Imaging and Unix Glossary

back to the main page


links to information for neuroimagers and visitors to Tucson



search the CNL site


How to write cds on our machines



printable forms of use to CNL researchers


tranferring files between cnl machines


image rotation, orientation, and translation


image analysis software info and links


Instructions and Maintenance information for CNL machines


all about our MR scanners


new calendar of scanner and equipment usage


how to capture screenshots on the sgis


transferring information and files between computers


an introduction to using unix


A discussion forum for CNL Researchers

Index:1 2 3 A B C D E F G H I J K L M N O P Q R S T U V W X Y Z


1D files Files generated by the Afni Waver program.


2D Anatomicals See Anatomical files


3dAFNItoANALYZE an image conversion utility that comes with the afni package

3dcalc An afni tool that allows you to perform algebraic operations on images (e.g., multiplying images together to create a mask. See example). See also imcalc.

3dDeconvolve Afni tool for deconvolving data, generally used to identify the HRF of individual voxels using the data and the stimulus information. See Deconvolution Models, Sample Data Analysis with 3DDeconvolve, Irregular Stimulus Timing: Analysis with 3dDeconvolve,

3dinfo An afni utility that shows you information about the history of your afni BRIK.

>3dinfo fred+orig

>3dinfo -verb fred+orig

(for even more information.)

3dIntracranial An afni command used for skull stripping.

3dMINCtoAFNI an image conversion utility that comes with the afni package

3D Anatomicals See Anatomical files


AAL Atlas A free atlas of anatomical regoins in the brain. Used by WFU Pickatlas, MRIcro and Marsar. Official AAL Site

AC-PC The line from the Anterior Commisure to the Posterior Commisure defines a plane of section used in the Talairach atlas and thus frequently desirable as the plane of section for MRIs. A simple graphical representation of how to find this line in the sagittal plane is provided (borrowed from Chris Rorden's Mricro page).

Activ 2000 fMRI analysis package for MS Windows. Activ 2000 Home page

Activation When a voxel responds positively to a condition, that is, the intensity of the signal in the voxel rises over time in response to the condition. InAFNI, activation and deactivation are represented with different color scales. In SPM, there is no simple way to identify the difference between deactivations and activations.
SPM Archives -- 2000 (#1330)
"You should also be aware that an "activation" or a "deactivation" is always relative to some baseline which may be more or less well defined. If you are using rapid stimulus presentation (short SOA) without null events it will be less well defined, and it will be very difficult to
determine between activations and deactivations. In that case the interpretation of a positive finding in the contrast [1 -1] can be larger activation in A than in B, or less deactivation in A than in B, or anything in betweeen.

ADC See Diffusion.

Adjusted Data In SPM, data adjusted for confounds (e.g., global flow) and high and low pass filtering.

Affine Transformations Mathematical transformations that effectively change the view of an image by transformations such as rotation, translation etc. Affine transformations are considered linear. Nonlinear transformations that alter the relative size of different parts of the brain are non-affine.

From Mathworld: An affine transformation is any transformation that preserves collinearity (i.e., all points lying on a line initially still lie on a line after transformation) and ratios of distances (e.g., the midpoint of a line segment remains the midpoint after transformation). While an affine transformation preserves proportions on lines, it does not necessarily preserve angles or lengths. Any triangle can be transformed into any other by an affine transformation, so all triangles are affine and, in this sense, affine is a generalization of congruent and similar.

If you have a partial brain or a very abnormal brain, you may minimize non-affine transformations:

spm99: These options are under Defaults->Spatial Normalization->Defaults for Parameter Estimation:
Nonlinear Basis functions can be set to None.
Nonlinear Iterations can be set to One (there isn't a None option)
Nonlinear regularization can be set to "Very Light"

spm2: These options are under "Defaults" Defaults->Spatial Normalization->Defaults for Parameter Estimation: Nonlinear Regularization: Very light regularization
# Nonlinear Iterations? One nonlinear iteration

Afni (Automated Functional Neuro-Imaging), a free unix based fmri image processing program, available as source code or binaries for several unix platforms. Start afni by typing >afni at any command prompt (installed on merlin). Afni uses the BRIK/HEAD file format, but now supports the MINC format. Using to3d, one can create an appropriate HEAD file for any uncompressed image data, including SPM (Analyze), DICOM etc. See Conversion. . See also nifti, subbrick, multibucket image, fim, fico, to3d, afni preprocessing scripts.

The standard citation for the Afni software is:

Cox, R. W. (1996). AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. Computers & Biomedical Research, 29, 162-173.

afnireg2bshort An image conversion program that comes with the mgh package. Very similar to grecon2bshort.

AIR (Automated Image Registration) Developed by Roger Woods, this program does an excellent job of realigning functional images that might otherwise be useless because of movement artifacts. AIR uses a file format compatible with Analyze.

Analyze A medical image processing program from the Mayo Clinic. With slight modifications, the image file header structure has been used for SPM. Visit the Analyze Home Page. See also Analyze File Format Specs, Conversion and Image.

analyze2genesis One of a suite of imaging tools from UCLA. This one converts an img/hdr pair back into a series of *.MR files. See UCLA Brain Imaging Center.

Anatomical files Usually 256x256 grayscale images of the structures in the brain. Our axial images are in radiological orientation by default. Each image represents a slice through the brain and thus has thickness (which means it consistes of voxels rather than pixels). These can be overlaid with the much lower resolution (64x64) functional images, so that the regions of activation in the functional image can be localized in the brain's anatomy. The anatomical images may also be used for morphometry. Our images are in Genesis format (the native format of the GE scanner).
We create 2 different series of these images, the "2D" series which is usually 17-19 images in axial orientation and the 3D series which is usually ~124 images, taken sagittally from left to right.
Our files normally have the 2 bytes of image depth (16 bits) or 65,536 levels of gray (i.e., 2^16). Many image processing programs use 8 bit depth or 256 levels of gray (i.e., 2^8).
If you want to read the image as a "raw" image in ScionImage, ImageJ etc., then you need to know the offset. See also Image. See rdgehdr (Read GE Header).

Anisotropic In diffusion weighted imaging, movement of water molecules that is impeded in some directions more than others (e.g., movement through a tube). Compare to isotropic.

Anterior Toward the face or front of the head. Compare to Posterior

AR(1) or AR(1) + w (or (AR(2), AR(3), etc.): Terms used to describe different models of autocorrelation in your fMRI data. See autocorrelation below for more info. AR stands for autoregression. AR models are used to estimate to what extent the noise at each time point in your data is influenced by the noise in the time point (or points) before it. The amount of autocorrelation of noise is estimated as a model parameter, just like a beta weight. The difference between AR(1), AR(2), AR(1) + w, etc., is in which parameters are estimated. An AR(1) model describes the autocorrelation function in your data by looking only at one time point before each moment. In other words, only the correlation of each time point to the first previous time point is considered. In an AR(2) model, the correlation of each time point to the first previous time point and the second previous time point is considered; in an AR(3) model, the three time points before each time point are considered as parameters, etc. The "w" in AR(1) + w stands for "white noise." An AR(1) + w model assumes the value of noise isn't solely a function of the previous noise; it also includes a random white noise parameter in the model. AR(1) + w models, which are used in SPM2 and other packages, seem to do a pretty good job describes the "actual" fMRI noise function. A good model can be used to remove the effects of noise correlation in your data, thus validating the assumptions of the general linear model. (From Gablab Wiki: Glossary)

Image artifacts are problems in the image created by metal, problems with the machine, poorly sheilded wires.

Autocorrelation (function, correction, etc.): One major problem in the statistical analysis of fMRI data is the shape of fMRI noise. Analysis with the general linear model assumes each timepoint is an independent observation, implying the noise at each timepoint is independent of the noise at the next timepoint. But several empirical studies have shown that in fMRI, that assumption's simply not true. Instead, the amount of noise at each timepoint is heavily correlated with the amount of noise at the timepoints before and after. fMRI noise is heavily "autocorrelated," i.e., correlated with itself. This means that each timepoint isn't an independent observation - the temporal data is essentially heavily smoothed, which means any statistical analysis that assumes temporal independence will give biased results. The way to deal with this problem is pretty well-established in other scientific domains. If you can estimate what the autocorrelation function is - in other words, what, exactly, is the degree of correlation of the noise from one timepoint to the next - than you can remove the amount of noise that is correlated from the signal, and hence render your noise "white," or random (rather than correlated). This strategy is called pre-whitening, and is referred to in some fMRI packages as autocorrelation correction. The models used to do this in fMRI are mostly AR(1) + w models, but sometimes more complicated ones are used." (From Gablab Wiki: Glossary)

averager One of a suite of imaging tools from UCLA. This program averages together a set of timepoints specified in a text file. See UCLA Brain Imaging Center.

Axial (Same as "transverse" and "horizontal")


B-spline, B-spline interpolation: A type of spline which is the generalization of the Bezier curve. MathWorld has this to say about them: B-Spline. Essentially, though, a B-spline is a type of easily describable and computable function which can take many locally smooth but globally arbitrary shapes. This makes them very nice for interpolation. SPM2 has ditched sinc interpolation in all of its resampling/interpolation functions (like normalization or coregistration - anything involving resampling and/or reslicing). Instead, it's now using B-spline interpolation, improving both computational speed and accuracy." (From Gablab Wiki: Glossary)

Backward Font
See Font
: A) The point from which deviations are measured. In a signal measure like % signal change, the baseline value is the answer to, "Percent signal change from what?" It's the zero point on a % signal change plot. B) A condition in your experiment that's intended to contain all of the cognitive tasks of your experimental condition - except the task of interest. In fMRI, you generally can only measure differences between two conditions (not anything absolute about one condition). So an fMRI baseline task is one where the person is doing everything you're not interested in, and not doing the thing you're interested in. This way you can look at signal during the baseline, subtract it from signal during the experimental condition, and be left with only the signal from the task of interest. Designing a good baseline is crucially important to your experiment. Resting with the eyes open is a common baseline for certain types of experiment, but inappropriate for others, where cognitive activity during rest may corrupt your results. In order to get good estimates of the shape of your HRF, you need to have a baseline condition (as opposed to several experimental conditions.) (From Gablab Wiki: Glossary)

Basis Function
(SPM) The hemodynamic response to each stimulus or epoch type is modeled in SPM as one or more basis functions. These are functions that extend over a relatively short (event-related) or long (epoch-related) period of time, and are convolved with the stimulus pulse functions to arrive at the linear regressor(s) representing what the brain response should look like at each voxel. If you ask SPM for "time derivatives", you get one extra basis function for each of the original basis functions. If there are multiple functions comprising the basis set, SPM adjusts them so that they are orthogonal.

In event-related designs, one possible basis function is a single "hrf" function. In SPM99, this is a pre-canned function equal to the sum of two beta functions and extending for 32 seconds, which is used by many investigators as a model of the Hemodynamic Response Function. The "hrf" basis function is fixed in shape, though you can add time and dispersion derivative functions to it to create a basis set that may be a more accurate model.

Definition taken from: The Stanford Gablab: StimBasisPlotting.html

Batch file
(See script below)

Bayesian Analysis Opening page for International Society for Bayesian Analysis website.
"Scientific inquiry is an iterative process of integrating accumulating information. Investigators assess the current state of knowledge regarding the issue of interest, gather new data to address remaining questions, and then update and refine their understanding to incorporate both new and old data. Bayesian inference provides a logical, quantitative framework for this process. It has been applied in a multitude of scientific, technological, and policy settings."

See also

beta image Also called a parameter image. It's a voxel-by-voxel summary of the beta weight for a given condition. Usually it's written as an actual image file or sub-dataset, so you could look at it just like a regular brain image, exploring the beta weight at each voxel. In SPM, you get one of these written out for every column in your design matrix - one for each experimental effect for which you're estimating parameter values. (From Gablab Wiki: Glossary)

beta weights Also called parameter weights, parameter values, etc. This is the value of the parameter estimated for a given effect / column in your design matrix. If you think of the general linear model as a multiple regression, the beta weight is the slope of the regression line for this effect. The parameter gets its name as a "beta" weight from the standard regression equation: Y = BX + E. Y is the signal, X is the design matrix, E is error, and B is a vector of beta weights, which estimate how much each column of the design matrix contributes to the signal. Beta weights can be examined, summed, and contrasted at the voxel-wise level for a standard analysis of fMRI results. They can also be aggregated across regions or correlated between subjects for a more region-of-interest-based analysis. (From Gablab Wiki: Glossary)

BIC (Brain Imaging Center) at MNI.

