the vector of initial values for theta

a list containing the sparse matrix elements Cmat, Gmat, and GtCinvG

the vector of response values

the sparse matrix of the data values

a sparse matrix of the prior precision found using the initial values of the hyperparameters

a sparse matrix representation of the basis function mapping the data locations to the mesh vertices

a precomputed matrix crossprod(X%*%Psi)

the number of columns for the random matrix used in the Hutchinson estimator

a value for the tolerance used for a stopping rule (compared to the squared norm of the differences between theta(s) and theta(s-1))

(logical) Should intermediate output be displayed?

a length-p vector where p is the AR order

(integer) the length of the time series that is being prewhitened

a scalar value of the residual variances of the AR model

a vector of length two containing the range and scale parameters kappa2 and phi, in that order

a list containing the sparse matrix elements Cmat, Gmat, and GtCinvG

the beta_hat estimates for a single task

the number of sessions

the stopping rule tolerance

(logical) Should intermediate output be displayed?

a scalar

a list with elements Cmat, Gmat, and GtCinvG

the integer number of sessions

The autocorrelated data

Demean y? Default: FALSE.

The model order limit. Default: 10.

fMRI timeseries data in CIFTI format ("*.dtseries.nii"). For single-session analysis this can be a file path to a CIFTI file or a "xifti" object from the ciftiTools package. For multi-session analysis this can be a vector of file paths or a list of "xifti" objects.

If BOLD is a "xifti" object(s), the surfaces, if any, will be used for the spatial model. However, if surfL and surfR are provided, they will override any surfaces in BOLD.

fMRI timeseries data in CIFTI format ("*.dtseries.nii"). For single-session analysis this can be a file path to a CIFTI file or a "xifti" object from the ciftiTools package. For multi-session analysis this can be a vector of file paths or a list of "xifti" objects.

If BOLD is a "xifti" object(s), the surfaces, if any, will be used for the spatial model. However, if surfL and surfR are provided, they will override any surfaces in BOLD.

Character vector indicating which brain structure(s) of BOLD to analyze: "left" cortex; "right" cortex; and/or "subcortical" structures. Or "all" to model all three. Default: c("left","right") (cortex only).

Which subcortical ROIs should be analyzed? Can be "all" to analyze all subcortex ROIs. See the ciftiTools_Name column of ciftiTools:::substructure_table() for a list of possible subcortical ROIs.

A numeric matrix or data.frame, or a "BayesfMRI_design" object from make_design. Can also be an array where the third dimension is the same length as the number of data locations, to model each location with its own design.

(Optional) A T \times N_{nuis} matrix of nuisance signals, where T is the number of timepoints and N is the number of nuisance signals, or a list of these for multi-session analysis. Nuisance signals are regressed from the fMRI data and design matrix prior to GLM computation. Nuisance signals can include motion regressors, HRF derivatives not being modeled as tasks, and other sources of noise.

Detrending/high-pass filtering is accomplished by adding DCT bases to the nuisance matrix; see the parameters hpf and DCT.

Do not add spike regressors for scrubbing to the nuisance matrix. Rather, provide these in scrub so that their corresponding timepoints are also removed from the BOLD data after nuisance regression.

(Optional) A T \times N_{scrub} matrix of spike regressors (one 1 value at the timepoint to scrub, and 0 for all other values), or a logical vector indicating the timepoints to scrub (TRUE to scrub, and FALSE to keep). For multi-session data, a session-length list of such matrices or logical vectors.

The spike regressors will be included in the nuisance regression, and afterwards the timepoints indicated in scrub will be removed from the BOLD data and design matrix.

Add DCT bases to nuisance to apply a temporal high-pass filter to the data, for detrending? hpf is the filter frequency. Use NULL to skip detrending. Detrending is strongly recommended for fMRI data, to help reduce the autocorrelation in the residuals, so NULL will induce a warning. Use "already" to disable the warning while skipping highpass filtering.

