Results 1  10
of
209
Advances in functional and structural mr image analysis and implementation as fsl
 NeuroImage
, 2004
"... The techniques available for the interrogation and analysis of neuroimaging data have a large influence in determining the flexibility, sensitivity and scope of neuroimaging experiments. The development of such methodologies has allowed investigators to address scientific questions which could not p ..."
Abstract

Cited by 274 (7 self)
 Add to MetaCart
(Show Context)
The techniques available for the interrogation and analysis of neuroimaging data have a large influence in determining the flexibility, sensitivity and scope of neuroimaging experiments. The development of such methodologies has allowed investigators to address scientific questions which could not previously be answered and, as such, has become an important research area in its own right. In this paper, we present a review of the research carried out by the Analysis Group at the Oxford Centre for Functional MRI of the Brain (FMRIB). This research has focussed on the development of new methodologies for the analysis of both structural and functional magnetic resonance imaging data. The majority of the research laid out in this paper has been implemented as freely available software tools within FMRIB’s Software Library (FSL). 1
Multilevel linear modelling for FMRI group analysis using Bayesian inference
 Neuroimage
, 2004
"... Functional magnetic resonance imaging studies often involve the acquisition of data from multiple sessions and/or multiple subjects. A hierarchical approach can be taken to modelling such data with a general linear model (GLM) at each level of the hierarchy introducing different random effects varia ..."
Abstract

Cited by 90 (6 self)
 Add to MetaCart
(Show Context)
Functional magnetic resonance imaging studies often involve the acquisition of data from multiple sessions and/or multiple subjects. A hierarchical approach can be taken to modelling such data with a general linear model (GLM) at each level of the hierarchy introducing different random effects variance components. Inferring on these models is nontrivial with frequentist solutions being unavailable. A solution is to use a Bayesian framework. One important ingredient in this is the choice of prior on the variance components and toplevel regression parameters. Due to the typically small numbers of sessions or subjects in neuroimaging, the choice of prior is critical. To alleviate this problem, we introduce to neuroimage modelling the approach of reference priors, which drives the choice of prior such that it is noninformative in an informationtheoretic sense. We propose two inference techniques at the top level for multilevel hierarchies (a fast approach and a slower more accurate approach). We also demonstrate that we can infer on the top level of multilevel hierarchies by inferring on the levels of the hierarchy separately and passing summary statistics of a noncentral multivariate t distribution between them.
I (2004) Resting fluctuations in arterial carbon dioxide induce significant low frequency variations in BOLD signal. Neuroimage 21:16521664. Supplemental Figure Legends Supplemental Figure 1. Graph analysis procedure. Figure illustrates the individual st
 for 0.40 ! T ! 0.5, to 10 for 0.2 ! T < 0.40 and to 5 for T < 0.2. Of all formed graphs G and , the
"... Carbon dioxide is a potent cerebral vasodilator. We have identified a significant source of lowfrequency variation in blood oxygen leveldependent (BOLD) magnetic resonance imaging (MRI) signal at 3 T arising from spontaneous fluctuations in arterial carbon dioxide level in volunteers at rest. Fluct ..."
Abstract

Cited by 60 (0 self)
 Add to MetaCart
(Show Context)
Carbon dioxide is a potent cerebral vasodilator. We have identified a significant source of lowfrequency variation in blood oxygen leveldependent (BOLD) magnetic resonance imaging (MRI) signal at 3 T arising from spontaneous fluctuations in arterial carbon dioxide level in volunteers at rest. Fluctuations in the partial pressure of endtidal carbon dioxide (PETCO 2)ofF1.1 mm Hg in the frequency range 0 –0.05 Hz were observed in a cohort of nine volunteers. Correlating with these fluctuations were significant generalized grey and white matter BOLD signal fluctuations. We observed a mean (Fstandard error) regression coefficient across the group of 0.110 F 0.033 % BOLD signal change per mm Hg CO2 for grey matter and 0.049 F 0.022 % per mm Hg in white matter. PET CO2related BOLD signal fluctuations showed regional differences across the grey matter, suggesting variability of the responsiveness to carbon dioxide at rest. Functional magnetic resonance imaging (fMRI) results were corroborated by transcranial
The Statistical Analysis of fMRI Data
, 2008
"... In recent years there has been explosive growth in the number of neuroimaging studies performed using functional Magnetic Resonance Imaging (fMRI). The field that has grown around the acquisition and analysis of fMRI data is intrinsically interdisciplinary in nature and involves contributions from ..."