Big Endian Describes a computer architecture (hardware) in which, within a given multi-byte numeric representation, the most significant byte has the lowest address (the word is stored 'big-end-first'). This is used on Suns, SGI's and MACs. (See also Byte Swapping and Little Endian)

Bitmap Image An image composesd of pixels. Check out the Beginner's Guide to Bitmaps. See also Image, voxel and image depth.

Blackman filter see filter

Block Design (also called "Boxcar" design) An experimental design in which stimuli are presented for fixed periods of time, regardless of subject response. Because a block is treated as a single indivisible unit for analysis, trials within a block should all belong to a single condition. Block design thus reduces the opportunity to randomize and mix trials or to use differences in speed or accuracy of subject response to analyze the data later on (Compare to Event-Related Design, See also Afni Block Design). Blocks are also called epochs.

Boxcar Design (see Block Design).

bfloat The bfloat and accompanying hdr file of the same name (e.g., fred.bfloat and fred.hdr) are an image format used for functional data in the MGH-fsfast software. bfloat files contain blocks of floating point numbers representing the image data, and have a small, very short, header that specifies the number of pixels in each of (only)three dimensions, usually interpreted as Y, X and time. Mutiple slice locations are generally represented by increasing the y dimension to make a vertical stack format. (UCLA Brain Mapping Center Image Format Page)

Brain Voyager
A commercial package from the Netherlands for the analysis and visualization of functional and structural magnetic resonance imaging data sets. BrainVoyager can do both standard and surface based analyses and runs on Windows and Unix machines. Brain Voyager uses *.VMR files but appears to be able to import files in Analyze format. Visit the Brain Voyager Home.

Brede Database A database where you input Talairach coordinates and it outputs studies that got activation there. See also

bshort The bshort and accompanying hdr file of the same name (e.g., fred.bshort and fred.hdr) are a format sometimes used for structural/anatomical files by the MGH-fsfast software. "bshort files contain blocks of unsigned short integers representing the image data, and have a small, very short, header that specifies the number of pixels in each of (only)three dimensions, usually interpreted as Y, X and time. Mutiple slice locations are generally represented by increasing the y dimension to make a vertical stack format." (UCLA Brain Mapping Center Image Format Page)

BRIK A BRIK is the afni file that holds images. A BRIK is accompanied by a HEAD file which contains header information about the such things as the size and number of images in the BRIK. Typically a BRIK holds either a small set of anatomical slices (e.g., 17-124) corresponding to a T1 or spgr image set, OR a set of thousands of functional images from a single functional run. See also afni and image, to3d. Bucket See Multibucket Image

Burn See CD Burning.

Byte Swapping The process of converting a file that uses little endian byte order to big endian byte order or vice-versa. The string 'UNIX' might look like 'NUXI' on a machine with a different 'byte order'. We sometimes need to worry about this for our images when we move them from the PC to sun or sgi (or vice-versa). See little endian and big endian. Also see the new afni preprocessing scripts (that help you deal with some of the byte swapping issues).

C-Programming A couple of useful sites: and

Canonical HRF A model of an "average" HRF. Intended to describe the shape of a generic HRF; given this shape and the design matrix, an analysis package will look for signals in the fMRI data whose shape matches the canonical HRF. The different analysis packages (SPM, AFNI, BrainVoyager, etc.) use slightly different canonical HRFs, but they all share the same basic features - a gradual rise up to a peak around six seconds, followed by a more gradual fall back to baseline. Some progams model a slight undershoot; some don't. (From Gablab Wiki: Glossary) (From Gablab Wiki: Glossary) Cantata Part of the Khoros package.

Capture Images See Screenshots

Caret (Computerized Anatomical Reconstruction and Editing Toolkit) is designed for interactively viewing, manipulating (flattening), and analyzing surface reconstructions of the cerebral cortex. You can access it on Merlin by typing >caret at the prompt. Caret is distributed as free standing binaries available for sgis, sun and linux systems. Caret's companion program, Surefit, is used to generate the anatomical files that Caret requires, and any endeavour to make flat maps should likely begin with SureFit and then move on to Caret. Several tutorial data sets are available. Caret is one of several cortical cartography programs available from the Van Essen labs. See also the Caret homepage.

cat A unix command (short for concatenate) which can be used to paste one file to the bottom of another file. In the example below, we concatenate file1 and file2 into file3. See also cut and paste (for side by side concatenation):

>cat file1 file2 >file3

Cavity A topological error in which an island of gray matter voxels is stranded in a sea of white matter. Such errors are important in the reconstruction of the gray matter surface. See also topolgy and handle.

CD Burning How to burn CDs on unix and linux machines.

Cell Array A cell array is a useful Matlab structure to know about if you want to work in SPM. A cell array can hold different sized vectors in each cell. In SPM, you can use the cell array to hold a vector of stimulus onsets for each of several conditions (e.g., the vector for the first condition is in cell 1. The vector for the second condition is in cell 2 etc.)
To create a cell array:
>>a{1} = [1 2 4 6 8]
>>a{2} = [ 5 77 89]
>>a{3} = [3 4 5 6 2 1 7 8 9 334]

You now have a cell array, a, that contains 3 row vectors (this is perfect for SPM stimulus onset times) in 3 cells.
To view a description of the cell array:
To see the contents of cell 1:
To alter the 4th value in cell 1 from 6 to 99:

See also pg 13 of the SPM99WorkbookStudy1.doc.

Client CNL_FMRI See listserv, and imaging-analysis listserv

CNR Contrast to Noise Ratio. Determines the differences between distinct types of tissues in medical images. Compare to SNR.

con image, contrast image A voxel-by-voxel summary of the value of some contrast you've defined. This is often created as a voxel-by-voxel weighted sum of beta images, with the weights given by the value of the contrast vector. In SPM, it's actually written out as a separate image file; in other programs, it's usually written as a separate sub-bucket or the equivalent. It shouldn't be confused with the statistic image, which is a voxel-by-voxel of the test statistic associated with each contrast value. (In SPM, those statistic images are labeled spmT or spmF images.) Only the contrast images - not the statistic images - are suitable for input to a second-level group analysis. (From Gablab Wiki: Glossary) Contact Phonelist

Contrast The actual signal in fMRI data is unfortunately kind of arbitrary. The numbers at each voxel in your functional images don't have a whole lot of connection to any physiological parameter, and so it's hard to look at a single functional image (or set of images) and know the state of the brain. On the other hand, you can easily look at two functional images and see what's different between them. If those functional images are taken during different experimental conditions, and the difference between them is big enough, then you know something about what's happening in the brain during those conditions, or at least you can probably write a paper claiming you do. Which is good! So the fundamental test in fMRI experiments is not done on individual signal values or beta weights, but rather on differences of those things. A contrast is a way of specifying which images you want to include in that difference. A given contrast is specified as a vector of weights, one for each experimental condition / column in your design matrix. The contrast values are then created by taking a weighted sum of beta weights at each voxel, where the weights are specified by the contrast vector. Those contrast values are then tested for statistical significance in a variety of ways. (From Gablab Wiki: Glossary) See also activation and F-contrast.

Coregistration The process of bringing two brain images into alignment Ideally, you'd like them lined up so that their edges line up and the point represented by a given voxel in one image represents the same point in the other image. Coregistration generally refers specifically to the problem of aligning two images of different modalities - say, T1 fMRI images and PET images, or anatomical MRI scans and functional MRI scans. It goes for some of the same goals as realignment, but it generally uses different algorithms to make it more robust.(From Gablab Wiki: Glossary)

Conversion (between Image formats) Afni allows several conversions...more all the time... between BRIK, IMA, img and mnc files. See Bob Cox's What's New page for the newest updates on these capabilities. MRIcro and ezDicom are also very useful for converting between image formats and displaying different formats (mostly raw, dicom and img). See also UCLA Brain Imaging Center and imconvert. See mri_convert. In this glossary, see, nifti, image and format.

Conversion Test Image The Left Lesion Test Data (This data has a big hole in the left front, so you can test your understanding of what is happening to right and left given a particular program or manipulation. There is a single functional image, and "2D" and "3D" structural images in spm format).




Sample Command





>3dAFNItoANALYZE fred test+orig

Converts an afni pair test+orig.BRIK/HEAD into the appropriate number of spm readable *.img/hdr pairs: one for an anatomical image and one for each time point for a BRIK built from a P-file.




>3dMINCtoAFNI test.mnc

Converts test.mnc to an afni BRIK/HEAD pair



reconstructed P-files

>from3d -input test+orig -prefix fred

deconstructs brik




>3dAFNItoMINC brain+orig.*

Creates a single output file, brain.mnc rather than a pair.



nifti *.nii

>3dAFNItoNIFTI brain+orig

Creates a single output file, brain.nii rather than a pair.




>to3d -epan -prefix name -time:tz 120 17 2000 seqplus 3Df:0:0:64:64:120:'test.bfloat'





>to3d -anat -prefix fred test*.bshort

Structurals:This will create fred+orig.BRIK and HEAD files from a series of test bshort files (structural slices). The to3d interface may require additional info.

P-file brik (a brik with a time dimension) start with deconstruction to make a series of bshorts, and then format them into bshorts.




>to3d test.hdr

to3d interface may require additional info


the output of from3d (for afnireg2bshort)

greconed P-files (for grecon2bshort)


>afnireg2bshort -i study1_3dreg -fgs 1 -nas 17 -nfs 80

>grecon2bshort -i P15872 -fgs 1 -nas 17 -nfs 80

-i=input, -fgs=first good slice, -nas number of anatomical slices, -nfs number of functional slices.
A bshort and header file will be produces for each anatomical slice (e.g., 17 files will be produced for these sample commands)




>MR2bshort -i E22078 -s 2 -fs 1 -ns 17 -o fred -slice3w

Post Sept, 2002 default data format (e.g., etc.):
>MR2bshort2 -i 42 -s 4 -fs 1 -ns 25 -o out -slice3w

-i <exam#> -s <series#> -fs <first slice> -ns <# of slices> -o <output prefix> -slice3w (this tells the program the numbering should be 3 characters wide). A bshort and hdr file will be produced for each MR file.

To add waveforms together. A convolution is an integral that expresses the amount of overlap of one function g as it is shifted over another function f. It therefore "blends" one function with another.
Example: convolve Vector A "1 2" with Vector B "2 3 4".
Row 1: Multiply the first element in A by each element in B.
Row 2: Shift right, multiply the second element in A by each element in B.
Row 3/Result: Add values in columns:

>>conv ([1 2] ,[2 3 4])


Row 1



1*4 +


Row 2





Row 3





copy files copy directories

The native file format used by freesurfer to store 3D structural image data. COR volumes always have 3 dimensions (no time dimension). Each dimension is 256 voxels, voxels are always 1 mm and isotropic. Voxel values are stored as unsigned bytes in coronal slices, one slice to eachfile, labelled COR-001 through COR-256. See also nifti.

In SPM, "coregistration" refers specifically to the process of aligning the functional image with a higher resolution anatomical image. See also Realignment.


Cut and Paste Check the link to see how to do it in a Unix shell window. See also cat.

Databases of brain areas and their apparent functions are becoming more common: See and the Brede database

Deactivation The inverse of activation. When a voxel deactivates, its intensities dip in response to a condition rather than rising in response to a condition. See Activation.

Deconvolution To take waveforms apart. See Convolution and 3dDeconvolve.

