Get function for fascicle evaluation structure
val = feGet(fe,param,varargin)
INPUTS:
Coords - Nx3 set of coordinates in image space
voxelIndices - Vector of 1's and 0's, there is a one for each
location the the connectome coordinates for which there is
a match in the coords
Copyright (2013-2014), Franco Pestilli & Brian Wandell,
Stanford University, pestillifranco@gmail.com.
---------- List of arguments ----
Name of the current fe structure.
name = feGet(fe,'name')
----------
The type of objes (always, fascicle evaluation)
type = feGet(fe,'type')
----------
Structure of parameters and results from LIFE analysis
life = feGet(fe,'life')
----------
Return the connectome (fiber group) in image coordinates.
fg = feGEt(fe,'fg img')
----------
return the connectome in acpc coordinates.
fg = feGet(fg,'fg acpc')
----------
Return the VOI comprised by the full connectome.
roi = feGet(fe,'roi')
----------
Transforms between coords for fg, roi, and dwi data
xform = feGet(fe,'xform')
----------
Load the diffusion weighted data.
dwi = feGet(fe,'dwi')
----------
Return the path to the diffusion weighted data file.
dwiFile = feGet(fe,'dwifile')
----------
Load the repeated-measure diffusion weighted data.
dwi = feGet(fe,'dwirepeat')
----------
Load the repeated measure of the diffusion weighted data.
dwiFile = feGet(fe,'dwirepeatfile')
----------
Directory where the LiFe structes are saved by defualt.
sdir = feGet(fe,'savedir');
----------
Diffusion directions.
val = feGet(fe,'bvecs');
----------
Indices to the diffusion directions in the DWi 4th Dimension.
val = feGet(fe,'bvecs indices');
----------
Number of B0's
val = feGet(fe,'n bvals');
----------
B0 Values.
bval = feGet(fe,'bvals')
----------
Returns a nVoxels X nBvecs array of measured diffusion signal
val = feGet(fe,'dsiinvox');
val = feGet(fe,'dsiinvox',voxelsIndices);
val = feGet(fe,'dsiinvox',coords);
----------
Returns a nVoxels X nBvecs array of demeaned diffusion signal
val =feGet(fe,'dsiinvoxdemeaned');
val =feGet(fe,'dsiinvoxdemeaned',voxelsIndices);
val =feGet(fe,'dsiinvoxdemeaned',coords);
----------
Get the diffusion signal at 0 diffusion weighting (B0) for this voxel
val = feGet(fe,'b0signalvoxel');
val = feGet(fe,'b0signalvoxel',voxelIndex);
val = feGet(fe,'b0signalvoxel',coords);
----------
Return a subset of fibers from the conenctome.
fg = feGet(fe,'fiberssubset',fiberList);
----------
Return the coordinates of the VOI comprised in the connectome.
Always in image space.
coords = feGet(fe,'roi coords')
----------
Number of voxels in the connectome/VOI.
nVoxels = feGet(fe,'n voxels')
----------
Indexes of actually used voxels.
These can be different than the number of voxels in the VOI in case some
voxels have no fibers in them.
TO BE DEPRECATED.
indxUsedVx = feGet(fe,'index of used voxels')
----------
Number of actually used voxels.
This can be less than the number of voxels in the VOI in case some
voxels have no fibers in them.
TO BE DEPRECATED.
nUsedVx = feGet(fe,'index of used voxels')
----------
Returns the first index to a voxel in the dSig vector.
The dSig vector is 1:nBvecs*nVoxels.
val = feGet(fe,'start rows')
----------
Return the rows corresponding to a set of voxels.
rows = feGet(fe,'voxel rows',idxVoxel)
----------
Return the model (M matrix), or a subset of it.
Mfiber = feGet(fe,'M fiber');
Mfiber = feGet(fe,'model',voxelsIndices);
Mfiber = feGet(fe,'model',coords);
----------
Return a subset of measurements from the fiber portion of M matrix
The rows of M are specified. Remember that the rows refer to a
combination of voxel and measurement direction and potentially in the
future a b-value. Hence, the rows are not specific to a voxel.
Mfiber = feGet(fe,'mfiber subset',rowsToKeep);
----------
Pairing of fibers and nodes in each voxel.
v2fnp = feGet(fe,'v2fnp');
----------
Tensors computed for each node and fiber in a set of voxels.
t = feGet(fe,'tensors')
t = feGet(fe,'tensors',voxelsIndices);
t = feGet(fe,'tensors',coords);
----------
Get all the tensors in a single voxel.
val = feGet(fe,'voxtensors',voxelIndex)
val = feGet(fe,'voxtensors',coord)
----------
Returns the model (M) with only a subset of columns (fibers).
fiberList is a vector of indexes, e.g, [1 10 100]
M = feGet(fe,'keepfibers',fiberList)
----------
Isotropic portion of M matrix. Returns the matrix for the full model or
for a subset of voxels.
Miso = feGet(fe,'M iso')
Miso = feGet(fe,'M iso',voxelsIndices)
Miso = feGet(fe,'M iso',coords)
----------
Total number of nodes in each voxel The voxel2FN pairs are row size
equals number of nodes and column size is always 2. The entries of
the first column are the fiber number. The entries in the second
column are the node of the fiber inside the voxel.
val = feGet(fe,'n nodes');
val = feGet(fe,'n nodes',voxelsIndices);
val = feGet(fe,'n nodes',coords);
----------
Return the total number of fibers for all the voxels or in a set of
voxels.
nFibers = feGet(fe,'uniquefnum'); % all the voxels
nFibers = feGet(fe,'uniquefnum',[1 2 3 4]); % for some the voxels,
% specified by indexes
nFibers = feGet(fe,'uniquefnum',coords); % for some the voxels,
% specified by coordinates
----------
Return the indexes of the fibers for all the voxels or in a set of
voxels.
idxFibers = feGet(fe,'uniquef'); % all the voxels
idxFibers = feGet(fe,'uniquef',[1 2 3 4]); % for some the voxels,
% specified by indexes
idxFibers = feGet(fe,'uniquef',coords); % for some the voxels,
% specified by coordinates
----------
Return the total number of fibers for all the voxels or in a set of
voxels
nFibers = feGet(fe,'totfnum'); % all the voxels
nFibers = feGet(fe,'totfnum',[1 2 3 4]); % for some the voxels,
% specified by indexes
nFibers = feGet(fe,'totfnum',coords); % for some the voxels,
% specified by coordinates
----------
Return the indexes of the fibers for all the voxels or in a set of
voxels
idxFibers = feGet(fe,'totf'); % all the voxels
idxFibers = feGet(fe,'totf',[1 2 3 4]); % for some the voxels,
% specified by indexes
idxFibers = feGet(fe,'totfibers',coords); % for some the voxels,
% specified by coordinates
----------
Return the number of fibers in the model.
nFibers = feGet(fe,'n fibers');
----------
Weights of the isotropic voxel signals, this is the mean signal in
each voxel.
val = feGet(fe,'iso weights');
val = feGet(fe,'iso weights'coords)
val = feGet(fe,'iso weights',voxelIndices)
----------
Weights of the fiber component. For all the fibers or a subset of
them.
w = feGet(fe,'fiber weights')
w = feGet(fe,'fiber weights',fiberIndices)
----------
Weights of the fiber component with a subset of the fibers' weights
set to zero. This can be used to test the loss in RMSE for the
connectome when a subset of fibers is removed, but the connectome is
not fitted again.
w = feGet(fe,'fiber weights test',fiberIndices)
----------
Predicted signal (demeaned) with a subset of fibers' weights set to 0.
This can be used to test the loss in RMSE for the connectome when a
subset of fibers is removed, but the connectome is not fitted again.
pSig = feGet(fe,'pSig fiber test',fiberIndices);
pSig = feGet(fe,'pSig fiber test',fiberIndices,voxelIndices);
pSig = feGet(fe,'pSig fiber test',fiberIndices,coords);
----------
Measured signal in VOI, this is the raw signal. not demeaned
dSig = feGet(fe,'dSig full')
---------
Measured signal in VOI, demeaned, this is the signal used for the
fiber-portion of the M model.
dSig = feGet(fe,'dsigdemeaned');
dSig = feGet(fe,'dsigdemeaned',[1 10 100]);
dSig = feGet(fe,'dsigdemeaned',coords);
---------
Get the demeaned signal for a subset of rows.
Useful for cross-validation.
dSig = feGet(fe,'dsigrowssubset',voxelsList);
---------
Predicted signal of fiber alone (demeaned).
pSig = feGet(fe,'pSig fiber');
pSig = feGet(fe,'pSig fiber',coords);
pSig = feGet(fe,'pSig fiber',voxelIndices);
---------
Predicted measured signal from both fiber and iso
pSig = feGet(fe,'pSig full')
pSig = feGet(fe,'pSig full',coords)
pSig = feGet(fe,'pSig full',voxelIndices)
---------
The woxels returned by a fit of LiFE by voxel/fiber
w = feGet(fe,'fiberweightsvoxelwise')
---------
Predict the diffusion signal for the fiber component
with the voxel-wise fit of LiFE
pSig = feGet(fe,'psigfvoxelwise')
pSig = feGet(fe,'psigfvoxelwise',coords)
pSig = feGet(fe,'psigfvoxelwise',voxelIndices)
---------
Predict the diffusion signal for the fiber component
with the voxel-wise fit of LiFE, return the an array of pSigXnVoxel
pSig = feGet(fe,'psigfvoxelwisebyvoxel')
pSig = feGet(fe,'psigfvoxelwisebyvoxel',coords)
pSig = feGet(fe,'psigfvoxelwisebyvoxel',voxelIndices)
---------
The fiber and isotropic weights as a long vector
w = feGet(fe,'fullweights')
---------
Return the global R2 (fraction of variance explained) of the full life
model.
R2 = feGet(fe,'total r2');
---------
Percent variance explained
R2 = feGet(fe,'explained variance')
---------
Root mean squared error of the LiFE fit to the whole data
rmse = feGet(fe,'rmse')
---------
Residual signal: (fiber prediction - measured_demeaned).
res = feGet(fe,'res sig fiber')
---------
Residual signal: (full model prediction - measured signal).
res = feGet(fe,'res sig full');
---------
Residual signal: (fiber model prediction - demeaned measured signal)
with added mean signal. Res is returned as a vector.
This is used to reconstruct an image (volume) to be used for the
refinement process.
res = feGet(fe,'fiber res sig with mean');
res = feGet(fe,'fiber res sig with mean',coords);
res = feGet(fe,'fiber res sig with mean',voxelIndices);
---------
Residual signal fiber model prediction - demeaned measured signal
with added mean signal. Res is returned as an array (nBvecs x nVoxel).
This is used to reconstruct an image (volume) to be used for the
refinement process.
res = feGet(fe,'fiber res sig with mean voxel');
res = feGet(fe,'fiber res sig with mean voxel',coords);
res = feGet(fe,'fiber res sig with mean voxel',voxelIndices);
---------
Predicted signal by the fiber model in a set of voxels.
Vox subfield stores per voxel within the VOI. The pSig has size of the
dwi. - It is stored as val = pSig(X,Y,Z,Theta)
pSig = feGet(fe, 'pSig fiber by voxel');
pSig = feGet(fe, 'pSig fiber by voxel',coords);
pSig = feGet(fe, 'pSig fiber by voxel',voxelIndex);
---------
Return a column vector of the proportion of variance explained in
each voxel.
R2byVox = feGet(fe,'voxr2');
R2byVox = feGet(fe,'voxr2',coords);
---------
Return a column vector of the proportion of variance explained in
each voxel. (Normalized to the squared mean diffusion signal in each voxel)
R2byVox = feGet(fe,'voxr2zero');
R2byVox = feGet(fe,'voxr2zero',coords);
---------
Return the percent of varince explained in each voxel.
