Figure 1b]. They show consistent and effective changes in the BOLD signals in the activated area over time, thus fitting the paradigm design of the functional task.

BOLD signal changes are larger in BOLD images corresponding to echo times 40, 45, and 50ms with averages of 1.4, 2.1 (the largest change), and 1.66%, respectively; indicating stronger T *2 -weighting in the former three scans [Figure 2a]. Also shown in the figure is the fluctuation of the BOLD signals, implying a better contrast-to-noise ratio and hence, better detection of brain activation for BOLD.

[Figure 2b] illustrates the BOLD activation maps and the average BOLD signal change; the maximal BOLD signal changes as a function of the echo times at lower and higher echo times, are low and very similar [Figure 2a] and [Figure 1b].

The number of activated voxels due to BOLD signal changes, and activation maps obtained by single-shot SE-SENSE scans for all subjects are detailed in [Table 1] and [Figure 3] respectively. An average of 196 29 activated voxels were detected at TE=55ms, which is significantly more than the 117 29 voxels detected at TE=45ms, and the 81 13 voxels detected at TE=35ms.

This difference in the sizes of the activated areas in the motor area may be explained, at least partially, by the following arguments. First, the echo time is still-defined for single-shot SE-EPI scans with significant durations of the acquisition window but the "effective" TE is actually varied. The TE difference might be less than what it intuitively appears to be (10vs 30ms). Second, although the activation experiments were performed in brain areas with relatively fewer susceptibility gradients, field inhomogeneity always existed due to imperfect shimming and other reasons, particularly at high fields, which could degrade the image quality. As the single-shot EPI sequence is relatively sensitive to field inhomogeneity, more artifacts (e.g., signal attenuation and geometrical distortion) associated with the imperfect field, would present in the BOLD images, leading to reduced SNR.

Finally, both maximal and average signal changes and peak BOLD signals occurred at an echo time of 45ms [Figure 2a] and [Figure 2b]. At a very short echo time (TE=30ms), the sequence was still sensitive to BOLD effects whereas at higher echo times (TE=60ms), a larger volume of activation was seen in spite of a smaller BOLD effect [Figure 2a],[Figure 2b] and [Figure 3]. At all the echo times studied, susceptibility artifacts did not destroy the BOLD signal [Figure 1a] and [Figure 1b]. Quadrature ghost artifacts that are associated with higher field strengths [Figure 1] and which are typically seen in single-shot EPI, were not visible and, the distribution of activation in the motor, premotor, and sensory regions [Figure 1a] and [Figure 3] was consistent with that reported in previously reported studies for this motor task. [29],[30]


We have found that single-shot spin-echo echo planar imaging using SENSE at high magnetic fields is a robust algorithm for obtaining functional maps of neuronal activity in the motor system, but it remains sensitive to the echo time used. Susceptibility artifacts associated with higher field fMRI are reduced by the use of SENSE. With the use of appropriate imaging parameters, single-shot SE-EPI with SENSE at higher field systems produces robust maps of functional activity with reduced susceptibility artifacts and better localization of the BOLD signal in the cortical microvascular bed.

We have studied single-shot SE-EPI with SENSE scans in the context of their sensitivity to BOLD signals and their vulnerability to susceptibility-induced artifacts at 3 Tesla. BOLD imaging can be achieved using single-shot SE-EPI with SENSE scanning, with improved BOLD contrast and reduced susceptibility artifacts. Functional experiments with sensorimotor activation in normal subjects demonstrated the advantages of single-shot SE-EPI with SENSE scanning.


