Kendrick's MATLAB Utilities

From VISTA LAB WIKI

Jump to: navigation, search

[Please note: these utilities are superseded by a newer repository available at https://github.com/kendrickkay/.] This is a collection of MATLAB utilities that I have written (Kendrick Kay, kendrick@post.harvard.edu, http://kendrickkay.net). The philosophy of the code is to maximize power (i.e. the ability to perform many different things) and generality (i.e. the ability to re-use code for many different situations). Please let me know if you have any comments or if you encounter bugs or insufficient documentation.

Detailed examples of some of the utilities are given at Kendrick's MATLAB Utility Examples, which is accessible at the wiki (http://white.stanford.edu/newlm/index.php/Kendrick's_MATLAB_Utilities).

Some external MATLAB toolboxes that are used by the utilities are included for your convenience: ba_interp3.zip and NIFTI_20110215.zip. You will need to manually unzip them and place them in the location of your choosing. You may have to mex the file in ba_interp3 for your own platform (try "which ba_interp3 -all").

Contents

[edit] How to get the code

Checkout the SVN repository "kendrick" and add it to your MATLAB path, e.g.:

 cd ~
 svn checkout https://white.stanford.edu/repos/kendrick
 echo "addpath(genpath('~/kendrick'));" >> ~/matlab/startup.m

The code is frequently updated, so you may want to run "svn update" periodically.

[edit] Basic information

The code assumes a MATLAB version of something between R2008a and R2009b. No guarantees on what happens outside of that range, although the bulk of the code will probably still work under other MATLAB versions. The code assumes the existence of several MATLAB toolboxes: stats, optim, and images. Also, some of the code assumes the existence of SPM5, FSL, ba_interp3, and NIFTI_20110215.

If you like, you can browse the code repository in HTML: http://white.stanford.edu/~knk/doc/. This HTML conversion was generated using m2html on Oct 30, 2011 at 2pm (so be aware it might be out-of-date).

Some of the code uses random-number generation, so it is recommended that you have something like

 rand('state',sum(100*clock));
 randn('state',sum(100*clock));

in your ~/matlab/startup.m file.

[edit] History of major code changes

Here we document major code changes that you should be aware of. Minor code changes and tweaks that have no significant impact are not mentioned here (e.g. things like cosmetic adjustments, code speed-ups, making functions more general by adding optional input arguments, fixing bugs that would have result in code crashes, etc.). Major changes include things that make the code not backwards compatible and things change actual analysis results.