R2byVox = feGet(fe,'var exp by voxel');
R2byVox = feGet(fe,'var exp by voxel',coords);
---------
Demeaned diffusion signal in each voxel.
dSigByVoxel = feGet(fe,'dsigdemeaned by Voxel');
dSigByVoxel = feGet(fe,'dsigdemeaned by Voxel',coords);
dSigByVoxel = feGet(fe,'dsigdemeaned by Voxel',vxIndex);
---------
Full (measured) signal in each voxel.
dSigByVoxel = feGet(fe,'dSig full by Voxel');
dSigByVoxel = feGet(fe,'dSig full by Voxel',coords);
dSigByVoxel = feGet(fe,'dSig full by Voxel',vxIndex);
---------
Predicted signal by the full model in a set of voxeles.
pSigByVoxel = feGet(fe, 'pSig full by voxel');
pSigByVoxel = feGet(fe, 'pSig full by voxel',coords);
pSigByVoxel = feGet(fe, 'pSig full by voxel',voxelIndex);
---------
A volume of RMSE values.
RMSE = feGet(fe,'vox rmse')
RMSE = feGet(fe,'vox rmse',coords)
RMSE = feGet(fe,'vox rmse',vxIndex)
---------
A volume of RMSE values with a subset of fibers' weights set to 0.
RMSE = feGet(fe,'vox rmse',fiberIndices)
RMSE = feGet(fe,'vox rmse',fiberIndices,coords)
RMSE = feGet(fe,'vox rmse',fiberIndices,voxelIndex)
---------
Fibers' residual signal by voxel.
res = feGet(fe,'res sig fiber vox')
res = feGet(fe,'res sig fiber vox',coords)
res = feGet(fe,'res sig fiber vox',vxIndex)
---------
Full (measured) residual signal by voxel.
res = feGet(fe,'res sig full vox')
res = feGet(fe,'res sig full vox',coords)
res = feGet(fe,'res sig full vox',vxIndex)
---------
Fibers' residual signal by voxel with mean signal added (with added
isotropic component).
This is used to compute the residual signal for the refinement.
res = feGet(fe,'fiber res sig with mean vox')
res = feGet(fe,'fiber res sig with mean vox',coords)
res = feGet(fe,'fiber res sig with mean vox',vxIndex)
---------
Residual signal full model prediction from a multi-voxel fit.
res = feGet(fe,'res sig full voxfit');
res = feGet(fe, 'res sig full voxfit',coords);
res = feGet(fe, 'res sig full voxfit',voxelIndex);
---------
Fiber density statistics.
Computes the fiber density (how many fibers in each voxel)
We compute the following values:
(1) The number of fibers in each voxel
(2) The number of unique fibers with non-zero weights
(3) The sum of the weights in each voxel
(4) The mean of the weights in each voxel
(5) The variance of the weigths in each voxel
pSig = feGet(fe,'fiberdensity');
---------
Given an VOI or a set of indices to voxels returns the indices of the matching voxels inside the big
volume, which ordinarily represents the full connectome.
voxelIndices = feGet(fe,'voxelsindices',coords)
IMPORTANT: Size(coords) must be Nx3;
voxelIndices = feGet(fe,'voxelsindices',voxelIndices)
IMPORTANT: Indices MUST be a column vector for this
code to work. Size(voxelIndices) = Nx1;
---------
Given an VOI finds the indices of the matching voxels inside the big
volume, which ordinarily represents the full connectome.
foundVoxels = feGet(fe,'find voxels',coords)
coords is a Nx3 set of coordinates in image space
foundVoxels is a vector of 1's and 0's, there is a one for each
location the the connectome coordinates for which there is a match in
the coords
---------
Given a set of VOI coords finds the row numbers of the Model matrix
(or equivalently the dSig vector) that represent the data for this
set of VOI coords.
foundVoxels = feGet(fe,'coords 2 rows',coords)
coords - a Nx3 set of coordinates in image space
foundVoxels - a binary vector that is 1 for each Model matrix row
that corresponds to at least one of the coords.
---------
Quaternian transformation from IMAGE space to ACPC space.
xform = feGet(fe,'xform')
---------
Quaternian transformation from ACPC space to IMAGE space.
xform = feGet(fe,'xform')
---------
Dimensions of the DW volume.
dim = feGet(fe,'dims')
---------
Dimensions of the maps of parameters and results.
dims = feGet(fe, 'mapsize')
---------
Path to the 3D Anatomy Volume.
anatomyfile = feGet(fe, 't1 file')
End of feGet.m parameters,
Copyright (2013-2014), Franco Pestilli, Stanford University, pestillifranco@gmail.com.0001 function val = feGet(fe,param,varargin) 0002 % Get function for fascicle evaluation structure 0003 % 0004 % val = feGet(fe,param,varargin) 0005 % 0006 % 0007 % INPUTS: 0008 % Coords - Nx3 set of coordinates in image space 0009 % voxelIndices - Vector of 1's and 0's, there is a one for each 0010 % location the the connectome coordinates for which there is 0011 % a match in the coords 0012 % 0013 % 0014 % Copyright (2013-2014), Franco Pestilli & Brian Wandell, 0015 % Stanford University, pestillifranco@gmail.com. 0016 % 0017 % 0018 %---------- List of arguments ---- 0019 % Name of the current fe structure. 0020 % name = feGet(fe,'name') 0021 %---------- 0022 % The type of objes (always, fascicle evaluation) 0023 % type = feGet(fe,'type') 0024 %---------- 0025 % Structure of parameters and results from LIFE analysis 0026 % life = feGet(fe,'life') 0027 %---------- 0028 % Return the connectome (fiber group) in image coordinates. 0029 % fg = feGEt(fe,'fg img') 0030 %---------- 0031 % return the connectome in acpc coordinates. 0032 % fg = feGet(fg,'fg acpc') 0033 %---------- 0034 % Return the VOI comprised by the full connectome. 0035 % roi = feGet(fe,'roi') 0036 %---------- 0037 % Transforms between coords for fg, roi, and dwi data 0038 % xform = feGet(fe,'xform') 0039 %---------- 0040 % Load the diffusion weighted data. 0041 % dwi = feGet(fe,'dwi') 0042 %---------- 0043 % Return the path to the diffusion weighted data file. 0044 % dwiFile = feGet(fe,'dwifile') 0045 %---------- 0046 % Load the repeated-measure diffusion weighted data. 0047 % dwi = feGet(fe,'dwirepeat') 0048 %---------- 0049 % Load the repeated measure of the diffusion weighted data. 0050 % dwiFile = feGet(fe,'dwirepeatfile') 0051 %---------- 0052 % Directory where the LiFe structes are saved by defualt. 0053 % sdir = feGet(fe,'savedir'); 0054 %---------- 0055 % Diffusion directions. 0056 % val = feGet(fe,'bvecs'); 0057 %---------- 0058 % Indices to the diffusion directions in the DWi 4th Dimension. 0059 % val = feGet(fe,'bvecs indices'); 0060 %---------- 0061 % Number of B0's 0062 % val = feGet(fe,'n bvals'); 0063 %---------- 0064 % B0 Values. 0065 % bval = feGet(fe,'bvals') 0066 %---------- 0067 % Returns a nVoxels X nBvecs array of measured diffusion signal 0068 % val = feGet(fe,'dsiinvox'); 0069 % val = feGet(fe,'dsiinvox',voxelsIndices); 0070 % val = feGet(fe,'dsiinvox',coords); 0071 %---------- 0072 % Returns a nVoxels X nBvecs array of demeaned diffusion signal 0073 % val =feGet(fe,'dsiinvoxdemeaned'); 0074 % val =feGet(fe,'dsiinvoxdemeaned',voxelsIndices); 0075 % val =feGet(fe,'dsiinvoxdemeaned',coords); 0076 %---------- 0077 % Get the diffusion signal at 0 diffusion weighting (B0) for this voxel 0078 % val = feGet(fe,'b0signalvoxel'); 0079 % val = feGet(fe,'b0signalvoxel',voxelIndex); 0080 % val = feGet(fe,'b0signalvoxel',coords); 0081 %---------- 0082 % Return a subset of fibers from the conenctome. 0083 % fg = feGet(fe,'fiberssubset',fiberList); 0084 %---------- 0085 % Return the coordinates of the VOI comprised in the connectome. 0086 % Always in image space. 0087 % coords = feGet(fe,'roi coords') 0088 %---------- 0089 % Number of voxels in the connectome/VOI. 0090 % nVoxels = feGet(fe,'n voxels') 0091 %---------- 0092 % Indexes of actually used voxels. 0093 % These can be different than the number of voxels in the VOI in case some 0094 % voxels have no fibers in them. 0095 % 0096 % TO BE DEPRECATED. 0097 % indxUsedVx = feGet(fe,'index of used voxels') 0098 %---------- 0099 % Number of actually used voxels. 0100 % This can be less than the number of voxels in the VOI in case some 0101 % voxels have no fibers in them. 0102 % 0103 % TO BE DEPRECATED. 0104 % nUsedVx = feGet(fe,'index of used voxels') 0105 %---------- 0106 % Returns the first index to a voxel in the dSig vector. 0107 % The dSig vector is 1:nBvecs*nVoxels. 0108 % val = feGet(fe,'start rows') 0109 %---------- 0110 % Return the rows corresponding to a set of voxels. 0111 % rows = feGet(fe,'voxel rows',idxVoxel) 0112 %---------- 0113 % Return the model (M matrix), or a subset of it. 0114 % Mfiber = feGet(fe,'M fiber'); 0115 % Mfiber = feGet(fe,'model',voxelsIndices); 0116 % Mfiber = feGet(fe,'model',coords); 0117 %---------- 0118 % Return a subset of measurements from the fiber portion of M matrix 0119 % The rows of M are specified. Remember that the rows refer to a 0120 % combination of voxel and measurement direction and potentially in the 0121 % future a b-value. Hence, the rows are not specific to a voxel. 0122 % Mfiber = feGet(fe,'mfiber subset',rowsToKeep); 0123 %---------- 0124 % Pairing of fibers and nodes in each voxel. 0125 % v2fnp = feGet(fe,'v2fnp'); 0126 %---------- 0127 % Tensors computed for each node and fiber in a set of voxels. 0128 % t = feGet(fe,'tensors') 0129 % t = feGet(fe,'tensors',voxelsIndices); 0130 % t = feGet(fe,'tensors',coords); 0131 %---------- 0132 % Get all the tensors in a single voxel. 0133 % val = feGet(fe,'voxtensors',voxelIndex) 0134 % val = feGet(fe,'voxtensors',coord) 0135 %---------- 0136 % Returns the model (M) with only a subset of columns (fibers). 0137 % fiberList is a vector of indexes, e.g, [1 10 100] 0138 % M = feGet(fe,'keepfibers',fiberList) 0139 %---------- 0140 % Isotropic portion of M matrix. Returns the matrix for the full model or 0141 % for a subset of voxels. 0142 % Miso = feGet(fe,'M iso') 0143 % Miso = feGet(fe,'M iso',voxelsIndices) 0144 % Miso = feGet(fe,'M iso',coords) 0145 %---------- 0146 % Total number of nodes in each voxel The voxel2FN pairs are row size 0147 % equals number of nodes and column size is always 2. The entries of 0148 % the first column are the fiber number. The entries in the second 0149 % column are the node of the fiber inside the voxel. 