Abstract

Cited by 35 (0 self)
 Add to MetaCart
In recent years there has been explosive growth in the number of neuroimaging studies performed using functional Magnetic Resonance Imaging (fMRI). The field that has grown around the acquisition and analysis of fMRI data is intrinsically interdisciplinary in nature and involves contributions from researchers in neuroscience, psychology, physics and statistics, among others. A standard fMRI study gives rise to massive amounts of noisy data with a complicated spatiotemporal correlation structure. Statistics plays a crucial role in understanding the nature of the data and obtaining relevant results that can be used and interpreted by neuroscientists. In this paper we discuss the analysis of fMRI data, from the initial acquisition of the raw data to its use in locating brain activity, making inference about brain connectivity and predictions about psychological or disease states. Along the way, we illustrate interesting and important issues where statistics already plays a crucial role. We also seek to illustrate areas where statistics has perhaps been underutilized and will have an increased role in the future.
Modeling the hemodynamic response function in fMRI: efficiency, bias and mismodeling. Neuroimage
, 2009
"... Most brain research to date have focused on studying the amplitude of evoked fMRI responses, though there has recently been an increased interest in measuring onset, peak latency and duration of the responses as well. A number of modeling procedures provide measures of the latency and duration of f ..."
Abstract

Cited by 34 (4 self)
 Add to MetaCart
(Show Context)
Most brain research to date have focused on studying the amplitude of evoked fMRI responses, though there has recently been an increased interest in measuring onset, peak latency and duration of the responses as well. A number of modeling procedures provide measures of the latency and duration of fMRI responses. In this work we compare several techniques that vary in their assumptions, model complexity, and interpretation. For each method, we introduce methods for estimating amplitude, peak latency, and duration and for performing inference in a multisubject fMRI setting. We then assess the techniques' relative sensitivity and their propensity for misattributing task effects on one parameter (e.g., duration) to another (e.g., amplitude). Finally, we introduce methods for quantifying model misspecification and assessing bias and powerloss related to the choice of model. Overall, the results show that it is surprisingly difficult to accurately recover true taskevoked changes in BOLD signal and that there are substantial differences among models in terms of power, bias and parameter confusability. Because virtually all fMRI studies in cognitive and affective neuroscience employ these models, the results bear on the interpretation of hemodynamic response estimates across a wide variety of psychological and neuroscientific studies. © 2008 Elsevier Inc. All rights reserved. Introduction Functional magnetic resonance imaging (fMRI) is based on studying the vascular response in the brain to neuronal activity and can be used to study mental activity. It is most commonly performed using blood oxygenation leveldependent (BOLD) contrast To date most fMRI studies have been primarily focused on estimating the amplitude of evoked HRFs across different tasks. However, there is a growing interest in studying the timetopeak and duration of activation as well In this paper, we focus on estimation of response amplitude/height (H), timetopeak (T), and fullwidth at halfmax (W) in the HRF as potential measures of response magnitude, latency and duration of neuronal activity. Ideally, the parameters of the HRF should be directly interpretable in terms of changes in neuronal activity, and should be estimated so that statistical power is maximized. An accurate estimate of the HRF shape may help prevent both false positive and negative results from arising due to illfitting constrained statistical models, as even minor amounts of mismodeling can lead to severe loss in power and validity The issue of interpretability is complex, and the problem can be divided into two parts, shown in In spite of these challenges, well over a thousand studies have to date demonstrated relationships between taskevoked changes in brain metabolic activity and measured BOLD responses. These studies treat the evoked BOLD response as the signal of interest, without attempting to make a direct quantitative link to neuronal activity. Early studies presented events with large separation in time (e.g., visual stimuli every 2030 s), so that taskevoked average BOLD responses could be recovered, and H, T, and W estimated directly. However, this design is highly inefficient, as very few stimuli can be presented in a session, and it has become common practice to present events rapidly enough so that the BOLD responses to different events overlap. The dominant analysis strategy is to assume that BOLD responses to events