Using at least two DCT bases is as sufficient for detrending as using linear and quadratic drift terms in the nuisance matrix. So if DCT detrending is being used here, there is no need to add linear and quadratic drift terms to nuisance.

Temporal resolution of the data, in seconds.

For cortex spatial model. Left and right cortex surface geometry in GIFTI format ("*.surf.gii"). These can be a file path to a GIFTI file or a "surf" object from ciftiTools.

Surfaces can alternatively be provided through the $surf metadata in BOLD if it is "xifti" data. If neither are provided, by default the HCP group-average fs_LR inflated surfaces included in ciftiTools will be used for the cortex spatial model.

For cortex spatial model. The number of vertices to which each cortical surface should be resampled, or NULL to not resample.

For computational feasibility, a value of 10000 (default) or lower is recommended for Bayesian spatial modeling. If Bayes=FALSE, resamp_res can be set to NULL for full-resolution classical modeling.

For volumetric model. What order neighborhood around data locations to keep? 0 for no neighbors, 1 for 1st-order neighbors, 2 for 1st- and 2nd-order neighbors, etc. Smaller values will provide greater computational efficiency at the cost of higher variance around the edge of the data.

For volumetric model. The number of extra voxel layers around the bounding box. Set to NULL for no buffer. (We recommend not changing buffer unless you know what you're doing. Instead, to reduce the number of boundary voxels, adjust nbhd_order).

The names of the task-fMRI BOLD sessions, for multi-session analysis. If not provided here, will be inferred from names(BOLD), inferred from names(design), or generated automatically, in that order.

Controls scaling the BOLD response at each location.

"mean":

Scale the data to percent local signal change.

"sd":

Scale the data by local standard deviation.

"none":

Center the data but do not scale it.

Perform spatial Bayesian modeling? Default: TRUE. If FALSE, only perform classical (massive univariate) modeling. (The classical GLM result is always returned, whether Bayes is TRUE or FALSE.)

Should informative or default non-informative hyperpriors be assumed on SPDE hyperparameters?

(For prewhitening) The order of the autoregressive (AR) model to use for prewhitening. If 0, do not prewhiten. Default: 6.

For multi-session modeling, note that a single AR model is used; its coefficients will be the average estimate from each session.

(For prewhitening) The FWHM parameter for spatially smoothing the coefficient estimates for the AR model to use for prewhitening. Recall that \sigma = \frac{FWHM}{2*sqrt(2*log(2)}. Set to 0 to not smooth the estimates. Default: 5.

(For prewhitening) Use the Akaike information criterion (AIC) to select AR model orders between 0 and ar_order? Default: FALSE.

The maximum number of threads to use for parallel computations: prewhitening parameter estimation, and the inla-program model estimation. Default: 4. Note that parallel prewhitening requires the parallel package.

Return the INLA model object? (It can be large.) Use "trimmed" (default) returns the results sufficient for activations and BayesGLM2; "minimal" returns enough for BayesGLM2 but not activations; "full" returns the full inla output.

1 (default) to print occasional updates during model computation; 2 for occasional updates as well as running INLA in verbose mode (if Bayes), or 0 for no printed updates.

Tolerance for mean and variance of each data location. Locations which do not meet these thresholds are masked out of the analysis. Default: 1e-6 for both.

Either (1) a length N list of "BGLM" objects, or (2) a length N character vector of files storing "BGLM" objects saved with saveRDS. "fit_bglm" objects also are accepted.

(Optional) A list of contrast vectors that specify the group-level summaries of interest. If NULL (DEFAULT), use contrasts that compute the average of each field (field HRF) across all subjects/sessions.

Each contrast vector is length KSN specifying a group-level summary of interest, where K is the number of fields in the first-level design matrices, S is the number of sessions, and N is the number of subjects. The vector is grouped by fields, then sessions, then subjects.

For a single session/subject, the contrast vector for the first field would be:

c0 <- c(1, rep(0, K-1)) #indexes the first field for a single session

so the full contrast vector for the group average over all sessions/subjects for the first field would be:

contrasts = rep(c0, S*N) /(S*N).