  • 2011/10/13 - in pton.m, now always generate a CLUT at 8-bit (256 rows)
  • 2011/10/09 - in fitprfstatic.m, <outputfcn> no longer accepts all those special inputs; it is now a usual OutputFcn.
  • 2011/09/22 - add required argument to copyxform_nii
  • 2011/09/16 - the default behavior in dicomloaddir has changed. we now match all files and perform numerical sorting.
  • 2011/09/13 - make the yellow color in cmapdistinct slightly darker
  • 2011/08/30 - in fitprf.m, we now use stepwise fitting in the nonlinear case (i.e. <mode> is not 0 nor {0}). this should improve the avoidance of local minima problems, but does change previous behavior!!
  • 2011/07/11 - in dicomloaddir.m, we now have the extra assumption that all files are DICOM files when <filenameformat> is supplied
  • 2011/07/01 - we now use nanmedian and nanmean in calcmdse.m
  • 2011/06/29 - in printnice.m, we now temporarily change PaperPositionMode to auto before printing. this should make rasterized image output better (in terms of getting the resolution right).
  • 2011/06/27 - in olsmatrix.m, we now explicitly handle the case of all-zero regressors. previously, NaNs would result. now, we just ensure that these all-zero regressors get returned a zero weight.
  • 2011/05/26 - in scatterb.m, we weren't drawing the main line. weird. we have changed it now to do the right thing. this modifies old behavior.
  • 2011/04/26 - in constructorientationfilter.m, fixed major bug. orientation was interpreted wrongly and filters were peaked at the wrong orientation.
  • 2011/04/02 - in matchfiles.m, time-sorting behavior did not work (it would always give alphabetical order). fixed now.
  • 2011/03/09 - in defineellipse3d.m, we now show a maximum of 25 slices.
  • 2011/03/08 - for alignvolumedata.m, make much faster (using ba_interp3), take less memory, remove spline interpolation option, ensure that the alignvolumedata_auto_outputfcn runs at optimization completion.
  • 2011/03/08 - smoothvolumes handles NaNs intelligently now. this changes old behavior.
  • 2011/03/06 - complete revamp of calcnoiseceiling.m.
  • 2011/02/24 - in matchfiles.m, we now properly match patterns that have spaces in them (since we now escape the spaces using \). previous behavior was to simply not match those cases.
  • 2011/02/02 - in zerodiv.m, in the case that y is not a scalar and wantcaution is set to 0, we were allowing division by 0 to result in Inf and *not* replaced with val as desired. big mistake. we have now fixed this. utilities that could have used this buggy behavior of zerodiv include calccod, calccorrelation, calcentropy, calcsparseness, calczscore, l1unitlength, unitlength.
  • 2011/01/21 - in matchfiles.m, we explicitly use MATLAB's sort function (for alphabetical ordering) to ensure consistency across platforms. this might change the alphabetical ordering of matches!
  • 2010/01/17 - completely change cmapdistinct.m; leave JVM on by default in runmatlabinbg.pl
  • 2010/01/05 - in scatterline.m, we now attempt to use nanmean and nanstd to allow more cases in. see code for details.
  • 2010/12/27 - in divnorm.m, we now explicitly handle cases where the numerator is Inf. the goal is to try to keep outputs finite when possible.
  • 2010/12/16 - circlewithin.m changed from binary (all within or not) to a graded response
  • 2010/12/06 - in fitprf.m, change initial seeds for certain basis function cases (in particular, we no longer force weights to 1. it seems that nonlinear fitting can operate successfully even if there is a gain ambiguity)
  • 2010/12/03 - in fitprf.m, changed the form of the input <hrfnormfun> (this requires changes to how the function is called);
  • 2010/12/02 - in fitprf.m and fitprfmulti.m, a number of significant changes: [<ar> input has been deprecated; <arfile> input has been removed; the <meanint>,<driftstd>,<signalrms>,<noiserms> outputs have been removed (I no longer think they are useful); fix bug --- the signal and drift outputs were not getting assigned in the re-do calls of the [A B C] case of <maxiter>]; removed evaluateprfmodel.m
  • 2010/11/24 - remove hrfsample1, remove hrfsample2 [these were obsolete]
  • 2010/11/23 - in calccod.m, change the output range to percentages (R^2). thus the range is (-Inf,100]. also, we removed the <wantr> input since it was dumb. also, please note that many functions that rely on calccod.m inherit this change to percentages. thus, this was a big global code change.
  • 2010/10/24 - we now impose assumption in rowfun.m that each result has the same size
  • 2010/10/22 - in fitprf.m, fix bug concerning exitflag and maxiters. if there were multiple resamplings, then we were checking only the last exitflag when deciding to re-do the fitting process. now, we do the right thing --- we check to see if any of the resamplings stopped because they reached the max number of iterations.
  • 2010/10/22 - in fitprf.m, change initial seed for prfmodel and hrfmodel basis functions to 0!
  • 2010/10/09 - remove showmulticlass.m
  • 2010/10/05 - fixed bugs in sortnumerical.m. previous calls may have crashed or failed to produce correct answer.
  • 2010/10/02 - fot defineellipse3d.m, change initial visualization state
  • 2010/10/02 - for fitgaussian3d.m, new default initial seed. more liberal tolerance. explicitly guard against nans.
  • 2010/10/02 - in defineellipse3d.m, the fitted Gaussian no longer has a dc fixed to 0 but has an exponent now.
  • 2010/09/27 - zerodiv.m no longer issues warnings when not <wantcaution>
  • 2010/08/25 - explicitly put /home/knk in queuedaemon. this should be revisited eventually...
  • 2010/08/25 - for outputfcnplot, we revamped this function [hack out ratio check; hack out <restol>; hack out NaN case for <numiters>]
  • 2010/08/11 - queuedaemon, runme have been revamped and are not backwards-compatible.
  • 2010/06/22 - unix_wrapper.m now reports status and result to the command window.
  • 2010/06/15 - start using calccod.m instead of calccorrelation.m where appropriate (see subspaceapprox, fitprf, fitprfmulti, fitgabor2d, fitorientedgaussian2d, fitrbf2d)

[edit] Summary of MATLAB functions

Given the large number of functions, it is nice to have a short summary of each one so that you can do a simple text search to see if there is a function that already does what you need.


