def test_write():
streams = []
out_f = StringIO()
tv.write(out_f, [], {})
yield assert_equal, out_f.getvalue(), tv.empty_header().tostring()
out_f.truncate(0)
# Write something not-default
tv.write(out_f, [], {'id_string':'TRACKb'})
# read it back
out_f.seek(0)
streams, hdr = tv.read(out_f)
yield assert_equal, hdr['id_string'], 'TRACKb'
# check that we can pass none for the header
out_f.truncate(0)
tv.write(out_f, [])
out_f.truncate(0)
tv.write(out_f, [], None)
# check that we check input values
out_f.truncate(0)
yield (assert_raises, tv.HeaderError,
tv.write, out_f, [],{'id_string':'not OK'})
yield (assert_raises, tv.HeaderError,
tv.write, out_f, [],{'version':2})
yield (assert_raises, tv.HeaderError,
tv.write, out_f, [],{'hdr_size':0})
def test_round_trip():
out_f = StringIO()
xyz0 = np.tile(np.arange(5).reshape(5, 1), (1, 3))
xyz1 = np.tile(np.arange(5).reshape(5, 1) + 10, (1, 3))
streams = [(xyz0, None, None), (xyz1, None, None)]
tv.write(out_f, streams, {})
out_f.seek(0)
streams2, hdr = tv.read(out_f)
yield assert_true, streamlist_equal(streams, streams2)
def test_round_trip():
out_f = StringIO()
xyz0 = np.tile(np.arange(5).reshape(5,1), (1, 3))
xyz1 = np.tile(np.arange(5).reshape(5,1) + 10, (1, 3))
streams = [(xyz0, None, None), (xyz1, None, None)]
tv.write(out_f, streams, {})
out_f.seek(0)
streams2, hdr = tv.read(out_f)
yield assert_true, streamlist_equal(streams, streams2)
def streams(self):
if 'streams' in self._cache:
return self._cache['streams']
trk_fname = get_onefile(self._glob_prefix + 'fiber_track_mni.trk')
if trk_fname is None:
return None
streams, hdr = tv.read(trk_fname)
self._cache['hdr'] = hdr
self._cache['streams'] = streams
return streams
def load_reduce_translate(fname, reduction=1):
""" Loads a trackvis file using DIPY io, only keeping a ratio of
1/reduction tracks. Performs trajectory approximation to reduce the
track lengths.
Parameter
---------
fname : str
The Trackvis file
reduction : int, optional
The reduction factor (keep only 1 line per set of reduction lines)
Returns
-------
tracks : list
A list of tracks
"""
lines, hdr = tv.read(fname)
print 'loaded,',
sys.stdout.flush()
#ras = tv.get_affine(hdr)
ras = tv.aff_from_hdr(hdr)
if not ras[:3,-1].any():
# dot(ras[:3,:3],md_vox) + t = (0,0,0) --> t = -dot(ras[:3,:3],md_vox)
md_vox = hdr['dim']/2
t = -np.dot(ras[:3,:3], md_vox)
ras[:3,-1] = t
tracks = [l[0] for l in lines[::reduction]]
del lines
ras_cmap = ni_api.AffineTransform.from_params('ijk', 'xyz', ras)
flat_tracks, breaks = flatten_tracks(tracks)
del tracks
flat_tracks_xyz = ras_cmap(flat_tracks)
tracks = recombine_flattened_tracks(flat_tracks_xyz, breaks)
print 'translated,',
sys.stdout.flush()
tracks_reduced = [tp.approximate_ei_trajectory(line, alpha=np.pi/4)
for line in tracks ]
print 'reduced,'
sys.stdout.flush()
return tracks_reduced
def init(self):
import dipy.io.trackvis as tv
lines,hdr = tv.read(self.fname)
ras = tv.aff_from_hdr(hdr)
self.affine=ras
tracks = [l[0] for l in lines]
if self.yellow_indices != None :
tracks = [t for t in tracks if tm.length(t) > 20]
print 'tracks loaded'
#self.data = [100*np.array([[0,0,0],[1,0,0],[2,0,0]]).astype(np.float32) ,100*np.array([[0,1,0],[0,2,0],[0,3,0]]).astype(np.float32)]#tracks[:20000]
if self.dummy_data:
self.data = [100*np.array([[0,0,0],[1,0,0],[2,0,0]]).astype(np.float32) ,100*np.array([[0,1,0],[0,2,0],[0,3,0]]).astype(np.float32)]
if self.data_subset!=None:
self.data = tracks[self.data_subset[0]:self.data_subset[1]]