To obtain the group average for the first field, for just the first session, input zeros for the remaining sessions:

c2 <- c(c0, rep(0, K*(S-1))) contrasts = rep(c2, N) /N.

To obtain the group mean difference between two sessions (S=2) for the first field:

c3 <- c(c0, -c0) contrasts = rep(c3, N) / N.

To obtain the mean over sessions of the first field, just for the first subject:

c4 <- rep(c0, S) c(c4, rep(0, K*S*(N-1))) / S.

(Optional) Vector of posterior quantiles to return in addition to the posterior mean.

(For inference only) The type of excursion function for the contrast (">", "<", "!="), or a vector thereof (each element corresponding to one contrast). If NULL, no inference performed.

(Optional) Names of contrasts.

(For inference only) Activation threshold for the excursion set, or a vector thereof (each element corresponding to one contrast). Default: 0.

(For inference only) Significance level for activation for the excursion set, or a vector thereof (each element corresponding to one contrast). Default: .05.

Number of theta values to sample from posterior. Default: 50.

Number of beta vectors to sample conditional on each theta value sampled. Default: 100.

The number of cores to use for sampling betas in parallel. If NULL (default), do not run in parallel.

1 (default) to print occasional updates during model computation; 2 for occasional updates as well as running INLA in verbose mode (if Bayes), or 0 for no printed updates.

See BayesGLM.

See BayesGLM.

See BayesGLM.

See BayesGLM.

See BayesGLM.

See BayesGLM.

See BayesGLM.

The list of "BGLM" objects, from fit_bayesglm.

See BayesGLM.

"classical" or "Bayesian"

The design argument input. Will be formatted to a nS-length list.

Scale the design matrix by dividing each column by its maximum and then subtracting the mean? Default: TRUE. If FALSE, the design matrix is centered but not scaled.

The expected number of sessions, if known.

The expected number of timepoints, if known. For multi-session data this is a session-length vector.

The expected number of designs, if known. For per-location modeling this is equal to nVd0, the initial number of data locations. For multi-session data this is a session-length vector.

FALSE if per-location modeling is not being performed (i.e. for multiGLM); TRUE if it is; or, NULL to infer based on the dimensions of design (TRUE if the design has three dimensions.)

The nuisance argument input. Will be formatted to a nS-length list.

The expected number of sessions, if known.

The expected number of timepoints, if known. For multi-session data this is a session-length vector.

The scrub argument input. Will be formatted to a nS-length list.

The expected number of sessions, if known.

The expected number of timepoints, if known. For multi-session data this is a session-length vector.

The design matrix/array

The nuisance matrix

The scrub matrix

Perform spatial Bayesian modeling? Default: TRUE. If FALSE, only perform classical (massive univariate) modeling. (The classical GLM result is always returned, whether Bayes is TRUE or FALSE.)

precision matrix

noise variance

list with X and y

data locations to mesh locations matrix

posterior mean of w

posterior covariance of w

crossprod(X%*%Psi)

(logical) Should the EM implementation of the Bayesian GLM be used? Default: FALSE. This method is still in development.

A vector of hyperparameter values at which to compute the posterior log density

A SPDE object from inla.spde2.matern() function, determines prior precision matrix

Posterior mean from combined subject-level models.

Posterior precision matrix from combined subject-level models.

Number of subjects

a list containing kappa2, phi, and sigma2, in that order

the spde object

the model_data object containing y and X

a conversion matrix (N by V) (or N by n)

number of covariates

The value for Matrix::crossprod(X%*%Psi) (saves time on computation)

parallelization cluster

The number of samples used to approximate traces using the Hutchinson estimator. If set to 0, the exact trace is found.

a vector containing kappa2, phi, and sigma2, in that order

the spde object

the model_data object containing y and X

a conversion matrix (N by V) (or N by n)

number of covariates

The value for Matrix::crossprod(X%*%Psi) (saves time on computation)

Needed for SQUAREM (it is an argument to the fixed-point functions)

The number of samples used to approximate traces using the Hutchinson estimator. If set to 0, the exact trace is found.

