segment
¶
Module: segment.benchmarks
¶
Module: segment.benchmarks.bench_quickbundles
¶
Benchmarks for QuickBundles
Run all benchmarks with:
import dipy.segment as dipysegment
dipysegment.bench()
If you have doctests enabled by default in nose (with a noserc file or environment variable), and you have a numpy version <= 1.6.1, this will also run the doctests, let’s hope they pass.
Run this benchmark with:
nosetests s –match ‘(?:^[b_.//])[Bb]ench’ bench_quickbundles.py
MDFpy 
Attributes 
Metric 
Computes a distance between two sequential data. 
QB_New 
alias of QuickBundles 
QB_Old 
alias of QuickBundles 
assert_array_equal (x, y[, err_msg, verbose]) 
Raises an AssertionError if two array_like objects are not equal. 
assert_arrays_equal (arrays1, arrays2) 

assert_equal 
Fail if the two objects are unequal as determined by the ‘==’ operator. 
bench_quickbundles () 

get_data ([name]) 
provides filenames of some test datasets or other useful parametrisations 
measure (code_str[, times, label]) 
Return elapsed time for executing code in the namespace of the caller. 
Module: segment.bundles
¶
BundleMinDistanceAsymmetricMetric ([num_threads]) 
Asymmetric Bundlebased Minimum distance 
BundleMinDistanceMetric ([num_threads]) 
Bundlebased Minimum Distance aka BMD 
BundleSumDistanceMatrixMetric ([num_threads]) 
Bundlebased Sum Distance aka BMD 
RecoBundles (streamlines[, cluster_map, ...]) 
Methods 
StreamlineLinearRegistration ([metric, x0, ...]) 
Methods 
Streamlines (*args, **kwargs) 
Attributes 
chain 
chain(*iterables) –> chain object 
apply_affine (aff, pts) 
Apply affine matrix aff to points pts 
bundles_distances_mam 
Calculate distances between list of tracks A and list of tracks B 
bundles_distances_mdf 
Calculate distances between list of tracks A and list of tracks B 
nbytes (streamlines) 

qbx_and_merge (streamlines, thresholds[, ...]) 
Run QuickBundlesX and then run again on the centroids of the last layer 
select_random_set_of_streamlines (...) 
Select a random set of streamlines 
set_number_of_points 
Change the number of points of streamlines 
time (() > floating point number) 
Return the current time in seconds since the Epoch. 
Module: segment.clustering
¶
ABCMeta 
Metaclass for defining Abstract Base Classes (ABCs). 
AveragePointwiseEuclideanMetric 
Computes the average of pointwise Euclidean distances between two sequential data. 
Cluster ([id, indices, refdata]) 
Provides functionalities for interacting with a cluster. 
ClusterCentroid (centroid[, id, indices, refdata]) 
Provides functionalities for interacting with a cluster. 
ClusterMap ([refdata]) 
Provides functionalities for interacting with clustering outputs. 
ClusterMapCentroid ([refdata]) 
Provides functionalities for interacting with clustering outputs that have centroids. 
Clustering 
Methods 
Identity 
Provides identity indexing functionality. 
Metric 
Computes a distance between two sequential data. 
MinimumAverageDirectFlipMetric 
Computes the MDF distance (minimum average directflip) between two sequential data. 
QuickBundles (threshold[, metric, ...]) 
Clusters streamlines using QuickBundles [Garyfallidis12]. 
QuickBundlesX (thresholds[, metric]) 
Clusters streamlines using QuickBundlesX. 
ResampleFeature 
Extracts features from a sequential datum. 
TreeCluster (threshold, centroid[, indices]) 
Attributes 
TreeClusterMap (root) 
Attributes 
abstractmethod (funcobj) 
A decorator indicating abstract methods. 
nbytes (streamlines) 

qbx_and_merge (streamlines, thresholds[, ...]) 
Run QuickBundlesX and then run again on the centroids of the last layer 
set_number_of_points 
Change the number of points of streamlines 
time (() > floating point number) 
Return the current time in seconds since the Epoch. 
Module: segment.mask
¶
applymask (vol, mask) 
Mask vol with mask. 
binary_dilation (input[, structure, ...]) 
Multidimensional binary dilation with the given structuring element. 
bounding_box (vol) 
Compute the bounding box of nonzero intensity voxels in the volume. 
clean_cc_mask (mask) 
Cleans a segmentation of the corpus callosum so no random pixels are included. 
color_fa (fa, evecs) 
Color fractional anisotropy of diffusion tensor 
crop (vol, mins, maxs) 
Crops the input volume. 
fractional_anisotropy (evals[, axis]) 
Fractional anisotropy (FA) of a diffusion tensor. 
generate_binary_structure (rank, connectivity) 
Generate a binary structure for binary morphological operations. 
median_filter (input[, size, footprint, ...]) 
Calculates a multidimensional median filter. 
median_otsu (input_volume[, median_radius, ...]) 
Simple brain extraction tool method for images from DWI data. 
multi_median (input, median_radius, numpass) 
Applies median filter multiple times on input data. 
otsu (image[, nbins]) 
Return threshold value based on Otsu’s method. 
segment_from_cfa (tensor_fit, roi, threshold) 
Segment the cfa inside roi using the values from threshold as bounds. 
warn 
Issue a warning, or maybe ignore it or raise an exception. 
Module: segment.metric
¶
ArcLengthFeature 
Extracts features from a sequential datum. 
AveragePointwiseEuclideanMetric 
Computes the average of pointwise Euclidean distances between two sequential data. 
CenterOfMassFeature 
Extracts features from a sequential datum. 
CosineMetric 
Computes the cosine distance between two vectors. 
EuclideanMetric 
alias of SumPointwiseEuclideanMetric 
Feature 
Extracts features from a sequential datum. 
IdentityFeature 
Extracts features from a sequential datum. 
Metric 
Computes a distance between two sequential data. 
MidpointFeature 
Extracts features from a sequential datum. 
MinimumAverageDirectFlipMetric 
Computes the MDF distance (minimum average directflip) between two sequential data. 
ResampleFeature 
Extracts features from a sequential datum. 
SumPointwiseEuclideanMetric 
Computes the sum of pointwise Euclidean distances between two sequential data. 
VectorOfEndpointsFeature 
Extracts features from a sequential datum. 
dist 
Computes a distance between datum1 and datum2. 
distance_matrix 
Computes the distance matrix between two lists of sequential data. 
mdf (s1, s2) 
Computes the MDF (Minimum average DirectFlip) distance [Garyfallidis12] between two streamlines. 
Module: segment.quickbundles
¶
QuickBundles (tracks[, dist_thr, pts]) 
Attributes 
bundles_distances_mdf 
Calculate distances between list of tracks A and list of tracks B 
downsample (xyz[, n_pols]) 
downsample for a specific number of points along the curve/track 
local_skeleton_clustering 
Efficient tractography clustering 
warn 
Issue a warning, or maybe ignore it or raise an exception. 
Module: segment.threshold
¶
otsu (image[, nbins]) 
Return threshold value based on Otsu’s method. 
upper_bound_by_percent (data[, percent]) 
Find the upper bound for visualization of medical images 
upper_bound_by_rate (data[, rate]) 
Adjusts upper intensity boundary using rates 
Module: segment.tissue
¶
ConstantObservationModel 
Observation model assuming that the intensity of each class is constant. 
IteratedConditionalModes 
Methods 
TissueClassifierHMRF ([save_history, verbose]) 
This class contains the methods for tissue classification using the Markov 
add_noise (signal, snr, S0[, noise_type]) 
Add noise of specified distribution to the signal from a single voxel. 
MDFpy
¶

class
dipy.segment.benchmarks.bench_quickbundles.
MDFpy
¶ Bases:
dipy.segment.metricspeed.Metric
Attributes
feature
Feature object used to extract features from sequential data is_order_invariant
Is this metric invariant to the sequence’s ordering Methods
are_compatible
(shape1, shape2)dist
(features1, features2)
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

are_compatible
(shape1, shape2)¶

dist
(features1, features2)¶

Metric
¶

class
dipy.segment.benchmarks.bench_quickbundles.
Metric
¶ Bases:
object
Computes a distance between two sequential data.
A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions). A feature object can be specified in order to calculate the distance between extracted features, rather than directly between the sequential data.
Parameters: feature : Feature object, optional
It is used to extract features before computing the distance.
Notes
When subclassing Metric, one only needs to override the dist and are_compatible methods.
Attributes
feature
Feature object used to extract features from sequential data is_order_invariant
Is this metric invariant to the sequence’s ordering Methods
are_compatible
Checks if features can be used by metric.dist based on their shape. dist
Computes a distance between two data points based on their features. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

are_compatible
()¶ Checks if features can be used by metric.dist based on their shape.
Basically this method exists so we don’t have to do this check inside the metric.dist function (speedup).
Parameters: shape1 : int, 1tuple or 2tuple
shape of the first data point’s features
shape2 : int, 1tuple or 2tuple
shape of the second data point’s features
Returns: are_compatible : bool
whether or not shapes are compatible

dist
()¶ Computes a distance between two data points based on their features.
Parameters: features1 : 2D array
Features of the first data point.
features2 : 2D array
Features of the second data point.
Returns: double
Distance between two data points.