else:
self.data = tracks
data_stats = np.concatenate(tracks)
self.min=np.min(data_stats,axis=0)
self.max=np.max(data_stats,axis=0)
self.mean=np.mean(data_stats,axis=0)
del data_stats
del lines
self.multiple_colors()
def test_write():
streams = []
out_f = StringIO()
tv.write(out_f, [], {})
yield assert_equal(out_f.getvalue(), tv.empty_header().tostring())
out_f.truncate(0)
# Write something not-default
tv.write(out_f, [], {'id_string': 'TRACKb'})
# read it back
out_f.seek(0)
streams, hdr = tv.read(out_f)
yield assert_equal(hdr['id_string'], 'TRACKb')
# check that we can pass none for the header
out_f.truncate(0)
tv.write(out_f, [])
out_f.truncate(0)
tv.write(out_f, [], None)
# check that we check input values
out_f.truncate(0)
yield assert_raises(tv.HeaderError, tv.write, out_f, [],
{'id_string': 'not OK'})
yield assert_raises(tv.HeaderError, tv.write, out_f, [], {'version': 3})
yield assert_raises(tv.HeaderError, tv.write, out_f, [], {'hdr_size': 0})
from dipy.io import trackvis as tv
from dipy.io import pickles as pkl
from dipy.core import track_learning as tl
from dipy.core import track_performance as pf
from dipy.core import track_metrics as tm
import time
import numpy as np
import os
fname = '/home/eg01/Data/PBC/pbc2009icdm/brain1/brain1_scan1_fiber_track_mni.trk'
C_fname = '/tmp/larch_tree.pkl'
appr_fname = '/tmp/larch_tracks.trk'
print 'Loading trackvis file...'
streams, hdr = tv.read(fname)
print 'Copying tracks...'
tracks = [i[0] for i in streams]
#tracks=tracks[:1000]
#print 'Deleting unnecessary data...'
del streams #,hdr
if not os.path.isfile(C_fname):
print 'Starting LARCH ...'
tim = time.clock()
C, atracks = tl.larch(tracks, [50.**2, 20.**2, 5.**2], True, True)
#tracks=[tm.downsample(t,3) for t in tracks]
from dipy.core import track_metrics as tm from dipy.viz import fos from dipy.io import trackvis as tv from dipy.io import pickles as pkl from dipy.core import track_performance as pf import time import numpy as np #fname='/home/eg01/Data/PBC/pbc2009icdm/brain1/brain1_scan1_fiber_track_mni.trk' fname='/home/eg309/Data/PBC/pbc2009icdm/brain1/brain1_scan1_fiber_track_mni.trk' print 'Loading file...' streams,hdr=tv.read(fname) print 'Copying tracks...' T=[i[0] for i in streams] print 'Representing tracks using only 3 pts...' tracks=[tm.downsample(t,3) for t in T] print 'Deleting unnecessary data...' del streams,hdr print 'Hidden Structure Clustering...' now=time.clock() C=pf.local_skeleton_clustering(tracks,d_thr=20) print 'Done in', time.clock()-now,'s.' print 'Reducing the number of points...' T=[pf.approximate_ei_trajectory(t) for t in T]
[8, 10853, 15572], [ 9, 13280, 8461], [ 0, 11275, 9224]]) ''' corr_mat_demo=np.array([[ 1, 10560, 3609],[ 2, 17872, 15377],[ 3, 6447, 3897], [4, 18854, 6409], [ 5, 14416, 4515], [ 7, 9956, 13913], [8, 10853, 15572], [ 9, 13280, 8461], [ 0, 11275, 9224]]) print 'Minimum track length', min_len, 'mm' print 'Number of segments for downsampling',down print 'Number of tracks for detection',rand_tracks print 'Minimum searched track length', min_search_len, 'mm' print 'Maximum searched track length', max_search_len, 'mm' tracks1,hdr1=tv.read(br1path) tracks2,hdr2=tv.read(br2path) tracks3,hdr3=tv.read(br3path) #Load only track points, no scalars or parameters. tracks1=[t[0] for t in tracks1] tracks2=[t[0] for t in tracks2] tracks3=[t[0] for t in tracks3] print 'Before thresholding' print len(tracks1) print len(tracks2) print len(tracks3)
