This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 79.129.229.122

This content was downloaded on 26/04/2015 at 10:10

Please note that terms and conditions apply.

Physiological measurements using ultra-high field fMRI: a review

View the table of contents for this issue, or go to the journal homepage for more

2014 Physiol. Meas. 35 R167

(http://iopscience.iop.org/0967-3334/35/9/R167)

Home Search Collections Journals About Contact us My IOPscience

R167

Physiological Measurement

Physiological measurements using ultra-high field fMRI: a review

Sue Francis and Rosa Sanchez Panchuelo1

Sir Peter Mansfield Magnetic Resonance Centre, School of Physics and Astronomy, University Park, University of Nottingham, Nottingham, NG7 2RD, UK

E-mail: rosa.panchuelo@nottingham.ac.uk

Received 1 November 2013, revised 2 May 2014Accepted for publication 4 June 2014Published 13 August 2014

AbstractFunctional MRI (fMRI) has grown to be the neuroimaging technique of choice for investigating brain function. This topical review provides an outline of fMRI methods and applications, with a particular emphasis on the recent advances provided by ultra-high field (UHF) scanners to allow functional mapping with greater sensitivity and improved spatial specificity. A short outline of the origin of the blood oxygenation level dependent (BOLD) contrast is provided, followed by a review of BOLD fMRI methods based on gradient-echo (GE) and spin-echo (SE) contrast. Phase based fMRI measures, as well as perfusion contrast obtained with the technique of arterial spin labelling (ASL), are also discussed. An overview of 7 T based functional neuroimaging is provided, outlining the potential advances to be made and technical challenges to be addressed.

Keywords: ultra-high field, fMRI, ASL, brain, haemodynamics, resting state, layer

(Some figures may appear in colour only in the online journal)

Topical Review

Printed in the UK

R167

Pmea

© 2014 Institute of Physics and engineering in medicine

2014

35

Physiol. meas.

Pmea

0967-3334

10.1088/0967-3334/35/9/R167

Topical Review

Physiological measurement

IOP

Institute of Physics and Engineering in Medicine

TOP

1Author to whom any correspondence should be addressed.

0967-3334/14/09R167+19$33.00 © 2014 Institute of Physics and Engineering in Medicine Printed in the UK

Physiol. Meas. 35 (2014) R167–R185 doi:10.1088/0967-3334/35/9/R167

Topical Review

R168

Physiol. Meas. 35 (2014) R167

1. Introduction

The technique of functional magnetic resonance imaging (fMRI) has grown to be the neuro-imaging technique of choice for investigating brain function. fMRI was developed in the early 1990s following key developments; the invention of echo planar imaging (EPI) (Mansfield 1977) which allowed MR images to be collected on a physiological time scale, and the obser-vation that an increase in oxygenation in capillaries and veins upon brain activation results in a detectable enhancement in the MRI signal, an effect termed as the blood oxygenation level dependent (BOLD) contrast (Ogawa et al 1990). Immediately after this discovery, real time vascular changes were monitored in a cat’s brain by modifying the inhaled oxygen concentration (Turner et al 1991). Two independent studies (Ogawa et al 1992; Kwong et al 1992) followed showing that the BOLD effect can be used to image human brain function. This development of functional MRI using BOLD contrast constituted an enormous advance in the field of imaging brain function. fMRI offers excellent spatial resolution, relatively high temporal resolution, and whole brain coverage, making it possible to identify the entire network of brain areas engaged in a particular task. Most importantly, fMRI is entirely non-invasive, allowing repeated, multiple scans within the same individual. This, together with the ever-increasing availability of MR scanners, led the neuroscience community to quickly adopt fMRI for a wide range of applications, from identifying active brain areas while per-forming sensorimotor tasks, to studies of the neural basis of cognition and behavior, and resting state networks.

fMRI does not measure neuronal activity directly, rather local changes in vascular proper-ties in response to neural activity which collectively give rise to the BOLD haemodynamic response. Figure 1 shows the MR methods to detect the haemodynamic and metabolic changes underlying the BOLD response: cerebral blood volume (CBV) alterations using dynamic susceptibility contrast agent (Smirnakis et al 2007), a method used in early fMRI studies (Villringer et al 1988; Belliveau et al 1991) and which has recently been applied to study activity across cortical layers (Goense et al 2012); cerebral blood flow (CBF) changes meas-ured with arterial spin labelling (ASL) (Williams et al 1992; Detre et al 1994; Kim 1995), a measure recently used to address the origin of the BOLD response (Goense et al 2012; Mullinger et al 2013; Mullinger et al 2014); and magnetic resonance spectroscopy (MRS) to directly assess metabolite changes (Mangia et al 2006; Schaller et al 2014).

This article describes the different contrast mechanisms used in fMRI studies. BOLD contrast based on gradient-echo (GE) and spin-echo (SE) measures and perfusion contrast obtained with ASL are reviewed, with an emphasis on high and ultra-high field (UHF) imag-ing. Both task induced functional changes and functional connectivity measures are discussed.

Figure 1. MR methods of measuring brain function.

Topical Review

R169

Physiol. Meas. 35 (2014) R167

2. Bold fMRI

Functional MRI based on BOLD contrast is now widely used to map brain function. BOLD contrast arises from the change in oxygenation resulting from changes in CBF, CBV and the cerebral metabolic rate of oxygen (CMRO2). In response to a local increase in neural activa-tion, the vasculature responds by increasing CBF (accompanied by an increase in CBV) to the active brain region in order to satisfy the increase in the rate of oxygen (CMRO2) and glucose (CMRglu) metabolism. In practice, the increase in local CBF exceeds the demand for increased CMRO2 in the active brain region, an effect first observed using PET (Fox and Raichle 1986), increasing the blood oxygenation in downstream capillaries and veins. The reason for this mismatch between CBF and neuronal activity remains to be completely understood and has been the subject of extensive research (Buxton et al 2004; Uludag et al 2004), as reviewed in (Kim and Ogawa 2012). Experimental studies comparing electrophysiological measurements with BOLD have shown that the BOLD effect is linearly correlated with local field potentials (Logothetis et al 2001) and evoked potentials (Arthurs and Boniface 2003), with a recent study addressing the coupling of EEG with underlying BOLD and CBF changes (Mullinger et al 2013). Current research suggests that neuronal activity is coupled to local blood flow changes through an intermediary, the astrocytes (for a review see (Petzold and Murthy 2011)).

The origin of BOLD contrast itself arises from the change in blood oxygenation on activa-tion, since deoxygenated haemoglobin is paramagnetic whilst oxygenated haemoglobin is diamagnetic (Pauling and Coryell 1936). At baseline, T2*-weighted MR images show a dark-ening in the vicinity of deoxygenated red blood cells due to the loss of signal caused by the difference in susceptibility between the paramagnetic deoxyhaemoglobin and surrounding tis-sue. On activation the oxygenation in capillaries and veins is increased to active brain region, reducing the concentration of deoxyhaemoglobin, resulting in an increase in the BOLD signal (Thulborn et al 1982). This change in the BOLD signal can be observed by acquiring a series of T2*-weighted images (typically with a 2D-EPI acquisition) while a subject performs a task or experiences some form of stimulation. The BOLD contrast depends on both intravascular (IV) and extravascular (EV) signal changes. As field strength is increased, the EV contribution dominates over the IV effects, and the percentage change in BOLD signal due to EV effects increases. Further, the EV BOLD signal originates more strongly from the dephasing around small vessels (microvasculature) compared to large vessels (macrovasculature) (Boxerman et al 1995). This has led to a rapidly growing interest in moving from high ( > 3 T) to ultra-high field ( > 7 T) MR scanners for fMRI.