feature
¶ Feature object used to extract features from sequential data

is_order_invariant
¶ Is this metric invariant to the sequence’s ordering

assert_array_equal¶

dipy.segment.benchmarks.bench_quickbundles.
assert_array_equal
(x, y, err_msg='', verbose=True)¶ Raises an AssertionError if two array_like objects are not equal.
Given two array_like objects, check that the shape is equal and all elements of these objects are equal. An exception is raised at shape mismatch or conflicting values. In contrast to the standard usage in numpy, NaNs are compared like numbers, no assertion is raised if both objects have NaNs in the same positions.
The usual caution for verifying equality with floating point numbers is advised.
Parameters: x : array_like
The actual object to check.
y : array_like
The desired, expected object.
err_msg : str, optional
The error message to be printed in case of failure.
verbose : bool, optional
If True, the conflicting values are appended to the error message.
Raises: AssertionError
If actual and desired objects are not equal.
See also
assert_allclose
 Compare two array_like objects for equality with desired relative and/or absolute precision.
assert_array_almost_equal_nulp
,assert_array_max_ulp
,assert_equal
Examples
The first assert does not raise an exception:
>>> np.testing.assert_array_equal([1.0,2.33333,np.nan], ... [np.exp(0),2.33333, np.nan])
Assert fails with numerical inprecision with floats:
>>> np.testing.assert_array_equal([1.0,np.pi,np.nan], ... [1, np.sqrt(np.pi)**2, np.nan]) ... <type 'exceptions.ValueError'>: AssertionError: Arrays are not equal (mismatch 50.0%) x: array([ 1. , 3.14159265, NaN]) y: array([ 1. , 3.14159265, NaN])
Use assert_allclose or one of the nulp (number of floating point values) functions for these cases instead:
>>> np.testing.assert_allclose([1.0,np.pi,np.nan], ... [1, np.sqrt(np.pi)**2, np.nan], ... rtol=1e10, atol=0)
assert_arrays_equal¶

dipy.segment.benchmarks.bench_quickbundles.
assert_arrays_equal
(arrays1, arrays2)¶
assert_equal¶

dipy.segment.benchmarks.bench_quickbundles.
assert_equal
(self, first, second, msg=None)¶ Fail if the two objects are unequal as determined by the ‘==’ operator.
get_data¶

dipy.segment.benchmarks.bench_quickbundles.
get_data
(name='small_64D')¶ provides filenames of some test datasets or other useful parametrisations
Parameters: name : str
the filename/s of which dataset to return, one of: ‘small_64D’ small region of interest nifti,bvecs,bvals 64 directions ‘small_101D’ small region of interest nifti,bvecs,bvals 101 directions ‘aniso_vox’ volume with anisotropic voxel size as Nifti ‘fornix’ 300 tracks in Trackvis format (from Pittsburgh
Brain Competition)
 ‘gqi_vectors’ the scanner wave vectors needed for a GQI acquisitions
of 101 directions tested on Siemens 3T Trio
‘small_25’ small ROI (10x8x2) DTI data (b value 2000, 25 directions) ‘test_piesno’ slice of N=8, K=14 diffusion data ‘reg_c’ small 2D image used for validating registration ‘reg_o’ small 2D image used for validation registration ‘cb_2’ two vectorized cingulum bundles
Returns: fnames : tuple
filenames for dataset
Examples
>>> import numpy as np >>> from dipy.data import get_data >>> fimg,fbvals,fbvecs=get_data('small_101D') >>> bvals=np.loadtxt(fbvals) >>> bvecs=np.loadtxt(fbvecs).T >>> import nibabel as nib >>> img=nib.load(fimg) >>> data=img.get_data() >>> data.shape == (6, 10, 10, 102) True >>> bvals.shape == (102,) True >>> bvecs.shape == (102, 3) True
measure¶

dipy.segment.benchmarks.bench_quickbundles.
measure
(code_str, times=1, label=None)¶ Return elapsed time for executing code in the namespace of the caller.
The supplied code string is compiled with the Python builtin
compile
. The precision of the timing is 10 milliseconds. If the code will execute fast on this timescale, it can be executed many times to get reasonable timing accuracy.Parameters: code_str : str
The code to be timed.
times : int, optional
The number of times the code is executed. Default is 1. The code is only compiled once.
label : str, optional
A label to identify code_str with. This is passed into
compile
as the second argument (for runtime error messages).Returns: elapsed : float
Total elapsed time in seconds for executing code_str times times.
Examples
>>> etime = np.testing.measure('for i in range(1000): np.sqrt(i**2)', ... times=times) >>> print("Time for a single execution : ", etime / times, "s") Time for a single execution : 0.005 s
BundleMinDistanceAsymmetricMetric
¶

class
dipy.segment.bundles.
BundleMinDistanceAsymmetricMetric
(num_threads=None)¶ Bases:
dipy.align.streamlinear.BundleMinDistanceMetric
Asymmetric Bundlebased Minimum distance
Methods
distance
(xopt)setup
(static, moving)Setup static and moving sets of streamlines 
__init__
(num_threads=None)¶ An abstract class for the metric used for streamline registration
If the two sets of streamlines match exactly then method
distance
of this object should be minimum.Parameters: num_threads : int
Number of threads. If None (default) then all available threads will be used. Only metrics using OpenMP will use this variable.

distance
(xopt)¶

BundleMinDistanceMetric
¶

class
dipy.segment.bundles.
BundleMinDistanceMetric
(num_threads=None)¶ Bases:
dipy.align.streamlinear.StreamlineDistanceMetric
Bundlebased Minimum Distance aka BMD
This is the cost function used by the StreamlineLinearRegistration
References
[Garyfallidis14] Garyfallidis et al., “Direct nativespace fiber bundle alignment for group comparisons”, ISMRM, 2014. Methods
setup(static, moving) distance(xopt) 
__init__
(num_threads=None)¶ An abstract class for the metric used for streamline registration
If the two sets of streamlines match exactly then method
distance
of this object should be minimum.Parameters: num_threads : int
Number of threads. If None (default) then all available threads will be used. Only metrics using OpenMP will use this variable.

distance
(xopt)¶ Distance calculated from this Metric
Parameters: xopt : sequence
List of affine parameters as an 1D vector,

setup
(static, moving)¶ Setup static and moving sets of streamlines
Parameters: static : streamlines
Fixed or reference set of streamlines.
moving : streamlines
Moving streamlines.
num_threads : int
Number of threads. If None (default) then all available threads will be used.
Notes
Call this after the object is initiated and before distance.

BundleSumDistanceMatrixMetric
¶

class
dipy.segment.bundles.
BundleSumDistanceMatrixMetric
(num_threads=None)¶ Bases:
dipy.align.streamlinear.BundleMinDistanceMatrixMetric
Bundlebased Sum Distance aka BMD
This is a cost function that can be used by the StreamlineLinearRegistration class.
Notes
The difference with BundleMinDistanceMatrixMetric is that it uses uses the sum of the distance matrix and not the sum of mins.
Methods
setup(static, moving) distance(xopt) 
__init__
(num_threads=None)¶ An abstract class for the metric used for streamline registration
If the two sets of streamlines match exactly then method
distance
of this object should be minimum.Parameters: num_threads : int
Number of threads. If None (default) then all available threads will be used. Only metrics using OpenMP will use this variable.

distance
(xopt)¶ Distance calculated from this Metric
Parameters: xopt : sequence
List of affine parameters as an 1D vector

RecoBundles
¶

class
dipy.segment.bundles.
RecoBundles
(streamlines, cluster_map=None, clust_thr=15, nb_pts=20, seed=42, verbose=True)¶ Bases:
object
Methods
recognize
(model_bundle, model_clust_thr[, ...])Recognize the model_bundle in self.streamlines 
__init__
(streamlines, cluster_map=None, clust_thr=15, nb_pts=20, seed=42, verbose=True)¶ Recognition of bundles
Extract bundles from a participants’ tractograms using model bundles segmented from a different subject or an atlas of bundles. See [Garyfallidis17] for the details.
Parameters: streamlines : Streamlines
The tractogram in which you want to recognize bundles.
cluster_map : QB map
Provide existing clustering to start RB faster (default None).
clust_thr : float
Distance threshold in mm for clustering streamlines
seed : int
Setup for random number generator (default 42).
nb_pts : int
Number of points per streamline (default 20)
Notes
Make sure that before creating this class that the streamlines and the model bundles are roughly in the same space. Also default thresholds are assumed in RAS 1mm^3 space. You may want to adjust those if your streamlines are not in world coordinates.
References
[Garyfallidis17] Garyfallidis et al. Recognition of white matter bundles using local and global streamlinebased registration and clustering, Neuroimage, 2017.

recognize
(model_bundle, model_clust_thr, reduction_thr=20, reduction_distance='mdf', slr=True, slr_metric=None, slr_x0=None, slr_bounds=None, slr_select=(400, 600), slr_method='LBFGSB', pruning_thr=10, pruning_distance='mdf')¶ Recognize the model_bundle in self.streamlines
Parameters: model_bundle : Streamlines
model_clust_thr : float
reduction_thr : float
reduction_distance : string
mdf or mam (default mam)
slr : bool
Use Streamlinebased Linear Registration (SLR) locally (default True)
slr_metric : BundleMinDistanceMetric
slr_x0 : array
(default None)
slr_bounds : array
(default None)
slr_select : tuple
Select the number of streamlines from model to neirborhood of model to perform the local SLR.
slr_method : string
Optimization method (default ‘LBFGSB’)
pruning_thr : float
pruning_distance : string
MDF (‘mdf’) and MAM (‘mam’)
Returns: recognized_transf : Streamlines
Recognized bundle in the space of the model tractogram
recognized_labels : array
Indices of recognized bundle in the original tractogram
recognized_bundle : Streamlines
Recognized bundle in the space of the original tractogram
References
[Garyfallidis17] Garyfallidis et al. Recognition of white matter bundles using local and global streamlinebased registration and clustering, Neuroimage, 2017.

StreamlineLinearRegistration
¶

class
dipy.segment.bundles.
StreamlineLinearRegistration
(metric=None, x0='rigid', method='LBFGSB', bounds=None, verbose=False, options=None, evolution=False, num_threads=None)¶ Bases:
object
Methods
optimize
(static, moving[, mat])Find the minimum of the provided metric. 
__init__
(metric=None, x0='rigid', method='LBFGSB', bounds=None, verbose=False, options=None, evolution=False, num_threads=None)¶ Linear registration of 2 sets of streamlines [Garyfallidis15].
Parameters: metric : StreamlineDistanceMetric,
If None and fast is False then the BMD distance is used. If fast is True then a faster implementation of BMD is used. Otherwise, use the given distance metric.
x0 : array or int or str
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 is 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])
method : str,
‘L_BFGS_B’ or ‘Powell’ optimizers can be used. Default is ‘L_BFGS_B’.
bounds : list of tuples or None,
If method == ‘L_BFGS_B’ then we can use bounded optimization. For example for the six parameters of rigid rotation we can set the bounds = [(30, 30), (30, 30), (30, 30),
(45, 45), (45, 45), (45, 45)]
That means that we have set the bounds for the three translations and three rotation axes (in degrees).
verbose : bool,
If True then information about the optimization is shown.
options : None or dict,
Extra options to be used with the selected method.
evolution : boolean
If True save the transformation for each iteration of the optimizer. Default is False. Supported only with Scipy >= 0.11.
num_threads : int
Number of threads. If None (default) then all available threads will be used. Only metrics using OpenMP will use this variable.
References
[Garyfallidis15] (1, 2) Garyfallidis et al. “Robust and efficient linear registration of whitematter fascicles in the space of streamlines”, NeuroImage, 117, 124–140, 2015 [Garyfallidis14] Garyfallidis et al., “Direct nativespace fiber bundle alignment for group comparisons”, ISMRM, 2014. [Garyfallidis17] Garyfallidis et al. Recognition of white matter bundles using local and global streamlinebased registration and clustering, Neuroimage, 2017.