''' Benchmarks for Zhang metrics '''
import os
from os.path import join as pjoin
import numpy as np
import dipy.io.trackvis as tv
import dipy.core.track_performance as pf
import dipy.core.track_metrics as tm
from numpy.testing import measure
_data_path = pjoin(os.path.dirname(__file__), 'data')
_track_tuples, _ = tv.read(pjoin(_data_path, 'tracks300.trk.gz'))
tracks300 = [_t[0] for _t in _track_tuples]
def bench_zhang():
print 'Zhang min'
print '=' * 10
#ref_time = measure('tm.most_similar_track_zhang(tracks300)')
#print 'reference time: %f' % ref_time
opt_time = measure('pf.most_similar_track_zhang(tracks300)')
print 'optimized time: %f' % opt_time
print
def bench_cut_plane():
print 'Cut plane'
print '=' * 10
opt_time = measure('pf.cut_plane(tracks300, tracks300[0])')
print 'optimized time: %f' % opt_time
print
''' Benchmarks for Zhang metrics '''
import os
from os.path import join as pjoin
import numpy as np
import dipy.io.trackvis as tv
import dipy.core.track_performance as pf
import dipy.core.track_metrics as tm
from numpy.testing import measure
_data_path = pjoin(os.path.dirname(__file__), 'data')
_track_tuples, _ = tv.read(pjoin(_data_path, 'tracks300.trk.gz'))
tracks300 = [_t[0] for _t in _track_tuples]
def bench_zhang():
print 'Zhang min'
print '=' * 10
#ref_time = measure('tm.most_similar_track_zhang(tracks300)')
#print 'reference time: %f' % ref_time
opt_time = measure('pf.most_similar_track_zhang(tracks300)')
print 'optimized time: %f' % opt_time
print
def bench_cut_plane():
print 'Cut plane'
print '=' * 10
opt_time = measure('pf.cut_plane(tracks300, tracks300[0])')
print 'optimized time: %f' % opt_time
frame = make_window_maker(trajs, colors)()
app.MainLoop()
del frame
del app
if __name__ == "__main__":
#trajs=[100*np.random.rand(10,3),100*np.random.rand(20,3)]
#colors=[np.random.rand(3,),np.random.rand(3,)]
from dipy.io import trackvis as tv
from dipy.core import track_performance
#from enthought.mayavi import mlab
#from enthought.tvtk.api import tvtk
import numpy as np
fname='/home/eg309/Data/PBC/pbc2009icdm/brain1/brain1_scan1_fiber_track_mni.trk'
lines, hdr = tv.read(fname)
pts = [p[0] for p in lines]
pts_reduced = [track_performance.approx_polygon_track(p) for p in pts]
red = np.array([1,0,0])
trajs=pts_reduced
colors=[np.array([1.,0,0]) for i in range(len(trajs))]
show(trajs)
def init(self):
br1path='/home/eg01/Data_Backup/Data/Eleftherios/CBU090133_METHODS/20090227_145404/Series_003_CBU_DTI_64D_iso_1000/dtk_dti_out/dti_FACT.trk'
br2path='/home/eg01/Data_Backup/Data/Eleftherios/CBU090134_METHODS/20090227_154122/Series_003_CBU_DTI_64D_iso_1000/dtk_dti_out/dti_FACT.trk'
br3path='/home/eg01/Data_Backup/Data/Eleftherios/CBU090133_METHODS/20090227_145404/Series_003_CBU_DTI_64D_iso_1000/dtk_dti_out/dti_RK2.trk'
min_len=20
down=20
rand_tracks=-1 #default 10 min_search_len=70
min_search_len=70
max_search_len=140
corr_mat_demo=np.array([[ 1, 10560, 3609],[ 2, 17872, 15377],[ 3, 6447, 3897], [4, 18854, 6409], [ 5, 14416, 4515], [ 7, 9956, 13913], [8, 10853, 15572], [ 9, 13280, 8461], [ 0, 11275, 9224]])
print 'Minimum track length', min_len, 'mm'
print 'Number of segments for downsampling',down
print 'Number of tracks for detection',rand_tracks
print 'Minimum searched track length', min_search_len, 'mm'
print 'Maximum searched track length', max_search_len, 'mm'
tracks1,hdr1=tv.read(br1path)
tracks2,hdr2=tv.read(br2path)
tracks3,hdr3=tv.read(br3path)
#Load only track points, no scalars or parameters.