BOLD timeseries in vector form (TVx1), result of sparse_and_PW

List of large sparse design matrices (each TVxV), one per regressor, result of sparse_and_PW

See fit_bayesglm.

See fit_bayesglm.

Has prewhitening been performed on the data and design?

Compute SE of model coefficients?

See fit_bayesglm.

See fit_bayesglm.

See fit_bayesglm.

scalar

spde object

See BayesGLM.

QC mask.

vector of min, max and initial value for prior on log kappa and log tau. Min and max are extreme quantiles, not hard constraints.

See BayesGLM.

QC mask.

vector of min, max and initial value for prior on log kappa and log tau. Min and max are extreme quantiles, not hard constraints.

Temporal resolution of the data, in seconds.

a scalar

spde object

Matrix of dimension nk by N_s found by solve(Sig_inv,Vh)

posterior mean

matrix of random variables with nrow(Sig_inv) rows and Ns columns

scalar

a vector

an spde object

Matrix of dimension nk by N_s found by solve(Sig_inv,Vh)

Matrix of dimension nk by nk inside the desired trace product

Matrix of dimension nk by N_s in which elements are -1 or 1 with equal probability.

Result of BayesGLM or fit_bayesglm model call, of class "BGLM" or "fit_bglm".

Use spatial Bayesian modeling to identify activations based on the joint posterior distribution? Default: TRUE. If FALSE, activations will be based on classical (massive univariate) GLM model, with multiple comparisons correction (see correction). Note that TRUE is only applicable if x includes Bayesian results (i.e. x <- BayesGLM(..., Bayes = TRUE) was run.)

Activation threshold, for example 1 for 1 percent signal change if scale_BOLD=="mean" during model estimation. Setting a gamma is required for the Bayesian method; NULL (default) will use a gamma of zero for the classical method.

Significance level for inference. Default: 0.05.

For the classical method only: Type of multiple comparisons correction: "FWER" (Bonferroni correction, the default), "FDR" (Benjamini Hochberg), or "none".

The field(s) to identify activations for. Give either the name(s) as a character vector, or the numerical indices. If NULL (default), analyze all fields.

The session(s) to identify activations for. Give either the name(s) as a character vector, or the numerical indices. If NULL (default), analyze the first session.

1 (default) to print occasional updates during model computation; 2 for occasional updates as well as running INLA in verbose mode (if Bayes), or 0 for no printed updates.

A BayesGLM object

See activations.

(Optional) An "inla.mesh" object (see make_mesh for surface data). Only necessary for computing surface areas of identified activations.

Result of BayesGLM, of class "BGLM".

See activations.

(For prewhitening) Use the Akaike information criterion (AIC) to select AR model orders between 0 and ar_order? Default: FALSE.

(For prewhitening) The order of the autoregressive (AR) model to use for prewhitening. If 0, do not prewhiten. Default: 6.

For multi-session modeling, note that a single AR model is used; its coefficients will be the average estimate from each session.

(For prewhitening) The FWHM parameter for spatially smoothing the coefficient estimates for the AR model to use for prewhitening. Recall that \sigma = \frac{FWHM}{2*sqrt(2*log(2)}. Set to 0 to not smooth the estimates. Default: 5.

A single sample of theta (hyperparameters) from q(theta|y)

A SPDE object from inla.spde2.matern() function.

A crossproduct of design matrix.

A crossproduct of design matrix and BOLD y.

A list of vectors of length M*K specifying the contrasts of interest.

Vector of posterior quantiles to return in addition to the posterior mean

Vector of excursion function type (">", "<", "!=") for each contrast

Vector of activation thresholds for each contrast

Significance level for activation for the excursion sets

The number of samples to draw from full conditional of beta given the current value of theta (p(beta|theta,y))

Character vector indicating which brain structure(s) of BOLD to analyze: "left" cortex; "right" cortex; and/or "subcortical" structures. Or "all" to model all three. Default: c("left","right") (cortex only).