optimize
(static, moving, mat=None)¶ Find the minimum of the provided metric.
Parameters: static : streamlines
Reference or fixed set of streamlines.
moving : streamlines
Moving set of streamlines.
mat : array
Transformation (4, 4) matrix to start the registration.
mat
is applied to moving. Default value None which means that initial transformation will be generated by shifting the centers of moving and static sets of streamlines to the origin.Returns: map : StreamlineRegistrationMap

Streamlines
¶

class
dipy.segment.bundles.
Streamlines
(*args, **kwargs)¶ Bases:
nibabel.streamlines.array_sequence.ArraySequence
Attributes
common_shape
Matching shape of the elements in this array sequence. data
Elements in this array sequence. is_array_sequence
total_nb_rows
Total number of rows in this array sequence. Methods
append
(element[, cache_build])Appends element to this array sequence. copy
()Creates a copy of this ArraySequence
object.extend
(elements)Appends all elements to this array sequence. finalize_append
()Finalize process of appending several elements to self append()
can be a lot faster if it knows that it is appending several elements instead of a single element.load
(filename)Loads a ArraySequence
object from a .npz file.save
(filename)Saves this ArraySequence
object to a .npz file.shrink_data
()
__init__
(*args, **kwargs)¶

append
(element, cache_build=False)¶ Appends element to this array sequence. Append can be a lot faster if it knows that it is appending several elements instead of a single element. In that case it can cache the parameters it uses between append operations, in a “build cache”. To tell append to do this, use
cache_build=True
. If you usecache_build=True
, you need to finalize the append operations withfinalize_append()
. Parameters ——— element : ndarrayElement to append. The shape must match already inserted elements shape except for the first dimension. cache_build : {False, True}
 Whether to save the build cache from this append routine. If True,
append can assume it is the only player updating self, and the
caller must finalize self after all append operations, with
self.finalize_append()
.
None Notes —– If you need to add multiple elements you should consider ArraySequence.extend.

extend
(elements)¶ Appends all elements to this array sequence. Parameters ——— elements : iterable of ndarrays or
ArraySequence
objectIf iterable of ndarrays, each ndarray will be concatenated along the first dimension then appended to the data of this ArraySequence. IfArraySequence
object, its data are simply appended to the data of this ArraySequence.None Notes —– The shape of the elements to be added must match the one of the data of this
ArraySequence
except for the first dimension.

finalize_append
()¶ Finalize process of appending several elements to self
append()
can be a lot faster if it knows that it is appending several elements instead of a single element. To tell the append method this is the case, usecache_build=True
. This method finalizes the series of append operations after a call toappend()
withcache_build=True
.

chain
¶

class
dipy.segment.bundles.
chain
¶ Bases:
object
chain(*iterables) –> chain object
Return a chain object whose .__next__() method returns elements from the first iterable until it is exhausted, then elements from the next iterable, until all of the iterables are exhausted.
Methods
from_iterable
chain.from_iterable(iterable) –> chain object 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

from_iterable
()¶ chain.from_iterable(iterable) –> chain object
Alternate chain() constructor taking a single iterable argument that evaluates lazily.

apply_affine¶

dipy.segment.bundles.
apply_affine
(aff, pts)¶ Apply affine matrix aff to points pts
Returns result of application of aff to the right of pts. The coordinate dimension of pts should be the last.
For the 3D case, aff will be shape (4,4) and pts will have final axis length 3  maybe it will just be N by 3. The return value is the transformed points, in this case:
res = np.dot(aff[:3,:3], pts.T) + aff[:3,3:4] transformed_pts = res.T
This routine is more general than 3D, in that aff can have any shape (N,N), and pts can have any shape, as long as the last dimension is for the coordinates, and is therefore length N1.
Parameters: aff : (N, N) arraylike
Homogenous affine, for 3D points, will be 4 by 4. Contrary to first appearance, the affine will be applied on the left of pts.
pts : (..., N1) arraylike
Points, where the last dimension contains the coordinates of each point. For 3D, the last dimension will be length 3.
Returns: transformed_pts : (..., N1) array
transformed points
Examples
>>> aff = np.array([[0,2,0,10],[3,0,0,11],[0,0,4,12],[0,0,0,1]]) >>> pts = np.array([[1,2,3],[2,3,4],[4,5,6],[6,7,8]]) >>> apply_affine(aff, pts) array([[14, 14, 24], [16, 17, 28], [20, 23, 36], [24, 29, 44]]...)
Just to show that in the simple 3D case, it is equivalent to:
>>> (np.dot(aff[:3,:3], pts.T) + aff[:3,3:4]).T array([[14, 14, 24], [16, 17, 28], [20, 23, 36], [24, 29, 44]]...)
But pts can be a more complicated shape:
>>> pts = pts.reshape((2,2,3)) >>> apply_affine(aff, pts) array([[[14, 14, 24], [16, 17, 28]], [[20, 23, 36], [24, 29, 44]]]...)
bundles_distances_mam¶

dipy.segment.bundles.
bundles_distances_mam
()¶ Calculate distances between list of tracks A and list of tracks B
Parameters: tracksA : sequence
of tracks as arrays, shape (N1,3) .. (Nm,3)
tracksB : sequence
of tracks as arrays, shape (N1,3) .. (Nm,3)
metric : str
‘avg’, ‘min’, ‘max’
Returns: DM : array, shape (len(tracksA), len(tracksB))
distances between tracksA and tracksB according to metric
bundles_distances_mdf¶

dipy.segment.bundles.
bundles_distances_mdf
()¶ Calculate distances between list of tracks A and list of tracks B
All tracks need to have the same number of points
Parameters: tracksA : sequence
of tracks as arrays, [(N,3) .. (N,3)]
tracksB : sequence
of tracks as arrays, [(N,3) .. (N,3)]
Returns: DM : array, shape (len(tracksA), len(tracksB))
distances between tracksA and tracksB according to metric
See also
dipy.metrics.downsample
qbx_and_merge¶

dipy.segment.bundles.
qbx_and_merge
(streamlines, thresholds, nb_pts=20, select_randomly=None, rng=None, verbose=True)¶ 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 maybe originally devided 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.
Parameters: streamlines : Streamlines
thresholds : sequence
List of distance thresholds for QuickBundlesX.
nb_pts : int
Number of points for discretizing each streamline
select_randomly : int
Randomly select a specific number of streamlines. If None all the streamlines are used.
rng : RandomState
If None then RandomState is initialized internally.
verbose : bool
If True print information in stdout.
Returns: clusters : obj
Contains the clusters of the last layer of QuickBundlesX after merging.
References
[Garyfallidis12] Garyfallidis E. et al., QuickBundles a method for tractography simplification, Frontiers in Neuroscience, vol 6, no 175, 2012. [Garyfallidis16] Garyfallidis E. et al. QuickBundlesX: Sequential clustering of millions of streamlines in multiple levels of detail at record execution time. Proceedings of the, International Society of Magnetic Resonance in Medicine (ISMRM). Singapore, 4187, 2016.
select_random_set_of_streamlines¶

dipy.segment.bundles.
select_random_set_of_streamlines
(streamlines, select)¶ Select a random set of streamlines
Parameters: streamlines : list
List of 2D ndarrays of shape[1]==3
select : int
Number of streamlines to select. If there are less streamlines than
select
thenselect=len(streamlines)
.Returns: selected_streamlines : list
Notes
The same streamline will not be selected twice.
set_number_of_points¶

dipy.segment.bundles.
set_number_of_points
()¶  Change the number of points of streamlines
 (either by downsampling or upsampling)
Change the number of points of streamlines in order to obtain nb_points1 segments of equal length. Points of streamlines will be modified along the curve.
Parameters: streamlines : ndarray or a list or
dipy.tracking.Streamlines
If ndarray, must have shape (N,3) where N is the number of points of the streamline. If list, each item must be ndarray shape (Ni,3) where Ni is the number of points of streamline i. If
dipy.tracking.Streamlines
, its common_shape must be 3.nb_points : int
integer representing number of points wanted along the curve.
Returns: new_streamlines : ndarray or a list or
dipy.tracking.Streamlines
Results of the downsampling or upsampling process.
Examples
>>> from dipy.tracking.streamline import set_number_of_points >>> import numpy as np
One streamline, a semicircle:
>>> theta = np.pi*np.linspace(0, 1, 100) >>> x = np.cos(theta) >>> y = np.sin(theta) >>> z = 0 * x >>> streamline = np.vstack((x, y, z)).T >>> modified_streamline = set_number_of_points(streamline, 3) >>> len(modified_streamline) 3
Multiple streamlines:
>>> streamlines = [streamline, streamline[::2]] >>> new_streamlines = set_number_of_points(streamlines, 10) >>> [len(s) for s in streamlines] [100, 50] >>> [len(s) for s in new_streamlines] [10, 10]
time¶

dipy.segment.bundles.
time
() → floating point number¶ Return the current time in seconds since the Epoch. Fractions of a second may be present if the system clock provides them.
ABCMeta
¶

class
dipy.segment.clustering.
ABCMeta
¶ Bases:
type
Metaclass for defining Abstract Base Classes (ABCs).
Use this metaclass to create an ABC. An ABC can be subclassed directly, and then acts as a mixin class. You can also register unrelated concrete classes (even builtin classes) and unrelated ABCs as ‘virtual subclasses’ – these and their descendants will be considered subclasses of the registering ABC by the builtin issubclass() function, but the registering ABC won’t show up in their MRO (Method Resolution Order) nor will method implementations defined by the registering ABC be callable (not even via super()).
Methods
__call__
Call self as a function. mro
(() > list)return a type’s method resolution order register
(subclass)Register a virtual subclass of an ABC. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

register
(subclass)¶ Register a virtual subclass of an ABC.
Returns the subclass, to allow usage as a class decorator.

AveragePointwiseEuclideanMetric
¶

class
dipy.segment.clustering.
AveragePointwiseEuclideanMetric
¶ Bases:
dipy.segment.metricspeed.SumPointwiseEuclideanMetric
Computes the average of pointwise Euclidean distances between two sequential data.
A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions). A feature object can be specified in order to calculate the distance between the features, rather than directly between the sequential data.
Parameters: feature : Feature object, optional
It is used to extract features before computing the distance.
Notes
The distance between two 2D sequential data:
s1 s2 0* a *0 \  \  1*   b *1  \ 2* \ c *2
is equal to \((a+b+c)/3\) where \(a\) is the Euclidean distance between s1[0] and s2[0], \(b\) between s1[1] and s2[1] and \(c\) between s1[2] and s2[2].
Attributes
feature
Feature object used to extract features from sequential data is_order_invariant
Is this metric invariant to the sequence’s ordering Methods
are_compatible
Checks if features can be used by metric.dist based on their shape. dist
Computes a distance between two data points based on their features. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

Cluster
¶

class
dipy.segment.clustering.
Cluster
(id=0, indices=None, refdata=<dipy.segment.clustering.Identity object>)¶ Bases:
object
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_map : ClusterMap object
Reference to the set of clusters this cluster is being part of.
id : int
Id of this cluster in its associated cluster_map object.
refdata : list (optional)
Actual elements that clustered indices refer to.
Notes
A cluster does not contain actual data but instead knows how to retrieve them using its ClusterMap object.
Methods
assign
(*indices)Assigns indices to this cluster. 
__init__
(id=0, indices=None, refdata=<dipy.segment.clustering.Identity object>)¶

assign
(*indices)¶ Assigns indices to this cluster.
Parameters: *indices : list of indices
Indices to add to this cluster.