Deformation Field A "map" of what stretching, squishing, moving and resizing operations need to be done to each voxel so that the individual brain you are normalizing can be fitted to the template brain. This deformation field is generated as an *sn.mat file by normalization in spm and can be used for vbm. Here we see a representation of a simple deformation field (left) and then we see the field applied to an image of a cross to warp it (Image from: Ashburner, J and Friston, K.J. "Spatial Normalization using Basis Functions" Chapter 3, Human Brain Function)


Depth See Image Depth

Design matrix A model of your experiment and what you expect the neuronal response to it to be. In general represented as a matrix (funnily enough), where each row represents a time point / TR / functional image and each column represents a different experimental effect. It becomes the model in a multiple regression, following the vector equation: Y = BX + E. Y is a vector of length a (equal to nframes from the scanner), usually representing the signal from a single voxel. B is a vector of b, representing the effect sizes for each of b experimental conditions. E is an error vector the same length as Y. X is your design matrix, of size a x b. (From Gablab Wiki: Glossary)

DICOM (Digital Imaging and Communications in Medicine) The DICOM image format is commonly used for transfer and storage of medical images. Visit Chris Rorden's Dicom page for information about the format and free software to view and manipulate it. See also Image, ezDicom and MRIcro.

Diffusion Weighted Imaging (DWI) MRI sequences weighted by the diffusion of water. Diffusion Weighted Imaging (DWI) measures the molecular mobility of water in tissue. Less attenuation of signal is expected in regions of less restriction (i.e., less compartmentalization of the water). Diffusion in biological systems is complex, but directly related to tissue microstructure, which differs in normal and diseased tissues. Diffusion weighted MRI can be used to diagnose diseases like stroke and multiple sclerosis.

Diffusion Tensor Imaging (DTI) refers to diffusion imaging using Tensors (A mathematical construct for describing multidimensional vector systems). Diffusion Tensor MRI can be used for mapping white matter tracts. The acronym ADC refers to the Apparent Diffusion Coefficien (a quantitative measure of diffusion based on Diffusion Tensor Imaging), a quantification of diffusion. See also isotropic and anisotropic. See DPTools. See also the following interesting links:

A master's thesis with good descriptions and images of the basics of diffusion. Particularly helpful are the contents links on the first page: Image Acquisition and Tensors.

A good diffusion introduction in Nature Reviews Neuroscience:

Other software that may be useful:, (seems to be for Siemen's images) (A variety of potentially useful programs, including one to convert ge images to dicom).

See also , an atlas of DTI images.

Dilation This operation gradually increases the area of foreground pixels in an image. See HIPR's Dilation Page. Surefit uses dilation to help correct topological defects in a reconstruction. See also erosion.

Dispersion derivative The derivative with respect to the dispersion parameter in a gamma function. In SPM, the dispersion derivative of the canonical HRF looks a lot like the HRF but can be used as a basis function, to model some uncertainty in how wide you expect the HRF to be at each voxel.(From Gablab Wiki: Glossary)

Distal Away from the core or center. Compare to proximal.

DMDX A program written by Jonathan and Ken Forster. The program runs under MSWindows and is used to present a series of precisely timed stimuli to a viewer. Check out the DMDX Updates Page for several useful links: See also Presentation.

If you use DMDX in your project, you should probably reference the following paper:

Forster, K.L., Forster, J.C. (2003). DMDX: a windows display program with millisecond accuracy. Behav Res Methods Instrum Comput.35(1):116-24.

Dorsal Toward the back. In the case of the brain/head, used to mean toward the top of the head, same as Superior. Compare to ventral or inferior.

Downloads Download links for this site can be found in several places. Here are some primary downloads:

DPTools MS Windows software designed to aid in diffusion and perfusion analyses. See the fmri tools Home page. See also Activ2000.

DTI See Diffusion. See also

DWI See Diffusion.


Epoch See Block Design

Erosion This operation erodes or shrinks foreground pixels causing holes within those areas to become larger. See HIPR's erosion page for more information. Erosion is used by Surefit to correct topological errors. See also dilation.

Errors (at the scanner and with the presentation equipment) To submit reports of errors during scanning, please use the following subject lines for, respectively, problems (and maybe solutions) on MR1, MR2 and MR3, and likewise problems and/or their solutions for the presentation system (including goggles, cables, stereo, hercules etc):

MR1 problem
MR2 problem
MR3 problem
Presentation problem

Euler Count An automated count of topological errors based on the algorithms of Leonhard Euler.

Event-Related Design An experimental design in which the time a stimulus is presented is not fixed and is often (but not necessarily) extremely brief. Also sometimes called Single trial design because each trial is statistically independent of the other trials. To be statistically independent, different trial types must be intermixed so that it is not possible to predict the next trial type from the previous one. (Compare to Block Design, SPM99WorkbookStudy1.doc See also Afni-MGH Event Related Analysis, MGH).

Event Related Tutorial: SPM99event.doc

EzDICOM The free ezDICOM software runs on Windows computers. It is able to display most types of DICOM image (many other viewers are limited to showing uncompressed grayscale DICOM images) and can automatically detect and open Analyze, DICOM, Genesis, Interfile, Magnetom, Somatom and NEMA images. If MRIcro doesn't open the image, you should try ezdicom. For example, a single slice from one of our P-files can be opened in ezDicom as follows:File=>Open Raw: Width 64, Height 64, Slice Frames 1, Offset 0, Bits per Pixel 16, Convert Raw to Analyze, Dicom or interfile (you choose).


F-contrast Simply put, a T-contrast tests a single linear constraint on your model - something like "The effect size (parameter weight) for condition A is greater than that for condition B." T-contrasts can involve more than two parameters, but they can only ever test a single sort of proposition. So a T-contrast can test "The sum of parameters A and B is greater than that for parameters C and D," but not any sort of AND-ing or OR-ing of propositions.

An F-contrast, by contrast (ha!), is used to test whether any of several linear constraints is true. An F-contrast can be thought of as an OR statement containing several T-contrasts, such that if any of the T-contrasts that make it up are true, the F-contrast is true. So you could specify an F-contrast like "parameter A is different than B; parameter C is different than D; parameter E is different than F," and if any of those linear contrasts were significant, the F-contrast would be significant. The utility of the F-contrast is highest when you're just trying to detect areas with any sort of activation, and you don't have a clear idea as to the shape of the response. They were designed to be used with something like a Fourier basis set model, where you want to know if any combination of your cosine basis functions is significantly correlated with the brain activation. Testing that set with a T-contrast wouldn't be correct; it would tell you whether the sum of those basis functions' parameters was significant, which isn't what you'd want. Testing individually whether any of those parameters is significant, though, tells you something.

The disadvantage of the F-test is that it doesn't tell you anything about which parameters are driving the effect - that is, which of the linear constraints might be individually significant. It also doesn't tell you what the direction of the effect; parameter A might be different than parameter B, but you don't know which one is greater. This isn't a problem if you're using a basis set where different parameters don't have much individual physiological meaning (such as a Fourier set), but oftentimes F-tests are followed up with t-tests to further isolate which parameters are driving the effect and what direction the effect is in. (From Gablab Wiki: Contrasts FAQ )

FFT Fast Fourier Transform A method for deconvolving (taking apart) any complex waveform into its component sinusoidal waveforms, which aids greatly in their analysis. Used in acoustic analysis, fMRI, seismic analysis and lots of other things.

Fiasco An free fMRI analysis package for unix developed at the Statisitics dept, University of Pittsburg. It is a series of shell scripts and executables for several unix platforms. Some associated tools have been developed in Java and fiasco interacts with R (It is not clear whether it is at all dependent upon R). Fiasco uses its own proprietary "pgh" file format, described here. See

FICO Afni data type: Functional Intensity+ Correlation data (Multibucket)

Field of View see FOV

File system The way in which an operating system organizes its files and directories

File Transfer A page describing machine specific file transfer protocols for the CNL

Filter Filters remove unwanted frequency components from signals. Some filters preserve low frequencies but remove high frequencies (low pass filters). Others do the opposite (high pass filters). A band pass filter passes selected frequencies while cutting out others (e.g. You could produce a bandpass filter by doing both lowpass and highpass filtering--as long as you let some frequencies through). Several different filter shapes are available for smoothing images. In general, the the center voxel is weighted more heavily in a smoothing operation. However, the shapes of the filters vary. Blackman filters are relatively leptokurtotic (narrow), Hanning filters are intermediate and Hamming filters are the fattest. Gaussian filters are typically lowpass filters, used for smoothing in spm, Cambridge Introduction to Smoothing. The width of the Gaussian is sigma (the FWHM).

FIM Afni type: Functional intensity Map. The file is actually 3d because time data is merged.

Find A powerful unix command with difficult syntax. It can be used to find files or directories by ownership, age, name etc. and do almost anything with them (change them, delete them etc.)

FIR (or Finite Impulse Response) model A type of design matrix which assumes nothing about the shape of the hemodynamic response function. With an FIR model, you don't convolve your design matrix with a canonical HRF or any basis functions. Instead, you figure out how long an HRF you'd like to estimate - maybe 10 or 15 TRs following your stimulus. You then have a separate column in your design matrix for every time point of the HRF for every different condition. You separately estimate beta weights for every time point, and then line them up to form the timecourse of your HRF. The advantage is that you can separately estimate an unbiased HRF at every voxel for every condition - tremendous flexibility. The disadvantage is that the confidence in any one of your estimates will drop, because you use so many more degrees of freedom in estimation. Full FIR models may not be useable for very complex experiments or certain types of designs. (From Gablab Wiki: Glossary)

Fitted Response (In SPM) Simple plot of mean (averaged over session) regressor across PST.

FLAIR Fluid Attenuated Inversion Recovery

Flat mapping The process of rendering all or part of the gray matter, whether it is folded into sulci or gyri, as a flat surface accessible in a single view. See also Surface Based Analysis, Mrgray, Surefit/Caret, Freesurfer and BrainVoyager.

fMRI (Functional Magnetic Resonance Imaging) offers the cognitive neuroscience community enormous promise for understanding both normal and pathological brain functioning. The technique provides measures of regional change in brain activity while subjects are engaged in various cognitive tasks, combining activation information from the functioning brain with the exquisite anatomical detail of high-resolution structural MRI. Unlike other functional brain imaging techniques, fMRI is a completely noninvasive procedure; subjects are not exposed to chemical agents, radioactive materials, or X-rays. The benign nature of the procedure allows the implementation of longitudinal research designs since the same subject can be tested and re-tested over multiple sessions; a distinct advantage in longitudinal research aimed at following normal and abnormal development of cognitive functioning across the life span. Other functional MRI techniques, such as perfusion arterial spinal labeling MRI, diffusion-weighted MRI, and MR spectroscopy, are currently being developed that will greatly enhance our ability to measure brain functioning, especially in the presence of neuropathology. MRI techniques such as these will undoubtedly result in important contributions to clinical sciences as well, in the areas of diagnosis, assessment of risk, treatment monitoring, and drug development. For these reasons, fMRI is likely to become the dominant tool for examining brain function in humans. See also functionals, and k-space.

Font Sometimes a specialized font is used for experiments. To use a font, you should put it in the font area on windows (go to your C drive and search for "font"). On Windows 2000 it is C:\WINNT\Fonts. You can download the backward font here. It is called ARIALBK.TTF. (TTF=True Type Font).

Format Image file formats used by programs that process medical images come in a somewhat depressing variety of flavors. Essentially, each program has its own format AND there are several competing standards. To begin with, MR scanners produce images in a variety of proprietary formats (see Genesis, the GE format). The following formats used by various medical image processing programs are also of particular interest:

  • bfloat/hdr (a format used for functional data in the MGH-fsfast software)
  • BRIK/HEAD (used by Afni),
  • bshort/hdr (a format sometimes used for structural/anatomical files by the MGH-fsfast software),
  • COR (the native freesurfer/mgh format for structural/anatomical file),
  • dcm (dicom format-an attempt at standardization of a medical image format),
  • img/hdr (Analyze format, sometimes called the AVW format, used by AIR, FSL, Medx, Spamalize, SPM, and VIDA with slight modifications...they "should" all be compatible),
  • mnc (the minc format-another attept at a standard, supported by afni and used by surefit)

See Image and Conversion and nifti.