1Belliveau JW, Kennedy DN Jr, McKinstry RC, Buchbinder BR, Weisskoff RM, Cohen MS, et al. Functional mapping of the human visual cortex by magnetic resonance imaging. Science 1991;254:716-9.
2Ogawa S, Tank DW, Menon R, Ellermann JM, Kim SG, Merkle H, et al. Intrinsic signal changes accompanying sensory stimulation: Functional brain mapping with magnetic resonance imaging. Proc Natl Acad Sci USA 1992;89:5951-5.
3Blamire AM, Ogawa S, Ugurbil K, Rothman D, McCarthy G, Ellermann JM, et al. Dynamic mapping of the human visual cortex by high-speed magnetic resonance imaging. Proc Natl Acad Sci USA 1992;89:11069-73.
4Kollias SS, Valavanis A, Golay X, Boesiger P, McKinnon GC. Functional magnetic resonance imaging of cortical activation. Int J Neuroradiol 1996;2:450-72.
5Yang YH, Glover GH, van Gelderen P, Pater AC, Mattay VS, Frank JA, et al. A comparison of fast MR scan techniques for cerebral activation studies at 1.5 Tesla. Magn Reson Med 1998;39:61-7.
6Kruger G, Kastrup A, Glover GH. Neuroimaging at 1.5 T and 3.0 T: comparison of oxygenation-sensitive magnetic resonance imaging. Magn Reson Med 2001;45:595-604.
7Krasnow B, Tamm L, Greicius MD, Yang TT, Glover GH, Reiss AL, et al. Comparison of fMRI activation at 3 and 1.5 T during perceptual, cognitive, and affective processing. Neuroimage 2003;18:813-26.
8Stroman PW, Krause V, Frankenstein UN., Malisza KL, Tomanek B. Spin-echo versus gradient-echo fMRI with short echo times. Magn Reson Imaging 2001;19:827-31.
9Stroman PW, Krause V, Frankenstein UN, Malisza KL, Tomanek B. Characterization of contrast changes in functional MRI of the human spinal cord at 1.5 T. Magn Reson Imaging 2001;19:833-8.
10Sodickson DK, Manning WJ. Simultaneous acquisition of spatial harmonics (SMASH): Fast imaging with radiofrequency coil arrays. Magn Reson Med 1997;38:591-603.
11Pruessmann KP, Weiger M, Scheidegger MB, Boesinger P. SENSE: Sensitivity encoding for fast MRI. Magn Reson Med 1999;42:952-62.
12Bammer R, Keeling SL, Augustin M, Pruessmann KP, Wolf R, Stollberger R, et al. Improved diffusion-weighted single-shot echo-planar imaging (EPI) in stroke using sensitivity encoding (SENSE). Magn Reson Med 2001;46:548-54.
13de Zwart JA, van Gelderen P, Kellman P, Duyn JH. Reduction of gradient acoustic noise in MRI using SENSE-EPI. Neuroimage 2002;16:1151-5.
14Weiger M, Pruessmann KP, Boesinger P. Cardiac real-time imaging using SENSE. Magn Reson Med 2000;43:177-84.
15Pruessmann KP, Weiger M, Boesinger P. Sensitivity encoded cardiac MRI. J Cardiovasc Magn Reson 2001;3:1-9.
16Weiger M, Pruessmann KP, Kassner A, Roditi G, Lawton T, Reid A, et al. Contrast-enhanced 3D MRA using SENSE. J Magn Reson Imaging 2000;12:671-7.
17Golay X, Pruessmann KP, Weiger M, Crelier GR, Folkers PJ, Kollias SS, Boesinger P. PRESTO-SENSE: An ultrafast whole-brain fMRI technique. Magn Reson Med 2000;43:779-86.
18Boxerman JL, Hamberg LM, Rosen BR, Weisskoff RM. MR contrast due to intravascular magnetic susceptibility perturbations. Magn Reson Med 1995;34:555-66.
19Griswold MA, Jakob PM, Chen Q, Goldfarb JW, Manning WJ, Edelman RR, et al. Resolution enhancement in single-shot imaging using simultaneous acquisition of spatial harmonics (SMASH). Magn Reson Med 1999;41:1236-45.
20Bammer R, Keeling SL, Augustin M, Pruessmann KP, Wolf R, Stollberger R, et al. Improved diffusion-weighted single-shot echo-planar imaging (EPI) in stroke using sensitivity encoding (SENSE). Magn Reson Med 2001;46:548-54.
21Hoogenraad FG, Pouwels PJ, Hofman MB, Rombouts SA, Lavini C, Leach MO, et al. High-resolution segmented EPI in a motor task fMRI study. Magn Reson Imaging 2000;18:405-9.
22Hoogenraad FG, Pouwels PJ, Hofman MB, Reichenbach JR, Sprenger M, Haacke EM. Quantitative differentiation between BOLD models in fMRI. Magn Reson Med 2001;45:233-46.
23Hoogenraad FG, Hofman MB, Pouwels PJ, Reichenbach JR, Rombouts SA, Haacke EM. Sub-millimeter fMRI at 1.5 Tesla: Correlation of high resolution with low resolution measurements. J Magn Reson Imaging 1999;9:475-82.
24Friston KJ, Holmes AP, Worsley KJ, Poline JP, Frith CD, Frackowiak RS. Statistical parametric maps in functional imaging: A general linear approach. Hum Brain Map 1995;2:189-210.
25Yang Y, Engelien W, Pan H, Xu S, Silbersweig DA, Stern E. A CBF based event-related brain activation paradigm: Characterization of impulse-response function and comparison to BOLD. Neuroimage 2000;12:287-97.
26Hutton C, Bork A, Josephs O, Deichmann R, Ashburner J, Turner R. Image distortion correction in fMRI: A quantitative evaluation. Neuroimage 2002;16:217-40.
27Gorno-Tempini ML, Hutton C, Josephs O, Deichmann R, Price C, Turner R. Echo time dependence of BOLD contrast and susceptibility artifacts. Neuroimage 2002;15:136-42.
28Deichmann R, Josephs O, Hutton C, Corfield DR, Turner R. Compensation of susceptibility-induced BOLD sensitivity losses in echo-planar fMRI imaging. Neuroimage 2002;15:120-35.
29Kollias SS, Alkadhi H, Jaermann T, Crelier G, Hepp-Reymond MC. Identification of multiple nonprimary motor cortical areas with simple movements. Brain Res Rev 2001;36:185-95.
30Alkadhi H, Crelier GR, Boendermaker SH, Golay X, Hepp-Reymond MC, Kollias SS. Reproducibility of primary motor cortex somatotopy under controlled conditions. Am J Neuroradiol 2002;23:1524-32.