ClusterCentroid
¶

class
dipy.segment.clustering.
ClusterCentroid
(centroid, id=0, indices=None, refdata=<dipy.segment.clustering.Identity object>)¶ Bases:
dipy.segment.clustering.Cluster
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_map : ClusterMapCentroid object
Reference to the set of clusters this cluster is being part of.
id : int
Id of this cluster in its associated cluster_map object.
refdata : list (optional)
Actual elements that clustered indices refer to.
Notes
A cluster does not contain actual data but instead knows how to retrieve them using its ClusterMapCentroid object.
Methods
assign
(id_datum, features)Assigns a data point to this cluster. update
()Update centroid of this cluster. 
__init__
(centroid, id=0, indices=None, refdata=<dipy.segment.clustering.Identity object>)¶

assign
(id_datum, features)¶ Assigns a data point to this cluster.
Parameters: id_datum : int
Index of the data point to add to this cluster.
features : 2D array
Data point’s features to modify this cluster’s centroid.

update
()¶ Update centroid of this cluster.
Returns: converged : bool
Tells if the centroid has moved.

ClusterMap
¶

class
dipy.segment.clustering.
ClusterMap
(refdata=<dipy.segment.clustering.Identity object>)¶ Bases:
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.
Parameters: refdata : list
Actual elements that clustered indices refer to.
Attributes
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. 
__init__
(refdata=<dipy.segment.clustering.Identity object>)¶

add_cluster
(*clusters)¶ Adds one or multiple clusters to this cluster map.
Parameters: *clusters : Cluster object, ...
Cluster(s) to be added in this cluster map.

clear
()¶ Remove all clusters from this cluster map.

clusters
¶

clusters_sizes
()¶ Gets the size of every cluster contained in this cluster map.
Returns: list of int
Sizes of every cluster in this cluster map.

get_large_clusters
(min_size)¶ Gets clusters which contains at least min_size elements.
Parameters: min_size : int
Minimum number of elements a cluster needs to have to be selected.
Returns: list of Cluster objects
Clusters having at least min_size elements.

get_small_clusters
(max_size)¶ Gets clusters which contains at most max_size elements.
Parameters: max_size : int
Maximum number of elements a cluster can have to be selected.
Returns: list of Cluster objects
Clusters having at most max_size elements.

refdata
¶

remove_cluster
(*clusters)¶ Remove one or multiple clusters from this cluster map.
Parameters: *clusters : Cluster object, ...
Cluster(s) to be removed from this cluster map.

size
()¶ Gets number of clusters contained in this cluster map.

ClusterMapCentroid
¶

class
dipy.segment.clustering.
ClusterMapCentroid
(refdata=<dipy.segment.clustering.Identity object>)¶ Bases:
dipy.segment.clustering.ClusterMap
Provides functionalities for interacting with clustering outputs that have centroids.
Allows to retrieve easely 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: refdata : list
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. 
__init__
(refdata=<dipy.segment.clustering.Identity object>)¶

centroids
¶

Clustering
¶

class
dipy.segment.clustering.
Clustering
¶ Bases:
object
Methods
cluster
(data[, ordering])Clusters data. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

cluster
(data, ordering=None)¶ Clusters data.
Subclasses will perform their clustering algorithm here.
Parameters: data : list of Ndimensional arrays
Each array represents a data point.
ordering : iterable of indices, optional
Specifies the order in which data points will be clustered.
Returns: ClusterMap object
Result of the clustering.

Identity
¶

class
dipy.segment.clustering.
Identity
¶ Bases:
object
Provides identity indexing functionality.
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.

__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

Metric
¶

class
dipy.segment.clustering.
Metric
¶ Bases:
object
Computes a distance between two sequential data.
A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions). A feature object can be specified in order to calculate the distance between extracted features, rather than directly between the sequential data.
Parameters: feature : Feature object, optional
It is used to extract features before computing the distance.
Notes
When subclassing Metric, one only needs to override the dist and are_compatible methods.
Attributes
feature
Feature object used to extract features from sequential data is_order_invariant
Is this metric invariant to the sequence’s ordering Methods
are_compatible
Checks if features can be used by metric.dist based on their shape. dist
Computes a distance between two data points based on their features. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

are_compatible
()¶ Checks if features can be used by metric.dist based on their shape.
Basically this method exists so we don’t have to do this check inside the metric.dist function (speedup).
Parameters: shape1 : int, 1tuple or 2tuple
shape of the first data point’s features
shape2 : int, 1tuple or 2tuple
shape of the second data point’s features
Returns: are_compatible : bool
whether or not shapes are compatible

dist
()¶ Computes a distance between two data points based on their features.
Parameters: features1 : 2D array
Features of the first data point.
features2 : 2D array
Features of the second data point.
Returns: double
Distance between two data points.

feature
¶ Feature object used to extract features from sequential data

is_order_invariant
¶ Is this metric invariant to the sequence’s ordering

MinimumAverageDirectFlipMetric
¶

class
dipy.segment.clustering.
MinimumAverageDirectFlipMetric
¶ Bases:
dipy.segment.metricspeed.AveragePointwiseEuclideanMetric
Computes the MDF distance (minimum average directflip) between two sequential data.
A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions).
Notes
The distance between two 2D sequential data:
s1 s2 0* a *0 \  \  1*   b *1  \ 2* \ c *2
is equal to \(\min((a+b+c)/3, (a'+b'+c')/3)\) where \(a\) is the Euclidean distance between s1[0] and s2[0], \(b\) between s1[1] and s2[1], \(c\) between s1[2] and s2[2], \(a'\) between s1[0] and s2[2], \(b'\) between s1[1] and s2[1] and \(c'\) between s1[2] and s2[0].
Attributes
feature
Feature object used to extract features from sequential data is_order_invariant
Is this metric invariant to the sequence’s ordering Methods
are_compatible
Checks if features can be used by metric.dist based on their shape. dist
Computes a distance between two data points based on their features. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

is_order_invariant
¶ Is this metric invariant to the sequence’s ordering

QuickBundles
¶

class
dipy.segment.clustering.
QuickBundles
(threshold, metric='MDF_12points', max_nb_clusters=2147483647)¶ Bases:
dipy.segment.clustering.Clustering
Clusters streamlines using QuickBundles [Garyfallidis12].
Given a list of streamlines, the QuickBundles algorithm 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: threshold : float
The maximum distance from a bundle for a streamline to be still considered as part of it.
metric : str or Metric object (optional)
The distance metric to use when comparing two streamlines. By default, the Minimum average DirectFlip (MDF) distance [Garyfallidis12] is used and streamlines are automatically resampled so they have 12 points.
max_nb_clusters : int
Limits the creation of bundles.
References
[Garyfallidis12] Garyfallidis E. et al., QuickBundles a method for tractography simplification, Frontiers in Neuroscience, vol 6, no 175, 2012. Examples
>>> from dipy.segment.clustering import QuickBundles >>> from dipy.data import get_data >>> from nibabel import trackvis as tv >>> streams, hdr = tv.read(get_data('fornix')) >>> streamlines = [i[0] for i in streams] >>> # Segment fornix with a treshold 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). >>> from dipy.segment.metric import ResampleFeature >>> from dipy.segment.metric import AveragePointwiseEuclideanMetric >>> 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]
Methods
cluster
(streamlines[, ordering])Clusters streamlines into bundles. 
__init__
(threshold, metric='MDF_12points', max_nb_clusters=2147483647)¶

cluster
(streamlines, ordering=None)¶ Clusters streamlines into bundles.
Performs quickbundles algorithm using predefined metric and threshold.
Parameters: streamlines : list of 2D arrays
Each 2D array represents a sequence of 3D points (points, 3).
ordering : iterable of indices
Specifies the order in which data points will be clustered.
Returns: ClusterMapCentroid object
Result of the clustering.

QuickBundlesX
¶

class
dipy.segment.clustering.
QuickBundlesX
(thresholds, metric='MDF_12points')¶ Bases:
dipy.segment.clustering.Clustering
Clusters streamlines using QuickBundlesX.
Parameters: thresholds : list of float
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.
metric : str or Metric object (optional)
The distance metric to use when comparing two streamlines. By default, the Minimum average DirectFlip (MDF) distance [Garyfallidis12] is used and streamlines are automatically resampled so they have 12 points.
References
[Garyfallidis12] Garyfallidis E. et al., QuickBundles a method for tractography simplification, Frontiers in Neuroscience, vol 6, no 175, 2012. [Garyfallidis16] Garyfallidis E. et al. QuickBundlesX: Sequential clustering of millions of streamlines in multiple levels of detail at record execution time. Proceedings of the, International Society of Magnetic Resonance in Medicine (ISMRM). Singapore, 4187, 2016. Methods
cluster
(streamlines[, ordering])Clusters streamlines into bundles. 
__init__
(thresholds, metric='MDF_12points')¶

cluster
(streamlines, ordering=None)¶ Clusters streamlines into bundles.
Performs QuickbundleX using a predefined metric and thresholds.
Parameters: streamlines : list of 2D arrays
Each 2D array represents a sequence of 3D points (points, 3).
ordering : iterable of indices
Specifies the order in which data points will be clustered.
Returns: TreeClusterMap object
Result of the clustering.