For volumetric model. The number of extra voxel layers around the bounding box. Set to NULL for no buffer. (We recommend not changing buffer unless you know what you're doing. Instead, to reduce the number of boundary voxels, adjust nbhd_order).

NULL or a numeric matrix

A numeric matrix with the same number of rows as mat_or_NULL

Is PARDISO required? Default: FALSE.

number of samples

mean vector

Cholesky decomposition of the precision (found via Matrix::Cholesky(Q))

List of contrast vectors to be passed to inla::inla.

an spde object

A numeric matrix or data.frame, or a "BayesfMRI_design" object from make_design. Can also be an array where the third dimension is the same length as the number of data locations, to model each location with its own design.

dgCMatrix

The column indices to set to zero

A session-length list of T \times V numeric BOLD data.

Tolerance for mean and variance of each data location. Locations which do not meet these thresholds are masked out of the analysis. Defaults: 1e-6.

Print messages counting how many locations are removed? Default: TRUE.

The stopping tolerance for the EM algorithm. Default: 1e-3.

An object of class "inla", a result of a call to inla.

Vector of fMRI session names

A string representing the posterior summary statistic to be returned

An F \times 3 matrix, where each row contains the vertex indices for a given triangular face in the mesh. F is the number of faces in the mesh.

(Optional) Names of fields represented in design matrix.

Session-length list of numeric matrices/arrays, each with volumes along the first dimension.

Session-length list of spike regressors: numeric matrices, with volumes along the first dimension, valued at 1 for scrubbed volumes and 0 otherwise.

Scrubbing is performed by incorporating spike regressors in the nuisance matrix during nuisance regression (in a simultaneous framework), and then removing the scrubbed timepoints from the resulting BOLD and design.

Gives the spatial information:

surf

A list of two: vertices V \times 3 numeric matrix of vertex locations in XYZ coordinate space, and faces, F \times 3 matrix of positive integers defining the triangular faces.

mask

Mask of locations with valid data.

For voxel data, a list of six:

label

3D array of labeled locations to include in the model.

trans_mat

Projection matrix to convert voxel indices to XYZ position. Can be NULL.

trans_units

XYZ units. Can be NULL.

nbhd_order

See documentation for BayesGLM.

buffer

See documentation for BayesGLM.

Controls scaling the BOLD response at each location.

"mean":

Scale the data to percent local signal change.

"sd":

Scale the data by local standard deviation.

"none":

Center the data but do not scale it.

Perform spatial Bayesian modeling? Default: TRUE. If FALSE, only perform classical (massive univariate) modeling. (The classical GLM result is always returned, whether Bayes is TRUE or FALSE.)

Should informative or default non-informative hyperpriors be assumed on SPDE hyperparameters?

(For prewhitening) The order of the autoregressive (AR) model to use for prewhitening. If 0, do not prewhiten. Default: 6.

For multi-session modeling, note that a single AR model is used; its coefficients will be the average estimate from each session.

(For prewhitening) The FWHM parameter for spatially smoothing the coefficient estimates for the AR model to use for prewhitening. Recall that \sigma = \frac{FWHM}{2*sqrt(2*log(2)}. Set to 0 to not smooth the estimates. Default: 5.

(For prewhitening) Use the Akaike information criterion (AIC) to select AR model orders between 0 and ar_order? Default: FALSE.

The maximum number of threads to use for parallel computations: prewhitening parameter estimation, and the inla-program model estimation. Default: 4. Note that parallel prewhitening requires the parallel package.

Return the INLA model object? (It can be large.) Use "trimmed" (default) returns the results sufficient for activations and BayesGLM2; "minimal" returns enough for BayesGLM2 but not activations; "full" returns the full inla output.

1 (default) to print occasional updates during model computation; 2 for occasional updates as well as running INLA in verbose mode (if Bayes), or 0 for no printed updates.

Tolerance for mean, variance and SNR of each data location. Locations which do not meet these thresholds are masked out of the analysis. Default: 1e-6 for mean and variance, 50 for SNR.