tracks1=[t[0] for t in tracks1]
tracks2=[t[0] for t in tracks2]
tracks3=[t[0] for t in tracks3]
print 'Before thresholding'
print len(tracks1)
print len(tracks2)
print len(tracks3)
print hdr1['dim']
print hdr2['dim']
print hdr3['dim']
#Apply thresholds
tracks1=[t for t in tracks1 if tm.length(t) > min_len]
tracks2=[t for t in tracks2 if tm.length(t) > min_len]
tracks3=[t for t in tracks3 if tm.length(t) > min_len]
print 'After thresholding'
print len(tracks1)
print len(tracks2)
print len(tracks3)
print 'Downsampling'
tracks1z=[tm.downsample(t,down) for t in tracks1]
tracks2z=[tm.downsample(t,down) for t in tracks2]
tracks3z=[tm.downsample(t,down) for t in tracks3]
print 'Detecting random tracks'
lt1=len(tracks1)
lt2=len(tracks2)
lt3=len(tracks3)
if rand_tracks==-1:
#use already stored indices
t_ind=corr_mat_demo[:,1]
t_ind=np.array(t_ind)
print 'Indices of tracks for detection', t_ind
print 'Finding corresponding tracks'
global track2track
track2track= self.corresponding_tracks(t_ind,tracks1z,tracks2z)
global track2track2
track2track2=self.corresponding_tracks(t_ind,tracks1z,tracks3z)
print 'First Correspondance Matrix'
print track2track
print 'Second Correspondance Matrix'
print track2track2
print 'first brain'
print track2track[:,1].T
print 'second brain'
print track2track[:,2].T
print 'third brain'
print track2track2[:,2].T
#fos.add(r,fos.line(tracks1,fos.red,opacity=0.01))
#fos.add(r,fos.line(tracks2,fos.cyan,opacity=0.01))
tracks1zshift = tracks1z
tracks2zshift = tracks2z
tracks3zshift = tracks3z
m1z=np.concatenate(tracks1zshift).mean(axis=0)
m2z=np.concatenate(tracks2zshift).mean(axis=0)
m3z=np.concatenate(tracks3zshift).mean(axis=0)
#tracks1zshift=[t+np.array([-70,0,0]) for t in tracks1z]
#tracks2zshift=[t+np.array([70,0,0]) for t in tracks2z]
tracks1zshift=[t-m1z for t in tracks1z]
tracks2zshift=[t-m2z for t in tracks2z]
tracks3zshift=[t-m3z for t in tracks3z]
global t1
#devel07
t1=tracks.Tracks(None,data_ext=tracks1zshift)
t1.angular_speed = 0.1
t1.brain_color=[1,0,0]
t1.manycolors=False
t1.opacity = 0.01
t1.orbit_demo=True
t1.orbit_anglez_rate = 0.
