model bundle streamlines used as a reference to extract similar
streamlines from input tractogram
model_clust_thrfloat
MDF distance threshold for the model bundles
reduction_thrfloat, optional
Reduce search space in the target tractogram by (mm) (default 10)
reduction_distancestring, optional
Reduction distance type can be mdf or mam (default mdf)
slrbool, optional
Use Streamline-based Linear Registration (SLR) locally
(default True)
num_threadsint, optional
Number of threads to be used for OpenMP parallelization. If None
(default) the value of OMP_NUM_THREADS environment variable is used
if it is set, otherwise all available threads are used. If < 0 the
maximal number of threads minus \(|num_threads + 1|\) is used (enter
-1 to use as many threads as possible). 0 raises an error.
slr_metricBundleMinDistanceMetric
slr_x0array or int or str, optional
Transformation allowed. translation, rigid, similarity or scaling
Initial parametrization for the optimization.
If 1D array with:
a) 6 elements then only rigid registration is performed with
the 3 first elements for translation and 3 for rotation.
b) 7 elements also isotropic scaling is performed (similarity).
c) 12 elements then translation, rotation (in degrees),
scaling and shearing are performed (affine).
Here is an example of x0 with 12 elements:
x0=np.array([0,10,0,40,0,0,2.,1.5,1,0.1,-0.5,0])
This has translation (0, 10, 0), rotation (40, 0, 0) in
degrees, scaling (2., 1.5, 1) and shearing (0.1, -0.5, 0).
If int:
6
x0=np.array([0,0,0,0,0,0])
7
x0=np.array([0,0,0,0,0,0,1.])
12
x0=np.array([0,0,0,0,0,0,1.,1.,1,0,0,0])
If str:
“rigid”
x0=np.array([0,0,0,0,0,0])
“similarity”
x0=np.array([0,0,0,0,0,0,1.])
“affine”
x0=np.array([0,0,0,0,0,0,1.,1.,1,0,0,0])
slr_boundsarray, optional
SLR bounds.
slr_selecttuple, optional
Select the number of streamlines from model to neighborhood of
model to perform the local SLR.
slr_methodstring, optional
Optimization method ‘L_BFGS_B’ or ‘Powell’ optimizers can be used.
(default ‘L-BFGS-B’)
pruning_thrfloat, optional
Pruning after reducing the search space.
pruning_distancestring, optional
Pruning distance type can be mdf or mam.
Returns:
recognized_transfStreamlines
Recognized bundle in the space of the model tractogram
recognized_labelsarray
Indices of recognized bundle in the original tractogram
Refine and recognize the model_bundle in self.streamlines
This method expects once pruned streamlines as input. It refines the
first output of RecoBundles by applying second local slr (optional),
and second pruning. This method is useful when we are dealing with
noisy data or when we want to extract small tracks from tractograms.
This time, search space is created using pruned bundle and not model
bundle.
See [1], [2] for
further details about the method.
Parameters:
model_bundleStreamlines
model bundle streamlines used as a reference to extract similar
streamlines from input tractogram
pruned_streamlinesStreamlines
Recognized bundle from target tractogram by RecoBundles.
model_clust_thrfloat
MDF distance threshold for the model bundles
reduction_thrfloat
Reduce search space by (mm) (default 14)
reduction_distancestring
Reduction distance type can be mdf or mam (default mdf)
slrbool
Use Streamline-based Linear Registration (SLR) locally.
slr_metricBundleMinDistanceMetric
Bundle distance metric.
slr_x0array or int or str
Transformation allowed. translation, rigid, similarity or scaling
Initial parametrization for the optimization.
If 1D array with:
6 elements then only rigid registration is performed with
the 3 first elements for translation and 3 for rotation.
7 elements also isotropic scaling is performed (similarity).
12 elements then translation, rotation (in degrees),
scaling and shearing are performed (affine).