MATRIX MANIPULATION:

  • addvectors - add vectors of unequal sizes
  • blob - sum over successive groups along a certain dimension of a matrix
  • catcell - concatenate elements of a cell matrix
  • catmatrix - concatenate matrices (possibly unequal in size)
  • cellfunfirst - apply function to first element of a cell matrix (or the matrix itself)
  • cropvalidvolume - extract a subvolume of a matrix based on valid values
  • equalizematrixdimensions - make two matrices the same dimensions by expanding
  • fillout - repeat a matrix to fill a desired size
  • filterout - remove elements of a matrix according to a simple boolean operation
  • flatten - make into a row vector
  • fillmatrix - take a vector and fill a matrix with it
  • flipdims - flip some dimensions of a matrix
  • placematrix - place a 2D matrix inside another 2D matrix at an arbitrary position
  • placematrix2 - place a matrix inside another matrix
  • placematrixmulti - repeatedly randomly place a 2D matrix inside another 2D matrix
  • reshape2D - get into 2D for convenient manipulation
  • reshape2D_undo - undo what reshape2D does
  • rotatematrix - 2D rotation for an arbitrary matrix
  • sizefull - get size of a matrix for a certain number of dimensions
  • splitmatrix - take slices of a matrix and put them into a cell vector
  • splitruns - find runs of non-NaN numbers in a vector
  • squeezedim - remove a specific dimension that has only one element
  • squish - squish together the first n dimensions of a matrix
  • upsamplematrix - change the dimensionality of a matrix by upsampling
  • vflatten - make into a column vector


STRUCT:

  • mergestructs - merge one struct into another


INDEXING:

  • calcposition - calculate the position of elements relative to a list
  • chunking - split a vector into chunks
  • firstel - return first element of a matrix
  • firstelc - return first element of a cell matrix
  • lastel - return last element of a matrix
  • linspacecircular - get equally spaced points in a circular domain
  • linspacefixeddiff - get equally spaced points given a fixed difference
  • linspacepixels - get equally spaced points that can be treated as centers of pixels
  • matrixindex - return different slices through a matrix based on an indexing matrix
  • resamplingindices - get indices corresponding to the process of resampling
  • slicematrix - return a slice from a matrix
  • subscript - get elements using a vector or cell vector of indices


RANDOMIZATION:

  • permutedim - shuffle a matrix along a certain dimension or globally
  • picksubset - return a random subset of the elements in a matrix


FORMAT:

  • enforcenan - set cases with all 0 to NaN
  • isfinitenum - evaluate whether each element is finite and numeric
  • isint - evaluate whether each element is an integer
  • isrowvector - evaluate whether matrix is 1 x n (where n >= 0)


PROGRAMMING:

  • allocatememory - force allocation of a certain amount of memory (to induce swapping)
  • alwaysone - always return 1
  • alwayszero - always return 0
  • assignseparatevars - split a matrix and assign it to multiple variables
  • assigntobaseworkspace - assign all variables to the base workspace
  • checkmemoryworkspace - figure how much memory is being used in the caller's workspace
  • checkpathconflicts - check for .m file conflicts
  • choose - functional form of if-else
  • copymatrix - make a copy of a matrix with a subscript assignment on-the-fly
  • decodenum - undo encodenum.m
  • defaultoptimset - return a default optimset options structure
  • encapsulate - evaluate a function at multiple values and encapsulate results into a cell vector
  • encodenum - encode a number as a string of uppercase letters
  • evalme - evaluate multi-line input in the MATLAB workspace
  • gethostname - get the hostname
  • getpid - get the process ID for this instance of MATLAB
  • GPUconv - convert to GPU format
  • GPUok - return whether we should try to use the GPU
  • gunziptemp - gunzip a file on the fly
  • identity - return an argument unchanged
  • infiniteloop - do nothing forever
  • inputmulti - take multi-line string input from stdin
  • maketempdir - make a new empty temporary directory and return the path to it.
  • mkdirquiet - make a directory, suppressing warnings
  • mtimesmulti - multiply arguments together
  • passmulti - pass multiple arguments to a function from a cell vector
  • playalarm - play an alarm to get the user's attention
  • plusmulti - add arguments together
  • processchunks - perform an operation on successive chunks of a big matrix to save on memory usage
  • promptforfile - get a .mat file name from the user
  • repeatuntil - call a function until a condition is satisfied
  • reportmemoryandtime - use a warning to report the current memory usage and the current time
  • rowfun - apply a function to each row of a 2D matrix
  • runmatlabinbg.pl - run MATLAB in the background for a single function/script call
  • setrandstate - induce randomness by (randomly) setting the seed of rand and randn
  • statusdots - write out dots (.) indicating progress of code execution
  • subsetfun - apply a function to successive subsets of a vector
  • swap - swap two arguments
  • unix_wrapper - a nice wrapper for unix.m