t1.orbit_anglex_rate = 0.
t1.orbit_angley_rate = .2
t1.init()
t1.position = np.array([-120,0,-30])
print 't1p',t1.position
global t2
#devel07
t2=tracks.Tracks(None,data_ext=tracks2zshift)
t2.angular_speed = 0.1
t2.brain_color=[0,1,1]
t2.manycolors=False
t2.opacity = 0.01
t2.orbit_demo=True
t2.orbit_anglez_rate = 0.
t2.orbit_anglex_rate = 0.
t2.orbit_angley_rate = .2
t2.init()
t2.position = np.array([0,0,-30])
print 't2p', t2.position
global t3
#devel07
t3=tracks.Tracks(None,data_ext=tracks3zshift)
t3.angular_speed = 0.1
t3.manycolors=False
t3.brain_color=[0,0,1]
t3.opacity = 0.01
t3.orbit_demo=True
t3.orbit_anglez_rate = 0.
t3.orbit_anglex_rate = 0.
t3.orbit_angley_rate = .2
t3.init()
#t3.position = -
#np.concatenate(tracks3zshift).mean(axis=0)+np.array([70,0,0])
t3.position = np.array([120,0,-30])
print 't3p', t3.position
self.slots={0:{'actor':t1,'slot':( 0, 800*MS ) },
1:{'actor':t2,'slot':( 0, 800*MS ) },
2:{'actor':t3,'slot':( 0, 800*MS ) }}
def __init__(self,fname,colormap=None, line_width=1., shrink=None, thinning = 0,
angle_table = None, manycolors = False, brain_color=[1,1,1]):
self.position = (0,0,0)
self.fname = fname
self.manycolors = manycolors
#self.color = monocolor
self.bbox = None
self.list_index = None
self.affine = None
self.data = None
self.list_index = None
self.rot_angle = 0
self.colormap = None
self.min = None
self.max = None
self.mean = None
self.material_color = False
self.fadeout = False
self.fadein = False
self.fadeout_speed = 0.
self.fadein_speed = 0.
self.min_length = 20.
self.angle = 0.
self.angular_speed = .5
self.line_width = line_width
self.opacity = 1.
self.opacity_rate = 0.
self.near_pick = None
self.far_pick = None
self.near_pick_prev = None
self.far_pick_prev = None
self.picked_track = None
self.pick_color = [1,1,0]
#self.brain_color = [1,1,1] # white
#self.brain_color = [.941,.862,.510] # buff
self.brain_color = brain_color
self.yellow_indices = None
self.dummy_data = False
self.data_subset = [0,20000]#None
self.orbit_demo = False
self.orbit_anglez =0.
self.orbit_anglez_rate = 10.
self.orbit_anglex = 0.
self.orbit_anglex_rate = 2.
self.angle_table = angle_table
if angle_table != None:
print 'Tracks angle_table shape %s' % str(self.angle_table.shape)
self.angle_table_index = 0
#print 'angle_table_index %d' % self.angle_table_index
self.shrink = shrink
self.picking_example = False
import dipy.io.trackvis as tv
lines,hdr = tv.read(self.fname)
ras = tv.aff_from_hdr(hdr
)
self.affine=ras
tracks = [l[0] for l in lines]
print 'tracks %d loaded' % len(tracks)
self.thinning = thinning
if self.yellow_indices != None :
tracks = [t for t in tracks if tm.length(t) > 20]
if self.thinning != 0:
tracks = [tracks[k] for k in range(0,len(tracks),self.thinning)]
print '%d tracks active' % len(tracks)
#self.data = [100*np.array([[0,0,0],[1,0,0],[2,0,0]]).astype(np.float32) ,100*np.array([[0,1,0],[0,2,0],[0,3,0]]).astype(np.float32)]#tracks[:20000]
if self.dummy_data:
self.data = [100*np.array([[0,0,0],[1,0,0],[2,0,0]]).astype(np.float32) ,100*np.array([[0,1,0],[0,2,0],[0,3,0]]).astype(np.float32)]
if self.data_subset!=None:
self.data = tracks[self.data_subset[0]:self.data_subset[1]]
else:
self.data = tracks
if self.shrink != None:
self.data = [ self.shrink*t for t in self.data]
data_stats = np.concatenate(tracks)
self.min=np.min(data_stats,axis=0)
self.max=np.max(data_stats,axis=0)
self.mean=np.mean(data_stats,axis=0)
del data_stats
del lines
def __init__(self,fname,ang_table=None,colormap=None, line_width=3., shrink=None,subset=None,tracks=None,text=None,text_color=np.array([1,0,0])):
self.position = (0,0,0)
self.fname = fname
self.manycolors = True
self.bbox = None
self.list_index = None
self.affine = None
self.data = None
self.list_index = None
self.rot_angle = 0
self.colormap = None
self.text = text
self.min = None
self.max = None
self.mean = None
self.material_color = False
self.fadeout = False
self.fadein = False
self.fadeout_speed = 0.