3. Limitations

The vascular origin of the BOLD response imposes constraints on the temporal and spatial resolution of BOLD fMRI studies. Although neuronal changes to stimulation occur on the order of milliseconds, fMRI signal changes are temporally blurred in the vascular domain, resulting in haemodynamic changes being delayed by several seconds with respect to the neuronal activity. The BOLD signal response can be considered as a convolution of the neuro-nal activity with a haemodynamic response function (HRF) (figure 2). The HRF peaks a few seconds after the onset of the stimulation, depending on the size of the vessels from which the signal is originating (de Zwart et al 2005). The positive component of the HRF is very robust and routinely used to produce maps of functional activity. The HRF is also proposed to display an initial decrease in signal intensity approximately two seconds after stimulus onset (Duong et al 2000; Yacoub et al 2001b), termed the ‘initial dip’, reflecting an increase in

Topical Review

R170

Physiol. Meas. 35 (2014) R167

oxygen consumption prior to the onset of increased blood flow. The existence of the initial dip has remained controversial (Hu and Yacoub 2012), but a recent study has highlighted the use of trial-by-trial variability to reveal a reliable initial dip (Watanabe et al 2013). After stimulus cessation, the BOLD signal decreases to a level below the initial baseline, from which it then recovers slowly to return to baseline, this is termed the ‘post-stimulus undershoot’. Even for a brief stimulus, the time from onset to the return of the signal intensity to baseline may be as long as 30 s. To model the underlying haemodynamics, it is possible to approximate the HRF to a summation of gamma variate functions whose parameters are optimized experimentally to form a canonical HRF (Friston et al 1998).

In an fMRI experiment, a dynamic scan consisting of a set of slices through the brain is acquired in a typical measurement period, TR, of a few seconds. However, high temporal resolution (<1 s) is important in fMRI experiments for optimal modelling of the haemody-namic function and is especially important in identifying functional connectivity between brain areas. Until recently, temporal resolution for whole brain coverage at a reasonable spa-tial resolution has been poor, but recent technical advances have overcome this, as highlighted in the Human Connectome Project (Ugurbil et al 2013).

The spatial resolution of fMRI is intrinsically limited by the volume over which haemo-dynamic regulation occurs. Oxygenation changes due to increased neuronal activity are ini-tiated in the capillary bed and propagate down the vasculature to large draining pial veins. Degradation of the specificity of the BOLD response can arise from the signals originating

Figure 2. (a) BOLD haemodynamic response function (HRF) to an event stimulus applied at time t = 0 s. (b) The modelled BOLD response (solid red line) is obtained by convolving the stimulus input (solid black line) with the HRF.

Topical Review

R171

Physiol. Meas. 35 (2014) R167

from these draining veins, which may lead to mis-localization of the source of activation by as much as a few millimetres from the site of elevated neuronal activity (Turner 2002). This prob-lem has been largely ignored in standard fMRI studies, where image resolution is relatively course (~3 mm isotropic voxels) compared with such spatial errors caused by draining veins, particularly in group studies where spatial smoothing is applied prior to normalizing sub-jects’ brains to a common MNI template. However, as high spatial resolution single-subject data becomes feasible and more popular, methods to suppress the contribution from draining veins are required. The identification of draining veins based on the phase of the MR signal (Menon 2002) or co-registration of fMRI data with a high resolution T2*-weighted scan on which veins are discernable (Hall et al 2002) has been used to remove draining vein contribu-tions. Alternatively, differential fMRI paradigm designs can be used to eliminate non-specific venous activation common to two or more task conditions (Menon et al 1997) yielding higher spatial resolution maps of activation, with spatial specificity on the millimetre to sub-milli-metre scale (Duong et al 2000; Yacoub et al 2001b). Spatial specificity can also be improved using spin echo rather than gradient echo based BOLD contrast, or by directly assessing CBF using ASL imaging techniques, as discussed later.

3.1. Improving the spatial resolution with ultra-high field

fMRI is limited in its spatial resolution by the available BOLD contrast-to-noise ratio (CNR), this is principally determined by the strength of the magnetic field. As magnetic field strength increases, BOLD signal changes increase (Gati et al 1997; Yacoub et al 2001a; van der Zwaag et al 2009). BOLD contrast depends on the ratio of the change in the transverse tissue relaxation rate on activation (ΔR2*) relative to the resting relaxation rate R2* (equal to 1/T2*) and is optimal when the echo time (TE) used matches the relax-ation time T2* of grey matter. A study comparing the BOLD signal change to a simple motor task across three field strengths (1.5, 3 and 7 T) (van der Zwaag et al 2009) showed a linear increase in ΔR2* with field strength, and an increase in ΔR2*/R2*, resulting in a substantial increase in the BOLD signal at ultra-high field (figure 3). This increase in BOLD contrast, in conjunction with the increased intrinsic SNR, yields an increase in BOLD CNR which can be exploited to improve the spatial resolution and/or decrease the number of repeats required to demonstrate robust activation in functional mapping. However, physiological noise, fluctuations associated with the cardiac cycle and respira-tion, has been shown to increase with magnetic field strength (Triantafyllou et al 2005; Kruger and Glover 2001). Nevertheless, at high spatial resolution (<1.5 mm isotropic) the physiological noise contribution at 7 T is significantly reduced to a level where fluctuations are thermal noise dominated (Triantafyllou et al 2005), and the benefit of UHF can be most fully realized. Furthermore, as the magnetic field strength increases, BOLD signal changes are more strongly weighted to the microvasculature signal from the capillary bed relative to signal from macrovasculature (Duong et al 2003); the EV BOLD signal from the capil-laries increases with the square of the magnetic field strength whilst large veins show a linear increase (Thulborn et al 1982; Gati et al 1997; Duong et al 2003). UHF also offers intrinsically improved spatial specificity when using echo times (TE) optimized for grey matter, since the IV signal is diminished due to the disproportional shortening of blood T2 compared to tissue T2 or T2* (Thulborn et al 1982; Gati et al 1997; Duong et al 2003). The width of the cortical point spread function (PSF) of the gradient-echo signal in grey matter ranges from 3.9 mm at 3 T (Parkes et al 2005) to less than 2 mm at 7 T (Shmuel et al 2007) and 1.6 mm at 9.4 T (Park et al 2004), reducing as contributions from large draining veins are removed. Recent studies of layer-specific activation using gradient-echo BOLD (Harel

Topical Review

R172

Physiol. Meas. 35 (2014) R167

et al 2006; Ress et al 2007; Koopmans et al 2010; Polimeni et al 2010) suggest that the intrinsic spatial resolution is at the sub-millimetre level if contributions from pial veins are excluded. The use of GE-BOLD fMRI at high field in combination with differential fMRI paradigms has allowed the mapping of ocular dominance columns in humans with sub-millimetre accuracy (Menon et al 1997; Cheng et al 2001).

3.2. Spin-echo BOLD

fMRI signal changes can also be observed using spin echo (SE) EPI T2-weighted BOLD contrast. A spin echo refocuses the static dephasing induced by field inhomogeneities around large vessels, thus reducing the functional contrast compared to gradient-echo BOLD con-trast by eliminating the EV contribution from large vessels to the BOLD effect. However, as a consequence, SE-BOLD contrast is sensitive to signal changes around small venules and capillaries (Kiselev and Posse 1999; Duong et al 2003; Yacoub et al 2003), offering a gain in spatial accuracy (Jochimsen et al 2004; Jochimsen and Moller 2008), with a sharper PSF of SE-BOLD compared to GE-BOLD (Olman et al 2004). The improvement in spatial accuracy is best realized at ultra-high field due to the accompanying reduction of the IV contribution to the BOLD signal. Improvements in the spatial specificity of the SE BOLD response with respect to GE BOLD contrast have been demonstrated in human visual cortex (Duong et al 2002; Duong et al 2003; Yacoub et al 2003), and motor cortex at 7 T (Harmer et al 2012) (figure 4) and 9.4 T (Budde et al 2014), and across the depth of the cortex (Siero et al 2013). The increased specificity of SE-BOLD at 7 T has been exploited to reveal robust and reproducible

Figure 3. Gradient echo BOLD time-course of a motor task at field strengths of 1.5, 3 and 7 T. Image adapted from van der Zwaag et al (2009) with permission. Copyright 2009 by Elsevier.

Topical Review

R173

Physiol. Meas. 35 (2014) R167

maps of ocular dominance (Yacoub et al 2007) and orientation columns (Yacoub et al 2008) in the human visual cortex, as well as layer specific activity (Olman et al 2012).

3.3. Technical challenges at ultra-high field

Despite the clear advances of UHF for fMRI studies, this is accompanied by technical chal-lenges related to the decrease in T2/T2* relaxation times, increased B0 and B1 inhomogeneity, and increased specific absorption rate (SAR) as a consequence of a quadratic increase in RF power.