OPTIMIZATION:

  • outputfcnplot - plot parameter history during an optimization
  • outputfcnsanitycheck - explicit check on the size of the residuals to prevent infinite computation


STRING:

  • cell2str - convert 2D cell matrix into string representation
  • cell2str2 - convert cell vector of strings into a single string that names each string
  • makedirid - return a string with elements of a path joined with '_'
  • makeletters - generate a cell vector of letter characters
  • mat2cellstr - convert a matrix of numbers into a cell matrix of strings
  • matchword - return the first word found in a string
  • randomword - generate a random word consisting of capital letters
  • sortnumerical - sort a cell vector of strings according to trailing integers
  • strsplit - split a string based on a pattern


INPUT/OUTPUT/FILE-MANIPULATION:

  • absolutepath - return the absolute path to a given file
  • getextension - get the extension (e.g. '.txt') of a filename
  • loadbinary - load numbers from a binary file
  • loadexcept - load variables from a .mat file, excluding certain variables
  • loadmulti - concatenate or merge a single variable from multiple .mat files
  • loadmulti2 - load multiple variables from one .mat file
  • loadtext - load strings from a text file
  • matchfiles - return strings containing paths to matched files and/or directories
  • multifile - apply a function to multiple files
  • niiload - load specific voxels from NIFTI files containing 4D data
  • raw2dload - load specific columns from binary files containing 2D data
  • savebinary - save numbers to a binary file
  • saveexcept - save all variables to a .mat file, excluding certain variables
  • savetext - save strings to a text file
  • stripfile - remove the file name from a string


FIGURE:

  • addlogticks - add logarithmic ticks to a figure whose data have already been log-transformed
  • axissquarify - make axis equal-aspect, square, and origin-centered and draw unity line
  • errorbar2 - draw error bars (as lines)
  • errorbar3 - draw error bars (as a polygon)
  • figureprep - make an invisible figure window
  • figurewrite - write the invisible figure window to an image file
  • getcolorchar - take an integer and get a color character
  • getfigurepos - return figure position (in normalized units)
  • hist1dimage - draw a histogram as a vertical 1D image
  • histrobust - run hist but ensuring a robust range
  • imagescmulti - make separate figure windows for multiple images
  • plotorientedbar - plot an oriented bar
  • plotrectangle - plot a rectangle
  • printnice - print figure window to file
  • setaxispos - set axis position (without mangling the Units setting)
  • setfigurepos - set figure position (without mangling the Units setting)
  • scatterb - do a binned scatterplot
  • scatterline - use local linear regression to summarize a scatterplot
  • scattersparse - do a scatterplot but limit the number of points
  • scatterimagesparse - do a image scatterplot and limit the number of points
  • squarify - figure out a roughly evenly balanced number of rows and columns necessary to fit a certain number of elements
  • straightline - draw horizontal or vertical lines


COLORMAP:

  • cmapdistinct - hue-based colormap for cases where you want maximally distinct colors
  • cmaplookup - take values and get colors from colormap
  • cmaphue - hue-based colormap suitable for circular ranges
  • cmapsign - blue-black-red colormap suitable for ranges like [-X X]
  • cmapsign2 - blue-white-red colormap suitable for ranges like [-X X]
  • cmapsign3 - blue-lightgray-red colormap suitable for ranges like [-X X]
  • colorinterpolate - make a colormap by interpolating between key colors
  • drawcolorbar - draw a colorbar on the current figure


SIMPLE MATH:

  • addnoise - add some noise to a matrix
  • allzero - check whether all elements are equal to zero within some tolerance
  • ang2complex - convert angles to unit-magnitude complex numbers
  • chunkfun - apply a function to successive chunks of a vector
  • chunknormalize - subtract mean of successive chunks of a matrix
  • calccod - calculate coefficient of determination (R^2)
  • calccodcell - calccod for cell vectors of matrices
  • calccorrelation - calculate Pearson's correlation coefficient (r)
  • calcmdse - calculate the median and the standard error on the median
  • calcmdsepct - calculate the median and percentiles on bootstraps of the median
  • calcsparseness - calculate sparseness
  • calczscore - convert to z-scores
  • circularavg - average numbers, respecting circularity
  • circulardiff - difference between two values defined in a circular domain
  • circularshiftandsum - take a vector, circularly shift it multiple times, and sum
  • clipoutliers - clip the outliers of a matrix
  • count - sum all elements of a matrix
  • countinstances - count how many 1s, 2s, etc. are in a matrix
  • discretize - round elements of a matrix to one of two values
  • divnorm - apply divisive normalization to a matrix
  • matchgain - figure out scalar factor to match one set of data to another
  • meancell - take the mean for data that are split into cell matrices
  • mod2 - like mod.m but return 0 outputs as the modulus itself
  • mod4 - like mod.m but restrict outputs to the range [-modulus/2,modulus/2]
  • mtimescell - do matrix multiplication but with an argument split into a cell vector
  • nanreplace - replace NaNs with a specified value
  • nantoclosest - replace NaNs with the nearest valid values
  • negreplace - replace negative values with a specified value
  • normalizerange - scale to fit within a specific range
  • normalizemax - scale such that max value is 1
  • posnegrect - positive- and negative-rectify
  • posrect - positive-rectify
  • restrictrange - truncate values to fit within a range
  • robustrange - calculate a robust range of a set of values
  • signedarraypower - take the array power but preserve sign
  • signforce - return +1 for positive or zero elements; -1 for negative elements
  • stdquartile - return something like std but in the percentile sense
  • unitlength - normalize to have unit length (L2-norm is 1)
  • unitlengthfast - normalize to have unit length (L2-norm is 1)
  • l1unitlength - normalize to have unit length (L1-norm is 1)
  • vectorlength - calculate vector length (L2-norm)
  • l1vectorlength - calculate vector length (L1-norm)
  • zerodiv - divide but with special handling for division by zero
  • zeromean - subtract off the mean


MATH/STATS:

  • applyfiltermultidim - apply filters for the case of images with multiple features
  • bootstrap - apply a function to bootstrap samples drawn from a vector (useful for computing standard error)
  • bootstrapdim - more general version of bootstrap
  • calcconfusionmatrix - calculate a confusion matrix (i.e. covariance, cross-product, etc.)
  • calcentropy - calculate entropy based on a matrix with counts
  • calcggaussianlikelihood - calculate likelihood of a set of points assuming independent generalized Gaussian distributions
  • calcintrinsicdim - calculate number of eigenvectors needed to achieve a certain amount of the variance in a matrix
  • calcmutualinformation - calculate mutual information between two matrices
  • calcmutualinformationcontinuous - calculate mutual information between two matrices in a continuous way
  • calcnoiseceiling - calculate how well a set of data can be predicted, taking into account the intrinsic noisiness of the data
  • calcoptimalhistbins - calculate the optimum number of bins to use for hist.m
  • calcrbflikelihood - calculate likelihood of a set of points given a collection of RBFs
  • calcrectlikelihood - calculate likelihood of a set of points given a collection of directions (half-squared responses)
  • calcsurfaceareanball - calculate surface area of a unit n-ball
  • centerofmass - calculate the center-of-mass of a matrix
  • constructdctmatrix - construct a matrix with DCT-II basis functions in the columns
  • constructdftmatrix - construct a matrix with DFT basis functions in the columns
  • constructfiltersubsample - construct 2D filters according to a subsampling scheme
  • constructfiltertiling - construct filters by tiling a single 2D filter
  • constructpolynomialmatrix - construct a matrix with polynomials in the columns
  • constructpolynomialmatrix2d - evaluate polynomial basis functions defined in 2D at some locations
  • constructpolynomialmatrix3d - evaluate polynomial basis functions defined in 3D at some locations
  • evaldivnorm - evaluate divisive-normalization function at some values
  • evalgeneralizedgaussian - evaluate generalized Gaussian pdf
  • filter2subsample - apply filter2 and then subsample the result
  • findlocal - find points that are close to a given point
  • fit2dpolynomialmodel - use polynomial basis functions to fit a surface defined in 2D, allowing scale factor for different cases
  • fit3dpolynomialmodel - use polynomial basis functions to fit a surface defined in 3D, allowing scale factor for different cases
  • fit3dpolynomialmodel2 - use polynomial basis functions to fit a surface defined in 3D, allowing DC offset for different cases
  • fitdivnorm - fit divisive-normalization function
  • fitgaussian1d - fit 1D Gaussian function
  • fitl1line - perform linear regression but use an L1-error metric
  • fitline2derror - fit a line to a set of 2D points using an error metric that is sensitive to both x and y
  • fitrbfpdf - fit a simplified mixture-of-Gaussians probability density function model using k-means
  • fitrectpdf - fit a rectified squared model using k-means and interpret as a PDF
  • fitrectdensity - fit a rectified squared model using k-means and interpret more like ANN
  • gradientdescent - perform gradient descent or forward stagewise regression
  • gradientdescent_wrapper - perform n-fold averaging technique for gradientdescent.m
  • localregression - given a degree, kernel, and bandwidth, use local regression to predict values (surface defined on one dimension)
  • localregression2d - given a degree, kernel, and bandwidth, use local regression to predict values (surface defined on two dimensions)
  • localregression3d - given a degree, kernel, and bandwidth, use local regression to predict values (surface defined on three dimensions)
  • localregression4d - given a degree, kernel, and bandwidth, use local regression to predict values (surface defined on four dimensions)
  • localregressionbandwidth - cross-validate to determine the optimal bandwidth for local regression
  • olscontrol - perform ordinary least-squares regression while controlling for some given regressors
  • olsmatrix - perform ordinary least-squares regression
  • performfreqwhitening - flatten the average amplitude spectra of a set of images
  • performpcawhitening - whiten a matrix in the PCA sense
  • projectionmatrix - project out a linear subspace using ordinary least-squares regression
  • randgg - generate random numbers from a generalized Gaussian
  • randnmulti - generate random numbers from a multivariate Gaussian
  • randomization - calculate p-values using randomization
  • randomorthogonalbasis - generate a random set of orthogonal basis functions
  • roundpvalue - round p-value up to have the form Xe-Y
  • runkmeans - run the k-means clustering algorithm
  • subspaceapprox - calculate correlation between a subspace and a vector