Matrix of faces in triangularization

Matrix of vertex locations in triangularization

(logical) Will this create the SPDE matrices for a surface or not?

spatial

An object of class "inla", a result of a call to inla()

The model used for the latent fields in the inla() call, an object of class "inla.spde"

Descriptive names of model regressors (fields).

An object of class "inla", a result of a call to inla()

Descriptive names of model regressors (fields).

Add DCT bases to nuisance to apply a temporal high-pass filter to the data, for detrending? hpf is the filter frequency. Use NULL to skip detrending. Detrending is strongly recommended for fMRI data, to help reduce the autocorrelation in the residuals, so NULL will induce a warning. Use "already" to disable the warning while skipping highpass filtering.

Using at least two DCT bases is as sufficient for detrending as using linear and quadratic drift terms in the nuisance matrix. So if DCT detrending is being used here, there is no need to add linear and quadratic drift terms to nuisance.

a vector c(kappa2,phi)

an spde object

vector

a vector c(kappa2,phi)

an spde object

vector

The list of "fit_bglm", "BGLM", or "act_BGLM" objects.

The object

scalar

scalar

an spde object

vector

See fit_bayesglm

See BayesGLM

The full-resolution "surf" object.

The downsampled "surf" object.

the hyperparameter theta vector of length K * 2 + 1, where the first K elements are the kappas, the next K elements are the phis, and the last element (unused here) corresponds to sigma2

a list containing three spde elements: Cmat, Gmat, and GtCinvG

The integer number of sessions

Vectorized BOLD data (all voxels, sessions, etc.)

List (length = number of sessions) of sparse design matrices size TVxVK from each session, each created using sparse_and_PW

List (length = number of fields) of beta objects from make_replicates

List (length = number of fields) of repl objects from make_replicates

A V \times 3 matrix, where each row contains the Euclidean coordinates at which a given vertex in the mesh is located. V is the number of vertices in the mesh

An F \times 3 matrix, where each row contains the vertex indices for a given triangular face in the mesh. F is the number of faces in the mesh.

The number of sessions sharing hyperparameters (can be different fields)

Vector of names for each field

Spatial info

See GLM_est_resid_var_pw.

A length V logical vector indicating if each vertex is within the input mask.

The maximum number of threads to use in the inla-program for model estimation. 0 (default) will use the maximum number of threads allowed by the system.

Tolerance for mean and variance of each data location. Locations which do not meet these thresholds are masked out of the analysis. Default: 1e-6 for both.

An "inla.mesh" object (see make_mesh for surface data)

An "inla.mesh" object (see make_mesh for surface data).

The maximum number of threads to use for parallel computations: prewhitening parameter estimation, and the inla-program model estimation. Default: 4. Note that parallel prewhitening requires the parallel package.

For volumetric model. What order neighborhood around data locations to keep? 0 for no neighbors, 1 for 1st-order neighbors, 2 for 1st- and 2nd-order neighbors, etc. Smaller values will provide greater computational efficiency at the cost of higher variance around the edge of the data.

scalar

an spde object

scalar

dgCMatrix

dgeMatrix

scalar

an spde object

scalar

Matrix of dimension nk by N_s found by solve(Sig_inv,Vh)

dgeMatrix

random matrix of -1 and 1 of dim dim(P)

scalar

an spde object

precomputed coefficient (scalar)

precomputed coefficient (scalar)

number of sessions (scalar)

scalar

a list of SPDE prior matrices from

precomputed coefficient (scalar)

precomputed coefficient (scalar)

number of sessions (scalar)

(Optional) A T \times N_{nuis} matrix of nuisance signals, where T is the number of timepoints and N is the number of nuisance signals, or a list of these for multi-session analysis. Nuisance signals are regressed from the fMRI data and design matrix prior to GLM computation. Nuisance signals can include motion regressors, HRF derivatives not being modeled as tasks, and other sources of noise.

Detrending/high-pass filtering is accomplished by adding DCT bases to the nuisance matrix; see the parameters hpf and DCT.