0150 % val = feGet(fe,'n nodes'); 0151 % val = feGet(fe,'n nodes',voxelsIndices); 0152 % val = feGet(fe,'n nodes',coords); 0153 %---------- 0154 % Return the total number of fibers for all the voxels or in a set of 0155 % voxels. 0156 % nFibers = feGet(fe,'uniquefnum'); % all the voxels 0157 % nFibers = feGet(fe,'uniquefnum',[1 2 3 4]); % for some the voxels, 0158 % % specified by indexes 0159 % nFibers = feGet(fe,'uniquefnum',coords); % for some the voxels, 0160 % % specified by coordinates 0161 %---------- 0162 % Return the indexes of the fibers for all the voxels or in a set of 0163 % voxels. 0164 % idxFibers = feGet(fe,'uniquef'); % all the voxels 0165 % idxFibers = feGet(fe,'uniquef',[1 2 3 4]); % for some the voxels, 0166 % % specified by indexes 0167 % idxFibers = feGet(fe,'uniquef',coords); % for some the voxels, 0168 % % specified by coordinates 0169 %---------- 0170 % Return the total number of fibers for all the voxels or in a set of 0171 % voxels 0172 % nFibers = feGet(fe,'totfnum'); % all the voxels 0173 % nFibers = feGet(fe,'totfnum',[1 2 3 4]); % for some the voxels, 0174 % % specified by indexes 0175 % nFibers = feGet(fe,'totfnum',coords); % for some the voxels, 0176 % % specified by coordinates 0177 %---------- 0178 % Return the indexes of the fibers for all the voxels or in a set of 0179 % voxels 0180 % idxFibers = feGet(fe,'totf'); % all the voxels 0181 % idxFibers = feGet(fe,'totf',[1 2 3 4]); % for some the voxels, 0182 % % specified by indexes 0183 % idxFibers = feGet(fe,'totfibers',coords); % for some the voxels, 0184 % % specified by coordinates 0185 %---------- 0186 % Return the number of fibers in the model. 0187 % nFibers = feGet(fe,'n fibers'); 0188 %---------- 0189 % Weights of the isotropic voxel signals, this is the mean signal in 0190 % each voxel. 0191 % val = feGet(fe,'iso weights'); 0192 % val = feGet(fe,'iso weights'coords) 0193 % val = feGet(fe,'iso weights',voxelIndices) 0194 %---------- 0195 % Weights of the fiber component. For all the fibers or a subset of 0196 % them. 0197 % w = feGet(fe,'fiber weights') 0198 % w = feGet(fe,'fiber weights',fiberIndices) 0199 %---------- 0200 % Weights of the fiber component with a subset of the fibers' weights 0201 % set to zero. This can be used to test the loss in RMSE for the 0202 % connectome when a subset of fibers is removed, but the connectome is 0203 % not fitted again. 0204 % w = feGet(fe,'fiber weights test',fiberIndices) 0205 %---------- 0206 % Predicted signal (demeaned) with a subset of fibers' weights set to 0. 0207 % This can be used to test the loss in RMSE for the connectome when a 0208 % subset of fibers is removed, but the connectome is not fitted again. 0209 % 0210 % pSig = feGet(fe,'pSig fiber test',fiberIndices); 0211 % pSig = feGet(fe,'pSig fiber test',fiberIndices,voxelIndices); 0212 % pSig = feGet(fe,'pSig fiber test',fiberIndices,coords); 0213 %---------- 0214 % Measured signal in VOI, this is the raw signal. not demeaned 0215 % 0216 % dSig = feGet(fe,'dSig full') 0217 %--------- 0218 % Measured signal in VOI, demeaned, this is the signal used for the 0219 % fiber-portion of the M model. 0220 % 0221 % dSig = feGet(fe,'dsigdemeaned'); 0222 % dSig = feGet(fe,'dsigdemeaned',[1 10 100]); 0223 % dSig = feGet(fe,'dsigdemeaned',coords); 0224 %--------- 0225 % Get the demeaned signal for a subset of rows. 0226 % Useful for cross-validation. 0227 % dSig = feGet(fe,'dsigrowssubset',voxelsList); 0228 %--------- 0229 % Predicted signal of fiber alone (demeaned). 0230 % pSig = feGet(fe,'pSig fiber'); 0231 % pSig = feGet(fe,'pSig fiber',coords); 0232 % pSig = feGet(fe,'pSig fiber',voxelIndices); 0233 %--------- 0234 % Predicted measured signal from both fiber and iso 0235 % pSig = feGet(fe,'pSig full') 0236 % pSig = feGet(fe,'pSig full',coords) 0237 % pSig = feGet(fe,'pSig full',voxelIndices) 0238 %--------- 0239 % The woxels returned by a fit of LiFE by voxel/fiber 0240 % w = feGet(fe,'fiberweightsvoxelwise') 0241 %--------- 0242 % Predict the diffusion signal for the fiber component 0243 % with the voxel-wise fit of LiFE 0244 % pSig = feGet(fe,'psigfvoxelwise') 0245 % pSig = feGet(fe,'psigfvoxelwise',coords) 0246 % pSig = feGet(fe,'psigfvoxelwise',voxelIndices) 0247 %--------- 0248 % Predict the diffusion signal for the fiber component 0249 % with the voxel-wise fit of LiFE, return the an array of pSigXnVoxel 0250 % pSig = feGet(fe,'psigfvoxelwisebyvoxel') 0251 % pSig = feGet(fe,'psigfvoxelwisebyvoxel',coords) 0252 % pSig = feGet(fe,'psigfvoxelwisebyvoxel',voxelIndices) 0253 %--------- 0254 % The fiber and isotropic weights as a long vector 0255 % w = feGet(fe,'fullweights') 0256 %--------- 0257 % Return the global R2 (fraction of variance explained) of the full life 0258 % model. 0259 % R2 = feGet(fe,'total r2'); 0260 %--------- 0261 % Percent variance explained 0262 % R2 = feGet(fe,'explained variance') 0263 %--------- 0264 % Root mean squared error of the LiFE fit to the whole data 0265 % rmse = feGet(fe,'rmse') 0266 %--------- 0267 % Residual signal: (fiber prediction - measured_demeaned). 0268 % res = feGet(fe,'res sig fiber') 0269 %--------- 0270 % Residual signal: (full model prediction - measured signal). 0271 % res = feGet(fe,'res sig full'); 0272 %--------- 0273 % Residual signal: (fiber model prediction - demeaned measured signal) 0274 % with added mean signal. Res is returned as a vector. 0275 % This is used to reconstruct an image (volume) to be used for the 0276 % refinement process. 0277 % res = feGet(fe,'fiber res sig with mean'); 0278 % res = feGet(fe,'fiber res sig with mean',coords); 0279 % res = feGet(fe,'fiber res sig with mean',voxelIndices); 0280 %--------- 0281 % Residual signal fiber model prediction - demeaned measured signal 0282 % with added mean signal. Res is returned as an array (nBvecs x nVoxel). 0283 % This is used to reconstruct an image (volume) to be used for the 0284 % refinement process. 0285 % res = feGet(fe,'fiber res sig with mean voxel'); 0286 % res = feGet(fe,'fiber res sig with mean voxel',coords); 0287 % res = feGet(fe,'fiber res sig with mean voxel',voxelIndices); 0288 %--------- 0289 % Predicted signal by the fiber model in a set of voxels. 0290 % Vox subfield stores per voxel within the VOI. The pSig has size of the 0291 % dwi. - It is stored as val = pSig(X,Y,Z,Theta) 0292 % 0293 % pSig = feGet(fe, 'pSig fiber by voxel'); 0294 % pSig = feGet(fe, 'pSig fiber by voxel',coords); 0295 % pSig = feGet(fe, 'pSig fiber by voxel',voxelIndex); 0296 %--------- 0297 % Return a column vector of the proportion of variance explained in 0298 % each voxel. 0299 % R2byVox = feGet(fe,'voxr2'); 0300 % R2byVox = feGet(fe,'voxr2',coords); 0301 %--------- 0302 % Return a column vector of the proportion of variance explained in 0303 % each voxel. (Normalized to the squared mean diffusion signal in each voxel) 0304 % R2byVox = feGet(fe,'voxr2zero'); 0305 % R2byVox = feGet(fe,'voxr2zero',coords); 0306 %--------- 0307 % Return the percent of varince explained in each voxel. 0308 % R2byVox = feGet(fe,'var exp by voxel'); 0309 % R2byVox = feGet(fe,'var exp by voxel',coords); 0310 %--------- 0311 % Demeaned diffusion signal in each voxel. 0312 % dSigByVoxel = feGet(fe,'dsigdemeaned by Voxel'); 0313 % dSigByVoxel = feGet(fe,'dsigdemeaned by Voxel',coords); 0314 % dSigByVoxel = feGet(fe,'dsigdemeaned by Voxel',vxIndex); 0315 %--------- 0316 % Full (measured) signal in each voxel. 0317 % dSigByVoxel = feGet(fe,'dSig full by Voxel'); 0318 % dSigByVoxel = feGet(fe,'dSig full by Voxel',coords); 0319 % dSigByVoxel = feGet(fe,'dSig full by Voxel',vxIndex); 0320 %--------- 0321 % Predicted signal by the full model in a set of voxeles. 0322 % pSigByVoxel = feGet(fe, 'pSig full by voxel'); 0323 % pSigByVoxel = feGet(fe, 'pSig full by voxel',coords); 0324 % pSigByVoxel = feGet(fe, 'pSig full by voxel',voxelIndex); 0325 %--------- 0326 % A volume of RMSE values. 0327 % RMSE = feGet(fe,'vox rmse') 0328 % RMSE = feGet(fe,'vox rmse',coords) 0329 % RMSE = feGet(fe,'vox rmse',vxIndex) 0330 %--------- 0331 % A volume of RMSE values with a subset of fibers' weights set to 0. 0332 % RMSE = feGet(fe,'vox rmse',fiberIndices) 0333 % RMSE = feGet(fe,'vox rmse',fiberIndices,coords) 0334 % RMSE = feGet(fe,'vox rmse',fiberIndices,voxelIndex) 0335 %--------- 0336 % Fibers' residual signal by voxel. 0337 % res = feGet(fe,'res sig fiber vox') 0338 % res = feGet(fe,'res sig fiber vox',coords) 0339 % res = feGet(fe,'res sig fiber vox',vxIndex) 0340 %--------- 0341 % Full (measured) residual signal by voxel. 0342 % res = feGet(fe,'res sig full vox') 0343 % res = feGet(fe,'res sig full vox',coords) 0344 % res = feGet(fe,'res sig full vox',vxIndex) 0345 %--------- 0346 % Fibers' residual signal by voxel with mean signal added (with added 0347 % isotropic component). 0348 % This is used to compute the residual signal for the refinement. 0349 % res = feGet(fe,'fiber res sig with mean vox') 0350 % res = feGet(fe,'fiber res sig with mean vox',coords) 0351 % res = feGet(fe,'fiber res sig with mean vox',vxIndex) 0352 %--------- 0353 % Residual signal full model prediction from a multi-voxel fit. 0354 % res = feGet(fe,'res sig full voxfit'); 0355 % res = feGet(fe, 'res sig full voxfit',coords); 0356 % res = feGet(fe, 'res sig full voxfit',voxelIndex); 0357 %--------- 0358 % Fiber density statistics. 