ResampleFeature
¶

class
dipy.segment.clustering.
ResampleFeature
¶ Bases:
dipy.segment.featurespeed.CythonFeature
Extracts features from a sequential datum.
A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions).
The features being extracted are the points of the sequence once resampled. This is useful for metrics requiring a constant number of points for all
streamlines.Attributes
is_order_invariant
Is this feature invariant to the sequence’s ordering Methods
extract
Extracts features from a sequential datum. infer_shape
Infers the shape of features extracted from a sequential datum. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

TreeCluster
¶

class
dipy.segment.clustering.
TreeCluster
(threshold, centroid, indices=None)¶ Bases:
dipy.segment.clustering.ClusterCentroid
Attributes
is_leaf
Methods
add
(child)assign
(id_datum, features)Assigns a data point to this cluster. update
()Update centroid of this cluster. 
__init__
(threshold, centroid, indices=None)¶

add
(child)¶

is_leaf
¶

TreeClusterMap
¶

class
dipy.segment.clustering.
TreeClusterMap
(root)¶ Bases:
dipy.segment.clustering.ClusterMap
Attributes
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_clusters
(wanted_level)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. iter_preorder
(node)remove_cluster
(*clusters)Remove one or multiple clusters from this cluster map. size
()Gets number of clusters contained in this cluster map. traverse_postorder
(node, visit)
__init__
(root)¶

get_clusters
(wanted_level)¶

iter_preorder
(node)¶

refdata
¶

traverse_postorder
(node, visit)¶

abstractmethod¶

dipy.segment.clustering.
abstractmethod
(funcobj)¶ A decorator indicating abstract methods.
Requires that the metaclass is ABCMeta or derived from it. A class that has a metaclass derived from ABCMeta cannot be instantiated unless all of its abstract methods are overridden. The abstract methods can be called using any of the normal ‘super’ call mechanisms.
Usage:
 class C(metaclass=ABCMeta):
@abstractmethod def my_abstract_method(self, ...):
...
qbx_and_merge¶

dipy.segment.clustering.
qbx_and_merge
(streamlines, thresholds, nb_pts=20, select_randomly=None, rng=None, verbose=True)¶ 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 maybe originally devided 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.
Parameters: streamlines : Streamlines
thresholds : sequence
List of distance thresholds for QuickBundlesX.
nb_pts : int
Number of points for discretizing each streamline
select_randomly : int
Randomly select a specific number of streamlines. If None all the streamlines are used.
rng : RandomState
If None then RandomState is initialized internally.
verbose : bool
If True print information in stdout.
Returns: clusters : obj
Contains the clusters of the last layer of QuickBundlesX after merging.
References
[Garyfallidis12] Garyfallidis E. et al., QuickBundles a method for tractography simplification, Frontiers in Neuroscience, vol 6, no 175, 2012. [Garyfallidis16] Garyfallidis E. et al. QuickBundlesX: Sequential clustering of millions of streamlines in multiple levels of detail at record execution time. Proceedings of the, International Society of Magnetic Resonance in Medicine (ISMRM). Singapore, 4187, 2016.
set_number_of_points¶

dipy.segment.clustering.
set_number_of_points
()¶  Change the number of points of streamlines
 (either by downsampling or upsampling)
Change the number of points of streamlines in order to obtain nb_points1 segments of equal length. Points of streamlines will be modified along the curve.
Parameters: streamlines : ndarray or a list or
dipy.tracking.Streamlines
If ndarray, must have shape (N,3) where N is the number of points of the streamline. If list, each item must be ndarray shape (Ni,3) where Ni is the number of points of streamline i. If
dipy.tracking.Streamlines
, its common_shape must be 3.nb_points : int
integer representing number of points wanted along the curve.
Returns: new_streamlines : ndarray or a list or
dipy.tracking.Streamlines
Results of the downsampling or upsampling process.
Examples
>>> from dipy.tracking.streamline import set_number_of_points >>> import numpy as np
One streamline, a semicircle:
>>> theta = np.pi*np.linspace(0, 1, 100) >>> x = np.cos(theta) >>> y = np.sin(theta) >>> z = 0 * x >>> streamline = np.vstack((x, y, z)).T >>> modified_streamline = set_number_of_points(streamline, 3) >>> len(modified_streamline) 3
Multiple streamlines:
>>> streamlines = [streamline, streamline[::2]] >>> new_streamlines = set_number_of_points(streamlines, 10) >>> [len(s) for s in streamlines] [100, 50] >>> [len(s) for s in new_streamlines] [10, 10]
time¶

dipy.segment.clustering.
time
() → floating point number¶ Return the current time in seconds since the Epoch. Fractions of a second may be present if the system clock provides them.
applymask¶

dipy.segment.mask.
applymask
(vol, mask)¶ Mask vol with mask.
Parameters: vol : ndarray
Array with \(V\) dimensions
mask : ndarray
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 appling the 3D mask to each 3D slice in vol (
vol[..., 0]
tovol[..., 1
).Returns: masked_vol : ndarray
vol multiplied by mask where mask may have been extended to match extra dimensions in vol
binary_dilation¶

dipy.segment.mask.
binary_dilation
(input, structure=None, iterations=1, mask=None, output=None, border_value=0, origin=0, brute_force=False)¶ Multidimensional binary dilation with the given structuring element.
Parameters: input : array_like
Binary array_like to be dilated. Nonzero (True) elements form the subset to be dilated.
structure : array_like, optional
Structuring element used for the dilation. Nonzero elements are considered True. If no structuring element is provided an element is generated with a square connectivity equal to one.
iterations : {int, float}, optional
The dilation is repeated iterations times (one, by default). If iterations is less than 1, the dilation is repeated until the result does not change anymore.
mask : array_like, optional
If a mask is given, only those elements with a True value at the corresponding mask element are modified at each iteration.
output : ndarray, optional
Array of the same shape as input, into which the output is placed. By default, a new array is created.
origin : int or tuple of ints, optional
Placement of the filter, by default 0.
border_value : int (cast to 0 or 1), optional
Value at the border in the output array.
Returns: binary_dilation : ndarray of bools
Dilation of the input by the structuring element.
See also
grey_dilation
,binary_erosion
,binary_closing
,binary_opening
,generate_binary_structure
Notes
Dilation [R424] is a mathematical morphology operation [R425] that uses a structuring element for expanding the shapes in an image. The binary dilation of an image by a structuring element is the locus of the points covered by the structuring element, when its center lies within the nonzero points of the image.
References
[R424] (1, 2) http://en.wikipedia.org/wiki/Dilation_%28morphology%29 [R425] (1, 2) http://en.wikipedia.org/wiki/Mathematical_morphology Examples
>>> from scipy import ndimage >>> a = np.zeros((5, 5)) >>> a[2, 2] = 1 >>> a array([[ 0., 0., 0., 0., 0.], [ 0., 0., 0., 0., 0.], [ 0., 0., 1., 0., 0.], [ 0., 0., 0., 0., 0.], [ 0., 0., 0., 0., 0.]]) >>> ndimage.binary_dilation(a) array([[False, False, False, False, False], [False, False, True, False, False], [False, True, True, True, False], [False, False, True, False, False], [False, False, False, False, False]], dtype=bool) >>> ndimage.binary_dilation(a).astype(a.dtype) array([[ 0., 0., 0., 0., 0.], [ 0., 0., 1., 0., 0.], [ 0., 1., 1., 1., 0.], [ 0., 0., 1., 0., 0.], [ 0., 0., 0., 0., 0.]]) >>> # 3x3 structuring element with connectivity 1, used by default >>> struct1 = ndimage.generate_binary_structure(2, 1) >>> struct1 array([[False, True, False], [ True, True, True], [False, True, False]], dtype=bool) >>> # 3x3 structuring element with connectivity 2 >>> struct2 = ndimage.generate_binary_structure(2, 2) >>> struct2 array([[ True, True, True], [ True, True, True], [ True, True, True]], dtype=bool) >>> ndimage.binary_dilation(a, structure=struct1).astype(a.dtype) array([[ 0., 0., 0., 0., 0.], [ 0., 0., 1., 0., 0.], [ 0., 1., 1., 1., 0.], [ 0., 0., 1., 0., 0.], [ 0., 0., 0., 0., 0.]]) >>> ndimage.binary_dilation(a, structure=struct2).astype(a.dtype) array([[ 0., 0., 0., 0., 0.], [ 0., 1., 1., 1., 0.], [ 0., 1., 1., 1., 0.], [ 0., 1., 1., 1., 0.], [ 0., 0., 0., 0., 0.]]) >>> ndimage.binary_dilation(a, structure=struct1,\ ... iterations=2).astype(a.dtype) array([[ 0., 0., 1., 0., 0.], [ 0., 1., 1., 1., 0.], [ 1., 1., 1., 1., 1.], [ 0., 1., 1., 1., 0.], [ 0., 0., 1., 0., 0.]])
bounding_box¶

dipy.segment.mask.
bounding_box
(vol)¶ Compute the bounding box of nonzero intensity voxels in the volume.
Parameters: vol : ndarray
Volume to compute bounding box on.
Returns: npmins : list
Array containg minimum index of each dimension
npmaxs : list
Array containg maximum index of each dimension
clean_cc_mask¶

dipy.segment.mask.
clean_cc_mask
(mask)¶ Cleans a segmentation of the corpus callosum so no random pixels are included.
Parameters: mask : ndarray
Binary mask of the coarse segmentation.
Returns: new_cc_mask : ndarray
Binary mask of the cleaned segmentation.
color_fa¶

dipy.segment.mask.
color_fa
(fa, evecs)¶ Color fractional anisotropy of diffusion tensor
Parameters: fa : arraylike
Array of the fractional anisotropy (can be 1D, 2D or 3D)
 evecs : arraylike
eigen vectors from the tensor model
Returns: rgb : Array with 3 channels for each color as the last dimension.
Colormap of the FA with red for the x value, y for the green value and z for the blue value.
ec{e})) imes fa
crop¶

dipy.segment.mask.
crop
(vol, mins, maxs)¶ Crops the input volume.
Parameters: vol : ndarray
Volume to crop.
mins : array
Array containg minimum index of each dimension.
maxs : array
Array containg maximum index of each dimension.
Returns: vol : ndarray
The cropped volume.
fractional_anisotropy¶

dipy.segment.mask.
fractional_anisotropy
(evals, axis=1)¶  Fractional anisotropy (FA) of a diffusion tensor.
Parameters: evals : arraylike
Eigenvalues of a diffusion tensor.
 axis : int
Axis of evals which contains 3 eigenvalues.
Returns: fa : array
Calculated FA. Range is 0 <= FA <= 1.
rac{1}{2} rac{(lambda_1lambda_2)^2+(lambda_1
lambda_3)^2+(lambda_2lambda_3)^2}{lambda_1^2+ lambda_2^2+lambda_3^2}}
generate_binary_structure¶