Do not add spike regressors for scrubbing to the nuisance matrix. Rather, provide these in scrub so that their corresponding timepoints are also removed from the BOLD data after nuisance regression.

An object of class "BGLM"

TRUE for plotting Bayesian results, FALSE for plotting classical GLM results. Default: NULL, which will use the Bayesian results if available and the classical results if not.

Which field should be plotted? Give the numeric indices or the names. NULL (default) will show all fields. This argument overrides the idx argument to view_xifti.

If NULL, the field names associated with idx will be used.

Which session should be plotted? NULL (default) will use the first.

Overrides the zlim argument for view_xifti. Default: c(-1, 1).

Additional arguments to view_xifti

An object of class "BGLM2"

Which contrast should be plotted? Give the numeric indices or the names. NULL (default) will show all contrasts. This argument overrides the idx argument to view_xifti.

Estimates of the "contrasts" (default), or their thresholded "activations".

Overrides the zlim argument for view_xifti. Default: c(-1, 1).

Additional arguments to view_xifti

An object of class "act_BGLM"

Which field should be plotted? Give the numeric indices or the names. NULL (default) will show all fields. This argument overrides the idx argument to view_xifti.

If NULL, the field names associated with idx will be used.

Which session should be plotted? NULL (default) will use the first.

Additional arguments to view_xifti

An object of class "prev_BGLM"

Which task should be plotted? Give the numeric indices or the names. NULL (default) will show all tasks. This argument overrides the idx argument to view_xifti.

Which session should be plotted? NULL (default) will use the first.

Color locations without any activation across all results (zero prevalence) the same color as the medial wall? Default: NULL to drop the zeros if only one idx is being plotted.

See view_xifti.

Additional arguments to view_xifti

an spde object

the mean

scale parameter

Matrix of dimension nk by N_s found by solve(Sig_inv,Vh)

random matrix of -1 and 1 of dim dim(P)

List of activations from activations. All should have the same sessions, fields, and brainstructures.

If activations at multiple thresholds were computed, which threshold should be used for prevalence? Default: the first (lowest).

For inference: the expected baseline rate of activation for all data locations, under the null hypothesis. Default: NULL (do not perform hypothesis testing).

Significance level for inference. Default: .05.

For the classical method only: Type of multiple comparisons correction: "FWER" (Bonferroni correction, the default), "FDR" (Benjamini Hochberg), or "none".

Estimated residuals in T \times V numeric matrix

Order of the AR model used to prewhiten the data at each location. If !aic (default), the order will be exactly ar_order. If aic, the order will be between zero and ar_order, as determined by the AIC.

See fit_bayesglm internal code.

A Vxp matrix of estimated AR coefficients, where V is the number of vertices and p is the AR model order

A vector length V containing the white noise variance estimates from the AR model

FWHM parameter for smoothing. Remember that \sigma = \frac{FWHM}{2*sqrt(2*log(2)}. Set to 0 or NULL to not do any smoothing. Default: 5.

number of samples

mean vector (length = p)

sparse p x p positive definite precision matrix (class = dgCMatrix)

For cortex spatial model. The number of vertices to which each cortical surface should be resampled, or NULL to not resample.

For computational feasibility, a value of 10000 (default) or lower is recommended for Bayesian spatial modeling. If Bayes=FALSE, resamp_res can be set to NULL for full-resolution classical modeling.

The activations object

The masks to be applied to each brain structure of x.

The "fit_bglm" result

The mask to be applied to x. It's relative to the full mesh/voxel array, not x.

The mesh

The mask to be applied to x (on top of any masks already applied to it.)

Return the INLA model object? (It can be large.) Use "trimmed" (default) returns the results sufficient for activations and BayesGLM2; "minimal" returns enough for BayesGLM2 but not activations; "full" returns the full inla output.

A vector consisting of all variables of interest for a single draw from a posterior distribution

A scale parameter used to determine at what distance cluster centers are considered to be the same.

An array of posterior samples (typically a matrix), in which the last margin corresponds to a single posterior sample

A scale parameter used to determine at what distance cluster centers are considered to be the same.

fMRI data as a locations by time (V \times T) numeric matrix.