0359 % Computes the fiber density (how many fibers in each voxel) 0360 % We compute the following values: 0361 % (1) The number of fibers in each voxel 0362 % (2) The number of unique fibers with non-zero weights 0363 % (3) The sum of the weights in each voxel 0364 % (4) The mean of the weights in each voxel 0365 % (5) The variance of the weigths in each voxel 0366 % pSig = feGet(fe,'fiberdensity'); 0367 %--------- 0368 % Given an VOI or a set of indices to voxels returns the indices of the matching voxels inside the big 0369 % volume, which ordinarily represents the full connectome. 0370 % voxelIndices = feGet(fe,'voxelsindices',coords) 0371 % IMPORTANT: Size(coords) must be Nx3; 0372 % voxelIndices = feGet(fe,'voxelsindices',voxelIndices) 0373 % IMPORTANT: Indices MUST be a column vector for this 0374 % code to work. Size(voxelIndices) = Nx1; 0375 %--------- 0376 % Given an VOI finds the indices of the matching voxels inside the big 0377 % volume, which ordinarily represents the full connectome. 0378 % 0379 % foundVoxels = feGet(fe,'find voxels',coords) 0380 % 0381 % coords is a Nx3 set of coordinates in image space 0382 % foundVoxels is a vector of 1's and 0's, there is a one for each 0383 % location the the connectome coordinates for which there is a match in 0384 % the coords 0385 %--------- 0386 % Given a set of VOI coords finds the row numbers of the Model matrix 0387 % (or equivalently the dSig vector) that represent the data for this 0388 % set of VOI coords. 0389 % foundVoxels = feGet(fe,'coords 2 rows',coords) 0390 % coords - a Nx3 set of coordinates in image space 0391 % foundVoxels - a binary vector that is 1 for each Model matrix row 0392 % that corresponds to at least one of the coords. 0393 %--------- 0394 % Quaternian transformation from IMAGE space to ACPC space. 0395 % xform = feGet(fe,'xform') 0396 %--------- 0397 % Quaternian transformation from ACPC space to IMAGE space. 0398 % xform = feGet(fe,'xform') 0399 %--------- 0400 % Dimensions of the DW volume. 0401 % dim = feGet(fe,'dims') 0402 %--------- 0403 % Dimensions of the maps of parameters and results. 0404 % dims = feGet(fe, 'mapsize') 0405 %--------- 0406 % Path to the 3D Anatomy Volume. 0407 % anatomyfile = feGet(fe, 't1 file') 0408 % 0409 % End of feGet.m parameters, 0410 % 0411 % Copyright (2013-2014), Franco Pestilli, Stanford University, pestillifranco@gmail.com. 0412 0413 val = []; 0414 0415 % Format the input parameters. 0416 param = lower(strrep(param,' ','')); 0417 0418 % Start sorting the input and computing the output. 0419 switch param 0420 case 'name' 0421 % Name of the current fe structure. 0422 % 0423 % name = feGet(fe,'name') 0424 val = fe.name; 0425 0426 case 'type' 0427 % The type of objes (always, fascicle evaluation) 0428 % 0429 % type = feGet(fe,'type') 0430 val = fe.type; % Always fascicle evaluation 0431 0432 % Set top level structure, not just single slot 0433 case 'life' 0434 % Structure of parameters and results from LIFE analysis 0435 % 0436 % life = feGet(fe,'life') 0437 val = fe.life; 0438 0439 case {'fgimg','fibersimg'} 0440 % Return the connectome (fiber group) in image coordinates. 0441 % 0442 % fg = feGet(fe,'fg img') 0443 val = fe.fg; % Fiber group candidate fascicles, uses fgSet/Get 0444 0445 case {'fgacpc','fibersacpc'} 0446 % return the connectome in acpc coordinates. 0447 % 0448 % fg = feGet(fg,'fg acpc') 0449 xform = feGet(fe,'img2acpcxform'); 0450 val = dtiXformFiberCoords(feGet(fe,'fgimg'),xform,'acpc'); 0451 0452 case 'roi' 0453 % Return the VOI comprised by the full connectome. 0454 % 0455 % roi = feGet(fe,'roi') 0456 val = fe.roi; 0457 0458 case 'roiacpc' 0459 % Return the VOI comprised by the full connectome. 0460 % 0461 % roi = feGet(fe,'roi acpc') 0462 name = sprintf('roi_%s',fe.fg.name); 0463 randColor = rand(1,3); 0464 val = dtiNewRoi(name,randColor,fefgGet(feGet(fe,'fg acpc'),'unique image coords')); 0465 0466 case 'xform' 0467 % Transforms between coords for fg, roi, and dwi data 0468 % 0469 % xform = feGet(fe,'xform') 0470 val = fe.life.xform; 0471 0472 case 'dwi' 0473 % Load the diffusion weighted data. 0474 % 0475 % dwi = feGet(fe,'dwi') 0476 val = dwiLoad(feGet(fe,'dwifile')); 0477 0478 case 'dwirepeat' 0479 % Load the diffusion weighted data. 0480 % 0481 % dwi = feGet(fe,'dwirepeat') 0482 val = dwiLoad(feGetRep(fe,'dwifile')); 0483 0484 case 'dwifile' 0485 % Load the diffusion weighted data. 0486 % 0487 % dwiFile = feGet(fe,'dwifile') 0488 val = fe.path.dwifile; 0489 0490 case 'dwifilerep' 0491 % Load the diffusion weighted data. 0492 % 0493 % dwiFile = feGet(fe,'dwifilerep') 0494 val = fe.path.dwifilerep; 0495 0496 case 'savedir' 0497 % Directory where the LiFe structes are saved by defualt. 0498 % 0499 % sdir = feGet(fe,'savedir'); 0500 val = fe.path.savedir; 0501 0502 case {'bvecs'} 0503 % Diffusion directions. 0504 % 0505 % val = feGet(fe,'bvecs'); 0506 val = fe.life.bvecs; 0507 0508 case {'bvecsindices'} 0509 % Indices to the diffusion directions in the DWi 4th Dimension. 0510 % 0511 % val = feGet(fe,'bvecs indices'); 0512 val = fe.life.bvecsindices; 0513 0514 case {'nbvecs','nbvals'} 0515 % Number of B0's 0516 % 0517 % val = feGet(fe,'n bvals'); 0518 val = length(feGet(fe,'bvals')); 0519 0520 case {'bvals'} 0521 % B0 Values. 0522 % 0523 % bval = feGet(fe,'bvals') 0524 val = fe.life.bvals; 0525 0526 case {'diffusionsignalinvoxel','dsiinvox','dsigvox','dsigmeasuredvoxel'} 0527 % Returns a nVoxels X nBvecs array of measured diffusion signal 0528 % 0529 % val = feGet(fe,'dsiinvox'); 0530 % val = feGet(fe,'dsiinvox',voxelsIndices); 0531 % val = feGet(fe,'dsiinvox',coords); 0532 val = fe.life.diffusion_signal_img(feGet(fe,'voxelsindices',varargin),:)'; 0533 0534 case {'diffusionsignalinvoxeldemeaned','dsiinvoxdemeaned'} 0535 % Returns a nVoxels X nBvecs array of demeaned diffusion signal 0536 % 0537 % val =feGet(fe,'dsiinvoxdemeaned'); 0538 % val =feGet(fe,'dsiinvoxdemeaned',voxelsIndices); 0539 % val =feGet(fe,'dsiinvoxdemeaned',coords); 0540 nBvecs = feGet(fe,'nBvecs'); 0541 voxelIndices = feGet(fe,'voxelsindices',varargin); 0542 val = fe.life.diffusion_signal_img(voxelIndices,:) - repmat(mean(fe.life.diffusion_signal_img(voxelIndices,:), 2),1,nBvecs); 0543 keyboard 0544 % THis seems to be wrong 0545 0546 case {'b0signalimage','b0vox'} 0547 % Get the diffusion signal at 0 diffusion weighting (B0) for this voxel 0548 % 0549 % val = feGet(fe,'b0signalvoxel'); 0550 % val = feGet(fe,'b0signalvoxel',voxelIndex); 0551 % val = feGet(fe,'b0signalvoxel',coords); 0552 val = fe.life.diffusion_S0_img(feGet(fe,'voxelsindices',varargin), :); 0553 0554 case {'fiberssubset','fsub','subsetoffibers','fgsubset'} 0555 % Return a subset of fibers from the conenctome. 0556 % 0557 % fg = feGet(fe,'fiberssubset',fiberList); 0558 val = feGet(fe,'fg img'); 0559 fiberList = varargin{1}; 0560 fibers = cell(size(fiberList)); 0561 for ff = 1:length(fiberList) 0562 fibers{ff} = val.fibers{fiberList(ff)}; 0563 if isfield(val,'pathwayInfo') 0564 pathInfo(ff) = val.pathwayInfo(fiberList(ff)); 0565 end 0566 end 0567 val.fibers = fibers; 0568 if isfield(val,'pathwayInfo') 0569 val.pathwayInfo = pathInfo; 0570 end 0571 val.name = sprintf('subsetFrom:%s',val.name); 0572 0573 case {'roivoxels','roicoords'} 0574 % Return the coordinates of the VOI comprised in the connectome. 0575 % Always in image space. 0576 % 0577 % coords = feGet(fe,'roi coords') 0578 val = fe.roi.coords; 0579 0580 case {'roicoordssubset'} 0581 % Return the coordinates of the VOI comprised in the connectome. 0582 % Always in image space. 0583 % 0584 % coords = feGet(fe,'roi coords') 0585 val = feGet(fe,'roi coords'); 0586 val = val(varargin{1},:); 0587 0588 case {'nroivoxels','nvoxels'} 0589 % Number of voxels in the connectome/VOI. 0590 % 0591 % nVoxels = feGet(fe,'n voxels') 0592 val = size(feGet(fe,'roivoxels'),1); 0593 0594 case {'usedvoxels','indexesofusedvoxels'} 0595 % Indexes of actually used voxels. 0596 % 0597 % These can be different than the number of voxels in the VOI in case some 0598 % voxels have no fibers in them. 0599 % 0600 % TO BE DEPRECATED. 0601 % 0602 % indxUsedVx = feGet(fe,'index of used voxels') 0603 val = find(feGet(fe,'nnodes')); 0604 0605 case {'nusedvoxels','numberofusedvoxels'} 0606 % Number of actually used voxels. 0607 % 0608 % This can be less than the number of voxels in the VOI in case some 0609 % voxels have no fibers in them. 0610 % 0611 % TO BE DEPRECATED. 0612 % 0613 % nUsedVx = feGet(fe,'index of used voxels') 0614 val = size(feGet(fe,'used voxels'),2); 0615 0616 case {'startrow','first index into the dsig vector for a voxel'} 0617 % Returns the first index to a voxel in the dSig vector. 0618 % 0619 % The dSig vector is 1:nBvecs*nVoxels. 0620 % 0621 % val = feGet(fe,'start rows') 0622 val = (varargin{1}-1)*feGet(fe,'nBvecs') + 1; 0623 0624 case {'voxelrows','all indexes into the dsig vector for a set of voxels'} 0625 % Return the rows corresponding to a set of voxels. 0626 % 0627 % rows = feGet(fe,'voxel rows',idxVoxel) 0628 n = feGet(fe,'n voxels'); 0629 nBvecs = feGet(fe,'n bvecs'); 0630 voxelsToKeep = zeros(n,1); voxelsToKeep(varargin{1}) = 1; 0631 val = logical(kron(voxelsToKeep(:),ones(nBvecs,1))); 0632 0633 case {'modeltensor'} 0634 val = fe.life.modelTensor; 0635 0636 case {'mfiber','m','model'} 0637 % Return the model (M matrix), or a subset of it. 0638 % 0639 % Mfiber = feGet(fe,'M fiber'); 0640 % Mfiber = feGet(fe,'model',voxelsIndices); 0641 % Mfiber = feGet(fe,'model',coords); 0642 0643 % idxVoxels is an integer list of the voxels to retain 0644 % If idxVoxels is not included, return the whole matrix 0645 % M contains only the rows for the specified voxels (all directions) 0646 if isempty(varargin) 0647 % Return the whole M matrix 0648 val = fe.life.Mfiber; 0649 else 0650 % Return only the rows for the specified voxels, including all 0651 % directions. 