TIME-SERIES PROCESSING:

  • calcpeak - calculate the peak of some time-series data (using interpolation)
  • constructbutterfilter1D - construct a Fourier-domain Butterworth filter
  • deconvolvevectors - deconvolve one vector from another vector
  • fouriertospace1D - convert magnitude filter in Fourier domain to the space domain
  • processmulti1D - apply a function that expects a single vector to multiple vectors
  • sincshift - shift time-series data using sinc interpolation and padding
  • tsfilter - filter time-series data with a Fourier-defined magnitude-based filter


IMAGE PROCESSING:

  • ba_interp3_wrapper - convenient wrapper for the 3D interpolation toolbox ba_interp3
  • calcamplitudespectrum - calculate a smooth function relating spatial frequency to amplitude
  • calccpfov - return number of cycles per FOV associated with fft2 after fftshifting
  • calccpfov1D - return number of cycles per FOV associated with fft after fftshifting
  • calcimagecoordinates - return coordinates of the pixels of an image
  • calcpositiondifferentfov - determine the coordinates of a new image with respect to an original image
  • calcunitcoordinates - return coordinates of points equally spaced within the square bounded by -.5 and .5
  • checkimagesize - inspect pixel dimensions of a batch of image files
  • chtoim - reshape image(s) from images x res*res to res x res x images
  • concatimages - concatenate multiple images together
  • constructbutterfilter - construct a Fourier-domain Butterworth filter
  • constructorientationfilter - construct a Fourier-domain orientation filter (von Mises distribution)
  • constructsmoothingfilter - construct a space-domain 2D or 3D Gaussian filter
  • evaldog2d - evaluate 2D Difference-of-Gaussians function at some coordinates
  • evalgabor2d - evaluate 2D Gabor function at some coordinates
  • evalgaussian1d - evaluate 1D Gaussian function at some coordinates
  • evalgaussian2d - evaluate 2D Gaussian function at some coordinates
  • evalgaussian3d - evaluate 3D Gaussian function at some coordinates
  • evalgrating2d - evaluate 2D sinusoidal grating function at some coordinates
  • evalorientedgaussian2d - evaluate oriented 2D Gaussian at some coordinates
  • evalrbf2d - evaluate 2D radial basis function at some coordinates
  • extractwindow - easily pull out different chunks of an image
  • fitgabor2d - fit 2D Gabor function
  • fitgaussian3d - fit 3D Gaussian function
  • fitorientedgaussian2d - fit oriented 2D Gaussian
  • fitrbf2d - fit 2D radial basis function
  • fftshift2 - apply fftshift along the first two dimensions
  • fouriertospace - convert magnitude filter in Fourier domain to the space domain
  • generatepinknoise - generate samples of pink noise
  • generaterandomphase - generate random phases suitable for multiplication with output of fft2
  • generatewinnermap - compress multidimensional data by using hue and brightness to represent the maximum value along a dimension
  • getsamplebrain - return a brain volume
  • getsampleimage - return a 500x500 grayscale natural image
  • ifftshift2 - apply ifftshift along the first two dimensions
  • imagesearch - search through image space to minimize some cost function
  • imreadmulti - load in multiple images
  • imresizedifferentfov - resize an image but use a different field-of-view
  • imwritemulti - write multiple images to multiple files
  • imtoch - reshape image(s) from res x res x images to images x res*res
  • makecircleimage - draw a circle as an image
  • makecolorimagestack - convert a 4D matrix containing multiple color images into a 3D matrix
  • makeconcentricgratings2d - make a concentric 2D sinusoidal grating
  • makedog2d - make a 2D Difference-of-Gaussians
  • makegabor2d - make a 2D Gabor
  • makegaussian2d - make a 2D Gaussian
  • makegaussian3d - make a 3D Gaussian
  • makegrating2d - make a 2D sinusoidal grating
  • makegratings2d - make a set of 2D sinusoidal gratings
  • makeimagestack - convert a 3D matrix containing multiple images into a 2D matrix
  • makeimagestack_wrapper - run makeimagestack for multiple 3D matrices
  • makeorientedgaussian2d - make an oriented 2D Gaussian
  • makesquareimage - draw a square as an image
  • placeimageintosquare - place an image anywhere within another image
  • processimages - perform some function on many image files
  • processmulti - apply a function that expects a single image to multiple images
  • smoothvolumes - use a Gaussian filter to smooth one or more 3D volumes
  • splitimages - split images into chunks
  • unitlengthfft2 - unit-length-normalize a Fourier domain filter
  • viewimage - do some visualizations for a given image
  • viewimages - visualize many images at once
  • viewmovie - view a series of images as a movie (or write a series of image files)