At UHF, the acquisition of high spatial resolution fMRI can be limited by the need for a short echo time (imposed by a grey matter T2* of ~ 25 ms) and increased vulnerability to susceptibility-induced geometric distortions and signal loss in EPI acquisitions. Parallel imag-ing techniques such as sensitivity encoding (SENSE) imaging (Pruessmann et al 1999) which uses multiple receive coils, can alleviate this problem, as the echo train length of EPI images is shortened thus reducing image distortion.

Magnetic field inhomogeneity can be minimized by using improved shimming capability (using higher order shim coils) and improved shimming methods. For example, image-based shimming requires the acquisition of a field map which is used to characterize field inhomoge-neity (ΔB0) and to determine the shim coil currents needed to minimize ΔB0 over the region of interest. Use of local image-based shimming (Wilson et al 2002; Poole and Bowtell 2008) has been shown to restrict geometric distortions to less than one voxel (1 mm) in GE-EPI data at 7 T (Sanchez-Panchuelo et al 2010). B1 inhomogeneity increases at UHF as the RF wavelength

Figure 4. Activation maps overlaid on corresponding echo planar images for GE and SE data and fMRI time series showing the BOLD percentage signal change for GE and SE data. Image adapted from Harmer et al (2012) with permission. Copyright 2011 by Wiley.

Topical Review

R174

Physiol. Meas. 35 (2014) R167

is short compared to the dimensions of the head, causing an inhomogeneous flip angle and central brightening in the images. Various methods have been proposed to reduce these strong transmit B1 inhomogeneities, such as multichannel parallel transmit for B1 shimming tech-niques (Adriany et al 2005; Adriany et al 2008; De Martino et al 2012) and the use of spatially targeted RF pulses with amplitude and phase modulation (Setsompop et al 2008; Zelinski et al 2008). High permittivity dielectric pads have also been shown to increase image quality at 7 T in regions of low radiofrequency transmit efficiency (Teeuwisse et al 2012), whilst tailored RF pulses, such as the optimized TR-FOCI (Hurley et al 2010), can be employed to improve the image quality for sequences that use inversion pulses, such as magnetization prepared rapid acquisition gradient echo structural scans and arterial spin labelling schemes.

SAR can be become limiting at UHF, particularly when using a SE-EPI acquisition for fMRI. A multi-slice SE-EPI acquisition with fat suppression can be severely restricted in num-ber of slices that can be acquired in a given repetition time compared to a multi-slice GE-EPI fMRI acquisition. To overcome this Slice Selective Gradient Reversal (SSGR) (Gomori et al 1988; Harmer et al 2012), in which the fat signal is sufficiently shifted relative to the excited water signal in the imaging slice, can be used. Standard 2D GE-EPI can also approach the SAR limit when a large number of slices are acquired. 3D GE-EPI, which uses a lower flip angle provides an attractive alternative, particularly at high spatial resolution for a large num-ber of slices, as the SNR of 3D GE-EPI is superior to 2D GE-EPI acquisitions, and parallel imaging allows acceleration in two spatial dimensions, significantly reducing the total volume acquisition time (Barry et al 2012; van der Zwaag et al 2012).

3.4. Applications

The majority of fMRI studies at high field to date have collected data with a spatial resolution of a few millimetres, and performed a group analysis across subjects to increase the statisti-cal power. In group analysis, individual subjects’ data is spatially normalized to a template brain (typically the MNI template), having first applied spatial smoothing of typically twice the voxel size to account for differences in brain morphology between subjects, limiting the spatial resolution to about 8–10 mm. However, as the need for clinical fMRI increases, the goal is to interpret the fMRI results on an individual subject basis.

The improvements in sensitivity and spatial specificity at UHF have improved the resolving power of fMRI to sub-millimetre isotropic resolution. Figure 5 illustrates 0.75 mm isotropic 3D GE-EPI data acquired at 7 T with a 32 channel receiver coil and SENSE acquisition factor of 2, showing excellent quality EP images which also reflect good structural detail. In such state-of-the-art data, the functional contrast-to-noise of the BOLD images is sufficiently high that spatial smoothing is not necessary or advisable. In fact, for such fMRI applications it is often crucial to not spatially smooth, to retain the high spatial resolution by reducing the par-tial volume effect of averaging voxels together.

The advantage of UHF fMRI has largely been exploited in targeted studies of partial coverage of the brain. The development of ‘travelling wave’ or ‘phase-encoding’ paradigm designs, which allow the identification of cortical mappings with respect to continuously varying parameters, has advanced precision of fMRI measures (Engel et al 1994; Engel 2012). This has been demonstrated in high resolution functional (retinotopic) and struc-tural (linked to myelination) parcellation of primary visual cortex V1 (Sanchez-Panchuelo et al 2012b), topographic mapping of the human tonotopic organization in the auditory cortex (Formisano et al 2003; Talavage et al 2004) and inferior colliculus (De Martino et al 2013a), and the somatotopic mapping of the digits of the hand (Sanchez-Panchuelo et al 2010) and fine scale within-finger somatotopic organization in the primary somatotopy

Topical Review

R175

Physiol. Meas. 35 (2014) R167

cortex (Sanchez-Panchuelo et al 2012a). In addition to cortical topographic maps, finger somatotopy has been mapped in the cerebellum (van der Zwaag et al 2013) and retinotopy has been mapped in the Lateral Geniculate Nucleus (LGN) (Chen et al 1999). Submillimetre columnar architectures, such as ocular dominance columns (Cheng et al 2001; Yacoub et al 2007) and orientation columns (Yacoub et al 2008) in the visual cortex have also recently been revealed with fMRI and laminar specific activations across the width of the cortical layer resolved (Koopmans et al 2010). fMRI of cortical layer activity is now possible (De Martino et al 2013b), using different conditions to preferentially activate input and output layers of the cortex, providing the possibility to study the brain at a level which so far remains relatively unexplored.

These high spatial resolution studies have, until recently, been confined to a small region of the brain. However, multiplex techniques where multiple slices are simultaneously excited and sampled, have recently been developed (Moeller et al 2010; Feinberg et al 2010), pro-viding whole brain coverage while maintaining high spatial resolution and short acquisition times. These method have facilitated high spatial and temporal investigations of functional connectivity networks (De Martino et al 2011), as outlined further below.

Recent developments have also been shown in fMRI analysis, such as the use of mul-tivariate pattern analysis (MVPA) to decode brain activity patterns (Norman et al 2006; Spiers and Maguire 2007). MVPA algorithms applied to relatively coarse spatial resolution 3 T fMRI data were shown to decode visual stimuli functionally segregated at the sub-millimetre resolution (Haxby et al 2001; Haynes and Rees 2005; Kamitani and Tong 2006), and reveal voxels that demonstrate preferential activation corresponding to a unique cortical column (such as an ocular dominance, orientation or direction of motion specific column, known to have dimensions of ~1 mm). It has been shown that high spatial resolution fMRI further improves the classification performance obtained with MVPA (Swisher et al 2010; Beckett et al 2012).

Figure 5. Statistical activation maps to vibrotactile stimulation of the left hand digits overlaid on 0.75 mm isotropic 3D GE-EPI data acquired at 7 T. The effect of spatial smoothing with a FWHM of 1.5 and 3 mm is also shown; the unsmoothed activation map is closely matched to the cortical ribbon, whilst for smoothed data the activated region extends into adjacent white matter.