add linearly Choices of HRF models range from the use of a single canonical HRF, the use of a basis set of smooth functions Thus, in sum, the nature of the underlying BOLD physiology limits the ultimate interpretability of the parameter estimates in terms of neuronal and metabolic function, but modeling taskevoked BOLD responses is useful, and is in fact critical for inference in virtually all neuroscientific fMRI studies. Because of the complexity in the relationship between neural activity and BOLD, we do not attempt to relate BOLD signal directly to underlying neuronal activity in this work. Instead, we concentrate on the second issue in In previous work we showed that with virtually all models of evoked BOLD responses, true changes in one parameter (e.g. T) can be mistaken for changes in others (e.g. H and W). This problem is independent from the issue of how neuronal activity actually leads to the BOLD response. The goal of this paper is to expand on our previous work assessing the validity and power of various hemodynamic modeling techniques by introducing techniques for performing inference on estimated H, T and W parameters in a multisubject fMRI setting, as well as methods for quantifying the amount of mismodeling each model gives rise to. We consider a number of BOLD response models, which vary significantly in their assumptions, model complexity and interpretation, under a range of different conditions, including variations in true BOLD amplitude, latency, and duration. Overall, the results reported here show that it is surprisingly difficult to accurately recover true taskevoked changes in BOLD H, T, and W parameters, and there are substantial differences among models in power, bias and parameter confusability. Because virtually all fMRI studies in cognitive and affective neuroscience employ these models, the results bear on the way HRFs are estimated in hundreds of neuroscientific studies published per year. Thus, the current results can inform the choice of BOLD response models used in these studies, until it becomes practical to incorporate more complete models of BOLD hemodynamics (including nonlinear neurovascular coupling) on a voxelbyvoxel basis in cognitive studies. Methods Modeling the hemodynamic response function The relationship between the stimulus and BOLD response is typically modeled using a linear time invariant (LTI) system, where the M.A. Lindquist et al. / NeuroImage 45 (2009) S187S198 signal at time t, y(t), is modeled as the convolution of a stimulus function s(t) and the hemodynamic response h(t), i.e. In many studies h(t) is assumed to take a fixed canonical shape. However, to increase the flexibility of the approach, h(t) is often modeled as a linear combination of B basis functions g i (t), where i = 1,….B. We can then write where the β i are unknown model parameters. Typically the vectors (s⁎g i )(t) are collated into the columns of a design matrix, X, and the model is expressed where β is a vector of regression coefficients, Y is a vector containing the observed data, and ε is a vector of unexplained error values. For most statistical analysis the use of a LTI system is considered a reasonable assumption that provides for valid statistical inference. Therefore, in this work we assume an LTI system, and our main focus will be finding flexible models for the impulse function in the LTI system, i.e. the HRF. A number of models, varying greatly in their flexibility, have been suggested in the literature. In the most rigid model, the shape of the HRF is completely fixed and the height (i.e., amplitude) of the response alone is allowed to vary In general, the more basis functions used in a linear model or the more free parameters in a nonlinear one, the more flexible the model is in measuring the parameters of interest. However, including more parameters also means more potential for error in estimating them, fewer degrees of freedom, and decreased power and validity if regressors are collinear. It is also easier and statistically more powerful to interpret differences between task conditions on a single parameter such as height than it is to test for differences in multiple parameters conditional, of course, on the interpretability of those parameter estimates. Together these problems suggest using a single, canonical HRF and it does in fact offer optimal power if the shape is specified exactly correctly. However, the shape of the HRF varies as a function of both task and brain region, and any fixed model will be misspecified for much of the brain HRF models In this work we study seven HRF models in a series of simulation studies and an application to real data. We briefly introduce each model below, but leave a more detailed description for Section A of the Appendix. The first model under consideration is SPMs canonical HRF (Here denoted GAM), which consists of a linear combination of two gamma functions (Eq. (A1) in the Appendix). To increase its ability to fit responses that are shifted in time or have extended activation durations, we