FOV Field of View, the size, in mm of the space (i.e, real distances) represented by the image (voxel/pixel size x number of voxels/pixels). For example, an image with 10 isotropic pixels, each 5 mm square, would have an FOV of 10x5 or 50 mm. Field of view can, of course, be different in each direction.

Freesurfer A free image manipulation program from the Massachucets General Hospital. For the most up-to-date and well-tested copy of freesurfer and the new mgh tools, use zoe, and see the mgh page for further information on setting up your environement and data to use the programs. Freesurfer uses the COR file format. See sample images at Marty Sereno's website, and MGH. . See also mgh, tkmedit, and tksurfer, and nifti.

from3d A file conversion program that comes with the afni package.


FSE Fast Spin Echo, a pulse sequence characterized by a series of rapidly applied 180 rephasing pulses and multiple echoes, changing the phase encoding gradient for each echo.

FSL The FMRIB Software Library (FSL) is a collection of functional and structural brain image analysis tools, written mainly by members of the Oxford Image Analysis group. These are stand alone binaries available for several platforms. Type >fsl on Merlin, Holly or Charlie, where the program is installed. fsl uses the AVW (Analyze img/hdr) format. See also nifti. Visit for download, online documentation of the tools, etc.

Note: Your .cshrc must have the following 2 items for fsl to run for you on merlin, holly or charlie:
/usr/local/bin/fsl in your .cshrc path
and the following environment variable line:
setenv FSLDIR /usr/local/bin/fsl;

FTP (File Transfer Protocol). See also specific protocols for transfer between campus and UMC or from the console to our UMC workstations. (see also transfer, special techniques for transferring images back to the console; See Trouble with permissions)

Functional files P-files (fMRI, "functional magnetic resonanace images") record the spatial and temporal coordinates of changes in the brain's blood oxygenation levels in k-space. A P-file is usually a series of thousands of images (shorts with 16 bit image depth, in our case). Each image is a single slice through the brain. Because each slice has thickness, the images are made up of voxels rather than pixels. Our voxels are typically about 3.5 mm x 3.5 mm x 5mm. Our image slices are usually in axial. Images have a radiological orientation. This relatively low resolution limits our ability to pinpoint the locations of activations. Each slice is 64x64 voxels and a set of approximately 17-19 slices will represent a brian volume (nas, number of anatomical slices). Depending on the length of the experiment, we may record from ~80 to ~250 brain volumes (nfs, number of functional slices; aka, ntr). During preprocessing, Afni treats these P-files as low resolution 4D anatomical files (although they contain functional information). As part of analysis, the files become 3d fim and fico files. See also Anatomical file, and ezDICOM. See rdgehdr (Read GE Header).

FWHM (Full Width half Maximum) A criterion for defining the width of a curve. The width of the curve is its width at half its maximum height. See also Half maximum.


Gablab Resources Check out Jeff Coopers Wiki pages at Log in as fmri, password fmri. The wiki is a collaborative effort to create a useful repository of shared information for Neuroimagers.

Garage Reservations See Reservations.

Gaussian Filter see filter

General Electric Medical Systems Home page

General Linear Model The general linear model is a statistical tool for quantifying the relationship between several independent and several dependent variables. It's a sort of extension of multiple regression, which is itself an extension of simple linear regression. The model assumes that the effects of different independent variables on a dependent variable can be modeled as linear, which sum in a standard linear-type fashion. THe standard GLM equation is Y = BX + E, where Y is signal, X is your design matrix, B is a vector of beta weights, and E is error unaccounted for by the model. Most neuroimaging software packages use the GLM as their basic model for fMRI data, and it has been a very effective tool at testing many effects. Other forms of discovering experimental effects exist, notably non-model-based methods like principal components analysis. (From Gablab Wiki: Glossary) generic2bshort One of a suite of imaging tools from UCLA. This one is used to convert a series of *.MR files into a bshort file. The tools are installed on Charlie, but not yet well tested. See UCLA Brain Imaging Center.

Genesis A file format (*.MR) produced by the General Electric MR Signa 5X. This is the native format saved by the Signa MR scanner. The data are stored as short integers (no fractional parts), with a single file for each image (a single location or time point). The first 7904 bytes of the file contain header information. (UCLA Brain Mapping Center Image Format Page)

genesis2analyze One of a suite of imaging tools from UCLA. This tool converts *.MR files to img/hdr files. The tools are installed on Charlie, but not yet well tested. See UCLA Brain Imaging Center.

Getting Started What you need to know to create or help run an fMRI experiment.

Ghostview and Ghostscript Programs for viewing and printing of postscript files. Ghostscript contains the necessary fonts etc to support Ghostview. Ghostview is the actual reader. The programs are available for unix, macintosh and PC platforms.

Global effects Any change in your fMRI signal that affects the whole brain (or whole volume) at once. Sources of these effects can be external (scanner drift, etc.) or physiological (motion, respiration, etc.). They are generally taken to be non-neuronal in nature, and so generally you'd like to remove any global effects from your signal, since it's extremely unlike to be caused by any actual neuronal firing. (From Gablab Wiki: Glossary)

Global scaling An analysis step in which the voxel values in every image are divided by the global mean intensity of that image. This effectively makes the global mean identical for every image in the analysis. In other words, it effectively removes any differences in mean global intensity between images. This is different than grand mean scaling! Global scaling (also called proportional scaling) was introduced in PET, where the signal could vary significantly image-to-image based on the total amount of cerebral blood flow, but it doesn't make very much sense to do generally in fMRI. The reason is because if your activations are large, the timecourse of your global means may correlate with your task - if you have a lot of voxels in the brain going up and down with your task, your global mean may well be going up and down with your task as well. So if you divide that variation out by scaling, you will lose those activations and possibly introduce weird negative activations! (see the Gablab Wiki PhysiologyFaq for some), considering that moment-to-moment global variations are very small in fMRI compared to PET. They can be quite large session-to-session, though, so grand mean scaling is generally a good idea (see below). (From Gablab Wiki: Glossary)

GNU Licensing Afni is now distributed under the GNU Open Source Free software license. Essentially, this license says that you can freely distribute a product and incorporate it into anything you build BUT, as soon as you incorporate it into something you build, you can no longer sell your product, but must also distribute it freely.

Goggles The new goggles from Resonance Technologies require different setup than the old goggles. See our first draft of instructions: new3tgoggles2.doc . See also the Machine page for updates on the goggles, and Resonance Technology (below) for contact information. See information on Hercules, the mobile presentation machine and Pharaoh, the 3T presentation machine.

One presentation system is permanantly installed on the 3T scanner. It includes a goggle/audio presentation system from Resonanace Technologies. The goggles are VisuaStim digital goggles with 3d stereoscopic capability, and an eye tracking system. The audio system is stereo with two way communication. The presentation computer is a Dell Dimension 8300 running Windows XP Prowith a 3.0 Ghz processor, 2 GB of main ram, a large SATA hard drive, an NVIDIA GeForce FX 5200 video card with 128 mb of ram onboard and 3d-stereo capability. The system uses custom hardware and a Measurement Computing PCI-DIO card to gather responses from two mice used by subjects in the scanner. In addition, this custom hardware setup allows researchers to automatically start the scanner sequence with a TTL pulse.


Grand mean scaling An analysis step in which the voxel values in every image are divided by the average global mean intensity of the whole session. This effectively removes any mean global differences in intensity between sessions. This is different than global scaling! This step makes a good deal of sense in fMRI, because differences between sessions can be substantial. By performing it at the first (within-subject) level, as well, it means you don't have to do it at the second (between-subject) level, since the between-subject differences are already removed as well. This step is performed by default by all the major analysis software packages.(From Gablab Wiki: Glossary)

Grant Proposal Writing See Grant Proposal Writing

Grecons This is a c program from Gary Glover at Stanford (we actually have several different versions for different sacanner output) that constructs images from our raw GE spiral fmri files (P-files). We have versions of grecons5x for the sgi and sun solaris. See spirec. See also Scanner Updates for information about the post September 2002 version of grecons AND information on the November, 2003 spiral code upgrade on the 3T. See the Grecons Table to determine which version of grecons you need to use. See prep for information of shell scripts that will make your life easier.

grecon2bshort An image conversion script that comes with the mgh package. Very similar to afnireg2bshort.

grhyp reconstructs spiral diffusion images acquired in 43 directions. It is installed on buddy, holly and merlin. The command to run it is grhyp [rawfile] [prefixforoutputfile] , e.g.,
>grhyp P12345 study
for help

Group Analysis For details of running group analysis in spm, see SPM99WorkbookStudy1.doc

Gui a graphical user interface on a computer, where you use the mouse to move around and click on things.

Gzip (also gunzip to decompress a gzipped file). Because gzip only acts on files, you may also find it useful to look at a find command that will recursively gzip files in multiple subdirectories. Software like Iceows from will work on Windows systems to ungzip and untar files that might otherwise be difficult to handle.

Handle A topological error, sometimes called a "crossover", in which a surface reconstruction is folded over onto itself. A "bridge" of gray matter across a sulcus is called an exohandle. A hole through the white matter between two sulci is an endohandle. Concern over topological errors arises in gray matter reconstruction routines such as those used in Surefit and Mrgray. See also cavity and topology.

Half Maximum (half max criteria) A notion frequently used in image processing to define an edge. The edge is said to occur when you are 1/2 way between the brightest and darkest points along a line that crosses that edge. (Keep in mind that you see an edge in a grayscale image because there is something bright next to something darker). See also FWHM

Hamming Filter see filter

Hanning Filter see filter

HDR Hemodynamic Response, the BOLD (blood oxygen level dependent) response caused by activity in the brain and measured by fMRI. (See also hrf)

Header Descriptors that accompany an image and define its format. See Image and Offset and rdgehdr (Read GE Header).

Help How to find help for unix

Hercules See also setup

High Pass filters see filter

History A listing of commands that you recently typed into a unix shell.

Horizontal (same as "transverse" and "axial")

HRF The hemodynamic response function. A mathematical function representing the hemodynamic response. May be used synonymously with hdr (they are very close).

Human Subjects To work with people, you must pass a test demonstrating that you know the rules of interacting with and experimenting on people. The main Human Subjects Protection Site for the U of A. Link to online test and other training materials.

IDL The Interactive Data Language, is software for data analysis, visualization, and cross-platform application development. Type >idl at the command prompt on Merlin where it is installed; or to start IDL with the GUI, type >idlde at the prompt on Merlin. See also and look online for tutorials (there is a lot out there). Useful Links: IDL Home, Program Library, Spamalize

IATR The goal for this site is to provide a centrally available listing of all image analysis tools that are available to the neuroscience community in order to facilitate the development, identification, and sharing of tools that are of use to the general community. It is hoped that this helps the "tool developers" to get their tools to a larger user community and to reduce redundancy (or at least utilize tool redundancy to facilitate optimal tool design) in tool development. This also helps "tool users" in identification of the existing tools for specific problems as they arise.

IID (Independent and Identically Distributed) The assumption of a statistical model that errors are indpendent and identically distributed.

Image The images we get from the scanner are all bitmap data. Such images consist of a set of numbers that specify the colors of each individual pixel (or voxel) in the image. A simple bitmap file consistes of a header and the raw bitmap data. The header may be stored internally to the image (as is the case for our raw MR files) or externally, as is the case for the Analyze-like format used by SPM (For each image there is a *.img and a *.hdr file). The header may contain information about the palette of colors used in the image (LUT), a file identifier (a code that tells the computer what kind of image it is, e.g., *.bmp, *.jpg etc), the number of lines per image, the number of pixels per line, the number of bits per pixel, compression type, x,y, and z origins etc. Image conversion usually involves changing the header information to match the expected order and position of elements in the desired image type. For details, see Anatomical Files, Functional Files, Offset, Pixel, Resolution and Voxel. Learn more about medical image file formats: Analyze, DICOM, MINC, NEMA, nifti, VTK, Afni Brik, MGH bshort, MGH bfloat. Look at the Medical Image format FAQ. Learn more about Image Processing and Image Properties. Learn about conversion between formats. Look at information about the mgh program mri_convert.