Here is an example of x0 with 12 elements:
x0=np.array([0,10,0,40,0,0,2.,1.5,1,0.1,-0.5,0])
This has translation (0, 10, 0), rotation (40, 0, 0) in
degrees, scaling (2., 1.5, 1) and shearing (0.1, -0.5, 0).
If int:
6
x0=np.array([0,0,0,0,0,0])
7
x0=np.array([0,0,0,0,0,0,1.])
12
x0=np.array([0,0,0,0,0,0,1.,1.,1,0,0,0])
If str:
“rigid”
x0=np.array([0,0,0,0,0,0])
“similarity”
x0=np.array([0,0,0,0,0,0,1.])
“affine”
x0=np.array([0,0,0,0,0,0,1.,1.,1,0,0,0])
slr_boundsarray
SLR bounds.
slr_selecttuple
Select the number of streamlines from model to neighborhood of
model to perform the local SLR.
slr_methodstring
Optimization method ‘L_BFGS_B’ or ‘Powell’ optimizers can be used.
pruning_thrfloat
Pruning after reducing the search space.
pruning_distancestring
Pruning distance type can be mdf or mam.
Returns:
recognized_transfStreamlines
Recognized bundle in the space of the model tractogram
recognized_labelsarray
Indices of recognized bundle in the original tractogram
Threshold controls
how much strictness user wants while calculating bundle adjacency
between two bundles. Smaller threshold means bundles should be strictly
adjacent to get higher BA score.
Extracted bundle from the whole brain tractogram (eg: AF_L)
expert_bundleStreamlines
Model bundle used as reference while extracting similar type bundle
from input tractogram
nb_ptsinteger, optional
Discretizing streamlines to have nb_pts number of points
thresholdfloat, optional
Threshold used for in computing bundle adjacency. Threshold controls
how much strictness user wants while calculating bundle adjacency
between two bundles. Smaller threshold means bundles should be strictly
adjacent to get higher BA score.
Calculates bundle shape similarity between two given bundles using
bundle adjacency (BA) metric
See [3], [2] for further
details about the method.
Parameters:
bundle1Streamlines
White matter tract from one subject (eg: AF_L)
bundle2Streamlines
White matter tract from another subject (eg: AF_L)
rngnp.random.Generator
Random number generator.
clust_thrarray-like, optional
list of clustering thresholds used in quickbundlesX
thresholdfloat, optional
Threshold used for in computing bundle adjacency. Threshold controls
how much strictness user wants while calculating shape similarity
between two bundles. Smaller threshold means bundles should be strictly
similar to get higher shape similarity score.
This can replace any class supporting indexing used for referencing
(e.g. list, tuple). Indexing an instance of this class will return the
index provided instead of the element. It does not support slicing.
Provides functionalities for interacting with a cluster.
Useful container to retrieve index of elements grouped together. If
a reference to the data is provided to cluster_map, elements will
be returned instead of their index when possible.
Parameters:
cluster_mapClusterMap object
Reference to the set of clusters this cluster is being part of.
idint, optional
Id of this cluster in its associated cluster_map object.
Provides functionalities for interacting with a cluster.
Useful container to retrieve the indices of elements grouped together and
the cluster’s centroid. If a reference to the data is provided to
cluster_map, elements will be returned instead of their index when
possible.
Parameters:
cluster_mapClusterMapCentroid object
Reference to the set of clusters this cluster is being part of.
idint, optional
Id of this cluster in its associated cluster_map object.
Provides functionalities for interacting with clustering outputs.
Useful container to create, remove, retrieve and filter clusters.
If refdata is given, elements will be returned instead of their
index when using Cluster objects.
Provides functionalities for interacting with clustering outputs
that have centroids.
Allows to retrieve easily the centroid of every cluster. Also, it is
a useful container to create, remove, retrieve and filter clusters.
If refdata is given, elements will be returned instead of their
index when using ClusterCentroid objects.
Parameters:
refdatalist
Actual elements that clustered indices refer to.