dipy.segment.mask.
generate_binary_structure
(rank, connectivity)¶ Generate a binary structure for binary morphological operations.
Parameters: rank : int
Number of dimensions of the array to which the structuring element will be applied, as returned by np.ndim.
connectivity : int
connectivity determines which elements of the output array belong to the structure, i.e. are considered as neighbors of the central element. Elements up to a squared distance of connectivity from the center are considered neighbors. connectivity may range from 1 (no diagonal elements are neighbors) to rank (all elements are neighbors).
Returns: output : ndarray of bools
Structuring element which may be used for binary morphological operations, with rank dimensions and all dimensions equal to 3.
See also
iterate_structure
,binary_dilation
,binary_erosion
Notes
generate_binary_structure can only create structuring elements with dimensions equal to 3, i.e. minimal dimensions. For larger structuring elements, that are useful e.g. for eroding large objects, one may either use iterate_structure, or create directly custom arrays with numpy functions such as numpy.ones.
Examples
>>> from scipy import ndimage >>> struct = ndimage.generate_binary_structure(2, 1) >>> struct array([[False, True, False], [ True, True, True], [False, True, False]], dtype=bool) >>> a = np.zeros((5,5)) >>> a[2, 2] = 1 >>> a array([[ 0., 0., 0., 0., 0.], [ 0., 0., 0., 0., 0.], [ 0., 0., 1., 0., 0.], [ 0., 0., 0., 0., 0.], [ 0., 0., 0., 0., 0.]]) >>> b = ndimage.binary_dilation(a, structure=struct).astype(a.dtype) >>> b array([[ 0., 0., 0., 0., 0.], [ 0., 0., 1., 0., 0.], [ 0., 1., 1., 1., 0.], [ 0., 0., 1., 0., 0.], [ 0., 0., 0., 0., 0.]]) >>> ndimage.binary_dilation(b, structure=struct).astype(a.dtype) array([[ 0., 0., 1., 0., 0.], [ 0., 1., 1., 1., 0.], [ 1., 1., 1., 1., 1.], [ 0., 1., 1., 1., 0.], [ 0., 0., 1., 0., 0.]]) >>> struct = ndimage.generate_binary_structure(2, 2) >>> struct array([[ True, True, True], [ True, True, True], [ True, True, True]], dtype=bool) >>> struct = ndimage.generate_binary_structure(3, 1) >>> struct # no diagonal elements array([[[False, False, False], [False, True, False], [False, False, False]], [[False, True, False], [ True, True, True], [False, True, False]], [[False, False, False], [False, True, False], [False, False, False]]], dtype=bool)
median_filter¶

dipy.segment.mask.
median_filter
(input, size=None, footprint=None, output=None, mode='reflect', cval=0.0, origin=0)¶ Calculates a multidimensional median filter.
Parameters: input : array_like
Input array to filter.
size : scalar or tuple, optional
See footprint, below
footprint : array, optional
Either size or footprint must be defined. size gives the shape that is taken from the input array, at every element position, to define the input to the filter function. footprint is a boolean array that specifies (implicitly) a shape, but also which of the elements within this shape will get passed to the filter function. Thus
size=(n,m)
is equivalent tofootprint=np.ones((n,m))
. We adjust size to the number of dimensions of the input array, so that, if the input array is shape (10,10,10), and size is 2, then the actual size used is (2,2,2).output : array, optional
The output parameter passes an array in which to store the filter output. Output array should have different name as compared to input array to avoid aliasing errors.
mode : {‘reflect’, ‘constant’, ‘nearest’, ‘mirror’, ‘wrap’}, optional
The mode parameter determines how the array borders are handled, where cval is the value when mode is equal to ‘constant’. Default is ‘reflect’
cval : scalar, optional
Value to fill past edges of input if mode is ‘constant’. Default is 0.0
origin : scalar, optional
The origin parameter controls the placement of the filter. Default 0.0.
Returns: median_filter : ndarray
Filtered array. Has the same shape as input.
Examples
>>> from scipy import ndimage, misc >>> import matplotlib.pyplot as plt >>> fig = plt.figure() >>> plt.gray() # show the filtered result in grayscale >>> ax1 = fig.add_subplot(121) # left side >>> ax2 = fig.add_subplot(122) # right side >>> ascent = misc.ascent() >>> result = ndimage.median_filter(ascent, size=20) >>> ax1.imshow(ascent) >>> ax2.imshow(result) >>> plt.show()
median_otsu¶

dipy.segment.mask.
median_otsu
(input_volume, median_radius=4, numpass=4, autocrop=False, vol_idx=None, dilate=None)¶ 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 ismedian_radius=4
,numpass=4
.Parameters: input_volume : ndarray
ndarray of the brain volume
median_radius : int
Radius (in voxels) of the applied median filter (default: 4).
numpass: int
Number of pass of the median filter (default: 4).
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. (default: False).
vol_idx : None or array, optional
1D array representing indices of
axis=3
of a 4D input_volume None (the default) corresponds to(0,)
(assumes first volume in 4D array).dilate : None or int, optional
number of iterations for binary dilation
Returns: maskedvolume : ndarray
Masked input_volume
mask : 3D ndarray
The binary brain mask
Notes
Copyright (C) 2011, the scikitimage 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.
multi_median¶

dipy.segment.mask.
multi_median
(input, median_radius, numpass)¶ Applies median filter multiple times on input data.
Parameters: input : ndarray
The input volume to apply filter on.
median_radius : int
Radius (in voxels) of the applied median filter
numpass: int
Number of pass of the median filter
Returns: input : ndarray
Filtered input volume.
otsu¶

dipy.segment.mask.
otsu
(image, nbins=256)¶ Return threshold value based on Otsu’s method. Copied from scikitimage to remove dependency.
Parameters: image : array
Input image.
nbins : int
Number of bins used to calculate histogram. This value is ignored for integer arrays.
Returns: threshold : float
Threshold value.
segment_from_cfa¶

dipy.segment.mask.
segment_from_cfa
(tensor_fit, roi, threshold, return_cfa=False)¶ Segment the cfa inside roi using the values from threshold as bounds.
Parameters: tensor_fit : TensorFit object
TensorFit object
roi : ndarray
A binary mask, which contains the bounding box for the segmentation.
threshold : arraylike
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_cfa : bool, optional
If True, the cfa is also returned.
Returns: mask : ndarray
Binary mask of the segmentation.
cfa : ndarray, 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.
ArcLengthFeature
¶

class
dipy.segment.metric.
ArcLengthFeature
¶ Bases:
dipy.segment.featurespeed.CythonFeature
Extracts features from a sequential datum.
A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions).
The feature being extracted consists of one scalar representing the arc length of the sequence (i.e. the sum of the length of all segments).
Attributes
is_order_invariant
Is this feature invariant to the sequence’s ordering Methods
extract
Extracts features from a sequential datum. infer_shape
Infers the shape of features extracted from a sequential datum. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

AveragePointwiseEuclideanMetric
¶

class
dipy.segment.metric.
AveragePointwiseEuclideanMetric
¶ Bases:
dipy.segment.metricspeed.SumPointwiseEuclideanMetric
Computes the average of pointwise Euclidean distances between two sequential data.
A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions). A feature object can be specified in order to calculate the distance between the features, rather than directly between the sequential data.
Parameters: feature : Feature object, optional
It is used to extract features before computing the distance.
Notes
The distance between two 2D sequential data:
s1 s2 0* a *0 \  \  1*   b *1  \ 2* \ c *2
is equal to \((a+b+c)/3\) where \(a\) is the Euclidean distance between s1[0] and s2[0], \(b\) between s1[1] and s2[1] and \(c\) between s1[2] and s2[2].
Attributes
feature
Feature object used to extract features from sequential data is_order_invariant
Is this metric invariant to the sequence’s ordering Methods
are_compatible
Checks if features can be used by metric.dist based on their shape. dist
Computes a distance between two data points based on their features. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

CenterOfMassFeature
¶

class
dipy.segment.metric.
CenterOfMassFeature
¶ Bases:
dipy.segment.featurespeed.CythonFeature
Extracts features from a sequential datum.
A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions).
The feature being extracted consists of one Ndimensional point representing the mean of the points, i.e. the center of mass.
Attributes
is_order_invariant
Is this feature invariant to the sequence’s ordering Methods
extract
Extracts features from a sequential datum. infer_shape
Infers the shape of features extracted from a sequential datum. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

CosineMetric
¶

class
dipy.segment.metric.
CosineMetric
¶ Bases:
dipy.segment.metricspeed.CythonMetric
Computes the cosine distance between two vectors.
A vector (i.e. a Ndimensional point) is represented as a 2D array with shape (1, nb_dimensions).
Notes
The distance between two vectors \(v_1\) and \(v_2\) is equal to \(\frac{1}{\pi} \arccos\left(\frac{v_1 \cdot v_2}{\v_1\ \v_2\}\right)\) and is bounded within \([0,1]\).
Attributes
feature
Feature object used to extract features from sequential data is_order_invariant
Is this metric invariant to the sequence’s ordering Methods
are_compatible
Checks if features can be used by metric.dist based on their shape. dist
Computes a distance between two data points based on their features. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

EuclideanMetric
¶

dipy.segment.metric.
EuclideanMetric
¶ alias of
SumPointwiseEuclideanMetric
Feature
¶

class
dipy.segment.metric.
Feature
¶ Bases:
object
Extracts features from a sequential datum.
A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions).
Parameters: is_order_invariant : bool (optional)
tells if this feature is invariant to the sequence’s ordering. This means starting from either extremities produces the same features. (Default: True)
Notes
When subclassing Feature, one only needs to override the extract and infer_shape methods.
Attributes
is_order_invariant
Is this feature invariant to the sequence’s ordering Methods
extract
Extracts features from a sequential datum. infer_shape
Infers the shape of features extracted from a sequential datum. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

extract
()¶ Extracts features from a sequential datum.
Parameters: datum : 2D array
Sequence of Ndimensional points.
Returns: 2D array
Features extracted from datum.

infer_shape
()¶ Infers the shape of features extracted from a sequential datum.
Parameters: datum : 2D array
Sequence of Ndimensional points.
Returns: int, 1tuple or 2tuple
Shape of the features.

is_order_invariant
¶ Is this feature invariant to the sequence’s ordering

IdentityFeature
¶

class
dipy.segment.metric.
IdentityFeature
¶ Bases:
dipy.segment.featurespeed.CythonFeature
Extracts features from a sequential datum.
A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions).
The features being extracted are the actual sequence’s points. This is useful for metric that does not require any preprocessing.
Attributes
is_order_invariant
Is this feature invariant to the sequence’s ordering Methods
extract
Extracts features from a sequential datum. infer_shape
Infers the shape of features extracted from a sequential datum. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

