Functional Magnetic Resonance Imaging Methods

1 REVIEWF unctional Magnetic Resonance Imaging MethodsJingyuan E. Chen1,2&Gary H. Glover1,2 Received: 24 May 2015 /Accepted: 28 July 2015 /Published online: 7 August 2015#Springer Science+Business Media New York 2015 AbstractSince its inception in 1992, Functional MagneticResonance Imaging (fMRI) has become an indispensible toolfor studying cognition in both the healthy and dysfunctionalbrain. FMRI monitors changes in the oxygenation of braintissue resulting from altered metabolism consequent to atask-based evoked neural response or from spontaneous fluc-tuations in neural activity in the absence of conscious menta-tion (theBresting state^).

2 Task-based studies have revealedneural correlates of a large number of important cognitiveprocesses, while fMRI studies performed in the resting statehave demonstrated brain-wide networks that result from brainregions with synchronized, apparently spontaneous this article, we review the Methods used to acquire andanalyze fMRI Magnetic Resonance Imaging (fMRI) is a neuro- Imaging tool that employs MRI to image dynamic changes inbrain tissue that are caused by changes in neural of neural activity may be caused by asking thesubject to perform a task designed to target a specific cognitiveprocess, or can occur spontaneously while the subject is rest-ing in the absence of conscious mentation ( , in theBrestingstate^).

3 Both types of studies- task-based and resting state,have become indispensible tools for studying cognition inhealthy as well as diseased brains, and tens of thousands ofstudies have been published (>150,000 listed ; brain^)inthe2+decades since nearly simultaneous introduction of thetechnique by three independent groups (Bandettini et ; Kwong et ; Ogawa et ).The MR contrast mechanism used for virtually all fMRIrelies on blood oxygenation level dependent (BOLD) changesin brain tissue, exhibited when a brain region experiencesaltered levels of oxygen consumption consequent to up- ordown-regulated metabolic activity caused, , by performinga cognitive task (Ogawa et ).

4 When there is a localincrease in neural (and glial) activity, concomitant increases inaerobic and anaerobic oxygen consumption trigger increaseddelivery of fully oxygenated hemoglobin through vasodilatory(Raichle et ; Roland and Larsen1976 ; Sokoloff et ; Fox et ; Malonek and Grinvald1996 ) processesthat increase Cerebral Blood Flow (CBF) to the region. Forreasons that are still not fully understood (Fox and Raichle1986 ; Frahm et ; Buxton et ), oxygen supplytransiently exceeds demand, which results in a net increase inlocal oxygenation for several seconds ( ).

5 Thus, the en-dogenous deoxyhemoglobin (Hb) is dynamically replacedwith oxyhemoglobin (HbO2), and is accompanied by a tran-sient increase in intravascular blood volume, resulting in achange in oxygenation state. Because Hb is paramagneticwhile HbO2 is diamagnetic, the change in state from paramag-netic to diamagnetic results in a decrease in R2 and R2*relaxivity rates (Thulborn et ; Ogawa et ).Thus, an MRI sequence with T2 (1/R2) or T2* (1/R2*)weighting can demonstrate BOLD contrast, and therefore*Jingyuan E. of Radiology, Department of Electrical Engineering,Stanford University, Stanford, CA 94305, USA2 Lucas MRI/S Center, MC 5488, 1201 Welch Road,Stanford, CA 94305-5488, USAN europsychol Rev (2015) 25:289 313 DOI neural activity changes through this hemodynamicallydriven microgradients in Magnetic field that surround vesselsand capillaries filled with Hb result in two forms of BOLD contrast (Bandettini et ; Weisskoff et ).

6 Thefirst is due to intravoxel dephasing, which is most prominentnear larger vessels, and which causes T2* weighted signalloss. This contrast increases linearly with Magnetic fieldstrength and is readily observed with gradient recalled echo(GRE) Imaging . The second type of contrast is due to diffu-sion of spins through the microgradients, causing a reductionin T2-weighted signal detected by spin echo (SE) MRI. Thediffusion mechanism is most prominent when the distance thespins diffuse during the signal acquisition is comparable to thespatial extent of the microgradients, which thereby tunes thismechanism to be most sensitive to detecting BOLD contrast incapillaries (Weisskoff et ).

7 Diffusioncontrastispro-portional to the square of the Magnetic field , as the field is increased, the weighting of T2 con-trast increases relative to T2* weighted contrast, with the re-sult that in fields of 4T and higher BOLD contrast is morelocalized to tissue than to the larger veins when SE acquisi-tions are employed (Yacoub et ). By contrast, withGRE acquisitions at 7T the T2* of veins is so short the venouscontribution becomes small and diffusion weighting from tis-sue microstructure dominates (Geissler et ). Becauseof this, SE acquisitions are to be preferred at 7T, although SEmethods have higher RF power deposition (SpecificAbsorption Rate, SAR), which may reduce the number ofslices that can be recalled acquisitionssuffer signal loss from staticmagnetic field distortions that are caused by Magnetic suscepti-bility differences at air-tissue interfaces, for example in frontalorbital or lateral parietal brain regions.

8 These gradients in mag-netic field (~9 ppm difference in susceptibility between air andbrain tissue) are large enough to cause signal dropout artifactsfrom intravoxel dephasing in GRE acquisitions. Spin echomethods refocus the static field heterogeneities, and thereforedo not have signal dropout. The relationships between contrast,artifacts and field strength are summarized in this time, 7 T and higher field magnets are not in wide-spread use, so that the majority of fMRI studies are performedat 3 T (in which T2 and T2*-weighted contrasts are compara-ble or T, which is mostly sensitive to BOLD contrast in thedraining veins (Kruger and Glover2001 ; ).)

9 Therefore, it would be wise to avoid T for neuropsycho-logical studies whenever possible, to obtain the most accuratedepiction of cognitive we have indicated, there are two primary types of fMRIstudies- those in which a cognitive task is used to modulatespecific neuronal activity, and resting state studies. In eithercase, a dynamic series of T2*-weighted scans is acquired,resulting in (Kruger et ) a time series of signals forevery brain voxel. These time series are submitted to variouslevels of correction and denoising (preprocessing steps) beforemodel- or data-driven analyses are applied to obtain maps ofactivity.

10 Because BOLD signals are tiny- typically a few per-cent or less- such analyses use statistical Methods to discernfalse from true activation at a given confidence article reviews the Methods employed to acquire andprocess BOLD fMRI data, with which to draw inferences regard-ing neural processes. We will not examine other Methods oftenused in conjunction with fMRI such as Diffusion Tensor Imaging (DTI) (Le Bihan et ), which can depict or summarizestructure of white matter, or Arterial Spin Labeling (ASL)(Williams et ), used to map the CBF either in stasis orduring task manipulation.