Topical Review

R176

Physiol. Meas. 35 (2014) R167

4. Phase based fMRI measures

Conventionally, BOLD fMRI is measured from the magnitude MR images. Only a few studies have investigated BOLD contrast using phase images (Menon 2002; Rowe et al 2007; Tomasi and Caparelli 2007; Hagberg et al 2008; Hahn et al 2009; Petridou et al 2009; Arja et al 2010; Hagberg et al 2012; Bianciardi et al 2014; Chen et al 2013). At the spatial resolution allowed by conventional magnetic fields, the BOLD phase effect is averaged out due to the orienta-tion dependence of microscopic field perturbation effects, hence substantial phase contrast can only be found near large veins of diameter comparable to the voxel dimensions. At high spatial resolution, the phase of the fMRI time-series have been used to identify the dominant non-local BOLD effects due to large veins and to remove their contribution from BOLD statis-tical maps (Menon 2002). The introduction of quantitative susceptibility maps (QSM) (Li and Leigh 2004; de Rochefort et al 2008; Shmueli et al 2009; Liu et al 2009; Wharton and Bowtell 2010; Schweser et al 2011) has opened up new possibilities to use phase images for fMRI studies. Bianciardi et al (2014) generated QSM for each volume in the phase time-series of 2.5 mm isotropic fMRI data acquired at 7 T in order to compute activation related susceptibil-ity change maps. They showed that functional, task related magnitude and phase changes can be detected with comparable sensitivity, and that these changes have the same BOLD origins. Balla et al (2014) used single orientation (SO) and multiple orientation (MO) experiments to assess the potential of functional QSM relative to standard magnitude BOLD fMRI data at 7 T.

5. Direct CBF fMRI measures

The direct change in CBF due to increased neuronal activation can be measured using arte-rial spin labelling (ASL) fMRI (Williams et al 1992). ASL uses labelled inflowing arterial blood as an endogenous tracer, allowing the non-invasive quantification of regional brain tissue perfusion (CBF). In ASL, the arterial blood upstream from the area of interest is mag-netically labelled by applying a single radiofrequency pulse (pulsed ASL) (Edelman et al 1994; Kim 1995; Kwong et al 1995), or a train of pulses (continuous or pseudocontinuous ASL (Wu et al 2007)) to invert the blood water magnetization. After a period of time—the post-label delay—to allow labelled spins to flow into the imaging plane, an image (label) is acquired. A second image (control) is acquired without labelling of inflowing arterial blood. Label images are then subtracted from control images to yield a perfusion weighted image that can be quantified in terms of cerebral blood flow in ml/100 g/min (Buxton et al 1998). The temporal resolution of perfusion based fMRI is lower than BOLD fMRI due to the time required to collect both a label and control pair, and the CNR of perfusion-based func-tional maps is lower than BOLD. However, as field strength increases the T1 of arterial blood increases, giving rise to an increase in the perfusion weighted difference signal and so perfu-sion CNR, which peaks at longer post-label delay. The feasibility, and advantages, of UHF ASL measurements have been demonstrated in pulsed ASL based fMRI studies conducted in the human brain at 7 T (Pfeuffer et al 2002; Gardener et al 2009), and the introduction of 3 D image acquisition strategies may provide improved resolution and coverage of per-fusion images for functional studies. Figure 6 shows an example of BOLD and perfusion weighted statistical maps acquired at 7 T in response to a finger tapping task. ASL fMRI is more spatially localized to the active tissue than BOLD fMRI (Silva et al 1999; Duong et al 2000), since the perfusion signal changes originate from the small arterioles and the parenchyma. Perfusion based fMRI has been shown to yield high-quality functional maps of orientation columns in the cat visual cortex (Duong et al 2001) at high field strength (4.7 T),

Topical Review

R177

Physiol. Meas. 35 (2014) R167

demonstrating the potential of this technique for high resolution mapping of sub-millimetre functional organization in humans.

Unlike BOLD, ASL permits measurements of baseline levels of CBF as well as activa-tion dependent changes in CBF. Since BOLD contrast reflects a complex interaction between haemodynamic changes in CBF, CBV, blood oxygenation and also depends on the baseline physiological state, this can make it difficult to interpret BOLD signal changes. Goense et al (Goense et al 2012) have used high resolution fMRI at 4.7 T in monkeys to study layer-specific fMRI signals, and shown that regions with positive BOLD responses have parallel increases in CBV and CBF, whereas areas with negative BOLD responses show a decrease in CBF but an increase in CBV, suggesting different mechanisms for neurovascular coupling for BOLD increases and decreases. Further, baseline CBF measures from ASL can be used to study physiological factors (e.g. breathing CO2 (Cohen et al 2002)) and pharmacological interven-tions, and how this is altered by vascular and neuronal diseases.

ASL is typically limited to a few brain slices and yields lower contrast-to-noise ratio than BOLD. However, more efficient labeling approaches are being developed to improve SNR while reducing RF power at ultra-high field, together with technical improvements such us 3D acquisitions, background suppression and multiplex excitation, this will allow whole brain coverage with increasing resolution. ASL offers the advantage of producing a quantifiable signal that is more directly related to neural activity, and may eventually replace BOLD as the technique of choice in certain applications, particularly in longitudinal studies and pharma-cological fMRI.

6. Functional connectivity using resting state fMRI

Until relatively recently, most fMRI studies had studied the response to a task or stimulus. However, brain activity continues in the absence of goal-directed neuronal action, and in recent years the importance of measuring connectivity between spatially separate but functionally related brain areas when at rest has become of considerable interest. This intrinsic neuronal activity manifests as spontaneous fluctuations in the fMRI time series (Gusnard and Raichle 2001), with this being suggested to be dominated by low frequency fluctuations (<0.1 Hz) and more recently by spontaneous events (Petridou et al 2013). Functional MRI can be applied to

Figure 6. Statistical maps of BOLD and ASL (perfusion) signal changes to a motor task overlaid on a BOLD weighted image and perfusion weighted image respectively. Improved spatial localization of ASL task-induced changes can be seen. BOLD and ASL data simultaneously acquired at 2 mm × 2 mm × 3 mm resolution using a double ac-quisition background suppression (DABS) sequence at 7 T, University of Nottingham.

Topical Review

R178

Physiol. Meas. 35 (2014) R167

identify functional connectivity (fc) between brain regions by measuring these spontaneous fluctuations. When spontaneous activity is strongly temporally correlated across two or more spatially segregated brain regions, these regions are assumed to be functionally connected. In resting state functional connectivity experiments, subjects are typically instructed to lie still in the scanner with their eyes closed (or opened while fixating on a cross) while fMRI data is collected (Fox et al 2005; Damoiseaux et al 2006).

There are a number of methods to study connectivity between brain regions and thus identify functional networks; seed-based correlation, where a seed voxel is pre-specified to determine brain areas which correlate temporally with this seed, or independent component analysis methods which allow identification of resting state networks based on spatial or temporal independence (Beckmann and Smith 2004) are commonly used. The first functional connectivity study was performed by Biswal et al (Biswal et al 1995), who demonstrated that under resting conditions the BOLD fMRI time series of a voxel in the sensorimotor cortex was strongly correlated with the time-series of other voxels in the contralateral sensorimo-tor cortex, suggesting functional communication between these regions during rest. Several groups have since employed resting-state fMRI to identify other resting state functional net-works (such as the visual network (Lowe et al 1998; Cordes et al 2000), auditory network (Cordes et al 2000), and default mode network (Gusnard et al 2001; Raichle et al 2001; Greicius et al 2003).

Resting state fMRI has become increasingly popular to study alterations in networks in the clinical setting. It offers several advantages for clinical use over task-related fMRI, given that

Figure 7. Functional connectivity maps (Pearson correlation coefficient images) for sensorimotor and default mode networks overlaid onto a standard MNI brain. The seed is located in right sensorimotor cortex and posterior cingulate cortex for SMN and DMN connectivity analysis respectively. Low spatial resolution (3 mm isotropic) 3 T data and high spatial resolution 3 T and 7 T data (1.5 × 1.5 × 3 mm3) shown for different spatial smoothing kernels (FWHM of 3, 4 and 5 mm) prior to correlation computation. Image adapted from Hale et al (2010) with permission. Copyright 2010 by Springer.

Topical Review

R179

Physiol. Meas. 35 (2014) R167

it does not require patients to perform a task nor hardware for stimulus presentation. Thus, resting state fMRI is very suitable to study patient populations who cannot accurately perform tasks due to physical impairment or cognitive dysfunction, eliminating confounds of differ-ences in the level of task performance. Resting state fMRI measures have been correlated to clinical variables in patient populations and controls, with applications in neurodegenerative brain diseases (Li et al, 2002; Lowe et al 2002; Greicius et al 2004; Wang et al 2007; Rocca et al 2010; Bonavita et al 2011), and neuropsychiatric disorders (Cherkassky et al 2006; Calhoun et al 2009; Church et al 2009).