also consider models using the canonical HRF plus its temporal derivative (TD) and the canonical HRF plus its temporal and dispersion derivatives (DD). The next class of models is based on the use of the finite impulse response (FIR) basis set, which is the most flexible basis set that can be applied directly in a linear regression framework. In this work, we study both the standard FIR model and a semiparametric smooth FIR model (sFIR). Finally, we also consider two models fit using nonlinear techniques. These include one with the same functional form as the canonical HRF but with 6 variable parameters (NL) and the inverse logit model (IL), which consists of the superposition of three separate inverse logit functions. Estimating parameters After modeling the HRF we seek to estimate its height (H), timetopeak (T) and width (W). Several of the models have closed form solutions describing the fits (e.g., the gamma based models and IL), and hence estimates of H, T and W can be derived analytically. A lack of closed form solution (e.g., for FIR models) does not preclude estimating values from the fitted HRFs, and procedures for doing so are described in Section B of the Appendix. However, when possible we use the parameter estimates to calculate H, T and W. Estimates for IL have been derived in . When using models that include the canonical HRF and its derivatives it is common to only use the nonderivative term as an estimate of the HRF amplitude. However, this solution will be biased and therefore for TD and DD we use a "derivative boost" to counteract anticipated delayinduced negative amplitude bias whereβ 1 andβ 2 are the regression parameters for the canonical HRF and first derivative term respectively. For DD it is whereβ 3 is the regression parameter corresponding to the second derivative term. Inference We also seek to compare several techniques for performing population inference on the estimated amplitude. Let H i be the estimated amplitude for subject i, i =1,….M, defined for hypothesis testing purposes to be the global extreme point for the HRF, i.e. either a minimum or a maximum. The goal is to test whether H significantly differs from 0 in the population. In this work we compare three statistical techniques: the standard summary statistics approach (Holmes and Friston, 1998), a bootstrap procedure Detecting model misspecification Each of the models presented in this paper differ in their ability to handle unexpected HRF shapes. Using an illfitting model will violate the assumptions (e.g., mean 0 noise) required for valid inference and even a small amount of mismodeling can result in severe power loss and inflate the false positive rate beyond the nominal value. Due to the massive amount of data, performing model diagnostics is challenging, and only limited attention has been given to this problem (e.g., Suppose r(i), i=1,…T are the whitened residuals and K(t) a kernel function. Let, be the moving average of w consecutive observations, starting at time t. Under the null hypothesis that the model is correct, Z w is mean 0 for all values of t. Thus a large value of Z w indicates that model misfit might be present and the statistic S = max Z w (t) measures the strongest evidence against the null hypothesis. Choosing a Gaussian kernel allows Gaussian random field theory to be used to determine the pvalue where p i is the pvalue for subject i. Under the null hypothesis of no effect, Q follows a chisquare distribution with 2M degrees of freedom. As a followup we have proposed techniques for determining whether there is taskrelated signal remaining in the residuals and for quantifying the amount of powerloss and bias directly attributable to model misspecification. Estimates of bias and powerloss can be computed from the residuals for each voxel, and bias and power loss maps can be constructed. The details of this procedure are beyond the scope of this paper, and we refer the interested reader to Comparing HRF models: simulation studies The simulations described below were designed to compare the performance of the HRF modeling methods, specifically with respect to the ability to model variations in stimulus onset and duration relative to the assumed experimental reference ("neuronal") signal (Eq. (1)). We also assess the validity and power of each method using different types of inference: the summary statistic, the bootstrap test, and the sign permutation test. Creation of "ground truth" data for simulation As shown in M.A. Lindquist et al. / NeuroImage 45 (2009) S187S198 group "random effects" analysis, we generated 15 subject datasets for each simulation, which consisted of the "true" BOLD time series at each voxel plus white noise, creating a plausible effect size (Cohen's d = 0.5) based on observed effect sizes in the visual and motor cortex Simulation 1 An eventrelated stimulus function with a single spike (see Simulation 2 Data were simulated for 15 subjects in the same manner as in Simulation 1. After fitting each of the 7 methods, the value of H was estimated for each