if __name__ == "__main__":
# datadir = '../data'
targnii = 'prob-face-object.nii.gz'
masknii = 'prob-face-object.nii.gz'
mask = nib.load(masknii)
mask = mask.get_data()
# ds = bpfmri_dataset(pjoin(datadir,targnii))
ds = bpfmri_dataset(targnii)
nbs = neighbor.volneighbors(mask, 3,26 )
nbs = nbs.compute_offsets()
map = GraphMapper(nbs)
nds = map(ds)
# assert_array_equal(data, map2nifti(nds).get_data()[...,0])
result = map2nifti(nds)
def compute_conn(self):
# get neigbors for each voxels
targ = np.reshape(self.targ.get_data(), (np.prod(self.dims[0:3]), -1))
if self.mask is not None:
mask = self.mask.get_data()
else:
mask = self.targ.get_data()
print np.shape(mask)
pdb.set_trace()
# tempdata = np.array([[[1,1,1],[2,2,2],[3,3,3]],[[4,4,4],[5,5,5],[6,6,6]],[[7,7,7],[8,8,8],[9,9,9]]])
# tempdata = np.array(
# get neigbors for each voxels
nbs = neighbor.volneighbors(mask, self.nb[0], self.nb[1])
# nbs = neighbor.volneighbors(tempdata, self.nb[0], self.nb[1])
vnum = nbs.voxnum
nbs = nbs.compute_offsets()
# compute connectivity matrx
I = np.zeros((vnum * 27))
J = np.zeros((vnum * 27))
V = np.zeros((vnum * 27))
count = 0
for v in range(vnum):
# print v
# print count
# pdb.set_trace()
vidx = nbs[v][0] # voxel index in image space
nbidx = nbs[v][1] # neighbor index in image space
# dist = distance.cdist(targ[vidx,:],targ[nbidx,:],self.metric)
# dist = np.exp(neurodist)
# dist =np.exp(- (nbs[v][2]**2+ nbs[v][3]**2))
dist = np.exp(-nbs[v][2])
# print np.shape(dist)
# print dist
# I = np.append(I,np.tile(nbs[v][0], nbidx.shape[0])) # in mask index
# J = np.append(J,nbs[v][1]) # in mask index
# V = np.append(V,dist)
nbs_len = len(nbidx)
# print nbs_len
I[count : count + nbs_len] = np.tile(nbs[v][0], nbidx.shape[0])
# print I[count:count + nbs_len]
J[count : count + nbs_len] = nbs[v][1]
# print J[count:count + nbs_len]
V[count : count + nbs_len] = dist
# print V[count:count + nbs_len]
# I.append(np.tile(nbs[v][0], nbidx.shape[0]))
# J.append(nbs[v][1])
# V.append(dist)
count = count + nbs_len
pdb.set_trace()
print "ok"
print np.shape(V)
print np.shape(I)
print np.shape(J)
dims = volgraph.get_dims(self)
# total = np.prod(dims)
total = np.prod(dims)
graph = sparse.csc_matrix((V, (I, J)), shape=(total, total))
return graph
def compute_conn(self):
# get neigbors for each voxels
targ = np.reshape(self.targ.get_data(), (np.prod(self.dims[0:3]), -1))
if self.mask is not None:
mask = self.mask.get_data()
else:
mask = self.targ.get_data()
print np.shape(mask)
pdb.set_trace()
# tempdata = np.array([[[1,1,1],[2,2,2],[3,3,3]],[[4,4,4],[5,5,5],[6,6,6]],[[7,7,7],[8,8,8],[9,9,9]]])
# tempdata = np.array(
# get neigbors for each voxels
nbs = neighbor.volneighbors(mask, self.nb[0], self.nb[1])
# nbs = neighbor.volneighbors(tempdata, self.nb[0], self.nb[1])
vnum = nbs.voxnum
nbs = nbs.compute_offsets()
#compute connectivity matrx
I = np.zeros((vnum * 27))
J = np.zeros((vnum * 27))
V = np.zeros((vnum * 27))
count = 0
for v in range(vnum):
# print v
# print count
# pdb.set_trace()
vidx = nbs[v][0] # voxel index in image space
nbidx = nbs[v][1] # neighbor index in image space
# dist = distance.cdist(targ[vidx,:],targ[nbidx,:],self.metric)
#dist = np.exp(neurodist)
# dist =np.exp(- (nbs[v][2]**2+ nbs[v][3]**2))
dist = np.exp(-nbs[v][2])
# print np.shape(dist)
# print dist
# I = np.append(I,np.tile(nbs[v][0], nbidx.shape[0])) # in mask index
# J = np.append(J,nbs[v][1]) # in mask index
# V = np.append(V,dist)
nbs_len = len(nbidx)
# print nbs_len
I[count:count + nbs_len] = np.tile(nbs[v][0], nbidx.shape[0])
# print I[count:count + nbs_len]
J[count:count + nbs_len] = nbs[v][1]
# print J[count:count + nbs_len]
V[count:count + nbs_len] = dist
# print V[count:count + nbs_len]
# I.append(np.tile(nbs[v][0], nbidx.shape[0]))
# J.append(nbs[v][1])
# V.append(dist)
count = count + nbs_len
pdb.set_trace()
print "ok"
print np.shape(V)
print np.shape(I)
print np.shape(J)
dims = volgraph.get_dims(self)
# total = np.prod(dims)
total = np.prod(dims)
graph = sparse.csc_matrix((V, (I, J)), shape=(total, total))
return graph
