RUMBA Software

The matrix class facilitates performing matrix math on fMRI data. A Matrix is actually a thin wrapper (adaptor) about a dManifold. Like dManifold, Matrix supports 2d container operations copy() Returns: Matrix Description: returns a deep copy size() Returns: int Description: returns number of items in container begin() Returns: dManifoldIterator Description: returns an iterator to the start of the container end() Returns: dManifoldIterator Description: returns an iterator that is "one off the end" of the container. __getitem__(int offset) Usage: M[offset] Returns: float Description: One dimensional indexing for container traversal __getitem__((int space, int time)) Usage: M[row,column] Returns: float Description: Access rows and columns __getitem__((Ellipsis, int y)) Usage: M[...,y] Returns: dSection Description: Returns the y_th column __getitem__((int x, Ellipsis)) Usage: M[x,...] Returns: dSection Description: Returns the x_th row __getitem__((int x,slice)) Usage: M[x,i:j] Returns: dManifold Description: Return a copy of the subsection [i,j) of the x_th column __getitem__((Ellipsis, slice)) Usage: M[...,i:j] Returns: type(M) Description: Return the sub matrix corresponding to columns i to j-1 __setitem__((int offset, double val)) Usage: M[offset] = val Returns: None Description: Sequential write access __setitem__((int row, int column, double val)) Usage: M[row,column] = val Returns: None Description: Write access In addition to the 2d container interface, Matrix supports the following constructors: Matrix(string) Load Matrix from a file Matrix(int rows,int cols) Create a matrix with a given number of rows and columns The following additional members are supported: int rows() Returns: int Description: number of rows cols() Returns: int Description: number of columnes toManifold() Returns: dManifold Description: returns underlying dManifold __getitem__((slice,int)) Usage: M[i:j,x] Returns: Matrix Description: Returns column matrix consisting of rows i...j-1 at column x __getitem__((slice,Ellipsis)) Usage: M[i:j,...] Returns: Matrix Description: Returns j-i x M.cols() matrix consisting of rows i...j-1 Matrix also supports the following arithmatic: __add__(Matrix N) Usage: M + N Returns: Matrix __iadd__(Matrix N) Usage: M += N Returns: Matrix __sub__(Matrix N) Usage: M-N Returns: Matrix __isub__(Matrix N) Usage: M -= N Returns: Matrix __mul__(Matrix N) Usage: M*N Returns: Matrix __mul__(float x) Usage: M*x Returns: Matrix __rmul__(float x) Usage: x*M Returns: Matrix __imul__(float x) Usage: M*=x Returns: Matrix

Methods defined here:

__add__(...)

__del__(...)

__getitem__(...)

__iadd__(...)

__imul__(...)

__init__(self, *args)

__isub__(...)

__mul__(...)

__rmul__(...)

__setitem__(...)

__str__(...)

__sub__(...)

Data and non-method functions defined here:

__doc__ = '\n\nThe matrix class facilitates performing mat...oat x)\n\tUsage: M*=x\n\tReturns: Matrix\n\n\n\n'

__module__ = 'RUMBA'

class SectionProxy

Methods defined here:

__init__(self, s, e)

__iter__(self)

next(self)

Data and non-method functions defined here:

__doc__ = None

__module__ = 'RUMBA'

class Splodge

Methods defined here:

__del__(...)

__init__(self, *args)

__str__(...)

Data and non-method functions defined here:

__doc__ = None

__module__ = 'RUMBA'

class dManifold

dManifold is a class used to represent neuro-imaging data. It also facilitates the loading and saving of data in many different formats. The main content of the class is a 4 dimensional data set. There are three ways that one mat to access the points of a dManifold: the more intuitive way is to use (x,y,z,t) coordinates. Another viewpoint is that each point has an index corresponding to an ordering on the points -- that is, you can view it as a 1 dimensional array. This is convenient for writing loops that traverse the data set. Finally, one can use (space,time) coordinates. This is useful for writing functions that need to iterate over time series. The Manifold class also facilitates sub Manifold extraction/drop-in. This discussion would not be complete without a description of the loading and saving which is useful in itself. These operations facilitate loading and saving files from different formats. The load constructor and save method decide appropriate formats from the filename given. Interface: The dManifold class implements the following 2 dimensional container interface. This interface is also implemented by the Matrix class. The main difference is that the 2 dimensional coordinates refer to space and time in the context of dManifold objects, and rows/columns in the context of matrices. copy() Returns: dManifold Description: returns a deep copy size() Returns: int Description: returns number of items in container begin() Returns: dManifoldIterator Description: returns an iterator to the start of the container end() Returns: dManifoldIterator Description: returns an iterator that is "one off the end" of the container. __getitem__(int offset) Usage: M[offset] Returns: float Description: One dimensional indexing for container traversal __getitem__((int space, int time)) Usage: M[space,time] Returns: float Description: Access data in space/time coordinates __getitem__((Ellipsis, int y)) Usage: M[...,y] Returns: dSection Description: Returns the y_th spatial volume. __getitem__((int x, Ellipsis)) Usage: M[x,...] Returns: dSection Description: Returns the x_th time series __getitem__((int x,slice)) Usage: M[x,i:j] Returns: dManifold Description: Return a copy of the subsection [i,j) of the x_th time series. __getitem__((Ellipsis, slice)) Usage: M[...,i:j] Returns: type(M) Description: Return the subset of points for which i <= t < j __setitem__((int offset, double val)) Usage: M[offset] = val Returns: None Description: Sequential write access __setitem__((int space, int time, double val)) Usage: M[space,time] = val Returns: None Description: Write access in terms of space,time coordinates Construction: The following consructors are supported: dManifold(int x,int y,int z,int t) Description: Construct a dManifold with dimensions x*y*z*t, where x,y and z are spatial dimensions, and t is the temporal extent dManifold(intPoint) Description: Convenience function. Same as above. dManifold(string) Description: Load a data set from a file. The string is the filename. If the data contains a header and an image file, the name of either file will suffice. Member functions for dealing with 4 dimensional data: The class supports a number of methods for accessing information about the 4 dimensional data structure. width(): Returns: int Description: Returns width height() Returns: int Description: Returns height depth() Returns: int Returns depth timepoints() Returns: int Returns number of timepoints. pixels() Returns total number of spatial locations. Same as M.width()*M.height()*M.depth() extent() Returns an intPoint containing space-time dimensions. __getitem__((int x, int y, int z, int t)) Returns: float Usage: M[x,y,z,t] __setitem__((int x, int y, int z, int t, double val)) Returns: None Usage: M[x,y,z,t] = val __getitem__(a0,a1,a2,a3) where a0,a1,a2,a3 are either ... , int, or int:int Returns: dManifold Usage: M[...,1,2,2:3] M[1,1,2,2:3] Description: Allows convenient selection of subsets of data. One can specify a range [i,j) with the notation i:j, or all data points in a given dimension with the notation "..." For example, M[... , 2, 3:6, 2] returns the subset of points for which y=2, 3<=z<6, t=2. The resulting data set is size M.width()*2*3*1 Accessing Headers: getHeader(string name) Returns: Splodge Description: Gets header with name. headerData() Returns: dictionary whose keys are strings, values are Splodges. Saving to disk: save(string filename) Returns: None Description: Saves data to filename. Realignment: Matrix matchRigid(dManifold standard, dManifold reslice, dManifold mask, int timebudget= 10 ); Matrix matchRigid(dManifold standard, dManifold reslice, int timebudget = 10 ); Matrix matchAffine(dManifold standard, dManifold reslice, dManifold mask, int timebudget= 10 ); Matrix matchAffine(dManifold standard, dManifold reslice, int timebudget = 10 ); Matrix matchMaskAffine(dManifold standard, dManifold reslice, dManifold mask, int timebudget= 10 ); Matrix matchMaskAffine(dManifold standard, dManifold reslice, int timebudget = 10 ); Matrix matchRatioAffine(dManifold standard, dManifold reslice, dManifold mask, int =256, int = 10); Matrix matchRatioAffine(dManifold standard, dManifold reslice, int partitions=256, int timebudget= 10); Matrix matchRatioAffine(Matrix start_params,dManifold standard, dManifold reslice, dManifold mask, int partitions=256, int timebudget= 10); Matrix matchRatioAffine(Matrix start_params,dManifold standard, dManifold reslice, int partitions=256, int timebudget= 10); Matrix matchPolynomial ( dManifold standard,dManifold reslice, dManifold mask, int degree, int timebudget= 10); Matrix matchPolynomial ( dManifold standard,dManifold reslice, int degree, int timebudget= 10); Compute parameters for realigning too data sets. matchRigid: use 6 parameter rigid body least squares model matchAffine: use 12 parameter affine least squares model matchPolynomial: use polynomial least squares model (30-168 parameters) matchRatioAffine: use 12 parameter model with AIR-like cost function. standard: reference data set reslice: data set to be resliced timebudget: units for computing. Each unit of timebudget corresponds to one complete scan of the data set. Since the data set is subsampled, a timebudget of 1 still allows many iterations of the algorithm. mask: spatial mask of pixels to ignore. degree: polynomial degree. partitions: number of partitions to segment data into. Reslicing: enum InterpolationType { Trilinear = 0 }; dManifold resliceRigid (Matrix params, dManifold data, InterpolationType type, intPoint outextent, doublePoint outdims ); dManifold resliceAffine (Matrix params, dManifold data, InterpolationType type, intPoint outextent, doublePoint outdims ); dManifold reslicePolynomial (Matrix params , dManifold data, InterpolationType type, intPoint outextent, doublePoint outdims ); dManifold resliceRigid (Matrix params , dManifold data , InterpolationType type); dManifold resliceAffine (Matrix params , dManifold data , InterpolationType type); dManifold reslicePolynomial (Matrix params , dManifold data , InterpolationType type); Utility functions: double maximum(dManifold); returns the maximum value in the data set minimum(dManifold); returns the minimum value in the data set normalize(double lower, double upper, dManifold data); range-normalizes dataset.