voxel and subject. Population inference was performed using the three testing procedures to determine whether the population height differed significantly from zero. The whole procedure was repeated 30 times and the number of times each voxel was deemed significant at the α = 0.001 level was recorded. Experimental procedures: thermal pain Participants (n = 20) provided informed consent and all procedures were approved by the Columbia University IRB. During fMRI scanning, 48 thermal stimuli, 12 at each of 4 temperatures, were delivered to the left forearm. Temperatures were calibrated individually for each participant before scanning to be warm, mildly painful, moderately painful, and near tolerance. Heat stimuli, preceded by a 2 s warning cue and 6 s anticipation period, lasted 10 s in duration followed by a 30 s intertrial interval. Functional T2⁎weighted EPIBOLD images (TR = 2 s, 3.5 × 3.5 × 4 mm voxels) were collected during functional runs of length 6 min. 8 s. Gradient artifacts were removed from reconstructed images prior to preprocessing. Images were slicetime corrected and adjusted for head motion using SPM5 software (http:// www.fil.ion.ucl.ac.uk/spm/). A highresolution anatomical image (T1weighted spoiledGRASS [SPGR] sequence, 1 × 1 × 1 mm voxels, TR = 19 ms) was coregistered to the mean functional image using a mutual information cost function, and segmented and warped to the MNI template. Warps were also applied to functional images, which were smoothed with a 6 mmFWHM Gaussian kernel, highpass filtered with a 120 s (.0083 Hz) discrete cosine basis set, and Winsorized at 3 standard deviations prior to analysis. Each of the 7 models were fit to data voxelwise in a single axial slice (z = −22 mm) covering several painrelated regions of interest, including the anterior cingulate cortex. Separate HRFs were estimated for stimuli of each temperature, though we focus on the responses to the highest heat level in the results. The misspecification statistic was calculated using a Gaussian kernel (8 s FWHM) and pvalues determined using Gaussian random field theory. Results Simulation studies Simulation 1 The results of Simulation 1 are summarized in The GAM model (first column) gives reasonable results for delayed onsets within 3 s and widths up to 3 s (squares in the upper lefthand corner), but underestimates amplitude dramatically as onset and duration increase. This is natural as the GAM model is correctly specified for the square in the upper lefthand corner (and thus optimal), but not well equipped to handle a large amount of model misspecification. Of special interest is the fact that there is no bias in the squares contained in the first two columns of the W map. This is true because in these cases the fixed width of the canonical HRF exactly coincides with the width of the simulated data. The same is true for the square in the upper lefthand corner of the T map. However, studying the results in the bottom row indicates severe mismodeling present for voxels in the lower righthand corner. The second and third column show equivalent results for TD and DD, which show that the inclusion of derivative terms provide a slight improvement over GAM for squares where there is a minor amount of mismodeling of the onset and duration. However, there is again a drastic decrease in model fit with delayed onsets greater than about 3 s or extended durations greater than 3 s. Interestingly enough, there appear to be only minor differences between the results for DD and TD, which indicates that the inclusion of the dispersion derivative does not lead to an improvement in the model robustness across onset and duration shifts. Also it is interesting to note that for each of the gammabased models (GAM, TD and DD) there is a consistent negative bias in the estimates of T and W (all voxels are blue). The estimation procedure was repeated using only the nonderivative term as an estimate of the HRF amplitude as is the common practice in the field. The results (not presented here) showed that the "derivative boost" resulted in a slight decrease in reported bias. Both models based on the use of FIR basis sets (FIR and sFIR) give rise to some bias in all three model parameters, with estimates tending to be negatively biased (e.g., shrunk towards zero for positive activations). The results for sFIR are consistent with similar simulations performed in . Of special interest is that for FIR, the W map shows a strong, systematic negative bias (all squares are blue), because the full response width is almost never captured due to the roughness of the FIR estimates. The sFIR model performs substantially better in estimating width, with substantial bias only with 45 s onset shifts, at a small cost in underestimating H. This cost is likely due to the fact that the Gaussian prior term leads to shrinkage of the amplitude of the fitted HRF. Both methods showed some bias in estimates of T, but without a clear, consistent directionality. Finally, it appears that the sFIR model has some minor (occasional) problems