def bundle_from_refs(brain,braind, refs, divergence_threshold=0.3, fibre_weight=0.7,far_thresh=25,zhang_thresh=15, end_thresh=10):
'''
'''
bundle = set([])
centres = []
indices = []
for ref in refs:
refd=tm.downsample(ref,3)
brain_rf, ind_fr = rm_far_tracks(refd,braind,dist=far_thresh,down=True)
brain_rf=[brain[i] for i in ind_fr]
#brain_rf,ind_fr = rm_far_tracks(ref,brain,dist=far_thresh,down=False)
heavy_weight_fibres, index, centre = refconc(brain_rf, ref, divergence_threshold, fibre_weight)
heavy_weight_fibres_z = [i for i in heavy_weight_fibres if pf.zhang_distances(ref,brain_rf[i],'avg')<zhang_thresh]
#heavy_weight_fibres_z_e = [i for i in heavy_weight_fibres_z if tm.max_end_distances(brain_rf[i],ref)>end_thresh]
hwfind = set([ind_fr[i] for i in heavy_weight_fibres_z])
bundle = bundle.union(hwfind)
bundle_med = []
for i in bundle:
minmaxdist = 0.
for ref in refs:
minmaxdist=min(minmaxdist,tm.max_end_distances(brain[i],ref))
if minmaxdist<=end_thresh:
bundle_med.append(i)
#centres.append(centre)
#indices.append(index)
#return list(bundle), centres, indices
return bundle_med
def detect_corresponding_tracks(indices, tracks1, tracks2):
''' Detect corresponding tracks from 1 to 2
Parameters
----------
indices : sequence
of indices of tracks1 that are to be detected in tracks2
tracks1 : sequence
of tracks as arrays, shape (N1,3) .. (Nm,3)
tracks2 : sequence
of tracks as arrays, shape (M1,3) .. (Mm,3)
Returns
-------
track2track : array
of int showing the correspondance
'''
li = len(indices)
track2track = np.zeros((li, 3))
cnt = 0
for i in indices:
rt = [pf.zhang_distances(tracks1[i], t, 'avg') for t in tracks2]
rt = np.array(rt)
track2track[cnt - 1] = np.array([cnt, i, rt.argmin()])
cnt += 1
return track2track.astype(int)
def detect_corresponding_tracks(indices,tracks1,tracks2):
''' Detect corresponding tracks from 1 to 2
Parameters
----------
indices : sequence
of indices of tracks1 that are to be detected in tracks2
tracks1 : sequence
of tracks as arrays, shape (N1,3) .. (Nm,3)
tracks2 : sequence
of tracks as arrays, shape (M1,3) .. (Mm,3)
Returns
-------
track2track : array
of int showing the correspondance
'''
li=len(indices)
track2track=np.zeros((li,3))
cnt=0
for i in indices:
rt=[pf.zhang_distances(tracks1[i],t,'avg') for t in tracks2]
rt=np.array(rt)
track2track[cnt-1]=np.array([cnt,i,rt.argmin()])
cnt+=1
return track2track.astype(int)
def bundle_from_refs(brain,braind, refs, divergence_threshold=0.3, fibre_weight=0.7,far_thresh=25,zhang_thresh=15, end_thresh=10):
'''
'''
bundle = set([])
# centres = []
# indices = []
for ref in refs:
refd=tm.downsample(ref,3)
brain_rf, ind_fr = rm_far_tracks(refd,braind,dist=far_thresh,down=True)
brain_rf=[brain[i] for i in ind_fr]
#brain_rf,ind_fr = rm_far_tracks(ref,brain,dist=far_thresh,down=False)
heavy_weight_fibres, index, centre = refconc(brain_rf, ref, divergence_threshold, fibre_weight)
heavy_weight_fibres_z = [i for i in heavy_weight_fibres if pf.zhang_distances(ref,brain_rf[i],'avg')<zhang_thresh]
#heavy_weight_fibres_z_e = [i for i in heavy_weight_fibres_z if tm.max_end_distances(brain_rf[i],ref)>end_thresh]
hwfind = set([ind_fr[i] for i in heavy_weight_fibres_z])
bundle = bundle.union(hwfind)
bundle_med = []
for i in bundle:
minmaxdist = 0.