IMAGE TRANSFORMATIONS:

  • constructcosinefilter - construct a Fourier-domain cosine-tapered filter
  • constructwhiteningfilter - construct a Fourier-domain whitening filter
  • detectedges - detect edges in an image using simple scheme
  • flattenspectra - apply spatial frequency filtering to flatten the spectra of some images
  • imagefilter - filter images with a Fourier-defined magnitude-based filter
  • phasescrambleimage - blend image with random-phase image
  • scrambleimage - scramble an image, preserving chunks of a certain size
  • varycontrast - vary the contrast of an image


IMAGE OPERATIONS/FILTERING:

  • applymultiscalegaborfilters - filter images with a set of multi-scale Gabor filters
  • applymultiscalegaussianfilters - filter images with a set of multi-scale Gaussian filters
  • applyv2filters - filter images generically [IN PROGRESS]
  • gabortransform - perform Gabor transformation on one or more images
  • maskmultiscalegaborfilters - modify results of applymultiscalegaborfilters according to a binary mask
  • visualizemultiscalegaborfilters* - visualize points associated with applymultiscalegaborfilters


INTERACTIVE GRAPHICS:

  • defineellipse3d - allow user to define a 3D ellipse on a 3D volume


GRAPHIC PROCESSING:

  • coordangle - calculate coordinates of an angle
  • coordpolygon - calculate coordinates of an equilateral polygon
  • drawarrow - draw an arrow (in two dimensions)
  • drawbar - draw an oriented bar
  • drawbars - draw a set of oriented bars
  • drawbartexture - draw a random array of oriented bars
  • drawcheckerboard - draw an oriented checkerboard
  • drawcheckerboards - draw a set of oriented checkerboards
  • drawclosedcontour - draw a generic closed contour shape
  • drawclosedcontours - render several closed contours
  • drawellipse - draw a complete or partial ellipse (includes circles and arcs as special cases)
  • drawrectangle - draw a rectangle
  • drawtext - draw a word
  • drawtexts - render several words
  • renderfigure - render the current figure into an image (assumed to be square)


GEOMETRY, POLYGON-LIKE:

  • calcdistpointline - calculate the distance between one or more points and a line
  • circlewithin - determine whether one or more circles are contained within a certain circle


3D SPATIAL TRANSFORMATION:

  • createrescalingmatrix - construct a transformation matrix that rescales with respect to (0.5,0.5,0.5)
  • xyzscale - construct a scaling transformation matrix
  • xyztranslate - construct a translating transformation matrix