Methods defined here:

__del__(...)

__getitem__(...)

__init__(self, *args)

__setitem__(...)

__str__(...)

Data and non-method functions defined here:

__doc__ = '\ndManifold is a class used to represent neuro-...Manifold data);\n\nrange-normalizes dataset.\n\n'

__module__ = 'RUMBA'

class dManifoldIterator

Methods defined here:

__del__(...)

__init__(self, *args)

Data and non-method functions defined here:

__doc__ = None

__module__ = 'RUMBA'

class dSection

Methods defined here:

__add__(...)

__del__(...)

__getitem__(...)

__init__(self, *args)

__mul__(...)

__setitem__(...)

__str__(...)

__sub__(...)

Data and non-method functions defined here:

__doc__ = None

__module__ = 'RUMBA'

class dSectionIterator

Methods defined here:

__del__(...)

__init__(self, *args)

Data and non-method functions defined here:

__doc__ = None

__module__ = 'RUMBA'

class dSectionPointer

Methods defined here:

__cmp__(...)

__del__(...)

__init__(self, *args)

Data and non-method functions defined here:

__doc__ = None

__module__ = 'RUMBA'

class doublePoint

Methods defined here:

__del__(...)

__init__(self, *args)

__str__(...)

Data and non-method functions defined here:

__doc__ = None

__module__ = 'RUMBA'

class intPoint

Methods defined here:

__del__(...)

__init__(self, *args)

__str__(...)

Data and non-method functions defined here:

__doc__ = None

__module__ = 'RUMBA'

Functions

Abs(...)

ArCoeff(...)

Divide(...)

DurbinWatson(...)

MatrixAdd(...)

MatrixCols(...)

MatrixIdentity(...)

MatrixInvert(...)

MatrixMultiply(...)

MatrixPseudoInvert(...)

MatrixRows(...)

MatrixSubtract(...)

MatrixTranspose(...)

Max(...)

MaxAbs(...)

Min(...)

Minus(...)

Multiply(...)

PREFIX(...)

Plus(...)

PyFunction(...)

Square(...)

Sum(...)

SumSquare(...)

accumulate(...)

applyAr(...)

arfilter(...)

bind(...)

fftPower(...)

fill(...)

firstSection(...)

firstVol(...)

for_each(...)

getRow(...)

histogram(...)

histogram2(...)

hrf(...)

identityMatrix(...)

inner_product(...)

inv(x)

invert(...)

jacobi(...)

lastSection(...)

lastVol(...)

light_svd(...)

loadBinaryFunction(...)

loadFunction(...)

loadUnaryFunction(...)

makeColumn(*x)

makeMatrix(...)

makeRow(*x)

maskMultiply(...)

matchAffine(...)

matchMaskAffine(...)

matchPolynomial(...)

matchRatioAffine(...)

matchRigid(...)

maximum(...)

minimum(...)

movingWindowsDetrend(...)

normalize(...)

regressionLine(...)

resliceAffine(...)

reslicePolynomial(...)

resliceRigid(...)

showheaders(m)

slowDftPower(...)

sort(...)

spacepoints(x)

squash(...)

svd(...)

toManifold(...)

tr(x)

trace(...)

transform(...)

transform2(...)

volumes(x)

Data
		ALPHA = 1 FLAG = 0 NUMERIC = 2 Trilinear = 0 __file__ = '/usr/lib/python2.2/site-packages/RUMBA.py' __name__ = 'RUMBA'

We would like to thank the McDonnell Foundation for their generous support of RUMBA through grants awarded in 2000 and 2003.
RUMBA is also partially funded by the National Science Foundation under Grant No. 0205178.
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the McDonell Foundation or the National Science Foundation.