for ref in refs:
minmaxdist=min(minmaxdist,tm.max_end_distances(brain[i],ref))
if minmaxdist<=end_thresh:
bundle_med.append(i)
#centres.append(centre)
#indices.append(index)
#return list(bundle), centres, indices
return bundle_med
def relabel_by_atlas_value_and_mam(atlas_tracks,atlas,tes,tracks,tracksd,zhang_thr):
emi=emi_atlas()
brain_relabeled={}
for e in range(1,9): #from emi:
print emi[e]['bundle_name']
indices=emi[e]['init_ref']+emi[e]['selected_ref']+emi[e]['apr_ref']
tmp=detect_corresponding_tracks(indices,atlas_tracks,tracks)
corresponding_indices=tmp[:,2]
corresponding_indices=list(set(corresponding_indices))
value_indices=[]
for value in emi[e]['value']:
value_indices+=track_indices_for_a_value_in_atlas(atlas,value,tes,tracks)
value_indices=list(set(value_indices))
print 'len corr_ind',len(corresponding_indices)
#check if value_indices do not have anything in common with corresponding_indices and expand
if list(set(value_indices).intersection(set(corresponding_indices)))==[]:
#value_indices=corresponding_indices
print 'len corr_ind',len(corresponding_indices)
for ci in corresponding_indices:
print 'koukou',ci
ref=tracksd[ci]
brain_rf, ind_fr = rm_far_tracks(ref,tracksd,dist=10,down=True)
value_indices+=ind_fr
value_indices=list(set(value_indices))
print 'len vi',len(value_indices)
value_indices_new=[]
#reduce value_indices which are far from every corresponding fiber
for vi in value_indices:
dist=[]
for ci in corresponding_indices:
dist.append(pf.zhang_distances(tracks[vi],tracks[ci],'avg'))
for d in dist:
if d <= zhang_thr[e-1]:
value_indices_new.append(vi)
value_indices=list(set(value_indices_new))
#store value indices
brain_relabeled[e]={}
brain_relabeled[e]['value_indices']=value_indices
brain_relabeled[e]['corresponding_indices']=corresponding_indices
brain_relabeled[e]['color']=emi[e]['color']
brain_relabeled[e]['bundle_name']=emi[e]['bundle_name'][0]
return brain_relabeled
def relabel_by_atlas_value_and_zhang(atlas_tracks,atlas,tes,tracks,tracksd,zhang_thr):
emi=emi_atlas()
brain_relabeled={}
for e in range(1,9): #from emi:
print emi[e]['bundle_name']
indices=emi[e]['init_ref']+emi[e]['selected_ref']+emi[e]['apr_ref']
tmp=detect_corresponding_tracks(indices,atlas_tracks,tracks)
corresponding_indices=tmp[:,2]
corresponding_indices=list(set(corresponding_indices))
value_indices=[]
for value in emi[e]['value']:
value_indices+=track_indices_for_a_value_in_atlas(atlas,value,tes,tracks)
value_indices=list(set(value_indices))
print 'len corr_ind',len(corresponding_indices)
#check if value_indices do not have anything in common with corresponding_indices and expand
if list(set(value_indices).intersection(set(corresponding_indices)))==[]:
#value_indices=corresponding_indices
print 'len corr_ind',len(corresponding_indices)
for ci in corresponding_indices:
print 'koukou',ci
ref=tracksd[ci]
brain_rf, ind_fr = rm_far_tracks(ref,tracksd,dist=10,down=True)
value_indices+=ind_fr
value_indices=list(set(value_indices))
print 'len vi',len(value_indices)
value_indices_new=[]
#reduce value_indices which are far from every corresponding fiber
for vi in value_indices:
dist=[]
for ci in corresponding_indices:
dist.append(pf.zhang_distances(tracks[vi],tracks[ci],'avg'))
for d in dist:
if d <= zhang_thr[e-1]:
value_indices_new.append(vi)
value_indices=list(set(value_indices_new))
#store value indices
brain_relabeled[e]={}
brain_relabeled[e]['value_indices']=value_indices
brain_relabeled[e]['corresponding_indices']=corresponding_indices
brain_relabeled[e]['color']=emi[e]['color']
brain_relabeled[e]['bundle_name']=emi[e]['bundle_name'][0]
return brain_relabeled
def test_zhang_distances():
xyz1 = np.array([[0,0,0],[1,0,0],[2,0,0],[3,0,0]])
xyz2 = np.array([[0,1,1],[1,0,1],[2,3,-2]])
# dm=array([[ 2, 2, 17], [ 3, 1, 14], [6, 2, 13], [11, 5, 14]])
# this is the distance matrix between points of xyz1
# and points of xyz2
zd = tm.zhang_distances(xyz1,xyz2)
# {'average_mean_closest_distance': 3.9166666666666665,
# 'maximum_mean_closest_distance': 5.333333333333333,
# 'minimum_mean_closest_distance': 2.5}
yield assert_almost_equal, zd[0], 1.76135602742
xyz1=xyz1.astype('float32')
xyz2=xyz2.astype('float32')
zd2 = pf.zhang_distances(xyz1,xyz2)
yield assert_almost_equal, zd2[0], 1.76135602742
def corresponding_tracks(indices,tracks1,tracks2):
''' Detect similar tracks in different brains
'''
li=len(indices)
track2track=np.zeros((li,3))
cnt=0
for i in indices:
rt=[pf.zhang_distances(tracks1[i],t,'avg') for t in tracks2]
rt=np.array(rt)
track2track[cnt-1]=np.array([cnt,i,rt.argmin()])
cnt+=1
return track2track.astype(int)
def test_zhang_distances():
xyz1 = np.array([[0, 0, 0], [1, 0, 0], [2, 0, 0], [3, 0, 0]])
xyz2 = np.array([[0, 1, 1], [1, 0, 1], [2, 3, -2]])
# dm=array([[ 2, 2, 17], [ 3, 1, 14], [6, 2, 13], [11, 5, 14]])
# this is the distance matrix between points of xyz1
# and points of xyz2
zd = tm.zhang_distances(xyz1, xyz2)
# {'average_mean_closest_distance': 3.9166666666666665,
# 'maximum_mean_closest_distance': 5.333333333333333,
# 'minimum_mean_closest_distance': 2.5}
yield assert_almost_equal, zd[0], 1.76135602742
xyz1 = xyz1.astype('float32')
xyz2 = xyz2.astype('float32')
zd2 = pf.zhang_distances(xyz1, xyz2)
yield assert_almost_equal, zd2[0], 1.76135602742
def corresponding_tracks(self,indices,tracks1,tracks2):
''' Detect similar tracks in different brains
'''
li=len(indices)
track2track=np.zeros((li,3))
cnt=0
for i in indices:
rt=[pf.zhang_distances(tracks1[i],t,'avg') for t in tracks2]
rt=np.array(rt)
track2track[cnt-1]=np.array([cnt,i,rt.argmin()])
cnt+=1
return track2track.astype(int)
def detect_corresponding_tracks_extended(indices, tracks1, indices2, tracks2):
""" Detect corresponding tracks from 1 to 2
Parameters:
----------------
indices: sequence
of indices of tracks1 that are to be detected in tracks2
tracks1: sequence
of tracks as arrays, shape (N1,3) .. (Nm,3)
indices2: sequence
of indices of tracks2 in the initial brain
tracks2: sequence
of tracks as arrays, shape (M1,3) .. (Mm,3)
Returns:
-----------
track2track: array of int
showing the correspondance
"""
li = len(indices)
track2track = np.zeros((li, 3))
cnt = 0
for i in indices:
rt = [pf.zhang_distances(tracks1[i], t, "avg") for t in tracks2]
rt = np.array(rt)
track2track[cnt - 1] = np.array([cnt, i, indices2[rt.argmin()]])
cnt += 1
return track2track.astype(int)