Recent studies at 7 T (Hale et al 2010; Newton et al 2012) have shown that, despite the lower signal-to-noise ratio of smaller voxels, the detectability of functional connectivity within the sensory-motor system increases as the spatial resolution increases, possibly due to decreased partial volume effects, demonstrating the potential benefits of UHF for detecting and analyzing functional connectivity. figure  7 shows the increased detectability of func-tional networks at 7 T compared to 3 T. Further, 7 T has allowed the study of resting state connectivity at both a fine and large scale within and across visual areas V1, V2 and V3 (Raemaekers et al 2014).

7. Summary

In summary, the field of fMRI has rapidly grown, with applications in basic sensory pro-cessing and cognitive neuroscience, as well as clinical applications. The recent advent of ultra-high field scanners, has led to significant advantages in terms of both the increase in the contrast to noise ratio (CNR) and the spatial specificity of functional MRI studies. This is an area which will continue to expand with improvements in hardware and analysis tech-niques, allowing functional measures of the human brain with even greater sensitivity and detailed specificity, to approach whole brain sub-millimetre spatial resolution acquisitions in future studies.

References

Adriany G, van de Moortele P F, Ritter J, Moeller S, Auerbach E J, Akgun C, Snyder C J, Vaughan T and Ugurbil  K 2008 A geometrically adjustable 16-channel transmit/receive transmission line array for improved RF efficiency and parallel imaging performance at 7 Tesla Magn. Reson. Med. 59 590–7

Adriany G, et al 2005 Transmit and receive transmission line arrays for 7 Tesla parallel imaging Magn. Reson. Med. 53 434–45

Arja S K, Feng Z, Chen Z, Caprihan A, Kiehl K A, Adali T and Calhoun V D 2010 Changes in fMRI magnitude data and phase data observed in block-design and event-related tasks Neuroimage 49 3149–60

Arthurs O J and Boniface S J 2003 What aspect of the fMRI BOLD signal best reflects the underlying electrophysiology in human somatosensory cortex? Clin. Neurophysiol. 114 1203–9

Balla D Z, Sanchez-Panchuelo R M, Wharton S, Hagberg G E, Scheffler K, Francis S and Bowtell R 2014 Functional quantitative susceptibility mapping (fQSM) Neuro Image at press (doi:10.1016/j.neuroimage.2014.06.011)

Barry R L, Strother S C and Gore J C 2012 Complex and magnitude-only preprocessing of 2D and 3D BOLD fMRI data at 7 T Magn. Reson. Med. 67 867–71

Beckett A, Peirce J W, Sanchez-Panchuelo R M, Francis, S and Schluppeck D 2012 Contribution of large scale biases in decoding of direction-of-motion from high-resolution fMRI data in human early visual cortex Neuroimage 63 1623–32

Beckmann  C F and Smith  S A 2004 Probabilistic independent component analysis for functional magnetic resonance imaging IEEE Trans. Med. Imag. 23 137–52

Topical Review

R180

Physiol. Meas. 35 (2014) R167

Belliveau J W, Kennedy D N Jr, Mckinstry R C, Buchbinder B R, Weisskoff R M, Cohen M S, Vevea J M, Brady T J and Rosen B R 1991 Functional mapping of the human visual cortex by magnetic resonance imaging Science 254 716–9

Bianciardi M, Van Gelderen P and Duyn J H 2014 Investigation of BOLD fMRI resonance frequency shifts and quantitative susceptibility changes at 7 T Hum. Brain Mapp. 35 2191–205

Biswal B, Yetkin F Z, Haughton V M and Hyde J S 1995 Functional connectivity in the motor cortex of resting human brain using echo-planar MRI Magn. Reson. Med. 34 537–41

Bonavita S et al 2011 Distributed changes in default-mode resting-state connectivity in multiple sclerosis Mult. Scler. J. 17 411–22

Boxerman J L, Bandettini P A, Kwong K K, Baker J R, Davis T L, Rosen B R and Weisskoff R M 1995 The intravascular contribution to fMRI signal change: Monte Carlo modeling and diffusion-weighted studies in vivo Magn. Reson. Med. 34 4–10

Budde J, Shajan G, Scheffler K and Pohmann R 2014 Ultra-high resolution imaging of the human brain using acquisition-weighted imaging at 9.4 T Neuroimage 86 592–8

Buxton R B, Uludag K, Dubowitz D J and Liu T T 2004 Modeling the hemodynamic response to brain activation Neuroimage 23 S220–33

Buxton R B, Wong E C and Frank L R 1998 Dynamics of blood flow and oxygenation changes during brain activation: the balloon model Magn. Reson. Med. 39 855–64

Calhoun V D, Eichele T and Pearlson G 2009 Functional brain networks in schizophrenia: a review Front. Hum. Neurosci. 3 17

Chen W, Zhu X H, Thulborn K R and Ugurbil K 1999 Retinotopic mapping of lateral geniculate nucleus in humans using functional magnetic resonance imaging Proc. Natl Acad. Sci. USA 96 2430–4

Chen Z, Liu J and Calhoun V D 2013 Susceptibility-based functional brain mapping by 3D deconvolution of an MR-phase activation map J. Neurosci. Methods 216 33–42

Cheng K, Waggoner R A and Tanaka K 2001 Human ocular dominance columns as revealed by high-field functional magnetic resonance imaging Neuron 32 359–74

Cherkassky V L, Kana R K, Keller T A and Just M A 2006 Functional connectivity in a baseline resting-state network in autism Neuroreport 17 1687–90

Church J A, Fair D A, Dosenbach N U F, Cohen A L, Miezin F M, Petersen S E and Schlaggar B L 2009 Control networks in paediatric Tourette syndrome show immature and anomalous patterns of functional connectivity Brain 132 225–38

Cohen E R, Ugurbil K and Kim S G 2002 Effect of basal conditions on the magnitude and dynamics of the blood oxygenation level-dependent fMRI response J. Cerebr. Blood Flow Metab. 22 1042–53

Cordes  D, Haughton  V M, Arfanakis  K, Wendt  G J, Turski  P A, Moritz  C H, Quigley  M A and Meyerand M E 2000 Mapping functionally related regions of brain with functional connectivity MR imaging Am. J. Neuroradiol. 21 1636–44

Damoiseaux J S, Rombouts S A R B, Barkhof F, Scheltens P, Stam C J, Smith S M and Beckmann C F 2006 Consistent resting-state networks across healthy subjects Proc. Natl Acad. Sci. USA 103 13848–53

De Martino F, Esposito F, Van de Moortele P F, Harel N, Formisano E, Goebel R, Ugurbil K and Yacoub E 2011 Whole brain high-resolution functional imaging at ultra high magnetic fields: an application to the analysis of resting state networks Neuroimage 57 1031–44

De Martino F, Moerel M, Van de Moortele P F, Ugurbil K, Goebel R, Yacoub E and Formisano E 2013a Spatial organization of frequency preference and selectivity in the human inferior colliculus Nature Commun 4 1386

De Martino F, Schmitter S, Moerel M, Tian J, Ugurbil K, Formisano E, Yacoub E and De Moortele P F 2012 Spin echo functional MRI in bilateral auditory cortices at 7 T: an application of B(1) shimming Neuroimage 63 1313–20

De Martino  F, Zimmermann  J, Muckli  L, Ugurbil  K, Yacoub  E and Goebel  R 2013b Cortical depth dependent functional responses in humans at 7T: improved specificity with 3D GRASE PLoS One 8 e60514

De Rochefort L, Brown R, Prince M R and Wang Y 2008 Quantitative MR susceptibility mapping using piece-wise constant regularized inversion of the magnetic field Magn. Reson. Med. 60 1003–9

De Zwart J A, Silva A C, Van Gelderen P, Kellman P, Fukunaga M, Chu R, Koretsky A P, Frank J A and Duyn  J H 2005 Temporal dynamics of the BOLD fMRI impulse response Neuroimage 24 667–77

Detre J A, Zhang W G, Roberts D A, Silva A C, Williams D S, Grandis D J, Koretsky A P and Leigh J S 1994 Tissue-specific perfusion imaging using arterial spin-labeling NMR Biomed. 7 75–82