Metric
¶

class
dipy.segment.metric.
Metric
¶ Bases:
object
Computes a distance between two sequential data.
A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions). A feature object can be specified in order to calculate the distance between extracted features, rather than directly between the sequential data.
Parameters: feature : Feature object, optional
It is used to extract features before computing the distance.
Notes
When subclassing Metric, one only needs to override the dist and are_compatible methods.
Attributes
feature
Feature object used to extract features from sequential data is_order_invariant
Is this metric invariant to the sequence’s ordering Methods
are_compatible
Checks if features can be used by metric.dist based on their shape. dist
Computes a distance between two data points based on their features. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

are_compatible
()¶ Checks if features can be used by metric.dist based on their shape.
Basically this method exists so we don’t have to do this check inside the metric.dist function (speedup).
Parameters: shape1 : int, 1tuple or 2tuple
shape of the first data point’s features
shape2 : int, 1tuple or 2tuple
shape of the second data point’s features
Returns: are_compatible : bool
whether or not shapes are compatible

dist
()¶ Computes a distance between two data points based on their features.
Parameters: features1 : 2D array
Features of the first data point.
features2 : 2D array
Features of the second data point.
Returns: double
Distance between two data points.

feature
¶ Feature object used to extract features from sequential data

is_order_invariant
¶ Is this metric invariant to the sequence’s ordering

MidpointFeature
¶

class
dipy.segment.metric.
MidpointFeature
¶ Bases:
dipy.segment.featurespeed.CythonFeature
Extracts features from a sequential datum.
A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions).
The feature being extracted consists of one Ndimensional point representing the middle point of the sequence (i.e. `nb_points//2`th point).
Attributes
is_order_invariant
Is this feature invariant to the sequence’s ordering Methods
extract
Extracts features from a sequential datum. infer_shape
Infers the shape of features extracted from a sequential datum. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

MinimumAverageDirectFlipMetric
¶

class
dipy.segment.metric.
MinimumAverageDirectFlipMetric
¶ Bases:
dipy.segment.metricspeed.AveragePointwiseEuclideanMetric
Computes the MDF distance (minimum average directflip) between two sequential data.
A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions).
Notes
The distance between two 2D sequential data:
s1 s2 0* a *0 \  \  1*   b *1  \ 2* \ c *2
is equal to \(\min((a+b+c)/3, (a'+b'+c')/3)\) where \(a\) is the Euclidean distance between s1[0] and s2[0], \(b\) between s1[1] and s2[1], \(c\) between s1[2] and s2[2], \(a'\) between s1[0] and s2[2], \(b'\) between s1[1] and s2[1] and \(c'\) between s1[2] and s2[0].
Attributes
feature
Feature object used to extract features from sequential data is_order_invariant
Is this metric invariant to the sequence’s ordering Methods
are_compatible
Checks if features can be used by metric.dist based on their shape. dist
Computes a distance between two data points based on their features. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

is_order_invariant
¶ Is this metric invariant to the sequence’s ordering

ResampleFeature
¶

class
dipy.segment.metric.
ResampleFeature
¶ Bases:
dipy.segment.featurespeed.CythonFeature
Extracts features from a sequential datum.
A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions).
The features being extracted are the points of the sequence once resampled. This is useful for metrics requiring a constant number of points for all
streamlines.Attributes
is_order_invariant
Is this feature invariant to the sequence’s ordering Methods
extract
Extracts features from a sequential datum. infer_shape
Infers the shape of features extracted from a sequential datum. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

SumPointwiseEuclideanMetric
¶

class
dipy.segment.metric.
SumPointwiseEuclideanMetric
¶ Bases:
dipy.segment.metricspeed.CythonMetric
Computes the sum of pointwise Euclidean distances between two sequential data.
A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions). A feature object can be specified in order to calculate the distance between the features, rather than directly between the sequential data.
Parameters: feature : Feature object, optional
It is used to extract features before computing the distance.
Notes
The distance between two 2D sequential data:
s1 s2 0* a *0 \  \  1*   b *1  \ 2* \ c *2
is equal to \(a+b+c\) where \(a\) is the Euclidean distance between s1[0] and s2[0], \(b\) between s1[1] and s2[1] and \(c\) between s1[2] and s2[2].
Attributes
feature
Feature object used to extract features from sequential data is_order_invariant
Is this metric invariant to the sequence’s ordering Methods
are_compatible
Checks if features can be used by metric.dist based on their shape. dist
Computes a distance between two data points based on their features. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

VectorOfEndpointsFeature
¶

class
dipy.segment.metric.
VectorOfEndpointsFeature
¶ Bases:
dipy.segment.featurespeed.CythonFeature
Extracts features from a sequential datum.
A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions).
The feature being extracted consists of one vector in the Ndimensional space pointing from one endpoint of the sequence to the other (i.e. S[1]S[0]).
Attributes
is_order_invariant
Is this feature invariant to the sequence’s ordering Methods
extract
Extracts features from a sequential datum. infer_shape
Infers the shape of features extracted from a sequential datum. 
__init__
()¶ Initialize self. See help(type(self)) for accurate signature.

dist¶

dipy.segment.metric.
dist
()¶ Computes a distance between datum1 and datum2.
A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions).
Parameters: metric : Metric object
Tells how to compute the distance between datum1 and datum2.
datum1 : 2D array
Sequence of Ndimensional points.
datum2 : 2D array
Sequence of Ndimensional points.
Returns: double
Distance between two data points.
distance_matrix¶

dipy.segment.metric.
distance_matrix
()¶ Computes the distance matrix between two lists of sequential data.
The distance matrix is obtained by computing the pairwise distance of all tuples spawn by the Cartesian product of data1 with data2. If data2 is not provided, the Cartesian product of data1 with itself is used instead. A sequence of Ndimensional points is represented as a 2D array with shape (nb_points, nb_dimensions).
Parameters: metric : Metric object
Tells how to compute the distance between two sequential data.
data1 : list of 2D arrays
List of sequences of Ndimensional points.
data2 : list of 2D arrays
Llist of sequences of Ndimensional points.
Returns: 2D array (double)
Distance matrix.
mdf¶

dipy.segment.metric.
mdf
(s1, s2)¶ Computes the MDF (Minimum average DirectFlip) distance [Garyfallidis12] between two streamlines.
Streamlines must have the same number of points.
Parameters: s1 : 2D array
A streamline (sequence of Ndimensional points).
s2 : 2D array
A streamline (sequence of Ndimensional points).
Returns: double
Distance between two streamlines.
References
[Garyfallidis12] Garyfallidis E. et al., QuickBundles a method for tractography simplification, Frontiers in Neuroscience, vol 6, no 175, 2012.
QuickBundles
¶

class
dipy.segment.quickbundles.
QuickBundles
(tracks, dist_thr=4.0, pts=12)¶ Bases:
object
Attributes
centroids
total_clusters
Methods
clusters
()clusters_sizes
()downsampled_tracks
()exemplars
([tracks])label2cluster
(id)label2tracks
(tracks, id)label2tracksids
(id)partitions
()points_per_track
()remove_cluster
(id)remove_clusters
(list_ids)remove_small_clusters
(size)Remove clusters with small size remove_tracks
()virtuals
()
__init__
(tracks, dist_thr=4.0, pts=12)¶ Highly efficient trajectory clustering [Garyfallidis12].
Parameters: tracks : sequence of (N,3) ... (M,3) arrays
trajectories (or tractography or streamlines)
dist_thr : float
distance threshold in the space of the tracks
pts : int
number of points for simplifying the tracks
References
[Garyfallidis12] Garyfallidis E. et al., QuickBundles a method for tractography simplification, Frontiers in Neuroscience, vol 6, no 175, 2012. Methods
clustering() returns a dict holding with the clustering result virtuals() gives the virtuals (track centroids) of the clusters exemplars() gives the exemplars (track medoids) of the clusters

centroids
¶

clusters
()¶

clusters_sizes
()¶

downsampled_tracks
()¶

exemplars
(tracks=None)¶

label2cluster
(id)¶

label2tracks
(tracks, id)¶

label2tracksids
(id)¶

partitions
()¶

points_per_track
()¶

remove_cluster
(id)¶

remove_clusters
(list_ids)¶

remove_small_clusters
(size)¶ Remove clusters with small size
Parameters: size : int, threshold for minimum number of tracks allowed

remove_tracks
()¶

total_clusters
¶

virtuals
()¶

bundles_distances_mdf¶

dipy.segment.quickbundles.
bundles_distances_mdf
()¶ Calculate distances between list of tracks A and list of tracks B
All tracks need to have the same number of points
Parameters: tracksA : sequence
of tracks as arrays, [(N,3) .. (N,3)]
tracksB : sequence
of tracks as arrays, [(N,3) .. (N,3)]
Returns: DM : array, shape (len(tracksA), len(tracksB))
distances between tracksA and tracksB according to metric
See also
dipy.metrics.downsample
downsample¶

dipy.segment.quickbundles.
downsample
(xyz, n_pols=3)¶ downsample for a specific number of points along the curve/track
Uses the length of the curve. It works in a similar fashion to midpoint and arbitrarypoint but it also reduces the number of segments of a track.
Parameters: xyz : arraylike shape (N,3)
array representing x,y,z of N points in a track
n_pol : int
integer representing number of points (poles) we need along the curve.
Returns: xyz2 : array shape (M,3)
array representing x,y,z of M points that where extrapolated. M should be equal to n_pols
Examples
>>> import numpy as np >>> # a semicircle >>> theta=np.pi*np.linspace(0,1,100) >>> x=np.cos(theta) >>> y=np.sin(theta) >>> z=0*x >>> xyz=np.vstack((x,y,z)).T >>> xyz2=downsample(xyz,3) >>> # a cosine >>> x=np.pi*np.linspace(0,1,100) >>> y=np.cos(theta) >>> z=0*y >>> xyz=np.vstack((x,y,z)).T >>> _= downsample(xyz,3) >>> len(xyz2) 3 >>> xyz3=downsample(xyz,10) >>> len(xyz3) 10
local_skeleton_clustering¶