with model misspecification, while the FIR shows no significant model misspecification. The NL model shows reasonable results with respect to bias, except for a strong tendency to underestimate duration, but shows severe problems with model misspecification. This can further be seen in Simulation 2 Simulation 2 assessed two nonparametric alternatives to the standard parametric summary statistics approach. The results are summarized in In summary, the results are consistent across the three inference techniques (each shown in one row in From Experiment The results of the pain experiment are summarized in Figs. 67. The location of the slice used and an illustration of areas of interest (rdACC and S2, two brain regions known to process pain intensity Discussion Though most brain research to date have focused on studying the amplitude of evoked activation, the onset and peak latencies of the HRF can provide information about the timing of activation for various brain areas and the width of the HRF provides information about the duration of activation. However, the independence of these parameter estimates has not been properly assessed, as it appears that even if basis functions are independent (or a nonlinear fitting procedure provides nominally independent estimates), the parameter estimates from real data may not be independent. The present study extends work originally presented in that seeks to bridge this gap in the literature. To assess independence, we determine the amount of confusability between estimates of height (H), timetopeak (T) and fullwidth at halfmaximum (W) and actual manipulations in the amplitude, timetopeak and duration of the stimulus. This was investigated using a simulation study that compares model fits across a variety of popular methods. Even models that attempt to account for delay such as a gamma function with nonlinear fitting (e.g., A key point of this paper is that model misspecification can result in bias in addition to loss in power. This bias may inflate the Type I error rate beyond the nominal α level, so that pvalues for the test are inaccurate. For example, a statistical parametric map thresholded at p b 0.001 may actually only control the false positive rate at, for example, p b 0.004. We find that even relatively minor model misspecification can result in substantial power loss. In light of our results, it seems important for studies that use a single canonical HRF or a highly constrained basis set to construct maps of bias and power loss, so that regions with low sensitivity or increased false positive rates may be identified. We discuss a procedure for detecting deviations in fMRI time series residuals. Using these ideas, it is possible to construct wholebrain bias and power loss maps due to systematic mismodeling. Matlab implementations of the IL model and a mismodeling toolbox can be obtained by contacting the authors. Conflict of interest The authors declare that there are no conflicts of interest.
Traininginduced functional activation changes in dualtask processing: an fMRI study
 Cerebral Cortex
, 2007
"... Although traininginduced changes in brain activity have been previously examined, plasticity associated with executive functions remains understudied. In this study, we examined trainingrelated changes in cortical activity during a dual task requiring executive control. Two functional magnetic res ..."
Abstract

Cited by 21 (2 self)
 Add to MetaCart
Although traininginduced changes in brain activity have been previously examined, plasticity associated with executive functions remains understudied. In this study, we examined trainingrelated changes in cortical activity during a dual task requiring executive control. Two functional magnetic resonance imaging (fMRI) sessions, one before training and one after training, were performed on both a control group and a training group. Using a regionofinterest analysis, we examined Time3 Group and Time3 Group3 Condition interactions to isolate trainingdependent changes in activation. We found that most regions involved in dualtask processing before training showed reductions in activation after training. Many of the decreases in activation were correlated with improved performance on the task. We also found an area in the dorsolateral prefrontal cortex that showed an increase in activation for the training group for the dualtask condition, which was also correlated with improved performance. These results are discussed in relation to the efficacy of training protocols for modulating attention and executive functions, dualtask processing, and fMRI correlates of plasticity.
Functional dissociations of risk and reward processing in the medial prefrontal cortex
 Cereb. Cortex
, 2009
"... Making a risky decision is a complex process that involves evaluation of both the value of the options and the associated risk level. Yet the neural processes underlying these processes have not so far been clearly identified. Using functional magnetic resonance imaging and a task that simulates ris ..."