Topical Review

R181

Physiol. Meas. 35 (2014) R167

Duong T Q, Kim D S, Ugurbil K and Kim S G 2000 Spatiotemporal dynamics of the BOLD fMRI signals: toward mapping submillimeter cortical columns using the early negative response Magn. Reson. Med. 44 231–42

Duong  T Q, Kim  D S, Ugurbil  K and Kim  S G 2001 Localized cerebral blood flow response at submillimeter columnar resolution Proc. Natl Acad. Sci. USA 98 10904–9

Duong T Q, Yacoub E, Adriany G, Hu X, Ugurbil K and Kim S G 2003 Microvascular BOLD contribution at 4 and 7 T in the human brain: gradient-echo and spin-echo fMRI with suppression of blood effects Magn. Reson. Med. 49 1019–27

Duong T Q, Yacoub E, Adriany G, Hu X, Ugurbil K, Vaughan J T, Merkle H and Kim S-G 2002 High-resolution, spin-echo BOLD, and CBF fMRI at 4 and 7 T Magn. Reson. Med.: Off. J. Soc. Magn. Reson. Med./Soc. Magn. Reson. Med. 48 589–93

Edelman R R, Siewert B, Darby D G, Thangaraj V, Nobre A C, Mesulam M M and Warach S 1994 Qualitative mapping of cerebral blood flow and functional localization with echo-planar MR imaging and signal targeting with alternating radio frequency Radiology 192 513–20

Engel  S A 2012 The development and use of phase-encoded functional MRI designs Neuroimage 62 1195–200

Engel S A, Rumelhart D E, Wandell B A, Lee A T, Glover G H, Chichilnisky E J and Shadlen M N 1994 fMRI of human visual cortex Nature 369 525

Feinberg D A, Moeller S, Smith S M, Auerbach E, Ramanna S, Glasser M F, Miller K L, Ugurbil K and Yacoub E 2010 Multiplexed echo planar imaging for sub-second whole brain FMRI and fast diffusion imaging Plos One 5 e15710

Formisano E, Kim D S, Di Salle F, Van de Moortele P F, Ugurbil K and Goebel R 2003 Mirror-symmetric tonotopic maps in human primary auditory cortex Neuron 40 859–69

Fox M D, Snyder A Z, Vincent J L, Corbetta M, Van Essen D C and Raichle M E 2005 The human brain is intrinsically organized into dynamic, anticorrelated functional networks Proc. Natl Acad. Sci. USA 102 9673–8

Fox P T and Raichle M E 1986 Focal physiological uncoupling of cerebral blood flow and oxidative metabolism during somatosensory stimulation in human subjects Proc. Natl Acad. Sci. USA 83 1140–4

Friston  K J, Fletcher  P, Josephs  O, Holmes  A, Rugg  M D and Turner  R 1998 Event-related fMRI: characterizing differential responses Neuroimage 7 30–40

Gardener A G, Gowland P A and Francis S T 2009 Implementation of quantitative perfusion imaging using pulsed arterial spin labeling at ultra-high field Magn. Reson. Med. 61 874–82

Gati J S, Menon R S, Ugurbil K and Rutt B K 1997 Experimental determination of the BOLD field strength dependence in vessels and tissue Magn. Reson. Med. 38 296–302

Goense  J, Merkle  H and Logothetis  N K 2012 High-resolution fMRI reveals laminar differences in neurovascular coupling between positive and negative BOLD responses Neuron 76 629–39

Gomori J M, Holland G A, Grossman R I, Gefter, W B and Lenkinski R E 1988 Fat suppression by section-select gradient reversal on spin-echo MR imaging. Work in progress Radiology 168 493–5

Greicius M D, Krasnow B, Reiss A L and Menon V 2003 Functional connectivity in the resting brain: a network analysis of the default mode hypothesis Proc. Natl Acad. Sci. USA 100 253–8

Greicius M D, Srivastava G, Reiss A L and Menon V 2004 Default-mode network activity distinguishes Alzheimer’s disease from healthy aging: evidence from functional MRI Proc. Natl Acad. Sci. USA 101 4637–42

Gusnard D, Akbudak E, Shulman G and Raichle M E 2001 Role of medial prefrontal cortex in a default mode of brain function Neuroimage 13 S414

Gusnard D A and Raichle M E 2001 Searching for a baseline: functional imaging and the resting human brain Nature Rev. Neurosci. 2 685–94

Hagberg G E, Bianciardi M, Brainovich V, Cassara A M and Maraviglia B 2008 The effect of physiological noise in phase functional magnetic resonance imaging: from blood oxygen level-dependent effects to direct detection of neuronal currents Magn. Reson. Imag. 26 1026–40

Hagberg  G E, Bianciardi  M, Brainovich  V, Cassara  A M and Maraviglia  B 2012 Phase stability in fMRI time series: effect of noise regression, off-resonance correction and spatial filtering techniques Neuroimage 59 3748–61

Hahn A D Nencka A S and Rowe D B 2009 Improving robustness and reliability of phase-sensitive fMRI analysis using temporal off-resonance alignment of single-echo timeseries (TOAST) Neuroimage 44 742–52

Hale  J R, Brookes  M J, Hall  E L, Zumer  J M, Stevenson  C M, Francis  S T and Morris  P G 2010 Comparison of functional connectivity in default mode and sensorimotor networks at 3 and 7T MAGMA 23 339–49

Topical Review

R182

Physiol. Meas. 35 (2014) R167

Hall D A, Goncalves M S, Smith S, Jezzard P, Haggard M P and Kornak J 2002 A method for determining venous contribution to BOLD contrast sensory activation Magn. Reson. Imag. 20 695–706

Harel  N, Lin  J, Moeller  S, Ugurbil  K and Yacoub  E 2006 Combined imaging-histological study of cortical laminar specificity of fMRI signals Neuroimage 29 879–87

Harmer  J, Sanchez-Panchuelo  R M, Francis  S and Bowtell  R 2012 Spatial location and strength of BOLD activation in high-spatial-resolution fMRI of the motor cortex: a comparison of spin echo and gradient echo fMRI at 7T NMR Biomed. 25 717–25

Haxby J V, Gobbini M I, Furey M L, Ishai A, Schouten J L and Pietrini P 2001 Distributed and overlapping representations of faces and objects in ventral temporal cortex Science 293 2425–30

Haynes  J D and Rees  G 2005 Predicting the orientation of invisible stimuli from activity in human primary visual cortex Nature Neurosci. 8 686–91

Hu X and Yacoub E 2012 The story of the initial dip in fMRI Neuroimage 62 1103–8Hurley A C Al-Radaideh A, Bai L, Aickelin U, Coxon R, Glover P and Gowland P A 2010 Tailored RF

pulse for magnetization inversion at ultrahigh field Magn. Reson. Med. 63 51–8Jochimsen T H and Moller H E 2008 Increasing specificity in functional magnetic resonance imaging by

estimation of vessel size based on changes in blood oxygenation Neuroimage 40 228–36Jochimsen T H Norris D G, Mildner T and Moller H E 2004 Quantifying the intra- and extravascular

contributions to spin-echo fMRI at 3 T Magn. Reson. Med. 52 724–32Kamitani Y and Tong F 2006 Decoding seen and attended motion directions from activity in the human

visual cortex Curr. Biol. 16 1096–102Kim S G 1995 Quantification of relative cerebral blood flow change by flow-sensitive alternating inversion

recovery (FAIR) technique: application to functional mapping Magn. Reson. Med. 34 293–301Kim S G and Ogawa S 2012 Biophysical and physiological origins of blood oxygenation level-dependent

fMRI signals J. Cerebr. Blood Flow Metab. 32 1188–206Kiselev V G and Posse S 1999 Analytical model of susceptibility-induced MR signal dephasing: effect

of diffusion in a microvascular network Magn. Reson. Med. 41 499–509Koopmans P J, Barth M and Norris D G 2010 Layer-specific BOLD activation in human V1 Hum. Brain