self.fadein_speed = 0.
self.min_length = 20.
self.angle = 0.
self.angular_speed = .5
self.line_width = line_width
self.opacity = 1.
self.near_pick = None
self.far_pick = None
self.near_pick_prev = None
self.far_pick_prev = None
self.picked_track = None
self.pick_color = [1,1,0]
self.brain_color = [1,1,1]
self.yellow_indices = None
self.dummy_data = False
self.tracks = tracks
if subset != None:
self.data_subset = subset #[0,20000]#None
else:
self.data_subset = None
self.orbit_demo = False
self.orbit_anglez = 0.
self.orbit_anglez_rate = 10.
self.orbit_anglex = 0.
self.orbit_anglex_rate = 2.
self.angle_table = ang_table
self.angle_table_index = 0
self.shrink = shrink
self.picking_example = False
self.partial_colors = False
import dipy.io.trackvis as tv
if self.tracks == None:
lines,hdr = tv.read(self.fname)
ras = tv.aff_from_hdr(hdr)
self.affine=ras
tracks = [l[0] for l in lines]
del lines
else:
tracks = self.tracks
if self.yellow_indices != None :
tracks = [t for t in tracks if tm.length(t) > 20]
print '%d tracks loaded' % len(tracks)
#self.data = [100*np.array([[0,0,0],[1,0,0],[2,0,0]]).astype(np.float32) ,100*np.array([[0,1,0],[0,2,0],[0,3,0]]).astype(np.float32)]#tracks[:20000]
if self.dummy_data:
self.data = [100*np.array([[0,0,0],[1,0,0],[2,0,0]]).astype(np.float32) ,100*np.array([[0,1,0],[0,2,0],[0,3,0]]).astype(np.float32)]
if self.data_subset!=None:
self.data = tracks[self.data_subset[0]:self.data_subset[1]]
else:
self.data = tracks
if self.shrink != None:
self.data = [ self.shrink*t for t in self.data]
data_stats = np.concatenate(tracks)
self.min=np.min(data_stats,axis=0)
self.max=np.max(data_stats,axis=0)
self.mean=np.mean(data_stats,axis=0)
del data_stats
app = wx.PySimpleApp()
frame = make_window_maker(trajs, colors)()
app.MainLoop()
del frame
del app
if __name__ == "__main__":
#trajs=[100*np.random.rand(10,3),100*np.random.rand(20,3)]
#colors=[np.random.rand(3,),np.random.rand(3,)]
from dipy.io import trackvis as tv
from dipy.core import track_performance
#from enthought.mayavi import mlab
#from enthought.tvtk.api import tvtk
import numpy as np
fname = '/home/eg309/Data/PBC/pbc2009icdm/brain1/brain1_scan1_fiber_track_mni.trk'
lines, hdr = tv.read(fname)
pts = [p[0] for p in lines]
pts_reduced = [track_performance.approx_polygon_track(p) for p in pts]
red = np.array([1, 0, 0])
trajs = pts_reduced
colors = [np.array([1., 0, 0]) for i in range(len(trajs))]
show(trajs)