0652 % voxelIndices = feGet(fe,'voxelsindices',varargin); 0653 % voxelRowsToKeep = feGet(fe,'voxel rows',voxelIndices); 0654 % val = fe.life.Mfiber(voxelRowsToKeep,:); 0655 val = fe.life.Mfiber(feGet(fe,'voxel rows',feGet(fe,'voxelsindices',varargin)),:); 0656 end 0657 0658 case {'mfibervoxelsubset','indexedafiber','mfiberindexed'} 0659 % Return a subset of measurements from the fiber portion of M matrix 0660 % The rows of M are specified. Remember that the rows refer to a 0661 % combination of voxel and measurement direction and potentially in the 0662 % future a b-value. Hence, the rows are not specific to a voxel. 0663 % 0664 % Mfiber = feGet(fe,'mfiber subset',rowsToKeep); 0665 0666 if isempty(varargin), error('Subset of rows must be specified'); end 0667 val = fe.life.Mfiber(varargin{1},:); 0668 0669 case {'msubsetfibers','mfiberssubset','mkeepfibers'} 0670 % Returns the model (M) with only a subset of columns (fibers). 0671 % 0672 % fiberList is a vector of indexes, e.g, [1 10 100] 0673 % 0674 % M = feGet(fe,'mkeepfibers',fiberList) 0675 val = fe.life.Mfiber(:,varargin{1}); 0676 0677 case {'miso'} 0678 % Isotropic portion of M matrix 0679 % returns the matrix for the full model or for a subset of voxels 0680 % 0681 % Miso = feGet(fe,'M iso') 0682 % Miso = feGet(fe,'M iso',voxelsIndices) 0683 % Miso = feGet(fe,'M iso',coords) 0684 if isempty(varargin) 0685 % Return the whole M iso 0686 val = feBuildSparseBlockDiag(feGet(fe,'nBvecs'),feGet(fe,'nVoxels')); 0687 else 0688 val = feBuildSparseBlockDiag(feGet(fe,'nBvecs'),length(feGet(fe,'voxelsindices',varargin))); 0689 end 0690 0691 case {'voxel2fnpair','voxel2fibernodepair','v2fnp'} 0692 % Pairing of fibers and nodes in each voxel. 0693 % 0694 % v2fnp = feGet(fe,'v2fnp'); 0695 if isempty(fe.life.voxel2FNpair) 0696 fprintf('[%s] fe.life.voxel2FNpair is empty, can be computed as: \nfe = feSet(fe,''v2fnp'',feGet(fe,''fg img''),''v2fn'',feGet(fe,''roi coords'')',mfilename); 0697 return 0698 end 0699 val = fe.life.voxel2FNpair; 0700 0701 case {'tensors','fibers tensors'} 0702 % Tensors computed for each node and fiber in a set of voxels. 0703 % 0704 % t = feGet(fe,'tensors') 0705 % t = feGet(fe,'tensors',voxelsIndices); 0706 % t = feGet(fe,'tensors',coords); 0707 if isempty(varargin) 0708 val = fe.life.fibers.tensors; 0709 else 0710 vv = feGet(fe,'voxelsindices',varargin); 0711 val = cell(size(vv)); 0712 for ff = 1:length(vv) 0713 val{ff} = fe.life.fibers.tensors{vv(ff)}; 0714 end 0715 end 0716 0717 case {'voxeltensors','voxtensors','voxq'} 0718 % Get all the tensors in a signle voxel 0719 % 0720 % val = feGet(fe,'voxtensors',voxelIndex) 0721 % val = feGet(fe,'voxtensors',coord) 0722 0723 % Index for the voxel 0724 vv = feGet(fe,'voxelsindices',varargin); 0725 0726 % The indexes of the voxeles used, to build LiFE. 0727 usedVoxels = feGet(fe,'usedVoxels'); 0728 voxIndex = usedVoxels(vv); 0729 nNodes = feGet(fe,'nNodes'); 0730 0731 % Get the tensors for each node in each fiber going through this voxel: 0732 val = zeros(nNodes(voxIndex), 9); % Allocate space for all the tensors (9 is for the 3 x 3 tensor components) 0733 for ii = 1:nNodes(voxIndex) % Get the tensors 0734 val(ii,:) = fe.life.fibers.tensors{fe.life.voxel2FNpair{voxIndex}(ii,1)} ... 0735 (fe.life.voxel2FNpair{voxIndex}(ii,2),:); 0736 end 0737 0738 case {'nnodes','numofnodes'} 0739 % Total number of nodes in each voxel The voxel2FN pairs are row size 0740 % equals number of nodes and column size is always 2. The entries of 0741 % the first column are the fiber number. The entries in the second 0742 % column are the node of the fiber inside the voxel. 0743 % 0744 % val = feGet(fe,'n nodes'); 0745 % val = feGet(fe,'n nodes',voxelsIndices); 0746 % val = feGet(fe,'n nodes',coords); 0747 if ~isempty(fe.life.voxel2FNpair) 0748 [val, ~] = cellfun(@size,fe.life.voxel2FNpair); 0749 val = val(feGet(fe,'voxelsindices',varargin)); 0750 end 0751 0752 case {'numberofuniquefibersbyvoxel','uniquefnum'} 0753 % Return the total number of fibers for all the voxels or in a set of 0754 % voxels 0755 % 0756 % nFibers = feGet(fe,'uniquefnum'); % all the voxels 0757 % nFibers = feGet(fe,'uniquefnum',[1 2 3 4]); % for some the voxels, 0758 % % specified by indexes 0759 % nFibers = feGet(fe,'uniquefnum',coords); % for some the voxels, 0760 % % specified by coordinates 0761 val = fe.life.fibers.unique.num(feGet(fe,'voxelsindices',varargin)); 0762 0763 case {'indextouniquefibersbyvoxel','uniquef'} 0764 % Return the indexes of the fibers for all the voxels or in a set of 0765 % voxels. 0766 % 0767 % idxFibers = feGet(fe,'uniquef'); % all the voxels 0768 % idxFibers = feGet(fe,'uniquef',[1 2 3 4]); % for some the voxels, 0769 % % specified by indexes 0770 % idxFibers = feGet(fe,'uniquef',coords); % for some the voxels, 0771 % % specified by coordinates 0772 vxIndex = feGet(fe,'voxels indices',varargin); 0773 val = cell(length(vxIndex),1); 0774 for ii = 1:length(vxIndex) 0775 val{ii} = fe.life.fibers.unique.index{vxIndex(ii)}; 0776 end 0777 0778 case {'numberoftotalfibersbyvoxels','totfnum'} 0779 % Return the total number of fibers for all the voxels or in a set of 0780 % voxels 0781 % 0782 % nFibers = feGet(fe,'totfnum'); % all the voxels 0783 % nFibers = feGet(fe,'totfnum',[1 2 3 4]); % for some the voxels, 0784 % % specified by indexes 0785 % nFibers = feGet(fe,'totfnum',coords); % for some the voxels, 0786 % % specified by coordinates 0787 val = fe.life.fibers.total.num(feGet(fe,'voxels indices',varargin)); 0788 0789 case {'indexoftotalfibersbuvoxel','totf'} 0790 % Return the indexes of the fibers for all the voxels or in a set of 0791 % voxels 0792 % 0793 % idxFibers = feGet(fe,'totf'); % all the voxels 0794 % idxFibers = feGet(fe,'totf',[1 2 3 4]); % for some the voxels, 0795 % % specified by indexes 0796 % idxFibers = feGet(fe,'totfibers',coords); % for some the voxels, 0797 % % specified by coordinates 0798 vxIndex = feGet(fe,'voxels indices',varargin); 0799 val = cell(length(vxIndex),1); 0800 for ii = 1:length(vxIndex) 0801 val{ii} = fe.life.fibers.total.index{vxIndex(ii)}; 0802 end 0803 0804 case {'nfibers'} 0805 % Return the number of fibers in the model. 0806 % 0807 % nFibers = feGet(fe,'n fibers') 0808 val = size(fe.life.Mfiber,2); 0809 0810 case {'isoweights','weightsiso','meanvoxelsignal'} 0811 % Weights of the isotropic voxel signals, this is the mean signal in 0812 % each voxel. 0813 % 0814 % val = feGet(fe,'iso weights'); 0815 % val = feGet(fe,'iso weights'coords) 0816 % val = feGet(fe,'iso weights',voxelIndices) 0817 val = feGet(fe,'Miso') \ feGet(fe,'dsig full')'; 0818 val = val(feGet(fe,'voxelsindices',varargin)); 0819 0820 case {'voxisoweights','voxweightsiso','voxmeanvoxelsignal'} 0821 % Weights of the isotropic voxel signals, this is the mean signal in 0822 % each voxel. 0823 % Organized nBvecs x nVoxels 0824 % 0825 % val = feGet(fe,'iso weights'); 0826 % val = feGet(fe,'iso weights'coords) 0827 % val = feGet(fe,'iso weights',voxelIndices) 0828 val = feGet(fe,'Miso') \ feGet(fe,'dsig full')'; 0829 val = repmat(val,1,feGet(fe,'nbvecs'))'; 0830 0831 if ~isempty(varargin) 0832 % voxelIndices = feGet(fe,'voxelsindices',varargin); 0833 % voxelRowsToKeep = feGet(fe,'voxel rows',voxelIndices); 0834 % val = val(voxelRowsToKeep,:); 0835 val = val(:,feGet(fe,'voxelsindices',varargin)); 0836 end 0837 0838 case {'dsigiso','isodsig'} 0839 % mean signalin each voxel, returned for each diffusion direction: 0840 % size(nBvecsxnVoxels, 1). 0841 % 0842 % val = feGet(fe,'iso disg'); 0843 % val = feGet(fe,'iso disg',coords) 0844 % val = feGet(fe,'iso dsig',voxelIndices) 0845 val = feGet(fe,'iso weights'); 0846 val = repmat(val,1,feGet(fe,'nbvecs'))'; 0847 val = val(:); 0848 0849 case {'fiberweights'} 0850 % Weights of the fiber component. For all the fibers or a subset of 0851 % them. 0852 % 0853 % w = feGet(fe,'fiber weights') 0854 % w = feGet(fe,'fiber weights',fiberIndices) 0855 0856 % If the model was not fit yet, fit it, install the fit and then return 0857 % the weights. 0858 if ~isfield(fe.life,'fit') 0859 fprintf('[%s] fe.life.fit is empty, can be computed as: \nfeSet(fe,''fit'',feFitModel(feGet(fe,''Mfiber''), feGet(fe,''dsigdemeaned''),''sgdnn''))',mfilename); 0860 return 0861 end 0862 0863 % Get the weights 0864 if ~isempty(varargin) % subselect the weights for the requested fibers 0865 val = fe.life.fit.weights(varargin{1}); 0866 else % Return all of them 0867 val = fe.life.fit.weights; 0868 end 0869 0870 case {'fiberweightstest'} 0871 % Weights of the fiber component with a subset of the fibers' weigth 0872 % set to zero. This can be used to test the loss in RMSE for the 0873 % connectome when a subset of fibers is removed, but the connectome is 0874 % not fitted again. 0875 % 0876 % w = feGet(fe,'fiber weights test',fiberIndices) 0877 0878 % If the model was not fit yet, fit it, install the fit and then return 0879 % the weights. 