Attributes:
centroids
clusters
refdata
Methods
add_cluster(*clusters)
Adds one or multiple clusters to this cluster map.
clear()
Remove all clusters from this cluster map.
clusters_sizes()
Gets the size of every cluster contained in this cluster map.
get_large_clusters(min_size)
Gets clusters which contains at least min_size elements.
get_small_clusters(max_size)
Gets clusters which contains at most max_size elements.
remove_cluster(*clusters)
Remove one or multiple clusters from this cluster map.
size()
Gets number of clusters contained in this cluster map.
Given a list of streamlines, the QuickBundles algorithm
[3] sequentially assigns each streamline to its
closest bundle in \(\mathcal{O}(Nk)\) where \(N\) is the number of streamlines
and \(k\) is the final number of bundles. If for a given streamline its
closest bundle is farther than threshold, a new bundle is created and the
streamline is assigned to it except if the number of bundles has already
exceeded max_nb_clusters.
Parameters:
thresholdfloat
The maximum distance from a bundle for a streamline to be still
considered as part of it.
metricstr or Metric object, optional
The distance metric to use when comparing two streamlines. By default,
the Minimum average Direct-Flip (MDF) distance
[3] is used and streamlines are
automatically resampled so they have 12 points.
>>> fromdipy.segment.clusteringimportQuickBundles>>> fromdipy.dataimportget_fnames>>> fromdipy.io.streamlineimportload_tractogram>>> fromdipy.tracking.streamlineimportStreamlines>>> fname=get_fnames(name='fornix')>>> fornix=load_tractogram(fname,'same',... bbox_valid_check=False).streamlines>>> streamlines=Streamlines(fornix)>>> # Segment fornix with a threshold of 10mm and streamlines resampled>>> # to 12 points.>>> qb=QuickBundles(threshold=10.)>>> clusters=qb.cluster(streamlines)>>> len(clusters)4>>> list(map(len,clusters))[61, 191, 47, 1]>>> # Resampling streamlines differently is done explicitly as follows.>>> # Note this has an impact on the speed and the accuracy (tradeoff).>>> fromdipy.segment.featurespeedimportResampleFeature>>> fromdipy.segment.metricspeedimportAveragePointwiseEuclideanMetric>>> feature=ResampleFeature(nb_points=2)>>> metric=AveragePointwiseEuclideanMetric(feature)>>> qb=QuickBundles(threshold=10.,metric=metric)>>> clusters=qb.cluster(streamlines)>>> len(clusters)4>>> list(map(len,clusters))[58, 142, 72, 28]
Thresholds to use for each clustering layer. A threshold represents the
maximum distance from a cluster for a streamline to be still considered
as part of it.
metricstr or Metric object, optional
The distance metric to use when comparing two streamlines. By default,
the Minimum average Direct-Flip (MDF) distance
[3] is used and streamlines are
automatically resampled so they have 12 points.
Run QuickBundlesX and then run again on the centroids of the last layer.
Running again QuickBundles at a layer has the effect of merging
some of the clusters that may be originally divided because of branching.
This function help obtain a result at a QuickBundles quality but with
QuickBundlesX speed. The merging phase has low cost because it is applied
only on the centroids rather than the entire dataset.
See [3] and [4] for
further details about the method.
Parameters:
streamlinesStreamlines
Streamlines.
thresholdssequence
List of distance thresholds for QuickBundlesX.
nb_ptsint
Number of points for discretizing each streamline
select_randomlyint
Randomly select a specific number of streamlines. If None all the
streamlines are used.
rngnumpy.random.Generator
If None then generator is initialized internally.
verbosebool, optional.
If True, log information. Default False.
Returns:
clustersobj
Contains the clusters of the last layer of QuickBundlesX after merging.
Binary mask. Has \(M\) dimensions where \(M <= V\). When \(M < V\), we
append \(V - M\) dimensions with axis length 1 to mask so that mask
will broadcast against vol. In the typical case vol can be 4D,
mask can be 3D, and we append a 1 to the mask shape which (via numpy
broadcasting) has the effect of applying the 3D mask to each 3D slice in
vol (vol[...,0] to vol[...,-1).