Abstract

Cited by 20 (3 self)
 Add to MetaCart
Making a risky decision is a complex process that involves evaluation of both the value of the options and the associated risk level. Yet the neural processes underlying these processes have not so far been clearly identified. Using functional magnetic resonance imaging and a task that simulates risky decisions, we found that the dorsal region of the medial prefrontal cortex (MPFC) was activated whenever a risky decision was made, but the degree of this activity across subjects was negatively correlated with their risk preference. In contrast, the ventral MPFC was parametrically modulated by the received gain/loss, and the activation in this region was positively correlated with an individual’s risk preference. These results extend existing neurological evidence by showing that the dorsal and ventral MPFC convey different decision signals (i.e., aversion to uncertainty vs. approach to rewarding outcomes), where the relative strengths of these signals determine behavioral decisions involving risk and uncertainty.
Overview of fMRI analysis
 British Journal of Radiology
"... fMRI (functional magnetic resonance imaging) is a powerful noninvasive tool in the study of the function of the brain, used, for example, by psychologists, psychiatrists and neurologists. fMRI can give high quality visualization of the location of activity in the brain resulting from sensory stimul ..."
Abstract

Cited by 18 (0 self)
 Add to MetaCart
(Show Context)
fMRI (functional magnetic resonance imaging) is a powerful noninvasive tool in the study of the function of the brain, used, for example, by psychologists, psychiatrists and neurologists. fMRI can give high quality visualization of the location of activity in the brain resulting from sensory stimulation or cognitive function. It therefore allows the study of how the healthy brain functions, how it is affected by different diseases, how it attempts to recover after damage and how drugs can modulate activity or postdamage recovery. After an fMRI experiment has been designed and carried out, the resulting data must be passed through various analysis steps before the experimenter can get answers to questions about experimentally related activations at the individual or multisubject level. This paper
Power calculation for group fMRI studies accounting for arbitrary design and temporal autocorrelation
 NeuroImage
, 2008
"... Abstract: When planning most scientific studies, one of the first steps is to carry out a power analysis to define a design and sample size that will result in a wellpowered study. There are limited resources for calculating power for group fMRI studies, due to the complexity of the model. Previous ..."
Abstract

Cited by 14 (1 self)
 Add to MetaCart
Abstract: When planning most scientific studies, one of the first steps is to carry out a power analysis to define a design and sample size that will result in a wellpowered study. There are limited resources for calculating power for group fMRI studies, due to the complexity of the model. Previous approaches for group fMRI power calculation simplify the study design and/or the variance structure in order to make the calculation possible. These approaches limit the designs that can be studied and may result in inaccurate power calculations. We introduce a flexible power calculation model that makes fewer simplifying assumptions, leading to a more accurate power analysis that can be used on a wide variety of study designs. Our power calculation model can be used to obtain region of interest (ROI) summaries of the mean parameters and variance parameters, which can be use to increase understanding of the data as well as calculate power for a future study. Our example illustrates that minimizing cost to achieve 80 % power is not as simple as finding the smallest sample size capable of achieving 80 % power, since smaller sample sizes require each subject to be scanned longer.
Individual differences in the effects of perceived controllability on pain perception: Critical role of the prefrontal cortex
, 2007
"... & The degree to which perceived controllability alters the way a stressor is experienced varies greatly among individuals. We used functional magnetic resonance imaging to examine the neural activation associated with individual differences in the impact of perceived controllability on selfrepo ..."
Abstract

Cited by 11 (3 self)
 Add to MetaCart
(Show Context)
& The degree to which perceived controllability alters the way a stressor is experienced varies greatly among individuals. We used functional magnetic resonance imaging to examine the neural activation associated with individual differences in the impact of perceived controllability on selfreported pain perception. Subjects with greater activation in response to uncontrollable (UC) rather than controllable (C) pain in the pregenual anterior cingulate cortex (pACC), periaqueductal gray (PAG), and posterior insula/SII reported higher levels of pain during the UC versus C conditions. Conversely, subjects with greater activation in the ventral lateral prefrontal cortex (VLPFC) in anticipation of pain in the UC versus C conditions reported less pain in response to UC versus C pain. Activation in the VLPFC was significantly correlated with the acceptance and denial subscales of the COPE inventory [Carver, C. S.,