Image Capture See Screenshots

Imaging_analysis listserv The listservs will be replaced by the webboard by the end of January 2005.

ImageJ A cross platform Java based version of NIH Image. Type >ImageJ on Merlin to access it.

Image Depth The number of colors or levels of grey scale that can be assigned to a single pixel or voxel. A binary (one bit) image can have two colors (usually black and white). You will also commonly encounter 8 bit images (2^8 or 256 colors or levels of grey) and 16 bit images (i.e., 2^16 or 65,536 colors or levels of gray). A 16 bit image takes twice as much room as an 8 bit image to store in a computer. A binary image takes very little room to store. See also Anatomical Files

imcalc A function in spm which allows you to perform basic algebraic operations on images. It is analogous to 3dcalc in afni. For example, imcalc will allow you to perform Matlab dot multiplication on two images with a command like this to imcalc:

i1 .* i2

(where i1 is the first image you select in the spm interface and i2 is the second image).

imconvert one of a suite of tools available from the UCLA Brain Imaging Center.

Inferior Toward the underside of the brain or head. Same as ventral (which means toward the stomach as opposed to the back)

inorm An MGH program that calculates the mean intensity data for the entire functional volume. The global mean of the fMRI signal inside the tissue is calculated by segmenting tissue from air. The number is later used to rescale the data so that when intersubject averaging is done, all subjects have the same global mean.

Interpolation The process of computing new intermediate data values between existing data values. (From Gitta Domik's Tutorial on Visualization). Interpolation may be used to smooth out jagged corners and edges on activation voxels, for details on doing this in Afni, click here.

Isotropic Exhibiting equal physical properties or actions (OED).

A characteristic of a pixel or voxel that is square or cubic respectively. Compare to non-isotropic. See the Caret (for Pfile/Functional files) and Surefit (for 2D and 3D anatomical files) pages for instructions on converting non-isotropic voxels to isotropic voxels.

In diffusion weighted imaging, isotropic movement of water molecules is movement that is unimpeded, or equal in all directions. Compare to anisotropic.

isxavg an mgh program

itk-snap A free image segmentation and roi creation tool.


Jitter "The practice of varying the timing of your TR relative to your stimulus presentation. It's also often connected to, or even identified as, the practice of varying your inter-trial interval. The idea in both of these practices is the same. If your TR is 2 seconds, and your stimulus is always presented exactly at the beginning of a TR and always 10 seconds long, then you'll sample the same point in your subject's BOLD response many times - but you might miss points in between those sampling points" (Gablab Wiki). By varying the relationship between the experimental condition and TR we can increase the success of deconvolving the FFT (signals) for each experimental condition from the other conditions. This becomes especially important when trials are very short (e.g., a few seconds). Optseq can be used to calculate how to add appropriate jitter to an experiment.

K-space Raw or Time Domain data used for raw functional files. A data map based on signal amplitude versus position rather than direct position within the subject. kspace.ppt (If you have powerpoint, you can watch a virtual reconstruction of k-space. Simply click the link, then each mouse click on the image will update the reconstruction). See the animated tutorials on kspace here MRI Q&A for Physicists: Kspace

Khoros and Cantata A high level programming interface in which one can attach widgets to one another in paths, modify their characteristics, and run signals (like images) through the entire path to filter or modify them.


Lab Manual Online (CNL Lab Manual)

Left Handed Coordinate System Left is right and right is left in viewed images. See radiological orientation. Left Lesion Test Image The Left Lesion Test Data (This data has a big hole in the left front, so you can test your understanding of what is happening to right and left given a particular program or manipulation. There is a single functional image, and "2D" and "3D" structural images in spm format).

Less A viewer, much easier to use and more flexible than more. Less is the default viewer for man pages on buddy, holly, merlin and charlie.

Linear Transformation See Affine Transformation

Link A link is like a shortcut under windows. You can make what is called a soft link by doing something like the following:

ln -s /data/harvey/run1/joe fred

This will create a link called fred in your current directory (the one you are in when you type the command). The link will point to /data/harvey/run1/joe. If you cd to fred, you will end up in /data/harvey/run1/joe. If you rm fred at some later time, fred will disappear, but /data/harvey/run1/joe will remain.


Listserv The listserv will be replaced by the webboard by the end of January 2005. See webboard in this glossary for more information.

Little Endian Describes a computer architecture in which, the least significant byte of a multibyte numeric representation is stored in the lowest-memory address, which is the address of the data. PCs use this format, whether thay are running Linux or Windows. See also Big Endian and Byte swapping.

Low pass filter see filter

LUT Lookup table. Pairs of numerical values that allow a program to match a meaningful value to one which specifies a color on the output device.



Magnet See Scanner

Marching Cubes A method of visualizing 3-D data structures by looking for level surfaces in a 3D-space comprised of a lattice of points. In contrast to volume rendering, where one can see the entire structure, marching cubes only allows a single surface to be rendered (From Gitta Domik's Tutorial on Visualization). Marching Cube Algorithms are used in Mrgray and Surefit to construct the gray matter surface.

Marsbar An SPM ROI Toolbox developed by Matthew Brett. See also SPM and MarsbarNotes.doc If you use Marsbar in a paper, use the following reference:Matthew Brett, Jean-Luc Anton, Romain Valabregue, Jean-Baptiste Poline. Region of interest analysis using an SPM toolbox [abstract] Presented at the 8th International Conferance on Functional Mapping of the Human Brain, June 2-6, 2002, Sendai, Japam. Available on CD-ROM in NeuroImage, Vol 16, No 2.

Mask In general, a mask is a file that filters out values from another file. The simplest way to do this is to draw shapes on an image and use the shapes to define what values are seen in some other image. Typically, the mask image has 1's inside the shapes and 0's outside. When you multiply an image by this 1,0 mask, only the values multiplied by 1 remain. The others all turn to 0's. Afni and SPM have associated mask drawing tools. Massachusetts General Hospital See also MGH and Freesurfer.

Matlab (Matrix Laboratory). A mathematical programming language and environment, optimized for matrix operations. Matrix operations are crucial to all kinds of signal processing for both sound and images. Matlab programs are *.m files, which are plain text scripts. Matlab matrix data is stored in *.mat files. type >matlab to start the gui interface, or >matlab -nojvm to start the command line version of matlab on Merlin, buddy or Holly. See also Matrix, Cell Array, and Vector.

Matrix MEDx A unix based commercial medical image processing application from Sensor Systems

mc-sess An mgh script that runs the afni motion correction.


MGH Massachusetts General Hospital produces a suite of image processing programs, primarily for event-related analysis. The new suite (see #2 below) includes afni, fsl, freesurfer and the fs-fast tools. See also freesurfer, tkmedit, and tksurfer. IMPORTANT : We use two distinct versions of the MGH utilities at the CNL:
1) The older version which does NOT include Afni, FSL, Minc tools or Freesurfer. See pages on afni preprocessing and afnievent analysis for use of the old mgh program.
2) The new (but not necessarily better) MGH program only runs on linux. It DOES include afni, freesurfer, minc tools and fsl. It is not something we have been able to use very successfully yet, though we have tried and have some documentation on installing and using it on the MGH page.

Citations for those using the mgh event related processing stream:

Glover, G. H. & Lee, A. T. (1995). Motion artifacts in fMRI: comparison of
2DFT with PR and spiral scan methods. Magnetic Resonance in Medicine, 33,

Dale, A. M. (1999). Optimal experimental design for event-related fMRI.
Human Brain Mapping, 8, 109-114.

Dale, Greve & Burdock ???

MINC (Medical Image NetCDF) is built on the general data format NetCDF (from the UCAR - University Corporation for Atmospheric Research). The MINC format was developed by the Montreal Neurological Institute. See also Surefit and Afni.

mkcontrast is an mgh program

MNI A template brain based on the average of a large set of brains (see also talairach). The template was developed by the Montreal Neurological Institute. SPM uses the MNI brain. See also BIC. For some interesting spm99 add on tools for viewing MNI and Talairach information, visit

mni2tal A freely available matlab script that converts mni coordinates to talairach coordinates:
Sample Input:
>>mni2tal([40 -16 -30; -45 -73 -12])

Output will appear on the screen and in a file called tal_out.txt

mni2talb was created to handle MGH Label files as input, hence it expects 5 values in each row rather than 3. It then strips off the first and fifth columns.

Both are available here. A discussion of the differences between talairach space and mni space, and links to matlab files to translate from one to the other.

Montreal Neurological Institute See also MINC and MNI.

Morphometry "Shape measurement", a term used to describe measurement of brain structures (like the hippocampus, amygdala etc.), generally for group comparisons.

Move move files move directories

Movie It is possible to make a simple movie (rotating brains etc.). For Afni, see Volume Rendering.

MR files (see Anatomical files)

MR1, MR2 and MR3 These are the 3 MRI scanners. Information about them can be found in two places: The Scanner Updates page (descriptions, safety, scheduling, analysis implications of using each machine, links to the MRITech Manual) and the Machine page (current status). See also transfer (special techniques for transferring images back to the console). To understand how number of slices interacts with number of repetitions and TR, see Spiralio Sequence Limitations.

MR2bshort (and MR2bshort2) CNL image conversion utilities that are based on a script that belongs to the mgh package.

mri_convert An MGH binary that does considerable image conversion. Can be found on Charlie...largely untested.

usage: mri_convert [options] <in volume> <out volume>


MrGray MrGray and MrFlatmesh work together to produce flattened represenatations of gray matter. MrGray is a standalone executable that runs on MSWindows. MrGray comes with a program to convertAnalyze/SPM headers to Mrgray's *.dat image files. MrFlatMesh is a set of Matlab routines that will run on any platform running Matlab.

MRI Magnetic Resonance Imaging See for a glossary of related terms. See the Links page for educational references. See also

MRICro A free Windows Image Viewing and manipulation program: See also Dicom and NEMA and MRIcron.

Multibucket image An image that has more than one value per pixel/voxel. e.g., In afni, an intensity fMRI image (fim) that has been correlated to behavioral changes (fico). Both the intensity and the correlation values are stored in each voxel. See See advice on splitting a multibucket image.

nas Number of anatomical slices in each brain volume of a P-file/Functional image.

NEMA (National Electrical Manufacturers Association) An organization associated with medical image standards,the ACR/NEMA 1.0 and 2.0 Standard is their earlier standard. The ACR/NEMA Dicom 3.0 is the more recent standard and is more accepted by the community. See also DICOM, MRIcro and Image.

Neurological Orientation Left is Left and right is right in a displayed image. This is also called the Right handed Coordinate system. (Compare to Radiological Orientation).

nfs Number of functional slices (number of brain volumes) in a P-file/Functional image. See "number of repetitions".

NIfTI Neuroimaging Informatics Technology Initiative An up and coming image format (as of 5/04/2004) being developed by a consortium including those who develop afni, spm and fsl, freesurfer (and others). The idea is to support a shared image format that will allow researchers to move data more easily between image processing programs and to have an optimal format for backup.

NIH Image A generalized free Image Processing program that was designed to run on the Macintosh. It has spawned offshoots: Scion Image, which runs under MS Windows; and Image J, a true Java based application that runs on any platform. Published research assisted by NIH Image should use a statement similar to the following in the materials and methods section "... analysis performed on a Macintosh computer using the public domain NIH Image program (developed at the U.S. National Institutes of Health and available on the Internet at".

Non-isotropic A characteristic of a pixel or voxel that has x,y,(z) dimensions that are not pqerfectly equivalent. Compare to isotropic.