0880 if ~isfield(fe.life,'fit') 0881 fprintf('[%s] fe.life.fit is empty, can be computed as: \nfeSet(fe,''fit'',feFitModel(feGet(fe,''Mfiber''), feGet(fe,''dsigdemeaned''),''sgdnn''))',mfilename); 0882 return 0883 end 0884 0885 % Get the weights 0886 if ~isempty(varargin) % subselect the weights for the requested fibers 0887 val = fe.life.fit.weights; 0888 val(varargin{1}) = 0; 0889 else % A set of fiber weights is required as input 0890 error('[%s] Indices to a subset of fiber-weights must be passed in w = feGet(fe,''fiber weights test'',fiberIndices))',mfilename); 0891 end 0892 0893 case {'fiberweightstestvoxelwise'} 0894 % Weights of the fiber component with a subset of the fibers' weigth 0895 % set to zero. This can be used to test the loss in RMSE for the 0896 % connectome when a subset of fibers is removed, but the connectome is 0897 % not fitted again. 0898 % 0899 % w = feGet(fe,'fiber weights test voxel wise',fiberIndices) 0900 0901 % If the model was not fit yet, fit it, install the fit and then return 0902 % the weights. 0903 if ~isfield(fe.life,'voxfit') 0904 fprintf('[%s] fe.life.fit is empty, can be computed as: \nfeSet(fe,''fit'',feFitModel(feGet(fe,''Mfiber''), feGet(fe,''dsigdemeaned''),''sgdnn''))',mfilename); 0905 return 0906 end 0907 0908 % Get the weights 0909 if ~isempty(varargin) % subselect the weights for the requested fibers 0910 val = fe.life.voxfit.weights; 0911 val(:,varargin{1}) = 0; 0912 else % A set of fiber weights is required as input 0913 error('[%s] Indices to a subset of fiber-weights must be passed in\nw = feGet(fe,''fiber weights test voxel wise'',fiberIndices))',mfilename); 0914 end 0915 0916 case {'dsigmeasured','dsigfull'} 0917 % Measured signal in VOI, this is the raw signal. not demeaned 0918 % 0919 % dSig = feGet(fe,'dSig full') 0920 val = fe.life.dSig; 0921 % Return a subset of voxels 0922 if ~isempty(varargin) 0923 % voxelIndices = feGet(fe,'voxelsindices',varargin); 0924 % voxelRowsToKeep = feGet(fe,'voxel rows',voxelIndices); 0925 % val = val(voxelRowsToKeep,:); 0926 val = val(feGet(fe,'voxel rows',feGet(fe,'voxelsindices',varargin))); 0927 end 0928 0929 case {'diffusionsignaldemeaned','dsigdemeaned'} 0930 % Measured signal in VOI, demeaned, this is the signal used for the 0931 % fiber-portion of the M model. 0932 % 0933 % dSig = feGet(fe,'dsigdemeaned'); 0934 % dSig = feGet(fe,'dsigdemeaned',[1 10 100]); 0935 % dSig = feGet(fe,'dsigdemeaned',coords); 0936 nVoxels = feGet(fe,'nVoxels'); 0937 nBvecs = feGet(fe,'nBvecs'); 0938 val = (fe.life.dSig - reshape(repmat( ... 0939 mean(reshape(fe.life.dSig, nBvecs, nVoxels),1),... 0940 nBvecs,1), size(fe.life.dSig)))'; 0941 % Return a subset of voxels 0942 if ~isempty(varargin) 0943 % voxelIndices = feGet(fe,'voxelsindices',varargin); 0944 % voxelRowsToKeep = feGet(fe,'voxel rows',voxelIndices); 0945 % val = val(voxelRowsToKeep,:); 0946 val = val(feGet(fe,'voxel rows',feGet(fe,'voxelsindices',varargin))); 0947 end 0948 0949 case {'dsigrowssubset','diffusionsignaldemeanedinsubsetofrows'} 0950 % Get the demeaned signal for a subset of rows. 0951 % Useful for cross-validation. 0952 % 0953 % dSig = feGet(fe,'dsigrowssubset',voxelsList); 0954 dSig = feGet(fe,'dsigdemeaned'); 0955 for vv = 1:length(voxelsList) 0956 val(feGet(fe,'voxel rows',vv)) = dSig(feGet(fe,'voxel rows',voxelList(vv))); 0957 end 0958 0959 case {'psigfiber', 'fiberpsig','fiberpredicted','dsigpredictedfiber'} 0960 % Predicted signal of fiber alone (demeaned). 0961 % 0962 % pSig = feGet(fe,'pSig fiber'); 0963 % pSig = feGet(fe,'pSig fiber',coords); 0964 % pSig = feGet(fe,'pSig fiber',voxelIndices); 0965 val = feGet(fe,'Mfiber')*feGet(fe,'fiber weights'); 0966 if ~isempty(varargin) 0967 % voxelIndices = feGet(fe,'voxelsindices',varargin); 0968 % voxelRowsToKeep = feGet(fe,'voxel rows',voxelIndices); 0969 % val = val(voxelRowsToKeep,:); 0970 val = val(feGet(fe,'voxel rows',feGet(fe,'voxelsindices',varargin))); 0971 end 0972 0973 case {'uniquefibersindicesinroi'} 0974 % Find the unique fibers indices in the FE roi. 0975 % Thisis necessary sometimes after changing the number of voxels in an 0976 % FE structure, as it is performed by feConnectomeReduceVoxels.m 0977 % 0978 % FibInRoi = feGet(fe,'uniquefibersinroi'); 0979 0980 % Get all the unique fibers in each voxel 0981 uniquefvx = fefgGet(feGet(fe,'fibers img'), ... 0982 'uniquefibersinvox', ... 0983 feGet(fe,'roi coords')); 0984 val = []; 0985 for ivx = 1:length(uniquefvx) 0986 val = [val; uniquefvx{ivx}]; 0987 end 0988 val = unique(val); 0989 0990 case {'weightsinroi'} 0991 % Find the weights of the fibers in a specified volume. 0992 % 0993 % We compute the following values: 0994 % (1) The number of unique fibers in the volume 0995 % (2) The weights for the fibers in the roi 0996 % 0997 % wFibInRoi = feGet(fe,'weightsinroi'); 0998 0999 % Get all the unique fibers in each voxel 1000 FibInRoi = feGet(fe,'uniquefibersindicesinroi'); 1001 1002 % extract the fber weights obtained in a LiFE fit 1003 w = feGet(fe,'fiber weights'); 1004 1005 % Return only the ones for the fibers in this roi 1006 val = w(unique( FibInRoi )); 1007 1008 case {'fiberdensity'} 1009 % Fiber density statistics. 1010 % 1011 % Computes the fiber density (how many fibers in each voxel) 1012 % 1013 % We compute the following values: 1014 % (1) The number of fibers in each voxel 1015 % (2) The number of unique fibers with non-zero weights 1016 % (3) The sum of the weights in each voxel 1017 % (4) The mean of the weights in each voxel 1018 % (5) The variance of the weigths in each voxel 1019 % 1020 % pSig = feGet(fe,'fiberdensity'); 1021 1022 % Get the unique fibers in each voxel 1023 uniquefvx = fefgGet(feGet(fe,'fibers img'), ... 1024 'uniquefibersinvox', ... 1025 feGet(fe,'roi coords')); 1026 1027 % extract the fber weights obtained in a LiFE fit 1028 w = feGet(fe,'fiber weights'); 1029 1030 % Compute the fiber density in three wasy: 1031 % (1) The number of fibers in each voxel 1032 % (2) The number of unique fibers with non-zero weights 1033 % (3) The sum of the weights in each voxel 1034 % (4) The mean of the weights in each voxel 1035 % (5) The variance of the weigths in each voxel 1036 val = nan(length(uniquefvx),5); 1037 for ivx = 1:length(uniquefvx) 1038 1039 % Number of fibers in each voxel 1040 val(ivx,1) = length(uniquefvx{ivx}); 1041 1042 % Number of fibers in ech voxel with non-zero weight 1043 val(ivx,2) = length(uniquefvx{ivx}(w(uniquefvx{ivx}) > 0)); 1044 1045 % Sum of fiber weights in each voxel 1046 val(ivx,3) = sum(w(uniquefvx{ivx})); 1047 1048 % Mean of fiber weights in each voxel 1049 val(ivx,4) = nanmedian(w(uniquefvx{ivx})); 1050 1051 % Var of fibers in each voxel 1052 val(ivx,5) = nanvar(w(uniquefvx{ivx})); 1053 end 1054 1055 case {'psigfibertest'} 1056 % Predicted signal (demeaned) with a subset of fibers' weights set to 0. 1057 % This can be used to test the loss in RMSE for the connectome when a 1058 % subset of fibers is removed, but the connectome is not fitted again. 1059 % 1060 % pSig = feGet(fe,'pSig fiber test',fiberIndices); 1061 % pSig = feGet(fe,'pSig fiber test',fiberIndices,voxelIndices); 1062 % pSig = feGet(fe,'pSig fiber test',fiberIndices,coords); 1063 if ~isempty(varargin) 1064 val = feGet(fe,'Mfiber')*feGet(fe,'fiber weights test',varargin{1}); 1065 1066 % Now select a subset of voxels if requested 1067 if (length(varargin) == 2) 1068 % voxelIndices = feGet(fe,'voxelsindices',varargin); 1069 % voxelRowsToKeep = feGet(fe,'voxel rows',voxelIndices); 1070 % val = val(voxelRowsToKeep,:); 1071 val = val(feGet(fe,'voxel rows',feGet(fe,'voxelsindices',varargin{2}))); 1072 end 1073 else 1074 error('[%s] A set f fiber indiced must be passed in. pSig = feGet(fe,''pSig fiber test'',fiberIndices))',mfilename); 1075 end 1076 1077 case {'psigfibertestvoxelwise'} 1078 % Predicted signal (demeaned) with a subset of fibers' weights set to 0. 1079 % This can be used to test the loss in RMSE for the connectome when a 1080 % subset of fibers is removed, but the connectome is not fitted again. 1081 % 1082 % pSig = feGet(fe,'pSig fiber test voxel wise',fiberIndices); 1083 % pSig = feGet(fe,'pSig fiber test voxel wise',fiberIndices,voxelIndices); 1084 % pSig = feGet(fe,'pSig fiber test voxel wise',fiberIndices,coords); 1085 1086 if ~isempty(varargin) 1087 nVoxels = feGet(fe,'nvoxels'); 1088 w = feGet(fe,'fiber weights test voxel wise',varargin{1}); % Weights with a subset set to 0 1089 val = cell(nVoxels,1); 1090 for ivox = 1:nVoxels 1091 % Predict the signal. 1092 val{ivox} = feGet(fe,'Mfiber',ivox)*w(ivox,:)'; 1093 end 1094 1095 feOpenLocalCluster 1096 % Reorganize the signal into a vector 1097 val = vertcat(val{:})'; 1098 1099 % Now select a subset of voxels if requested 1100 if (length(varargin) == 2) 1101 % voxelIndices = feGet(fe,'voxelsindices',varargin); 1102 % voxelRowsToKeep = feGet(fe,'voxel rows',voxelIndices); 1103 % val = val(voxelRowsToKeep,:); 1104 val = val(feGet(fe,'voxel rows',feGet(fe,'voxelsindices',varargin{2}))); 1105 end 1106 else 1107 error('[%s] A set f fiber indiced must be passed in. pSig = feGet(fe,''pSig fiber test'',fiberIndices))',mfilename); 1108 end 1109 1110 case {'predictedsignalfull','psigfull'} 1111 % Predicted measured signal from both fiber and iso 1112 % 1113 % pSig = feGet(fe,'pSig full') 1114 % pSig = feGet(fe,'pSig full',coords) 1115 % pSig = feGet(fe,'pSig full',voxelIndices) 1116 val = [feGet(fe,'M fiber'), feGet(fe,'M iso')] * feGet(fe,'full weights'); 1117 if ~isempty(varargin) 1118 % voxelIndices = feGet(fe,'voxelsindices',varargin); 1119 % voxelRowsToKeep = feGet(fe,'voxel rows',voxelIndices); 1120 % val = val(voxelRowsToKeep,:); 1121 val = val(feGet(fe,'voxel rows',feGet(fe,'voxelsindices',varargin))); 1122 end 1123 1124 case {'fullweights'} 1125 % The fiber and isotropic weights as a long vector 1126 % 1127 % w = feGet(fe,'fullweights') 1128 val = [feGet(fe,'fiber weights'); feGet(fe,'iso weights')]; 1129 1130 case {'fiberweightsvoxelwise'} 1131 % The woxels returned by a fit of LiFE by voxel/fiber 1132 % 1133 % w = feGet(fe,'fiberweightsvoxelwise') 1134 val = fe.life.voxfit.weights; 1135 1136 case {'psigfvoxelwise'} 1137 % Predict the diffusion signal for the fiber component 1138 % with the voxel-wise fit of LiFE 1139 % 1140 % pSig = feGet(fe,'psigfvoxelwise') 1141 % pSig = feGet(fe,'psigfvoxelwise',coords) 1142 % pSig = feGet(fe,'psigfvoxelwise',voxelIndices) 1143 1144 if ~isfield(fe.life,'voxfit'), 1145 error('[%s] Cannot find voxel-wise fit.