STIMULI

  • constructstimuli - examples of how to construct various types of stimuli


PSYCHTOOLBOX:

  • pton - initialize PsychToolbox stuff
  • ptoff - uninitialize PsychToolbox stuff
  • ptviewimage - show one or more images and allow user to position the images
  • ptviewmovie - show a stimulus movie
  • ptviewmoviecheck - check the results of ptviewmovie.m


EXPERIMENTAL DESIGN:

  • expdesignefficiency - calculate efficiency of an experimental design


HRF:

  • calchrfpeak - estimate the peak (positive or negative) value of an HRF
  • constructhrfmodeldct - return a matrix of DCT-II basis functions for modeling HRFs
  • constructhrfmodeldelta - return a matrix of delta basis functions for modeling HRFs
  • constructhrfmodelspline - return an spline-based HRF model
  • evaldoublegamma - evaluate double-gamma function at some x-values
  • getcanonicalhrf - return a canonical HRF (based on empirical data) for arbitrary duration and TR
  • getsamplehrf - return some sample HRFs estimated from actual fMRI data


MRI and fMRI:

  • computetemporalsnr - compute the stability of a series of 3D volumes over time
  • dicomloaddir - load volumes and voxel-size information from a DICOM directory
  • fmriquality - write figures to inspect the spatial quality of fMRI volumes
  • homogenizevolumes - homogenize a set of 3D volumes by dividing by fitted 3D polynomials
  • preconditionvolume - precondition a 3D volume by clipping outliers and removing low-frequency signal variations
  • undistortvolumes - resample volumes in order to correct for motion and distortion
  • preprocessfmri - perform complete pre-processing of fMRI data
  • preprocessfmri_standard* - example script that wraps around preprocessfmri
  • write3dstack - write out extracted slices in three orientations from a 3D matrix


NIFTI:

  • copyxform_nii - propagate transformation-related fields from one NII file to others
  • settr_nii - set the TR in the header information of an NII structure


Spiral MRI :

  • reconstructspiralfieldmap - reconstruct spiral fieldmap from raw P-file
  • f_getspiralhdr - helper function used by reconstructspiralfieldmap


SPM (COREGISTRATION, MOTION CORRECTION):

  • coregistervolumes - coregister one volume to another
  • createspmmatrix - return the default SPM transformation matrix
  • motioncorrectvolumes - perform motion correction for some volumes
  • writespmfiles - write a series of volumes to individual SPM/ANALYZE files


PRF, TIME-SERIES, MODELING:

  • constructprfmodel2Dgaussian - return an isotropic 2D Gaussian PRF model
  • nonlinearprfmodel - compute response of PRF model to some stimuli
  • fitprf - fit PRF model to some time-series data
  • fitprffast - fit PRF model simultaneously to many different time-series
  • fitprffast_amountofdata - wrapper for fitprffast for evaluating the effect of the amount of data
  • fitprfgui - GUI for fitprf
  • fitprfmulti - [IN PROGRESS]
  • fitprfstatic - [IN PROGRESS]
  • fitprfstaticmulti - [IN PROGRESS]


GRECONS:

  • adjustgrecons - a grecons wrapper that implements parameter smoothing
  • greconsmultifreq - a grecons wrapper that smooths fieldmaps and performs multi-frequency reconstruction
  • greconsevaluateb0 - a grecons wrapper that performs many B0 reconstructions
  • makellbands [contributed by Jon Winawer] - specify bandwidths for smoothing fieldmaps


QUEUEING (SGE):

  • sgerun - run MATLAB code using the Sun Grid Engine (SGE)
  • sgestat - print out some summary information regarding SGE jobs


ALIGNVOLUMEDATA:

  • alignvolumedata - manually align two 3D volumes (reference volume, target volume)
  • alignvolumedata_auto - automatically adjust transformation parameters
  • alignvolumedata_exporttransformation - report the current transformation parameters
  • alignvolumedata_preferences - query and set preferences
  • extractslices - extract slices from the reference or target volume
  • maketransformation - given a set of parameters, make a transformation struct
  • matrixtotransformation - convert 4x4 transformation matrix to a transformation struct
  • transformationtomatrix - convert transformation struct to a 4x4 transformation matrix


QUEUEING [OBSOLETE]:

  • queuedaemon - dispatch MATLAB jobs created by runme.m
  • runme - take multi-line input and run it in the background
Personal tools