Nonlinear Transformation See Affine Transformation

Non-sphericity See sphericity

Normalization In the MGH program (inorm), that we use with Afni, "normalization" refers to signal intensity normalization. That is, artifactual variation in the strength of the signal through time (and space?) is effectively removed or compensated for by an averaging process. When the term "normalisation" is used in relation to SPM, however, it refers to the process of warping the shape of the brain to a standardized template, essentially "talairaching" (see also Talairach and MNI)

ntr See Number of Repetitions

Number of Repetitions (a.k.a ntr, nfs "number of functional slices"), number of time frames [on the console itself]) The number of times the brain gets scanned in an fMRI experiment (the number of tr's). Typically, at the CNL, the brain is scanned every second or two for 2 to 15 minutes (depending on the length of the experiment). We thus end up with the number of scans varying from about 80 to about 450. If you try to scan the brain more than once per second, you have to sacrifice spatial resolution. If you scan for a long time (more than 15 minutes), the resultant image files (P-files) are huge. P-files typically run between about 20 and 50 mb. Experimenters often collect several separate P-files from a subject in the magnet, for example, one file before some manipulation and another file afterwards. See functional file. To understand how number of slices interacts with number of repetitions and TR, see Spiralio Sequence Limitations.


OCF The scanner schedule. See Scanning schedule for more information.

od (octal dump) a unix command that allows you to see every character in a file -including annoying hidden ones like nl (new line) and sp (space). This is valuable because the hidden character can sometimes destroy the functionality of otherwise good shell scripts. See detailed example here.
If you want to read the image as a "raw" image in ScionImage, ImageJ etc., then you need to know the offset. The offset (or "skip") in bytes is the number of bytes to skip in the file because they are header bytes (i.e., they contain information about the kind of file in the image, rather than the actual image data). To calculate the offset, take the size of the file and subtract the size of the image. The remainder must be header information. Our files are 138976 bytes. Image size for our files 256x256x2 or 131072 (xdim x ydim x imagedepth). 138976-131072=7904. Offset is 7904 bytes. See also Anatomical files.

Operating System The software that controls the computer hardware (Linux, Irix, Windows 2000 etc.)

Optimal Presentation Order see optseq

Optseq optseq is a program included in the old mgh tools (written by Doug Greve), and thus is installed on buddy, holly and merlin (simply type optseq to access it). It can be used to generate an optimal sequence (hence the name) for conditions in an event related study. See further explanation here. SPM99 appears to have similar capability as part of Model Specification (described in section 3.2.1 of the May 2001 SPM Manual). For detailed advice on using optseq, consult Siobhan Hoscheidt's excellent optseq.doc. optseq2 is installed on zoe and charlie. To access optseq2, type:


optseq2 uses double dashes (e.g., --ntp) before option flags; whereas optseq uses only a single dash (e.g., -ntp). optseq2 also uses lower case tr as a flag, whereas optseq uses uppercase TR). [Thanks to Chun yu Lin for noting these differences]. Doug Greve's optseq webpage. See also jitter.

Orientation (For Subjects)

outlier One of a suite of imaging tools from UCLA. This tool provides a way to evaluate outliers. The tools are installed on Charlie, but not yet well tested. See UCLA Brain Imaging Center.


Overlay An overlay is an image (for fMRI, usually of activations) that is viewed on top of a high resolution anatomical image (usually called the underlay).
P-file A file containing a series of raw fMRI images. See functional file and ezDICOM. See rdgehdr (Read GE Header).

Parametric vs Non-parametric statistics In general, parametric procedures have more stringent assumptions about the distribution and character of the variables than nonparametric statistics. Non-Parametric tests are often used in place of their parametric counterparts when certain assumptions about the underlying population are questionable. For example, when comparing two independent samples, the Wilcoxon Mann-Whitney test does not assume that the difference between the samples is normally distributed whereas its parametric counterpart, the two sample t-test does. Non-Parametric tests may be, and often are, more powerful in detecting population differences when certain assumptions are not satisfied. All tests involving ranked data, i.e. data that can be put in order, are non-parametric.

Parfile (parameter file--used by mgh tools) A text file necessary for event-related analysis. The file provides information about the stimulus schedule for each run, that is, details about the times at which each stimulus was presented for a particular subject (in an event-related paradigm, the timing is dependent on subject responses). Our parfiles are generally based on an output file generated by dmdx and then manipulated in a spreadsheet to have the correct format for the mgh programs. The parfiles are created by the FS-FAST command optseq.

Paste Unix command for concatenating text files vertically. See cut and paste. See also cat.


Perfusion weighted MRI An imaging technique used for measuring blood flow through the brain, specifically, the volume of blood supplying a unit mass of tissue per unit time. Perfusion images can be obtained using a bolus of contrast agent injected into the patient OR arterial blood can be tagged by magnetic inversion. Images of the tagged blood flowing into the brain are alternated with untagged images, and tagged and non-tagged images are subtracted to remove the signal from brain tissue, leaving the blood signal. See also Introduction to Perfusion

Peristimulus time (PST) In SPM, the peristimulus time for a given event is simply defined as the time that has passed since the last presentation of that particular event type.


Phonelist The CNL phonelist requires a username and password to access. It is in Excel format.

phx_tospm A Matlab program (actually, several different versions of the program) that take greconed, renamed files from our GE scanners and make proper spm img and hdr files. Unfortunately, the program is prone to being a little cranky if you don't get everything just right. Get versions of phx_tospm here.

Pilot Projects Link to the standard CNL form you need to fill out to apply for pilot project funding. See the Getting Started page too.


Pixel A "picture element", the smallest nondivisible component of a 2D image. It will have an x and y dimension and a color. (See also bitmap, voxel and image depth).

Pharaoh 3T presentation machine.

Planes (of section) The angle of slices. There are 3 standard angles and as many oblique angles as you can imagine.

PLS (Partial Least Squares Analysis) The PLS analysis software is free, runs under Matlab and requires at least the Matlab statistics toolbox (and possibly image processing and signal processing). It is used to identify task dependent changes in activity, changes in relations between brain and behavior and functional connectivity between brain regions. PLS is available for download (after a brief registration process) from the Rotman Research Institute PLS can be used for analysis of PET, ERP or fMRI data. It presumes that the data have been preprocessed in spm99 or spm2. It assumes the MNI coordinate system and prefers whole brain scans.

Posterior Toward the back of the head. Compare to Anterior.

Postscript A page description language developed by Adobe to allow entire pages to be sent describing graphics and text to a printer. The flexibility allows more than just normal ASCII characters to be sent and is the defacto standard in high end printing. Postscript documents may have the extension *.pdf (Adobe Acrobat files) or *.ps. You can read Postscript files using xpsview on the sgis or Ghostview (if it is installed) on most Unix or PC machines.

Prep Scripts
I have created several unix shell scripts (in /usr/local/bin) to facilitate data preparation and afni preprocessing. All are available here in the tools area on Merlin. They are also available on buddy, holly, merlin, charlie, pdw_cortex and tommy. Simply type the script name (assuming it is in the path) and it will tell you how to use it. Alternatively, you can look at the script in a text editor. "How-to" info is at the top and various comments appear throughout to help you understand what the script is doing. Because these scripts call other scripts and programs, those programs must also be available. To learn more about versions of grecons, see the grecons table.

  • There are 8 different scripts (prep, prep_v2, prepb, prepb_v2, prepio, prepio_v2, prepbio, prepbio_v2). prep is the most basic script.
  • Versions with "b" appended are for big data sets (more than 9999 images; i.e., ntr x nas > 9999).
  • Versions with "io" appended are more complex and can only be run on linux (because we don't have program binaries for the sgis). They are for the spiral in-out sequence.
  • Versions with "_v2" appended run version 2 of grecons (grecons_v2) , and where appropriate (on linux, for spiral in-out sequences, version 2 of expandonefile (expandonefile_v2)).

prep runs (and thus is dependent upon the presence of) grecons and spiral_rename. It also does some cleaning up (removes B0 files and puts your original P-file in a subdirectory called Orig).

>prep P12345.7

prep_v2 is the same as prep, but runs grecons_v2

prepb is the same as prep, except it is for big data sets (i.e., over 9999 images, nfs*nas > 9999). prepb calls grecons and spiral_renameb.

>prepb P12345.7

prepb_v2 is the same as prep, but runs version 2 of grecons_v2.

prepio depends on grecons, sprlioadd, expandonefile, and spiral_rename. It assumes 64*64 data, but you could alter the script if you needed something else (change the 64 in the expandonefile command to whatever is appropriate). Because sprlioadd and expandonefile only run on linux and sun, prepio is useless on the sgis. Use this only if you run the spiralio (spiral-in-out) sequence rather than the regular spiral sequence. It assumes byte order for the sgi's

Usage: prepio <Pfile> <nfs> <nas> <output name>
Example: prepio P12345.7 80 17 tap

prepio_v2 is the same as prepio, but runs grecons_v2 and expandonefile_v2

prepbio is like prepio except it runs spiral_renameb.

Usage: prepbio <Pfile> <nfs> <nas> <output name>
Example: prepbio P12345.7 80 17 tap

prepbio_v2 is the same as prepbio, but runs grecons_v2 and expandonefile_v2

super runs prepio_v2 followed by to3d to make a BRIK and HEAD and then 3dAFNItoANALYZE to create correctly oriented img and hdr files assumes images in axial orientation. Available on linux

superb Same as super, but for 10,000 or more files. Available on linux.

super12 runs grecons12 code (post Sept 15, 2005) but is otherwise the same as super. Available on linux. Fine for both the quad coil and 8 channel coil.

super12_8ch runs grecons12 code (post Sept 15, 2005) like super12. Uses -G 0.5 to set the gain for the 8 channel coil to avoid wrap around (causing some washed out images) Available on linux. Unless you encounter gain related problems, you don't need this....just use super12.

superb12 runs grecons12 code but is otherwise the same as superb . Fine for both the quad coil and 8 channel coil.

superb12_8ch runs grecons12 code like superb12. Uses -G 0.5 to set the gain for the 8 channel coil to avoid wrap around (causing some washed out images) Available on linux. Unless you encounter gain related problems, you don't need this....just use super12.

Image processing techniques applied to raw images to remove "noise" (See registration, realignment, normalization, smoothing, Afni Preprocessing). Note that protocols for the earliest parts of preprocessing (reconstruction) can vary with scanner type or scanner upgrades. See Scanner Updates for information about preprocessing data gathered after September, 2002.

Presentation A (currently free) DMDX-like program for presenting stimuli to subjects. See the NeuroBehavioral Systems homepage for information and download. See also FMRIB Technical Development.

Presentation Equipment See information on Hercules, the mobile presentation machine; and Pharaoh, the 3T presentation machine. See also goggles.

Printing (CCIT) You can print *.pdf, *.ps (Postscript) and *.txt files to CCIT and then go pick them up later (usually within 30 minutes). This is pretty fast and efficient, especially if you want to print a large manual.
You must transfer the files to your account using scp. Check the size of the remote file against your local one to make sure that the whole file transfered. If there is a lot of stuff on your account, sometimes only part of the file will transfer, your account will fill up and the transfer will be truncated. Such a file will never print, although it appears to be there. Follow the online instructions. I like to use the 3-hole punch double sided printing, so, to print a file called, I would type in the following command:

>a2ps -P3hole

You can call Operations at 621-2971 to see if they have completed your print job before you walk over there.

Private Library

Proton Density Image (PD) an image that is halfway between a T1 weighted and the T2 weighted image, also called a "spin density" image. It has a long TR and short TE. See also TR, TE, T1, T2, T2*. If you want more detail or background, visit the MRI section of the links page.

Proximal Toward the center or core. Compare to distal.

PSD A GE pulse sequence files. (Pulse Sequence Diagram). They're the executable files we run on the scanner hardware; e.g., sprl836 and sprl836.psd.o are PSD files for a spiral

PSD Download Error See Troubleshooting .

PST See Peristimulus time.

Publications CNL Publications


R Language A flexible data analysis language. It is free, similar to matlab (but more Lisp/Scheme-like),runs on most platforms and includes many extensions. R homepage, Cran Archive, Online Help.


Radiological Orientation Left is Right in a displayed image. This is also known as the Left handed coordinate system (Compare to Neurological orientation)

Raster Image Same as Bitmap Image.

rdgehdr (Read GE Header) Stanford has binaries posted for linux, sun and sgi that support reading header files from GE MRI structural (i.e., anatomical) files and from Pfiles.