Mapp. 31 1297–304Kruger  G and Glover  G H 2001 Physiological noise in oxygenation-sensitive magnetic resonance

imaging Mag. Reson. Med.: Off. J. Soc. Magn Reson. Med./Soc. Magn. Reson. Med. 46 631–7Kwong K K et al 1992 Dynamic magnetic resonance imaging of human brain activity during primary

sensory stimulation Proc. Natl Acad. Sci. USA 89 5675–9Kwong K K, Chesler D A, Weisskoff R M, Donahue K M, Davis T L, Ostergaard L, Campbell T A and

Rosen B R 1995 MR perfusion studies with T1-weighted echo planar imaging Magn. Reson. Med. 34 878–87

Li L and Leigh J S 2004 Quantifying arbitrary magnetic susceptibility distributions with MR Magn. Reson. Med. 51 1077–82

Li S J, Li Z, Wu G H, Zhang M J, Franczak M and Antuono P G 2002 Alzheimer disease: Evaluation of a functional MR imaging index as a marker Radiology 225 253–9

Liu T, Spincemaille P, De Rochefort L, Kressler B and Wang Y 2009 Calculation of susceptibility through multiple orientation sampling (COSMOS): a method for conditioning the inverse problem from measured magnetic field map to susceptibility source image in MRI Magn. Reson. Med. 61 196–204

Logothetis N K, Pauls J, Augath M, Trinath T and Oeltermann A 2001 Neurophysiological investigation of the basis of the fMRI signal Nature 412 150–7

Lowe M J, Mock B J and Sorenson J A 1998 Functional connectivity in single and multislice echoplanar imaging using resting-state fluctuations Neuroimage 7 119–132

Lowe M J, Phillips M D, Lurito J T, Mattson D, Dzemidzic M and Mathews V P 2002 Multiple sclerosis: low-frequency temporal blood oxygen level-dependent fluctuations indicate reduced functional connectivity—initial results Radiology 224 184–92

Mangia S, Tkac I, Gruetter R, Van de Moortele P F, Giove F, Maraviglia B and Ugurbil K 2006 Sensitivity of single-voxel 1H-MRS in investigating the metabolism of the activated human visual cortex at 7 T Magn. Reson. Imag. 24 343–8

Mansfield P 1977 Multi-planar image formation using NMR spin echoes J. Phys. C 10 55–8Menon R S 2002 Postacquisition suppression of large-vessel BOLD signals in high-resolution fMRI

Magn. Reson. Med. 47 1–9Menon R S, Ogawa S, Strupp J P and Ugurbil K 1997 Ocular dominance in human V1 demonstrated by

functional magnetic resonance imaging J. Neurophysiol. 77 2780–7

Topical Review

R183

Physiol. Meas. 35 (2014) R167

Moeller  S, Yacoub  E, Olman  C A, Auerbach  E, Strupp  J, Harel  N and Ugurbil  K 2010 Multiband multislice GE-EPI at 7 Tesla, with 16-fold acceleration using partial parallel imaging with application to high spatial and temporal whole-brain FMRI Magn. Reson. Med. 63 1144–53

Mullinger K J, Mayhew S D, Bagshaw A P, Bowtell R and Francis S T 2013 Poststimulus undershoots in cerebral blood flow and BOLD fMRI responses are modulated by poststimulus neuronal activity Proc. Natl Acad. Sci. 110 13636–41

Mullinger K J, Mayhew S D, Bagshaw A P, Bowtell R and Francis S T 2014 Evidence that the negative BOLD response is neuronal in origin: a simultaneous EEG-BOLD-CBF study in humans Neuroimage 94 263–74

Newton  A T, Rogers  A P, Gore  J C and Morgan  V L 2012 Improving measurement of functional connectivity through decreasing partial volume effects at 7 T Neuroimage 59 2511–7

Norman K A, Polyn S M, Detre G J and Haxby J V 2006 Beyond mind-reading: multi-voxel pattern analysis of fMRI data Trends Cogn. Sci. 10 424–30

Ogawa S, Lee T M, Kay A R and Tank D W 1990 Brain magnetic resonance imaging with contrast dependent on blood oxygenation Proc. Natl Acad. Sci. USA 87 9868–72

Ogawa S, Tank D W, Menon R, Ellermann J M, Kim S G, Merkle H and Ugurbil K 1992 Intrinsic signal changes accompanying sensory stimulation: functional brain mapping with magnetic resonance imaging Proc. Natl Acad. Sci. USA 89 5951–5

Olman C A, Harel N, Feinberg D A, He S, Zhang P, Ugurbil K and Yacoub E 2012 Layer-specific fMRI reflects different neuronal computations at different depths in human V1 PLoS One 7 e32536

Olman C A, Van de Moortele P F and Urgubil K 2004 Point spread function for gradient echo and spin echo BOLD fMRI at 7 Tesla Proc. ISMRM (Kyoto, May 2004) p 1066

Park J C, Ronen I, Toth L J, Urgubil K and Kim D S 2004 Comparison of point spread functions of BOLD and ASL fMRI at an ultra-high magnetic field, 9.4 T Proc. ISMRM (Kyoto, May 2004) vol 11

Parkes L M, Schwarzbach J V, Bouts A A, Deckers R H R, Pullens P, Kerskens C M and Norris D G 2005 Quantifying the spatial resolution of the gradient echo and spin echo BOLD response at 3 T Magn. Reson. Med. 54 1465–72

Pauling L and Coryell C D 1936 The magnetic properties and structure of hemoglobin, oxyhemoglobin and carbonmonoxyhemoglobin Proc. Natl Acad. Sci. USA 22 210–6

Petridou  N, Gaudes  C C, Dryden  I L, Francis  S T and Gowland  P A 2013 Periods of rest in fMRI contain individual spontaneous events which are related to slowly fluctuating spontaneous activity Hum. Brain Mapp. 34 1319–29

Petridou N, Schafer A, Gowland P and Bowtell R 2009 Phase vs. magnitude information in functional magnetic resonance imaging time series: toward understanding the noise Magn. Reson. Imag. 27 1046–57

Petzold G C and Murthy V N 2011 Role of astrocytes in neurovascular coupling Neuron 71 782–97Pfeuffer J, Adriany G, Shmuel A, Yacoub E, Van de Moortele P F, Hu X and Ugurbil K 2002 Perfusion-

based high-resolution functional imaging in the human brain at 7 Tesla Magn. Reson. Med. 47 903–11Polimeni J R, Fischl B, Greve D N and Wald L L 2010 Laminar analysis of 7T BOLD using an imposed

spatial activation pattern in human V1 Neuroimage 52 1334–46Poole M and Bowtell R 2008 Volume parcellation for improved dynamic shimming MAGMA 21 31–40Pruessmann K P, Weiger M, Scheidegger M B and Boesiger P 1999 SENSE: sensitivity encoding for fast

MRI Magn. Reson. Med. 42 952–62Raemaekers M, Schellekens W, Van Wezel R J, Petridou N, Kristo G and Ramsey N F 2014 Patterns of

resting state connectivity in human primary visual cortical areas: a 7T fMRI study Neuroimage 84 911–21Raichle M E, Macleod A M, Snyder A Z, Powers W J, Gusnard D A and Shulman G L 2001 A default

mode of brain function Proc. Natl Acad. Sci. USA 98 676–82Ress D, Glover G H, Liu J and Wandell B 2007 Laminar profiles of functional activity in the human brain

Neuroimage 34 74–84Rocca M A et al 2010 Default-mode network dysfunction and cognitive impairment in progressive MS

Neurology 74 1252–9Rowe D B, Meller C P and Hoffmann R G 2007 Characterizing phase-only fMRI data with an angular

regression model J. Neurosci. Methods 161 331–41Sanchez-Panchuelo R M, Besle J, Beckett A, Bowtell R, Schluppeck D and Francis S 2012a Within-

digit functional parcellation of brodmann areas of the human primary somatosensory cortex using functional magnetic resonance imaging at 7 Tesla J. Neurosci. 32 15815–22

Sanchez-Panchuelo R M, Francis S, Bowtell R and Schluppeck D 2010 Mapping human somatosensory cortex in individual subjects with 7T functional MRI J. Neurophysiol. 103 2544–56