Returns:
masked_volndarray
vol multiplied by mask where mask may have been extended to match
extra dimensions in vol
Simple brain extraction tool method for images from DWI data.
It uses a median filter smoothing of the input_volumes vol_idx and an
automatic histogram Otsu thresholding technique, hence the name
median_otsu.
This function is inspired from Mrtrix’s bet which has default values
median_radius=3, numpass=2. However, from tests on multiple 1.5T
and 3T data from GE, Philips, Siemens, the most robust choice is
median_radius=4, numpass=4.
Parameters:
input_volumendarray
3D or 4D array of the brain volume.
vol_idxNone or array, optional
1D array representing indices of axis=3 of a 4D input_volume.
None is only an acceptable input if input_volume is 3D.
median_radiusint, optional
Radius (in voxels) of the applied median filter.
numpass: int, optional
Number of pass of the median filter.
autocrop: bool, optional
if True, the masked input_volume will also be cropped using the
bounding box defined by the masked data. Should be on if DWI is
upsampled to 1x1x1 resolution.
dilateNone or int, optional
number of iterations for binary dilation
finalize_maskbool, optional
Whether to remove potential holes or islands.
Useful for solving minor errors.
Returns:
maskedvolumendarray
Masked input_volume
mask3D ndarray
The binary brain mask
Notes
Copyright (C) 2011, the scikit-image team
All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are
met:
Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in
the documentation and/or other materials provided with the
distribution.
Neither the name of skimage nor the names of its contributors may be
used to endorse or promote products derived from this software without
specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE AUTHOR “AS IS” AND ANY EXPRESS OR
IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
Segment the cfa inside roi using the values from threshold as bounds.
Parameters:
tensor_fitTensorFit object
TensorFit object
roindarray
A binary mask, which contains the bounding box for the segmentation.
thresholdarray-like
An iterable that defines the min and max values to use for the
thresholding.
The values are specified as (R_min, R_max, G_min, G_max, B_min, B_max)
return_cfabool, optional
If True, the cfa is also returned.
Returns:
maskndarray
Binary mask of the segmentation.
cfandarray, optional
Array with shape = (…, 3), where … is the shape of tensor_fit.
The color fractional anisotropy, ordered as a nd array with the last
dimension of size 3 for the R, G and B channels.
It calculates the image intensity histogram, and based on the rate value it
decide what is the upperbound value for intensity normalization, usually
lower bound is 0. The rate is the ratio between the amount of pixels in
every bins and the bins with highest pixel amount
Parameters:
datafloat
Input intensity value data
ratefloat
representing the threshold whether a specific histogram bin that should
be count in the normalization range
This method uses the Maximum a posteriori - Markov Random Field approach for segmentation by using the Iterative Conditional Modes and Expectation Maximization to estimate the parameters.
This method uses the Maximum a posteriori - Markov Random Field
approach for segmentation by using the Iterative Conditional Modes
and Expectation Maximization to estimate the parameters.
Parameters:
imagendarray,
3D structural image.
nclassesint,
Number of desired classes.
betafloat,
Smoothing parameter, the higher this number the smoother the
output will be.
tolerance: float, optional
Value that defines the percentage of change tolerated to
prevent the ICM loop to stop. Default is 1e-05.
If you want tolerance check to be disabled put ‘tolerance = 0’.
max_iterint, optional
Fixed number of desired iterations. Default is 100.
This parameter defines the maximum number of iterations the
algorithm will perform. The loop may terminate early if the
change in energy sum between iterations falls below the
threshold defined by tolerance. However, if tolerance is
explicitly set to 0, this early stopping mechanism is disabled,
and the algorithm will run for the specified number of
iterations unless another stopping criterion is met.
Returns:
initial_segmentationndarray,
3D segmented image with all tissue types specified in nclasses.
final_segmentationndarray,
3D final refined segmentation containing all tissue types.