\nTo fit the model voxel-wise run:\nfe = feFitModelByVoxel(fe)\n',mfilename); 1146 end 1147 val = fe.life.voxfit.psig; 1148 1149 % Get a subset of voxels. 1150 if ~isempty(varargin) 1151 % voxelIndices = feGet(fe,'voxelsindices',varargin); 1152 % voxelRowsToKeep = feGet(fe,'voxel rows',voxelIndices); 1153 % val = val(voxelRowsToKeep,:); 1154 val = val(feGet(fe,'voxel rows',feGet(fe,'voxelsindices',varargin))); 1155 end 1156 1157 case {'psigfvoxelwisebyvoxel'} 1158 % Predict the diffusion signal for the fiber component 1159 % with the voxel-wise fit of LiFE, return the an array of pSigXnVoxel 1160 % 1161 % pSig = feGet(fe,'psigfvoxelwisebyvoxel') 1162 % pSig = feGet(fe,'psigfvoxelwisebyvoxel',coords) 1163 % pSig = feGet(fe,'psigfvoxelwisebyvoxel',voxelIndices) 1164 nBvecs = feGet(fe,'nBvecs'); 1165 nVoxels = feGet(fe,'n voxels'); 1166 val = reshape(feGet(fe,'psigfvoxelwise'), nBvecs, nVoxels); 1167 1168 case {'totalr2'} 1169 % Return the global R2 (fraction of variance explained) of the full life 1170 % model. 1171 % 1172 % R2 = feGet(fe,'total r2'); 1173 1174 % measured = feGet(fe,'dsigdemeaned'); 1175 % predicted = feGet(fe,'fiber p sig'); 1176 % val = (1 - (sum((measured - predicted).^2 ) ./ ... 1177 % sum((measured - mean(measured)).^2) )); 1178 val = (1 - (sum((feGet(fe,'diffusion signal demeaned') - ... 1179 feGet(fe,'fiber predicted')).^2 ) ./ ... 1180 sum((feGet(fe,'diffusion signal demeaned') - ... 1181 mean(feGet(fe,'diffusion signal demeaned'))).^2) )); 1182 1183 case {'totalr2voxelwise'} 1184 % Return the global R2 (fraction of variance explained) of the full life 1185 % model from a voxel-wise fit. 1186 % 1187 % R2 = feGet(fe,'total r2 vocel wise'); 1188 1189 % measured = feGet(fe,'dsigdemeaned'); 1190 % predicted = feGet(fe,'p sig f voxel wise'); 1191 % val = (1 - (sum((measured - predicted).^2 ) ./ ... 1192 % sum((measured - mean(measured)).^2) )); 1193 val = (1 - (sum((feGet(fe,'diffusion signal demeaned') - ... 1194 feGet(fe,'psigfvoxelwise')').^2 ) ./ ... 1195 sum((feGet(fe,'diffusion signal demeaned') - ... 1196 mean(feGet(fe,'diffusion signal demeaned'))).^2) )); 1197 1198 case {'totpercentvarianceexplained','totpve'} 1199 % Percent variance explained 1200 % 1201 % R2 = feGet(fe,'explained variance') 1202 val = 100 * feGet(fe,'total r2'); 1203 1204 case {'rmsetotal','totalrmse'} 1205 % Root mean squared error of the LiFE fit to the whole data 1206 % 1207 % rmse = feGet(fe,'rmse') 1208 val = sqrt(mean((feGet(fe,'diffusion signal demeaned') - ... 1209 feGet(fe,'pSig fiber')).^2)); 1210 1211 case {'totalrmsevoxelwise'} 1212 % Root mean squared error of the LiFE fit to the whole data from a 1213 % vocel-wise fit 1214 % 1215 % rmse = feGet(fe,'rmse') 1216 val = sqrt(mean((feGet(fe,'diffusion signal demeaned') - ... 1217 feGet(fe,'pSig f voxelwise')').^2)); 1218 1219 case {'ressigfiber'} 1220 % Residual signal fiber prediction - measured_demeaned. 1221 % 1222 % res = feGet(fe,'res sig fiber') 1223 val = (feGet(fe,'dsigdemeaned') - feGet(fe,'psig fiber')); 1224 if ~isempty(varargin) 1225 % voxelIndices = feGet(fe,'voxelsindices',varargin); 1226 % voxelRowsToKeep = feGet(fe,'voxel rows',voxelIndices); 1227 % val = val(voxelRowsToKeep,:); 1228 val = val(feGet(fe,'voxel rows',feGet(fe,'voxelsindices',varargin))); 1229 end 1230 1231 case {'ressigfull'} 1232 % Residual signal full model prediction - measured signal. 1233 % 1234 % res = feGet(fe,'res sig full'); 1235 % res = feGet(fe, 'res sig full',coords); 1236 % res = feGet(fe, 'res sig full',voxelIndex); 1237 val = (feGet(fe,'dsig full') - feGet(fe,'psig full')'); 1238 if ~isempty(varargin) 1239 % voxelIndices = feGet(fe,'voxelsindices',varargin); 1240 % voxelRowsToKeep = feGet(fe,'voxel rows',voxelIndices); 1241 % val = val(voxelRowsToKeep,:); 1242 val = val(feGet(fe,'voxel rows',feGet(fe,'voxelsindices',varargin))); 1243 end 1244 1245 case {'ressigfullvoxelwise'} 1246 % Residual signal full model prediction from a multi-voxel fit. 1247 % 1248 % res = feGet(fe,'res sig full voxfit'); 1249 % res = feGet(fe, 'res sig full voxfit',coords); 1250 % res = feGet(fe, 'res sig full voxfit',voxelIndex); 1251 %tic,val = feGet(fe,'dsig demeaned') - feGet(fe,'psigfvoxelwise') + feGet(fe,'dsigiso');toc 1252 val = feGet(fe,'dsig full') - feGet(fe,'psigfvoxelwise')'; 1253 if ~isempty(varargin) 1254 % voxelIndices = feGet(fe,'voxelsindices',varargin); 1255 % voxelRowsToKeep = feGet(fe,'voxel rows',voxelIndices); 1256 % val = val(voxelRowsToKeep,:); 1257 val = val(feGet(fe,'voxel rows',feGet(fe,'voxelsindices',varargin))); 1258 end 1259 1260 case {'fiberressigwithmean'} 1261 % Residual signal fiber model prediction - demeaned measured signal 1262 % with added mean signal. VECTOR FORM. 1263 % 1264 % This is used to reconstruct an image (volume) to be used for the 1265 % refinement process. 1266 % 1267 % res = feGet(fe,'fiber res sig with mean'); 1268 1269 % predicted = feGet(fe,'psig fiber'); 1270 % measured = feGet(fe,'dsigdemeaned'); 1271 % val = (measured_full - predicted demeaned); 1272 val = (feGet(fe,'dsig full')' - feGet(fe,'psig fiber')); 1273 1274 if ~isempty(varargin) 1275 % voxelIndices = feGet(fe,'voxelsindices',varargin); 1276 % voxelRowsToKeep = feGet(fe,'voxel rows',voxelIndices); 1277 % val = val(voxelRowsToKeep,:); 1278 val = val(feGet(fe,'voxel rows',feGet(fe,'voxelsindices',varargin))); 1279 end 1280 1281 case {'predictedfibersignalvoxel','psigfvox'} 1282 % Predicted signal by the fiber model in a set of voxels. 1283 % 1284 % Vox subfield stores per voxel within the VOI 1285 % The pSig has size of the dwi. 1286 % It is stored as val = pSig(X,Y,Z,Theta) 1287 % 1288 % pSig = feGet(fe, 'pSig fiber by voxel'); 1289 % pSig = feGet(fe, 'pSig fiber by voxel',coords); 1290 % pSig = feGet(fe, 'pSig fiber by voxel',voxelIndex); 1291 nBvecs = feGet(fe,'nBvecs'); 1292 nVoxels = feGet(fe,'n voxels'); 1293 val = reshape(feGet(fe,'pSig fiber'), nBvecs, nVoxels); 1294 val = val(:,feGet(fe,'return voxel indices',varargin)); 1295 1296 case {'predictedfullsignalvoxel','psigfullvox'} 1297 % Predicted signal by the full model in a set of voxeles. 1298 % 1299 % Vox subfield stores per voxel within the VOI 1300 % The pSig has size of the dwi. 1301 % It is stored as val = pSig(X,Y,Z,Theta) 1302 % 1303 % pSig = feGet(fe, 'pSig full by voxel'); 1304 % pSig = feGet(fe, 'pSig full by voxel',coords); 1305 % pSig = feGet(fe, 'pSig full by voxel',voxelIndex); 1306 nBvecs = feGet(fe,'nBvecs'); 1307 nVoxels = feGet(fe,'n voxels'); 1308 val = reshape(feGet(fe,'pSig full'), nBvecs, nVoxels); 1309 1310 case {'voxelr2','r2vox','voxr2','r2byvoxel'} 1311 % Return a column vector of the proportion of variance explained in 1312 % each voxel. 1313 % 1314 % R2byVox = feGet(fe,'voxr2'); 1315 % R2byVox = feGet(fe,'voxr2',coords); 1316 measured = feGet(fe,'dSig demeaned by voxel'); 1317 predicted = feGet(fe,'pSig f vox'); 1318 nBvecs = feGet(fe,'nBvecs'); 1319 val = (1 - (sum((measured - predicted).^2 ) ./ ... 1320 sum((measured - repmat(mean(measured), nBvecs,1)).^2 ) )); 1321 val = val(feGet(fe,'return voxel indices',varargin)); 1322 1323 case {'voxelr2zero','r2voxzero','voxr2zero','r2byvoxelzero'} 1324 % Return a column vector of the proportion of variance explained in 1325 % each voxel. (Normalized to the squared diffusion signal in each voxel) 1326 % 1327 % R2byVox = feGet(fe,'voxr2zero'); 1328 % R2byVox = feGet(fe,'voxr2zero',coords); 1329 measured = feGet(fe,'dSig demeaned by voxel'); 1330 predicted = feGet(fe,'pSig f vox'); 1331 val = (1 - (sum((measured - predicted).^2 ) ./ sum(measured.^2))); 1332 val = val(feGet(fe,'return voxel indices',varargin)); 1333 1334 case {'voxelvarianceexplained','varexpvox','voxvarexp','varexpbyvoxel'} 1335 % Return the percent of varince explained in each voxel. 1336 % 1337 % R2byVox = feGet(fe,'var exp by voxel'); 1338 % R2byVox = feGet(fe,'var exp by voxel',coords); 1339 val = 100.*feGet(fe,'voxr2'); 1340 val = val(feGet(fe,'return voxel indices',varargin)); 1341 1342 case {'voxelr2voxelwise','r2voxvoxelwise','voxr2voxelwise'} 1343 % Return a column vector of the proportion of variance explained in 1344 % each voxel. 1345 % 1346 % R2byVox = feGet(fe,'voxr2voxelwise'); 1347 % R2byVox = feGet(fe,'voxr2voxewise',coords); 1348 measured = feGet(fe,'dSig demeaned by voxel'); 1349 predicted = feGet(fe,'psig f voxel wise by voxel'); 1350 nBvecs = feGet(fe,'nBvecs'); 1351 val = (1 - (sum((measured - predicted).^2 ) ./ ... 1352 sum((measured - repmat(mean(measured), nBvecs,1)).^2 ) )); 1353 val = val(feGet(fe,'return voxel indices',varargin)); 1354 1355 case {'voxelr2zerovoxelwise','voxr2zerovoxelwise'} 1356 % Return a column vector of the proportion of variance explained in 1357 % each voxel. (Normalized to the squared diffusion signal in each voxel) 1358 % 1359 % R2byVox = feGet(fe,'voxr2zerovoxelwise'); 1360 % R2byVox = feGet(fe,'voxr2zerovoxelwise',coords); 1361 measured = feGet(fe,'dSig demeaned by voxel'); 1362 predicted = feGet(fe,'psig f voxel wise by voxel'); 1363 val = (1 - (sum((measured - predicted).