Topical Review

R184

Physiol. Meas. 35 (2014) R167

Sanchez-Panchuelo R M, Schluppeck D, Francis S and Bowtell R 2012b Correspondence of human visual areas identified using functional and anatomical MRI in vivo at 7 T JMRI 35 287–99

Schaller  B, Xin  L, O’brien  K, Magill  A and Gruetter  R 2014 Are glutamate and lactate increases ubiquitous to physiological activation? A 1H functional MR spectroscopy study during motor activation in human brain at 7 Tesla Neuroimage 93 138–45

Schweser F, Deistung A, Lehr B W and Reichenbach J R 2011 Quantitative imaging of intrinsic magnetic tissue properties using MRI signal phase: an approach to in vivo brain iron metabolism? Neuroimage 54 2789–807

Setsompop K, Alagappan V, Zelinski A C, Potthast A, Fontius U, Hebrank F, Schmitt F, Wald L L and Adalsteinsson E 2008 High-flip-angle slice-selective parallel RF transmission with 8 channels at 7 T J. Magn. Reson. 195 76–84

Shmuel A, Yacoub E, Chaimow D, Logothetis N K and Ugurbil K 2007 Spatio-temporal point-spread function of fMRI signal in human gray matter at 7 Tesla Neuroimage 35 539–52

Shmueli K, De Zwart J A, Van Gelderen P, Li T Q, Dodd S J and Duyn J H 2009 Magnetic susceptibility mapping of brain tissue in vivo using MRI phase data Magn. Reson. Med. 62 1510–22

Siero J C, Ramsey N F, Hoogduin H, Klomp D W, Luijten P R and Petridou N 2013 BOLD specificity and dynamics evaluated in humans at 7 T: comparing gradient-echo and spin-echo hemodynamic responses PLoS One 8 e54560

Silva A C, Lee S P, Yang G, Iadecola C and Kim S G 1999 Simultaneous blood oxygenation level-dependent and cerebral blood flow functional magnetic resonance imaging during forepaw stimulation in the rat J. Cerebr. Blood F Met. 19 871–9

Smirnakis  S M, Schmid  M C, Weber  B, Tolias  A S, Augath  M and Logothetis  N K 2007 Spatial specificity of BOLD versus cerebral blood volume fMRI for mapping cortical organization J. Cerebr. Blood Flow Metab. 27 1248–61

Spiers H J and Maguire E A 2007 Decoding human brain activity during real-world experiences Trends Cogn. Sci. 11 356–65

Swisher J D, Gatenby J C, Gore J C, Wolfe B A, Moon C H, KimS G and Tong F 2010 Multiscale pattern analysis of orientation-selective activity in the primary visual cortex J. Neurosci. 30 325–30

Talavage T M, Sereno M I, Melcher  J R, Ledden, P J, Rosen, B R and Dale A M 2004 Tonotopic organization in human auditory cortex revealed by progressions of frequency sensitivity J. Neurophysiol. 91 1282–96

Teeuwisse W M, Brink W M, Haines K N and Webb A G 2012 Simulations of high permittivity materials for 7 T neuroimaging and evaluation of a new barium titanate-based dielectric Magn. Reson. Med. 67 912–8

Thulborn K R, Waterton  J C, Matthews P M and Radda G K 1982 Oxygenation dependence of the transverse relaxation time of water protons in whole blood at high field Biochim. Biophys. Acta 714 265–70

Tomasi  D G and Caparelli  E C 2007 Macrovascular contribution in activation patterns of working memory J. Cerebr. Blood Flow Metab. 27 33–42

Triantafyllou C, Hoge R D, Krueger G, Wiggins C J, Potthast A, Wiggins G C and Wald L L 2005 Comparison of physiological noise at 1.5 T, 3 T and 7 T and optimization of fMRI acquisition parameters Neuroimage 26 243–50

Turner R 2002 How much cortex can a vein drain? Downstream dilution of activation-related cerebral blood oxygenation changes Neuroimage 16 1062–7

Turner R, Le Bihan D, Moonen C T, Despres D and Frank J 1991 Echo-planar time course MRI of cat brain oxygenation changes Magn. Reson. Med. 22 159–66

Ugurbil K et al 2013 Pushing spatial and temporal resolution for functional and diffusion MRI in the human connectome project Neuroimage 80 80–104

Uludag K, Dubowitz D J, Yoder E J, Restom K, Liu T T and Buxton R B 2004 Coupling of cerebral blood flow and oxygen consumption during physiological activation and deactivation measured with fMRI Neuroimage 23 148–55

Van der Zwaag W, Francis S, Head K, Peters A, Gowland P, Morris P and Bowtell R 2009 fMRI at 1.5, 3 and 7 T: characterising BOLD signal changes Neuroimage 47 1425–34

Van der Zwaag W, Kusters R, Magill A, Gruetter R, Martuzzi R, Blanke O and Marques J P 2013 Digit somatotopy in the human cerebellum: A 7T fMRI study Neuroimage 67 354–62

Van der Zwaag W, Marques J P, Kober T, Glover G, Gruetter R and Krueger G 2012 Temporal SNR characteristics in segmented 3D-EPI at 7T Magn. Reson. Med. 67 344–52

Villringer A, Rosen B R, Belliveau J W, Ackerman J L, Lauffer R B, Buxton R B, Chao Y S, Wedeen V J and Brady T J 1988 Dynamic imaging with lanthanide chelates in normal brain—contrast due to magnetic-susceptibility effects Magn. Reson. Med. 6 164–74

Topical Review

R185

Physiol. Meas. 35 (2014) R167

Wang K, Liang M, Wang L, Tian L X, Zhang X Q, Li K C and Jiang T Z 2007 Altered functional connectivity in early Alzheimer’s disease: a resting-state fMRI study Hum. Brain Mapp. 28 967–78

Watanabe M, Bartels A, Macke J H, Murayama Y and Logothetis N K 2013 Temporal jitter of the BOLD signal reveals a reliable initial dip and improved spatial resolution Curr. Biol. 23 2146–50

Wharton  S and Bowtell  R 2010 Whole-brain susceptibility mapping at high field: a comparison of multiple- and single-orientation methods Neuroimage 53 515–25

Williams D S, Detre J A, Leigh J S and Koretsky A P 1992 Magnetic-resonance-imaging of perfusion using spin inversion of arterial water Proc. Natl Acad. Sci. USA 89 212–6

Wilson  J L, Jenkinson M, de Araujo  I, Kringelbach M L, Rolls E T and Jezzard P 2002 Fast, fully automated global and local magnetic field optimization for fMRI of the human brain Neuroimage 17 967–76

Wu W-C, Fernandez-Seara M, Detre J A, Wehrli F W and Wang J 2007 A theoretical and experimental investigation of the tagging efficiency of pseudocontinuous arterial spin labeling Magn. Reson. Med.: Off. J. Soc. Magn. Reson. Med./Soc. Magn. Reson. Med. 58 1020–7

Yacoub E, Duong T Q, Van de Moortele P F, Lindquist M, Adriany G, Kim S G, Ugurbil K and Hu X 2003 Spin-echo fMRI in humans using high spatial resolutions and high magnetic fields Magn. Reson. Med. 49 655–64

Yacoub E, Harel N and Ugurbil K 2008 High-field fMRI unveils orientation columns in humans Proc. Natl Acad. Sci. USA 105 10607–12

Yacoub E, Shmuel A, Logothetis N and Ugurbil K 2007 Robust detection of ocular dominance columns in humans using hahn spin echo BOLD functional MRI at 7 Tesla Neuroimage 37 1161–77

Yacoub E et al 2001a Imaging brain function in humans at 7 Tesla Magn. Reson. Med.: Off. J. Soc. Magn. Reson. Med./Soc. Magn. Reson. Med. 45 588–94

Yacoub E, Shmuel A, Pfeuffer J, Van de Moortele P F, Adriany G, Ugurbil K and Hu X 2001b Investigation of the initial dip in fMRI at 7 Tesla NMR Biomed. 14 408–12

Zelinski A C, Wald L L, Setsompop K, Alagappan V, Gagoski B A, Goyal V K and Adalsteinsson E 2008 Fast slice-selective radio-frequency excitation pulses for mitigating B+1 inhomogeneity in the human brain at 7 Tesla Magn. Reson. Med. 59 1355–64