dipy.segment.quickbundles.
local_skeleton_clustering
()¶ Efficient tractography clustering
Every track can needs to have the same number of points. Use dipy.tracking.metrics.downsample to restrict the number of points
Parameters: tracks : sequence
of tracks as arrays, shape (N,3) .. (N,3) where N=points
d_thr : float
average euclidean distance threshold
Returns: C : dict
Clusters.
See also
Notes
The distance calculated between two tracks:
t_1 t_2 0* a *0 \  \  1*   b *1  \ 2* \ c *2
is equal to \((a+b+c)/3\) where \(a\) the euclidean distance between
t_1[0]
andt_2[0]
, \(b\) betweent_1[1]
andt_2[1]
and \(c\) betweent_1[2]
andt_2[2]
. Also the same with t2 flipped (sot_1[0]
compared tot_2[2]
etc).Visualization:
It is possible to visualize the clustering C from the example above using the fvtk module:
from dipy.viz import fvtk r=fvtk.ren() for c in C: color=np.random.rand(3) for i in C[c]['indices']: fvtk.add(r,fvtk.line(tracks[i],color)) fvtk.show(r)
Examples
>>> tracks=[np.array([[0,0,0],[1,0,0,],[2,0,0]]), ... np.array([[3,0,0],[3.5,1,0],[4,2,0]]), ... np.array([[3.2,0,0],[3.7,1,0],[4.4,2,0]]), ... np.array([[3.4,0,0],[3.9,1,0],[4.6,2,0]]), ... np.array([[0,0.2,0],[1,0.2,0],[2,0.2,0]]), ... np.array([[2,0.2,0],[1,0.2,0],[0,0.2,0]]), ... np.array([[0,0,0],[0,1,0],[0,2,0]])] >>> C = local_skeleton_clustering(tracks, d_thr=0.5)
otsu¶

dipy.segment.threshold.
otsu
(image, nbins=256)¶ Return threshold value based on Otsu’s method. Copied from scikitimage to remove dependency.
Parameters: image : array
Input image.
nbins : int
Number of bins used to calculate histogram. This value is ignored for integer arrays.
Returns: threshold : float
Threshold value.
upper_bound_by_percent¶

dipy.segment.threshold.
upper_bound_by_percent
(data, percent=1)¶ Find the upper bound for visualization of medical images
Calculate the histogram of the image and go right to left until you find the bound that contains more than a percentage of the image.
Parameters: data : ndarray
percent : float
Returns: upper_bound : float
upper_bound_by_rate¶

dipy.segment.threshold.
upper_bound_by_rate
(data, rate=0.05)¶ Adjusts upper intensity boundary using rates
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: data : float
Input intensity value data
rate : float
representing the threshold whether a spicific histogram bin that should be count in the normalization range
Returns: high : float
the upper_bound value for normalization
ConstantObservationModel
¶

class
dipy.segment.tissue.
ConstantObservationModel
¶ Bases:
object
Observation model assuming that the intensity of each class is constant. The model parameters are the means \(\mu_{k}\) and variances \(\sigma_{k}\) associated with each tissue class. According to this model, the observed intensity at voxel \(x\) is given by \(I(x) = \mu_{k} + \eta_{k}\) where \(k\) is the tissue class of voxel \(x\), and \(\eta_{k}\) is a Gaussian random variable with zero mean and variance \(\sigma_{k}^{2}\). The observation model is responsible for computing the negative loglikelihood of observing any given intensity \(z\) at each voxel \(x\) assuming the voxel belongs to each class \(k\). It also provides a default parameter initialization.
Methods
initialize_param_uniform
Initializes the means and variances uniformly negloglikelihood
Computes the gaussian negative loglikelihood of each class at each voxel of image assuming a gaussian distribution with means and variances given by mu and sigmasq, respectively (constant models along the full volume). prob_image
Conditional probability of the label given the image seg_stats
Mean and standard variation for N desired tissue classes update_param
Updates the means and the variances in each iteration for all the labels. update_param_new
Updates the means and the variances in each iteration for all the labels. 
__init__
()¶ Initializes an instance of the ConstantObservationModel class

initialize_param_uniform
()¶ Initializes the means and variances uniformly
The means are initialized uniformly along the dynamic range of image. The variances are set to 1 for all classes
Parameters: image : array,
3D structural image
nclasses : int,
number of desired classes
Returns: mu : array,
1 x nclasses, mean for each class
sigma : array,
1 x nclasses, standard deviation for each class. Set up to 1.0 for all classes.

negloglikelihood
()¶ Computes the gaussian negative loglikelihood of each class at each voxel of image assuming a gaussian distribution with means and variances given by mu and sigmasq, respectively (constant models along the full volume). The negative loglikelihood will be written in nloglike.
Parameters: image : ndarray,
3D gray scale structural image
mu : ndarray,
mean of each class
sigmasq : ndarray,
variance of each class
nclasses : int
number of classes
Returns: nloglike : ndarray,
4D negloglikelihood for each class in each volume

prob_image
()¶ Conditional probability of the label given the image
Parameters: img : ndarray,
3D structural grayscale image
nclasses : int,
number of tissue classes
mu : ndarray,
1 x nclasses, current estimate of the mean of each tissue class
sigmasq : ndarray,
1 x nclasses, current estimate of the variance of each tissue class
P_L_N : ndarray,
4D probability map of the label given the neighborhood.
Previously computed by function prob_neighborhood
Returns: P_L_Y : ndarray,
4D probability of the label given the input image

seg_stats
()¶ Mean and standard variation for N desired tissue classes
Parameters: input_image : ndarray,
3D structural image
seg_image : ndarray,
3D segmented image
nclass : int,
number of classes (3 in most cases)
Returns: mu, std: ndarrays,
1 x nclasses dimension Mean and standard deviation for each class

update_param
()¶ Updates the means and the variances in each iteration for all the labels. This is for equations 25 and 26 of Zhang et. al., IEEE Trans. Med. Imag, Vol. 20, No. 1, Jan 2001.
Parameters: image : ndarray,
3D structural grayscale image
P_L_Y : ndarray,
4D probability map of the label given the input image computed by the expectation maximization (EM) algorithm
mu : ndarray,
1 x nclasses, current estimate of the mean of each tissue class.
nclasses : int,
number of tissue classes
Returns: mu_upd : ndarray,
1 x nclasses, updated mean of each tissue class
var_upd : ndarray,
1 x nclasses, updated variance of each tissue class

update_param_new
()¶ Updates the means and the variances in each iteration for all the labels. This is for equations 25 and 26 of the Zhang et al. paper
Parameters: image : ndarray,
3D structural grayscale image
P_L_Y : ndarray,
4D probability map of the label given the input image computed by the expectation maximization (EM) algorithm
mu : ndarray,
1 x nclasses, current estimate of the mean of each tissue class.
nclasses : int,
number of tissue classes
Returns: mu_upd : ndarray,
1 x nclasses, updated mean of each tissue class
var_upd : ndarray,
1 x nclasses, updated variance of each tissue class

IteratedConditionalModes
¶

class
dipy.segment.tissue.
IteratedConditionalModes
¶ Bases:
object
Methods
icm_ising
Executes one iteration of the ICM algorithm for MRF MAP estimation. initialize_maximum_likelihood
Initializes the segmentation of an image with given prob_neighborhood
Conditional probability of the label given the neighborhood Equation 2.18 of the Stan Z. 
__init__
()¶

icm_ising
()¶ Executes one iteration of the ICM algorithm for MRF MAP estimation. The prior distribution of the MRF is a Gibbs distribution with the Potts/Ising model with parameter beta:
https://en.wikipedia.org/wiki/Potts_model
Parameters: nloglike : ndarray,
4D shape, nloglike[x,y,z,k] is the negative log likelihood of class k at voxel (x,y,z)
beta : float,
positive scalar, it is the parameter of the Potts/Ising model. Determines the smoothness of the output segmentation.
seg : ndarray,
3D initial segmentation. This segmentation will change by one iteration of the ICM algorithm
Returns: new_seg : ndarray,
3D final segmentation
energy : ndarray,
3D final energy

initialize_maximum_likelihood
()¶  Initializes the segmentation of an image with given
 negloglikelihood
Initializes the segmentation of an image with negloglikelihood field given by nloglike. The class of each voxel is selected as the one with the minimum negloglikelihood (i.e. maximumlikelihood segmentation).
Parameters: nloglike : ndarray,
4D shape, nloglike[x,y,z,k] is the likelihhood of class k for voxel (x, y, z)
Returns: seg : ndarray,
3D initial segmentation

prob_neighborhood
()¶ Conditional probability of the label given the neighborhood Equation 2.18 of the Stan Z. Li book (Stan Z. Li, Markov Random Field Modeling in Image Analysis, 3rd ed., Advances in Pattern Recognition Series, Springer Verlag 2009.)
Parameters: seg : ndarray,
3D tissue segmentation derived from the ICM model
beta : float,
scalar that determines the importance of the neighborhood and the spatial smoothness of the segmentation. Usually between 0 to 0.5
nclasses : int,
number of tissue classes
Returns: PLN : ndarray,
4D probability map of the label given the neighborhood of the voxel.

TissueClassifierHMRF
¶

class
dipy.segment.tissue.
TissueClassifierHMRF
(save_history=False, verbose=True)¶ Bases:
object
This class contains the methods for tissue classification using the Markov Random Fields modeling approach
Methods
classify
(image, nclasses, beta[, tolerance, ...])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. 
__init__
(save_history=False, verbose=True)¶

classify
(image, nclasses, beta, tolerance=None, max_iter=None)¶ 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: image : ndarray,
3D structural image.
nclasses : int,
number of desired classes.
beta : float,
smoothing parameter, the higher this number the smoother the output will be.
tolerance: float,
value that defines the percentage of change tolerated to prevent the ICM loop to stop. Default is 1e05.
max_iter : float,
fixed number of desired iterations. Default is 100. If the user only specifies this parameter, the tolerance value will not be considered. If none of these two parameters
Returns: initial_segmentation : ndarray,
3D segmented image with all tissue types specified in nclasses.
final_segmentation : ndarray,
3D final refined segmentation containing all tissue types.
PVE : ndarray,
3D probability map of each tissue type.

add_noise¶

dipy.segment.tissue.
add_noise
(signal, snr, S0, noise_type='rician')¶ Add noise of specified distribution to the signal from a single voxel.
Parameters: signal : 1d ndarray
The signal in the voxel.
snr : float
The desired signaltonoise ratio. (See notes below.) If snr is None, return the signal asis.
S0 : float
Reference signal for specifying snr.
noise_type : string, optional
The distribution of noise added. Can be either ‘gaussian’ for Gaussian distributed noise, ‘rician’ for Ricedistributed noise (default) or ‘rayleigh’ for a Rayleigh distribution.
Returns: signal : array, same shape as the input
Signal with added noise.
Notes
SNR is defined here, following [R426], as
S0 / sigma
, wheresigma
is the standard deviation of the two Gaussian distributions forming the real and imaginary components of the Rician noise distribution (see [R427]).References
[R426] (1, 2) Descoteaux, Angelino, Fitzgibbons and Deriche (2007) Regularized, fast and robust qball imaging. MRM, 58: 497510 [R427] (1, 2) Gudbjartson and Patz (2008). The Rician distribution of noisy MRI data. MRM 34: 910914. Examples
>>> signal = np.arange(800).reshape(2, 2, 2, 100) >>> signal_w_noise = add_noise(signal, 10., 100., noise_type='rician')