Realignment In SPM this term refers to motion correction. All of the images in functional set are reoriented to a single image in the set. In Afni, this process is called "reregistration" or "registration". (See also coregistration)

Reference slice A term used in slice timing correction to denote the slice of the brain that no correction is done on. All other slices of each functional image will have their voxels' timecourses slightly shifted in the temporal domain so that they take on the values they "would have had" if the whole brain had been sampled at the same moment as the reference slice. See SliceTimingFaq for more, and for how to choose a reference slice. (From Gablab Wiki: Glossary)

Refresh Rate The number of times per second that the pixels on a screen are redrawn. It has to be faster than our flicker-fusion rate for us to see a steady image on the monitor. Usually 60 Hz (i.e., 60 times per second) is enough for human beings. Your computer monitor can be set to different rates. The flicker-fusion rate is different for different species...bats, for example, need MANY more cycles per second than do humans to reach the impression of fusion. This means that bats can see...and fly between...the blades of a moving ceiling fan...even when that ceiling fan is going fast enough to be just a blur for us humans.

Setting and Testing Refresh Rates: TimeDX (part of the DMDX disstribution)->Basic Tests->RefreshRate can be used to test the refresh rate. TimeDX produces the cycle length. 1000/Cycle length=Hz (Cycles per Second).

Cycle (ms)

Rate (Hz)









Some video card drivers allow you to set refresh rate (Control Panel->Display), but if these drivers seem difficult (or simply don't seem to work), freeware is available that will also do the job (e.g., Multires).

Registration or Reregistration See Realignment.

Remove remove files remove directories

rename_anat a unix shell script to rename (renumber) the E*.MR series so that the operating system correctly understands the ordering of the images. See Scanner Updates for information about new image reconstruction protocols (and a new rename_anat) for images gathered post Sept 2002.

rename_spiral a unix shell script to rename (renumber) the "functional" P-files series so that the operating system correctly understands the ordering of the images. A variant, rename_spiral2 can handle 10,000 or more files (available in the same area). spiral_rename and spiral_rename2 are the same program (but designed to be included in batch files). See prep and spiral_rename. Unfortunately, rename_spiral2 (and spiral_rename2) do not appear to work on linux machines.

Render The process of converting the polygonal or data specification of an image to the image itself, including color and opacity information (From Gitta Domik's Tutorial on Visualization). See also Volume Rendering.

Repetitions See Number of Repetitions

Reservations (Garage) reservations for visitors in parking garages may be transacted on-line:
 Your logon to the system will be the same as your Net ID name and password. This new system will enable you to make reservations for visitors on line without having to call Parking and Transportation Services. The Reservation System will be available each weekday from 6:00am - 8:00pm. Once you log onto the system, it will step you through the transaction process.  You will receive an email confirmation and transaction # for each reservation that you make.

Reservations requiring 25 or more spaces, or separate reservations totaling more than 25 spaces for a specific garage each day with the same FRS #, will have to be transacted through Visitor Parking, 621-3710.   We can then review your requirements and assure space availability for your chosen area on that day. Reservations will be charged automatically to the FRS account number used for
the transaction.  You will no longer need to submit an IBF. If you have questions pertaining to this change, please call Visitor Parking at
621-3710. Resolution A measure of an optical system's ability to distinguish objects. A system with higher resolution can distinguish smaller objects that are closer together than a system with lower resolution. Resolution corresponds to the amount of real space represented by each individual pixel or voxel in an image. An image with few and/or large voxels or pixels has lower resolution. An image with more voxels/pixels that each represent a smaller area of the image has higher resolution. See also Functional files, Anatomical files and SNR.

Resonance Technologies The company that produces the goggles we use in the MRI scanner. See also goggles and Hercules.

Contact info:
Resonance Technologies, Inc
Mokhtar Ziarati,
18121 Parthenia St.

Suite A
NorthRidge, CA, 91325

RF (Radio Frequency): RF energy may leak into the MR room if the room is not properly sheilded or may leak from improperly shielded cables or equipment (like goggles) in the room. This causes a line of noise (a light colored line) across the scan. See also the page on artifacts.

RGB This is an acronym for the Red Green Blue. It is also an SGI image file format, and should be readable with imgview on the SGIs, or with programs like gphoto or ImageMagick on Linux machines. A technical speciafication of the format can be found here. Tkmedit, an MGH program, exports images to this format (for printing, editing etc.)

Right Handed Coordinate System Left is left and right is right. See Neurological Orientation

ROI Region of Interest, usually two dimensional. Compare to VOI. See Russian SPM VOI Tool. See Afni ROI Tutorial. See Marsbar.




Safety (at the MR scanner)


Scanners We originally had a GE 1.5 Tesla Signa 5x Wholebody Echospeed Horizon System. See also Genesis. See the Scanner page for information about scanner hardware, facilities, the implications of updated software, troubleshooting, safety, scheduling, links to instructions for running the scanner, etc. See also PSD Download Error and System Disk Full Error.

To submit reports of errors during scanning, please use the following subject lines for, respectively, problems (and maybe solutions) on MR1, MR2 and MR3, and likewise problems and/or their solutions for the presentation system (including goggles, cables, stereo, hercules etc):

MR1 problem
MR2 problem
MR3 problem
Presentation problem

Scanning Schedule: The new schedule allows scheduling on each of the three magnets independently. Each individual (not just each lab) will want to sign up because the scheduling program provides you with a personal calendar. You'll need a valid frs number (grant number) to do so, but even without one, you can look around a little.

SIGNING UP: If you follow the "Scanning Schedule" link above, you'll see a gray Scheduler Login box in the middle of a white screen. Above the username and password is a link to New User Sign up. If you have a valid frs number, (and you intend to sign up for magnets), you should go ahead and click there and fill out the form. For affiliation everyone will be "in-state University". If the correct department and PI are not available on the schedule, choose "other" for now, and let Ted Trouard know what needs to be added. You can alter your profile later to change these details.

LOOKING AROUND: Even without logging in, you can check out the resource schedule tab (first one on left). You'll have to scroll down to find the Magnetic Resonance Facilities, and under that, the Clinical Research MRI Systems. Click any one of the three magnets to get to a relatively familiar calendar interface.

If you log in, then you can use the personal schedule (click on the personal calendar). You can add all kinds of events here, like a palm pilot desktop or PIM (Personal Information Management) tool. Under your User profile (see links at the bottom of any page), you can choose the box to make your personal schedule available to everyone (or not). Preferences (again at the bottom of any page) lets you set color, language and other fun display options for your data.
IMPORTANT: After you sign up, you will not actually be able to schedule yourself until you have administrative approval. You can contact the administrator with your username to get the approval (
Jan. 11th, 2007: Currently Scott Squire). Check the Scanner page for general information about magnet availability. For hardware issues, including issues with the goggles and presentation machines, check the machine page.

Scheduling See scheduling information for general information about the scanner availablility and also see the OCF Scanning Schedule

Scion Image A Windows version of NIH Image. You should register to download the software, but it only takes a moment (see also NIH Image and ImageJ)

SCP (Secure Copy) See Trouble with permissions

Screenshots Capturing them on the SGI

Scripts See Shell Script

sdcopen A file format used on the scanner console. To transfer files back to the console (for example, if a radiologist needs to look at them) see transfer for more details.

Segmentation partitioining the brain into GM (gray matter), WM (white matter) and CSF (Cerebrospinal fluid).

selxavg An MGH program that computes the average signal intensity maps for each condition.

Server An overused term, generally applies to a computer that provides services, or to the services provided.

Setup See Hercules and Forms.

Shell A shell is a window that you use to type in commands (the term is used primarily under Unix). Shells have a variety of characteristics that you may find useful, including history information and filename completion.

Shell Script A script is a simple program. We use shell scripts on the unix machines to perform a variety of tasks. These are essentially a series of commands that we could type into the shell, but all put together into a single file which is made executable and then called (by typing its path and name at the prompt). Shell scripts differ depending on the shell that is called at the top of the file. Each shell offers a simplified programming language to the script writer. Most scripts you'll see on our machines are sh scripts (bourne shell scripts) or csh scripts (c-shell scripts). Several examples of local scripts are available here. See also information about prep and afni scripts here.

Skip The distance between slices. See slice.

Size Determining the size of a file or directory can be a daunting task. This link can help.

Skull Stripping Virtual removal of the skull, meninges and nonbrain material from the MRI brain image. This is useful for volume rendering.

Single Trial Design See Event Related Design.

Slice A slice is a virtual slice through a 3D object. Typically, slices are parallel to one another and if you move from one slice to the next you are moving in the z direction. Slices may be contiguous, overlapping or actually skip some intervening tissue. Typical slice thickness varies from just under 1 mm to about 5 mm. To understand how number of slices interacts with number of repetitions and TR, see Spiralio Sequence Limitations.

Smoothing In SPM, Gaussian smoothing (aka filtering; a process that averages data from neighboring voxels) prepares data for warping into a standardized spatial template (SPM's normalisation) by de-emphasizing individual differences. Smoothing can increase signal to noise ratio. Smoothing is not generally done in Afni, and the appropriateness of the technique is a matter of contention among researchers. Cambridge Introduction to Smoothing.

SNR Signal to Noise Ratio. Comparison of the strength of information to the strength of noise in an image. The higher the SNR, the more signal you have, which is good. However, if you increase resolution, SNR goes down (i.e., There is no free lunch). See also Resolution, and CNR.

SONA Systems The experiment signup system for the Psychology Dept at the University of Arizona.

Spamalize A PET, Spect and MRI viewing and analysis package by Terry Oakes. It uses IDL (which you should have installed) and runs on Unix, Linux, Windows 98 and Macintosh platforms. It uses the Analyze img/hdr format. You can access it on Merlin by typing >spam

spiral_rename (and spiral_rename2). spiral_rename is a unix shell script for renaming up to 9999 images. It differs from rename_spiral in that it takes an argument rather than requesting user input spiral_rename2 is for renaming more than 9999 images. Download spiral_rename and spiral_rename2. See also rename_spiral.

spiralio A sequence for gathering functional images. The sequence can produce better quality images than the regular spiral sequence, though it requires some special processing. See Spiralio Sequence Limitations.

Spirec This is a c program that constructs raw images from our raw GE perfusion P-files. It is available locally as source code and also as a binary compiled for the sgi. See also grecon.

SPGR SPoiled Gradient Recalled A type of image we use for our 3D anatomicals (a.k.a. 3D structurals).

Sphericity The assumption of identically and independently distributed errors. SPM2 allows for departures from sphericity while spm99 does not allow for such departures. Departures from sphericity can take the form of:
i) non-identical distributions (e.g. heterogeneity of variance among conditions or groups, known as heteroscedasticy), or
ii) departure from independence (correlations amongst the errors that may be induced by observing different things in the same subject).
The two most pertinent sources of non-sphericity in neuroimaging are heteroscedasticy (e.g. when entering the coefficients of different basis functions into a group analysis), and serial correlations in fMRI. The second reason why non-sphericity estimation is important is that it enables multivariate analyses to be implemented within a univariate framework. (from in the spm2 distribution)


SPM Statistical Parametric Mapping, a free Unix-Matlab or MSWindows-Matlab based Medical Image Processing program out of the United Kingdom. Its purpose and functionality is similar to Afni. It uses a file format compatible with Analyze. . See also Marsbar, nifti, The Russian SPM VOI Tool, MNI. See our nice tutorials: CNL SPM Tutorial, Univ. of Arizona

SSH (Secure Shell)

Statistics A complicated area of mathematics with way too many confusing terms and acronyms. See this very useful site for some translations:

Structural files (see Anatomical files)

stxgrinder An MGH program that computes the actual contrast. Statistical, significance and parametric maps can also be computed.

Subbrick In Afni a functional 4D image is made up of 3D "subbricks", one for each functional slice (the number of subbricks = number of repetitions)

Subject Information

Subject Orientation

super and superb See the discussion of prep scripts.

Superior Toward the top of the head. Same as dorsal. Compare to inferior or ventral.

SureFit This program is a companion to Caret. SureFit should be used to produce the anatomical data file format that Caret requires. Because Surefit and Freesurfer both use different versions of the TCL_Library, they are in conflict with one another and aliases have been set up on Merlin to circumvent this problem. Type >surefit to start the program on Merlin. Tutorial data sets and a manual are available here. Surefit uses the MINC file format.