Option for scaling the BOLD response.

Original means of the BOLD data. ONLY provide if data has already been centered.

\code{"mean"} scaling will scale the data to percent local signal change.

\code{"sd"} scaling will scale the data by local standard deviation.

\code{"none"} will only center the data, not scale it.

Controls scaling the BOLD response at each location.

"mean":

Scale the data to percent local signal change.

"sd":

Scale the data by local standard deviation.

"none":

Center the data but do not scale it.

The original (unscaled) design matrix that is T x K, where T is the number of time points, and k is the number of field covariates

(Optional) A T \times N_{scrub} matrix of spike regressors (one 1 value at the timepoint to scrub, and 0 for all other values), or a logical vector indicating the timepoints to scrub (TRUE to scrub, and FALSE to keep). For multi-session data, a session-length list of such matrices or logical vectors.

The spike regressors will be included in the nuisance regression, and afterwards the timepoints indicated in scrub will be removed from the BOLD data and design matrix.

Random seed (optional). Default: NULL.

The names of the task-fMRI BOLD sessions, for multi-session analysis. If not provided here, will be inferred from names(BOLD), inferred from names(design), or generated automatically, in that order.

See fit_bayesglm.

See fit_bayesglm.

See fit_bayesglm.

A scalar

A scalar

An object containing the information about the mesh structure for the SPDE prior

Object of class "BGLM".

further arguments passed to or from other methods.

Object of class "summary.BGLM".

Object of class "BGLM2".

further arguments passed to or from other methods.

Object of class "summary.BGLM2".

Object of class "act_BGLM".

further arguments passed to or from other methods.

Object of class "summary.act_BGLM".

Object of class "act_fit_bglm".

further arguments passed to or from other methods.

Object of class "summary.act_fit_bglm".

Object of class "fit_bglm".

further arguments passed to or from other methods.

Object of class "summary.fit_bglm".

Object of class "fit_bglm2".

further arguments passed to or from other methods.

Object of class "summary.fit_bglm2".

Object of class "prev_BGLM".

further arguments passed to or from other methods.

Object of class "summary.prev_BGLM".

Object of class "prev_fit_bglm".

further arguments passed to or from other methods.

Object of class "summary.prev_fit_bglm".

For cortex spatial model. Left and right cortex surface geometry in GIFTI format ("*.surf.gii"). These can be a file path to a GIFTI file or a "surf" object from ciftiTools.

Surfaces can alternatively be provided through the $surf metadata in BOLD if it is "xifti" data. If neither are provided, by default the HCP group-average fs_LR inflated surfaces included in ciftiTools will be used for the cortex spatial model.

(logical) should the INLA_model_obj within the result be trimmed to only what is necessary to use activations? Default: TRUE.

An object of class "inla".

Just keep the two parameters needed for BayesGLM2? Default: FALSE. !minimal is required for activations, but minimal is sufficient for BayesGLM2.

The matrix to unmask

The input mask, and the modeled locations mask, as logical vectors. The latter should be a subset of the former. They should have the same length.

The fill value. Default: NA.

spatial

1 (default) to print occasional updates during model computation; 2 for occasional updates as well as running INLA in verbose mode (if Bayes), or 0 for no printed updates.

An "inla.mesh" object (see make_mesh for surface data).

A V \times 3 matrix, where each row contains the Euclidean coordinates at which a given vertex in the mesh is located. V is the number of vertices in the mesh

An array of 0s and 1s representing a volumetric mask

The spatial resolution in each direction, in mm. For example, c(2,2,2) indicates 2mm isotropic voxels.

For volumetric data, what order neighborhood around data locations to keep? (0 = no neighbors, 1 = 1st-order neighbors, 2 = 1st- and 2nd-order neighbors, etc.). Smaller values will provide greater computational efficiency at the cost of higher variance around the edge of the data.

For volumetric data, size of extra voxels layers around the bounding box, in terms of voxels. Set to NULL for no buffer.