fmristat Release Notes

* 1 October, 2004: Pca_image now does a PCA (or PLS) on correlations rather than covariances by default.

• Fmrilm: The fwhm of the spatial smoothing of the autocorrelation parameters is now chosen by default to achieve 100 df or 90% of the residual df of each contrast. The df's are now different for different contrasts - see

Worsley, K.J. (2004). Spatial smoothing of autocorrelations to control the degrees of freedom in fMRI analysis. NeuroImage?, submitted.

• To make the use of FMRISTAT simpler, these dfs and fwhms are now written into the image file headers, so that in most cases you can leave these parameters empty in the calls to most functions.

• The which_stats parameter of fmrilm and multistat: can now be a string describing the statistics you want, rather than 1's and 0's.

• New theoretical p-values based on discrete local maxima are now calculated by stat_summary - they can halve the p-values - see

Worsley, K.J. (2003). An improved theoretical P-value for SPMs based on discrete local maxima. NeuroImage?, submitted.

* 19 July, 2004: Stat_threshold has been extended to compute thresholds and p-values for multivariate test statistics such as Roy's maximum root, maximum canonical correlations, and cross- and auto- correlations - see

Cao, J. and Worsley, K.J. (1999). The geometry of correlation fields with an application to functional connectivity of the brain. Annals of Applied Probability, 9, 1021-1057

Worsley, K.J., Taylor, J.E. & Tomaiuolo, F. (2004). Roy's maximum root and maximum canonical correlation SPMs from multivariate multiple regression analysis of imaging data. NeuroImage?, 22:WE393.

* 28 May, 2004: New connectivity analyses added.

* 2 September, 2003: Darn it, more little bugs fixed ...

* 1 July, 2003: Complete revision of most functions, including bug fixes for dealing with ANALYZE format files:

• Fmrilm : Several adjustments have been made to allow for longer runs. Temporal trends are now modeled as cubic splines, rather than polynomials. The N_TEMPORAL parameter (formerly N_POLY) is now the number of spline terms per 6 minutes of scanner time, so for the old 6 minute runs, the number of terms remains the same, but more will be added for longer runs. Moreover for N_TEMPORAL=3, a cubic spline is just a cubic, so for the old 6 minute runs, N_TEMPORAL=3 will give the same results as before! It also adjusts the amount of smoothing of the AR coefficients for longer runs; if there are more frames, the AR coefficients are smoothed less, reducing the bias.

• Multistat : Now chooses the amount of smoothing to target a specified df (default: 100), so that if there are more runs/sessions/subjects, less smoothing will be needed, reducing the bias.
• Locmax : Now ouputs a volume of clusters labelled by an index, to make it easier to find clusters in visualization tools such as 'register'.
• Blob_brain is a new function that produces an isosurface 3D 'blob' plot that can be rotated. Useful for looking at activated regions coloured by effect size, clusters, and confidence regions.
• Stat_summary produces voxel and world (for minc files) coordinates, and a blob-plot. as well a an SPM-stlye summary of clusters, peaks and their P-values and Q-values.
• Finding the exact resels of a search region with mask_resels: A new section for finding exact non-isotropic resels of arbitrarily shaped search regions (up to now, fmristat has assumed that search regions are spherical). This gives more precise P-values for peaks, but these are not much different from the usual P-values that assume a spherical search region, so this is only of interest for the perfectionist.

* 25 October, 2002: I have noticed that you get a huge improvement in fmrilm by converting each volume of the data to percent of the whole volume. On the test data set here, I noticed a 20-80% increase in T-statistics, further enhanced by adding the whole volume average as a covariate to the linear model. This is now the default in fmrilm . To retrieve the old analysis, set n_poly (now a vector n_trends - see the help of fmrilm ) to [3 0 0]; for conversion to percent only (no extra covariate), set n_poly to [3 1 0]. Note that if you convert to percent of the time course (using the previous fmri2pcnt, now abandoned) then this has no effect on T-statistics. Nearly all the figures have been redrawn, using an improved view_slices . World coordinates in mm are now output from stat_summary to make it easier to report results.

* 5 August, 2002: A few bug fixes in reading and writing analyze format files. A new program efficiency to calculate efficiencies and optimum designs - see Efficiency and choosing the best design.

* 25 April, 2002: Fmrilm has been extended to allow for different confounds for different slices. This is useful for effective connectivity, where a new function fmri_interp has been added to resample a single fmri time series at different frame times and slice times. See Effective connectivity of all voxels with a reference.

* 12 April, 2002: Fmristat now reads ANALYZE format files created on any machine, and the file separator can be / or \.

* 8 April, 2002: Fmristat now allows for non-isotropic (non-constant FWHM) data. Many of the FWHM parameters can now be replaced by FWHM images. Fmristat now works for 2D data (one slice) as well as 3D data (more than one slice).

• Fmrilm now reads multiple 3D image files as well as a single 4D file - useful for ANALYZE format data.
• Fmrilm and multistat have been extended to estimate the effective FWHM of the data both from fixed and random effects. This allows for non-isotropic inference.
• Multistat has been extended to estimate mixed effects conjunctions. It is now very fast if no sd files are specified, making it suitable for analyzing PET data.
• Locmax now finds clusters as well as local maxima.
• Stat_threshold has been extended to find P-values and thresholds for conjunctions, and it now adjusts P-values and thresholds for clusters if the cluster extent is measured in resels and the resels are themselves estimated by fmrilm or multistat.