Surface Based Analyses Analyses of the cortical surface, represented as fiducial volume-rendered 3D objects, inflated lissencephalic representations, spheres, ellipsoids or flat maps. Such analyses are emerging as alternatives to traditional analyses based on stereotaxic coordinate systems (Talairach, MNI). Several applications provide relevant tools: Mrgray, Surefit/Caret, Freesurfer and BrainVoyager.

System disk is too full, applications will not be automatically started When you start up the scanner, and log in as signa, you may get confronted with the above message. The only button is "OK", so hit that, and you are dumped to a desktop, with no buttons/etc available.
Right click, go to system tools, and hit "Open Unix Shell". Type:
and you should see the /usr disk is mostly full.
>cd /usr/g/mrraw
Hopefully, this directory is full of Pfiles and links. The links are annoying, but don't take up much space at all. Remove all the Pfiles, or move them to mrisun, the /res disk, etc. Right click again, and go to "shutdown". Now next time you start it up, all will be well.

See also Troubleshooting .


T Test See F-contrast

T1 images An image with a short TR and short TE. See also TR, TE, T2, T2* and Proton Density. If you want more detail or background, visit the MRI section of the links page.

T2 images An image with a long TR and long TE. See also TR, TE, T1, T2* and Proton Density. If you want more detail or background, visit the MRI section of the links page.

T2* images T2 images are an idealization. In reality, the magnetic signal decays faster because there are imperfections in the magnetic field that result from flaws in magnet manufacture, tissue differences (bone, grey matter, white matter etc.) or even metal in the subject. "The sum total of all of these random and fixed effects is called T2*" (T - Two star). See Moriel's "Simply Physics". If you want more detail or background, visit the MRI section of the links page.

Talairach Talairach and Torneaux created a standard 3D spatial coordinate system based on a single human brain. The standard citation for the Talairach atlas is:

Talairach, J., & Tournoux, P. (1988). Co-planar stereotaxic atlas of the human brain. New York: Times Medical Publishing.

The citation for the online talairach daemon is:

Lancaster JL, Woldorff MG, Parsons LM, Liotti M, Freitas CS, Rainey L, Kochunov PV, Nickerson D, Mikiten SA, Fox PT, "Automated Talairach Atlas labels for functional brain mapping". Human Brain Mapping 10:120-131, 2000.

Many brain atlases use these stereotaxic coordinates to locate anatomical features in the 3D volume of the brain. Since the original Talairach coordinate system was developed, others have worked to produce improved sets of coordinates. See especially MNI. However, note also that new Surface Based coordinate systems may replace the earlier stereotaxic systems, especially for identification of points on the cortical surface.

In order to combine functional data across individuals, researchers apply talairaching techniques to warp each brain into a standard space. This process is done in Afni by identifying key anatomical anchors by eye and then warping the brain into the standardized space (see the entire page on Talairaching). In SPM, a similar process is automated and called "normalisation" (see above).

Interesting Links:

Tar (Tape archiving) A method of bundling up a whole directory into a single file. Software like Iceows from will work on Windows systems to ungzip and untar files that might otherwise be difficult to handle.

TE (Echo Time) The time between the initial 90 degree RF pulse and the echo. See also TR, T1, T2, and Proton Density. If you want more detail or background, visit the MRI section of the links page.

Telnet (see also transfer)

Temporal derivative Derivative of a function with respect to time. In SPM, the temporal derivative of the canonical HRF looks something like the canonical but can be used as a basis function, to model a degree of uncertainty as to the exact onset of the HRF. At least one empirical study has found that including the temporal derivative significantly reduces power in the study. (From Gablab Wiki: Glossary) Tensor See Diffusion

Test Image The Left Lesion Test Data (This data has a big hole in the left front, so you can test your understanding of what is happening to right and left given a particular program or manipulation. There is a single functional image, and "2D" and "3D" structural images in spm format).

An MGH program for viewing anatomical images. Online Guide

TkSurfer An MGH program for viewing surfaces. Online Guide

to3d A particularly versatile file conversion program that comes with the afni package. See the official Afni "How To". See also our CNL Afni Preprocessing page about troubles with to3d.

Topology A branch of mathematics concerned with properties of figures and surfaces which are independent of size and shape and are unchanged by any deformation that is continuous, neither creating new points nor fusing existing ones; hence, with those of abstract spaces that are invariant under homomorphic transformations. (OED online) . The field of topology is relevant to surface based reconstructions. See handle and cavity.
TR (Repetition Time) The time between consecutive 90 degree RF (Radio Frequency) pulses. See also TE, T1, T2, and Proton Density. If you want more detail or background, visit the MRI section of the links page. To understand how number of slices interacts with number of repetitions and TR, see Spiralio Sequence Limitations.

Transfer (see also Telnet, FTP, SSH, SCP) Sometimes people need to transfer images back to the consoles (to make films etc. as when there is an image you want a radiologist to look over):

1) Rename your image files to whatever.sdcopen. There is a script, rename_to_sdcopen, on mrisun, tommy and trouardo2 to do this. The following examples assumes filenames that begin with "3579":

>rename_to_sdcopen 3579

Thus, your output should be filenames like

2) ftp these files to /export/home1/sdc_image_pool/import on the scanner where they were collected (Unfortunately, you can't easily transfer the files to a different scanner via ftp because some of the header information for the images contains scanner specifc information; however, see step #4 below for a way around this). e.g.,

>ftp mr1 (you'll have to know the login and password)
>mput *sdcopen /export/home1/sdc_image_pool/import

3) When the images disappear from that directory, the scanner has imported them.
Usually this only takes a minute or so.

4) You can send images from MR3 to MR1 (e.g., if your images need to be filmed), through the networking button on the console. To do this, view the console window's control panel (upper right corner of the screen; see picture):

  • In the control panel, select the top right button (looks like a monitor with some images displayed on it). This will take you to the browser that lists all of the examinations that are currently available on that console. Highlight the examination/series you wish to transfer. Select Network from the items available on the top menu bar.
  • Under Network is listed Selected remote host. The default host is most often set for the Storage2DICOM system that holds radiology images. If you select Selected remote host it will display other possible remote hosts.
  • From that menu you can select mr1.
  • After you have selected mr1, go to Network again and select Send examination OR Send series depending on what you wish to transfer. The images will then be transferred to the other console.

<>Transverse (same as "axial" or "horizontal")

Type I and Type II error Two different types of statistical error. Type I error mistakenly declares a difference to be significant. Type II error mistakenly declares a difference to be nonsignificant.


Underlay See Overlay


VBM Voxel Based Morphometry. See Merlin's VBM section for a list of references we've collected so far. Also check out the VLSM program and documentation here for the analysis of lesioned brains.

Vector A row or column of number in Matrix math. See Matlab and Matrix.

Toward the stomach, in the case of the brain/head, the underside of the brain or head. Same as inferior.

VIDA (Volumetric Image Display and Analysis). This is a general purpose medical imaging processing program similar to Analyze. The program uses an image format compatible with Analyze. Visit the VIDA Homepage

VLSM VLSM is a technique for analyzing the relationship between lesion data and continuous behavioral measures. The set of MATLAB functions which implement VLSM are also referred to as VLSM. Check out the VLSM program and documentation here for the analysis of lesioned brains.

VOI Volume of Interest, 3 dimensional. Compare to ROI. See Russian SPM VOI Tool.

Volume Rendering Rendering in Afni is accomplished by using the "Render" plugin (Datamode->Plugins->Render). It requires a talairached brain. You may want to do skull stripping. The volume rendering plugin allows the user to save an animation sequence (of rotation, cut aways etc.) and then play those back.

Voxel a "volume element" or 3-D pixel. The 3 dimensions (x,y,z) can all be equivalent (isotropic) or they might differ from one another (non-isotropic).

VTK The Visual Toolkit is an open source, freely available software system for 3D computer graphics, image processing, and visualization. It is used in the Visible Human Project and in Surefit/Caret. VTK has its own file formats (see the pdf file). Visit the VTK Home.


UCLA Brain Imaging Center The Brain Imaging Center makes several tools freely available for image conversion, including analyze2genesis, averager, generic2bshort, genesis2analyze, and imconvert. The tools are compiled and installed on charlie. Access each ones help by typing its name at the prompt. imconvert is installed on merlin.

Umask A command you can use in your .cshrc to set your default permissions. See Trouble with permissions


Upgrades See Scanner Updates

Waver files
Waver files are Afni text files full of numbers representing hemodynamic responses for behavioral contrasts of interest and zeroing out or removing events that are not to be compared. By correlating a fim file to a waver file, we create a fico file that shows intensity peaks correlated to the behaviors of interest. The waver files functionally resemble SPM's model specification files.

Webboard-ASP Style or Webboard-Native Style A private forum and chat application available to members of the CNL for discussion of research related issues. ASP Style: Allows everyone access to RSS feeds (news), opportunity to post a picture with your profile (and see other people's images); polling (for managers), access to a calendar (still sketchy). Says "Options" on the menu instead of "More". Native Style: Relies on a different technology, uses the cnl colors. If I learn more about the differences, I'll let you know. So far it seems like ASP style offers you more options except that it does not explicitly list the mailing list addresses.

Contact the administrator to be added to the webboard. We expect the webboard to replace our listservs by the end of January 2005. If Dianne added you to the webboard, you got an automated welcome message. Log on, click "Options (or More)" (depends on which webboard style you login to)" on the menu at the top and click "My Profile" (set a better password, add info etc.).

Click "Options" again. Click "My Mailing Lists" and look at the different "Conferences" that are available to you. Check any conferences that are interesting enough to you that you want regular email from them. For example, if you are scanning, you may wish to be subscribed to

If you simply want to keep track of upcoming meetings and events, then you may wish to subscribe to

Other conferences exist (AFNI, SPM, DMDX etc.). You can always get on the webboard and search for topics in any of the conferences, as you have access to them whether or not you receive email from them.

In addition, you can ask the administrator to set up a private conference for you (only accessible to a select subgroup of users you designate). For example, I've created two private conferences (which most of you won't see): ADRisk (a research project involving about 10 people from several labs) AND HealthDatabaseCleanup (A administrative project involving error checking and organizing a large database of information on subjects).

Each conference is set up with an associated chat room. The chat room will allow you to have a virtual meeting with a group of people, and even use a small whiteboard to draw pictures. The transcripts of conferences are saved for you to refer back to later (Enter the chat room, choose the "Room Options" button at the top->Room registration->Transcripts->Show transcripts (this will allow you to search for transcripts)).

Wiki Wiki means "quick" in Hawaiian. It refers to a collaborative community web resource and the software used to create it. Everyone can contribute and feel responsible for the content. Check out the Wikipedia. The neuroimaging community now has its own wiki, run by Jeff Cooper at Stanford: Log in as fmri, password fmri. Please feel free to contribute and to browse the resources already available.

X A database of Talairach, MNI and anatomy that identifies articles about function for a requested area. Like the Brede database
xjView (for matlab 6.5 and spm2) a VERY useful viewing program that is small and easy to use. p-value slider, id of anatomy with a single mouse click, etc.

xv an older image manipulation and conversion program available for linux and solaris. Type >xv to start the program on any machine that has it installed. It has, unfortunately, lost support.

However, Stanford has binaries posted for linux, sun and sgi that have added support for GE MRI structural (i.e., anatomical) files.

yakview An MGH program


Yoke Two images are "yoked" together when moving the cursor in one, also moves the cursor in the other to the analogous location. This can be useful, for example, when viewing images of the same brain in different planes of section.


Zilverter A java program for reformatting zil files into csv files suitable for Excel.

Zip (see gzip): a method of reducing the size of a file. In general "zip" is used on Windows and gzip is used on Unix. Some unix machine (like linux) recognize the zip command. This should not be confused with Zip drives from Iomega. Zip drives simply store data and do not compress it. Software like Iceows from will work on Windows systems to ungzip and untar files that might otherwise be difficult to handle.