^2 ) ./ sum(measured.^2))); 1364 val = val(feGet(fe,'return voxel indices',varargin)); 1365 1366 case {'voxelvarianceexplainedvoxelwise','varexpvoxvoxelwise'} 1367 % Return the percent of varince explained in each voxel. 1368 % 1369 % R2byVox = feGet(fe,'var exp by voxel'); 1370 % R2byVox = feGet(fe,'var exp by voxel',coords); 1371 val = 100.*feGet(fe,'voxr2voxelwise'); 1372 val = val(feGet(fe,'return voxel indices',varargin)); 1373 1374 case {'dsigdemeanedbyvoxel','dsigdemeanedvox'} 1375 % Demeaned diffusion signal in each voxel 1376 % 1377 % dSigByVoxel = feGet(fe,'dsigdemeaned by Voxel'); 1378 % dSigByVoxel = feGet(fe,'dsigdemeaned by Voxel',coords); 1379 % dSigByVoxel = feGet(fe,'dsigdemeaned by Voxel',vxIndex); 1380 nBvecs = feGet(fe,'nBvecs'); 1381 nVoxels = feGet(fe,'n voxels'); 1382 val = reshape(feGet(fe,'dsigdemeaned'), nBvecs, nVoxels); 1383 val = val(:,feGet(fe,'return voxel indices',varargin)); 1384 1385 case {'dsigfullbyvoxel','dsigfullvox','voxdsigfull'} 1386 % Full (measured) signal in each voxel 1387 % 1388 % dSigByVoxel = feGet(fe,'dSig full by Voxel'); 1389 % dSigByVoxel = feGet(fe,'dSig full by Voxel',coords); 1390 % dSigByVoxel = feGet(fe,'dSig full by Voxel',vxIndex); 1391 nBvecs = feGet(fe,'nBvecs'); 1392 nVoxels = feGet(fe,'n voxels'); 1393 val = reshape(feGet(fe,'dSig full'), nBvecs, nVoxels); 1394 val = val(:,feGet(fe,'return voxel indices',varargin))'; 1395 1396 case {'voxelrmse','voxrmse'} 1397 % A volume of RMSE values 1398 % 1399 % RMSE = feGet(fe,'vox rmse') 1400 % RMSE = feGet(fe,'vox rmse',coords) 1401 % RMSE = feGet(fe,'vox rmse',vxIndex) 1402 measured = feGet(fe,'dsigdemeaned by voxel'); 1403 predicted = feGet(fe,'pSig f vox'); 1404 val = sqrt(mean((measured - predicted).^2,1)); 1405 val = val(feGet(fe,'voxelsindices',varargin)); 1406 1407 case {'voxelrmsevoxelwise','voxrmsevoxelwise'} 1408 % A volume of RMSE values optained with the voxel-wise (voxelwise) fit. 1409 % 1410 % RMSE = feGet(fe,'vox rmse voxelwise') 1411 % RMSE = feGet(fe,'vox rmse voxelwise',coords) 1412 % RMSE = feGet(fe,'vox rmse voxelwise',vxIndex) 1413 measured = feGet(fe,'dsigdemeaned by voxel'); 1414 predicted = feGet(fe,'pSig f voxel wise by voxel'); 1415 1416 val = sqrt(mean((measured - predicted).^2,1)); 1417 val = val(feGet(fe,'voxelsindices',varargin)); 1418 1419 case {'voxelrmsetest','voxrmsetest'} 1420 % A volume of RMSE values with a subset of fibers' weights set to 0. 1421 % 1422 % RMSE = feGet(fe,'vox rmse',fiberIndices) 1423 % RMSE = feGet(fe,'vox rmse',fiberIndices,coords) 1424 % RMSE = feGet(fe,'vox rmse',fiberIndices,voxelIndex) 1425 measured = feGet(fe,'dsigdemeaned by voxel'); 1426 % Reshape the predicted signal by voxles 1427 predicted = reshape(feGet(fe,'pSig fiber test',varargin{1}), feGet(fe,'nBvecs'), feGet(fe,'nVoxels')); 1428 val = sqrt(mean((measured - predicted).^2,1)); 1429 1430 if length(varargin) == 2 1431 val = val(feGet(fe,'voxelsindices',varargin)); 1432 end 1433 1434 case {'residualsignalfibervoxel','resfibervox'} 1435 % Fibers' residual signal by voxel 1436 % 1437 % res = feGet(fe,'res sig fiber vox') 1438 % res = feGet(fe,'res sig fiber vox',coords) 1439 % res = feGet(fe,'res sig fiber vox',vxIndex) 1440 nBvecs = feGet(fe,'nBvecs'); 1441 nVoxels = feGet(fe,'n voxels'); 1442 val = reshape(feGet(fe,'res sig fiber'), nBvecs, nVoxels); 1443 val = val(:,feGet(fe,'return voxel indices',varargin))'; 1444 1445 case {'residualsignalfullvoxel','resfullvox'} 1446 % Full (measured) residual signal by voxel 1447 % 1448 % res = feGet(fe,'res sig full vox') 1449 % res = feGet(fe,'res sig full vox',coords) 1450 % res = feGet(fe,'res sig full vox',vxIndex) 1451 nBvecs = feGet(fe,'nBvecs'); 1452 nVoxels = feGet(fe,'n voxels'); 1453 val = reshape(feGet(fe,'res sig full'), nBvecs, nVoxels); 1454 val = val(:,feGet(fe,'return voxel indices',varargin))'; 1455 1456 case {'fiberressigwithmeanvoxel'} 1457 % Residual signal fiber model prediction - demeaned measured signal 1458 % with added mean signal. VOXEL FORM (nBvecs x nVoxel). 1459 % 1460 % This is used to reconstruct an image (volume) to be used for the 1461 % refinement process. 1462 % 1463 % res = feGet(fe,'fiber res sig with mean'); 1464 1465 % predicted = feGet(fe,'psig fiber'); 1466 % measured = feGet(fe,'dsigdemeaned'); 1467 % val = (measured - predicted) + (measured_full - measured); 1468 val = feGet(fe,'fiberressigwithmean'); 1469 nBvecs = feGet(fe,'nBvecs'); 1470 nVoxels = feGet(fe,'n voxels'); 1471 val = reshape(val, nBvecs, nVoxels); 1472 1473 if ~isempty(varargin) 1474 % voxelIndices = feGet(fe,'voxelsindices',varargin); 1475 % voxelRowsToKeep = feGet(fe,'voxel rows',voxelIndices); 1476 % val = val(voxelRowsToKeep,:); 1477 val = val(:, feGet(fe,'return voxel indices',varargin)); 1478 end 1479 1480 case {'fiberressigwithmeanvox','resfibermeanvox'} 1481 % Fibers' residual signal by voxel with mean signal added (with added 1482 % isotropic component). 1483 % 1484 % This is used to compute the residual signal for the refinement. 1485 % 1486 % res = feGet(fe,'fiber res sig with mean vox') 1487 % res = feGet(fe,'fiber res sig with mean vox',coords) 1488 % res = feGet(fe,'fiber res sig with mean vox',vxIndex) 1489 nBvecs = feGet(fe,'nBvecs'); 1490 nVoxels = feGet(fe,'n voxels'); 1491 val = reshape(feGet(fe,'fiber res sig with mean'), nBvecs, nVoxels); 1492 val = val(:,feGet(fe,'return voxel indices',varargin))'; 1493 1494 case {'voxelsindices','returnvoxelindices','voxelindexes''returnvoxelindexes'} 1495 % Given an VOI or a set of indices to voxels returns the indices of the matching voxels inside the big 1496 % volume, which ordinarily represents the full connectome. 1497 % 1498 % voxelIndices = feGet(fe,'voxelsindices',coords) 1499 % voxelIndices = feGet(fe,'voxelsindices',voxelIndices) 1500 % IMPORTANT: Indices MUST be a column vector for this 1501 % code to work. 1502 % 1503 % coords is a Nx3 set of coordinates in image space 1504 % voxelIndices is a vector of 1's and 0's, there is a one for each 1505 % location the the connectome coordinates for which there is a match in 1506 % the coords 1507 % 1508 varargin = varargin{1}; 1509 if ( ~isempty(varargin) ) 1510 if ( size(varargin{1},2) == 3 ) 1511 % If a set of coordinates were passed in we return the indices. 1512 val = logical(feGet(fe,'find voxels',varargin{1})); 1513 else 1514 % If indices were passed in, we sort and return them. 1515 val = sort(varargin{1}); 1516 end 1517 else 1518 % If no coordinates were passed at all, we return all the indices. 1519 val = true(feGet(fe,'n voxels'),1); 1520 end 1521 1522 case {'findvoxels','findvoxelindexes','findvoxelsinconnectome','findvoxelsinconnectomeroi', ... 1523 'voxel2index','coords2index'} 1524 % Given an VOI finds the indices of the matching voxels inside the big 1525 % volume, which ordinarily represents the full connectome. 1526 % 1527 % foundVoxels = feGet(fe,'find voxels',coords) 1528 % 1529 % coords is a Nx3 set of coordinates in image space 1530 % foundVoxels is a vector of 1's and 0's, there is a one for each 1531 % location the the connectome coordinates for which there is a match in 1532 % the coords 1533 1534 % Compute the values for a subset of voxels 1535 if isempty(varargin) 1536 % If no coordinates were passed at all, we return all the indices. 1537 val = ones(feGet(fe,'n voxels'),1); 1538 else 1539 % The stored coordinates are at a resolution of the image, typically 1540 % 2mm isotropic. The VOI should also be in image resolution. 1541 % ---- Franco: 1542 % This is how I had it. I think it is worng: 1543 % val = ismember(feGet(fe,'roi coords'), varargin{1}, 'rows'); % This is slow 1544 % This is hwo I think it should be: 1545 [~,val] = ismember(varargin{1}, feGet(fe,'roi coords'),'rows'); % This is slow 1546 end 1547 1548 case {'voxelcoords2voxelrows','coords2rows'} 1549 % Given a set of VOI coords finds the row numbers of the Model matrix 1550 % (or equivalently the dSig vector) that represent the data for this 1551 % set of VOI coords. 1552 % 1553 % foundVoxels = feGet(fe,'coords 2 rows',coords) 1554 % 1555 % coords - a Nx3 set of coordinates in image space 1556 % foundVoxels - a binary vector that is 1 for each Model matrix row 1557 % that corresponds to at least one of the coords. 1558 val = feGet(fe,'voxel rows',find(feGet(fe,'find voxels',varargin{1}))); 1559 1560 % ------ Spatial coordinate transforms for voxels and fg to coordinate frames 1561 case {'xformimg2acpc','xform2acpc','img2acpc','img2acpcxform'} 1562 % Quaternian transformation from IMAGE space to ACPC space. 1563 % 1564 % xform = feGet(fe,'xform') 1565 val = fe.life.xform.img2acpc; 1566 case {'xformacpc2img','xform2img','acpc2img','acpc2imgxform'} 1567 % Quaternian transformation from ACPC space to IMAGE space. 1568 % 1569 % xform = feGet(fe,'xform') 1570 val = fe.life.xform.acpc2img; 1571 case {'volumesize','dims','dim','imagedim'} 1572 % Dimensions of the DW volume. 1573 % 1574 % dim = feGet(fe,'dims') 1575 val = fe.life.imagedim; 1576 case {'mapsize'} 1577 % Dimensions of the maps of parameters and results. 1578 % 1579 % dims = feGet(fe, 'mapsize') 1580 val = fe.life.imagedim(1:3); 1581 1582 case {'anatomyfile'} 1583 % Path to the 3D Anatomy Volume. 1584 % anatomyfile = feGet(fe, 'anatomy file') 1585 val = fe.path.anatomy; 1586 case 'dtfile' 1587 % Diffusion weighted file used for testing results 1588 % dtfile = feGet(fe,'dtfile') 1589 val = fe.path.dtfile; 1590 1591 otherwise 1592 help('feGet') 1593 fprintf('[feGet] Unknown parameter << %s >>...\n',param); 1594 keyboard 1595 end 1596 1597 end % END MAIN FUNCTION