-
Notifications
You must be signed in to change notification settings - Fork 2
/
skeletonize.py
208 lines (131 loc) · 4.68 KB
/
skeletonize.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
# Import necessary packages
from skeletonization import *
from postprocess import *
def combination_pairs(n):
pairs = np.array([])
for i in range(n):
for j in range(n-i-1):
pairs = np.concatenate((pairs,np.array([i,i+j+1])))
pairs = np.reshape(pairs,[pairs.shape[0]/2,2])
pairs = pairs.astype('uint8')
return pairs
def chunk_points(p, chunk_size=512, overlap=100):
min_p = np.min(p, 0)
max_p = np.max(p, 0)
min_bound = np.floor(min_p/256)*256
max_bound = np.ceil((max_p-min_bound)/chunk_size)*chunk_size + min_bound
n_chunk = (max_bound - min_bound)/chunk_size
n_chunk = n_chunk.astype('uint16')
p_list = []
filled = np.array([])
bound_list = np.zeros([n_chunk[0]*n_chunk[1]*n_chunk[2],2,3])
c = 0
for i in range(n_chunk[0]):
for j in range(n_chunk[1]):
for k in range(n_chunk[2]):
chunk_range = np.array([[i*chunk_size+min_bound[0],j*chunk_size+min_bound[1],k*chunk_size+min_bound[2]],
[(i+1)*chunk_size+min_bound[0],(j+1)*chunk_size+min_bound[1],(k+1)*chunk_size+min_bound[2]]])
chunk_range[0,:] = chunk_range[0,:] - overlap
chunk_range[1,:] = chunk_range[1,:] + overlap
bound_list[c,:,:] = chunk_range
valid = get_valid(p,chunk_range)
if np.sum(valid) < 50:
p_list.append(np.array([]))
c = c + 1
continue
p_valid = p[valid,:]
p_list.append(p_valid)
filled = np.concatenate((filled,np.array([i,j,k])))
c = c + 1
filled = np.reshape(filled,[filled.shape[0]/3,3])
n_filled = filled.shape[0]
chunk_pairs = combination_pairs(n_filled)
adjacent_chunks = np.array([])
for i in range(chunk_pairs.shape[0]):
chunk1_coord = filled[chunk_pairs[i,0],:]
chunk2_coord = filled[chunk_pairs[i,1],:]
loc_diff = np.abs(chunk1_coord - chunk2_coord)
if np.max(loc_diff) == 1:
chunk1_idx = chunk1_coord[2] + n_chunk[2]*chunk1_coord[1] + n_chunk[2]*n_chunk[1]*chunk1_coord[0]
chunk2_idx = chunk2_coord[2] + n_chunk[2]*chunk2_coord[1] + n_chunk[2]*n_chunk[1]*chunk2_coord[0]
adjacent_chunks = np.concatenate((adjacent_chunks,np.array([chunk1_idx,chunk2_idx])))
adjacent_chunks = np.reshape(adjacent_chunks,[adjacent_chunks.shape[0]/2,2])
p_list.insert(0, adjacent_chunks)
p_list.insert(0, bound_list)
return p_list
def crop_skeleton(skeleton, bound):
nodes = skeleton.nodes
edges = skeleton.edges
radii = skeleton.radii
nodes_valid_mask = get_valid(nodes, bound)
nodes_valid_idx = np.where(nodes_valid_mask)[0]
edges_valid_mask = np.isin(edges, nodes_valid_idx)
edges_valid_idx = edges_valid_mask[:,0]*edges_valid_mask[:,1]
edges_valid = edges[edges_valid_idx,:]
skeleton.edges = edges_valid
skeleton = consolidate_skeleton(skeleton)
return skeleton
def skeletonize_cell(points_list, parameters):
bound_list = points_list[0]
skeletons = []
for i in range(len(points_list)-2):
p = points_list[i+2]
if p.shape[0] != 0:
skeleton = skeletonize(p,1,[1,1,1],parameters)
if skeleton.nodes.shape[0] != 0:
bound = np.zeros([2,3])
bound[0,:] = bound_list[i,0,:] + 50
bound[1,:] = bound_list[i,1,:] - 50
skeleton = crop_skeleton(skeleton, bound)
skeletons.append(skeleton)
else:
skeletons.append(skeleton)
else:
skeleton = Skeleton()
skeletons.append(skeleton)
return skeletons
def skeletonize_cell_dist(points_list, parameters, n_core=None):
t0 = time()
pool = Pool(n_core)
# Skeletonize
points_chunk = points_list[2:]
try:
skeletons = pool.map(partial(skeletonize, object_id=1, dsmp_resolution=[1,1,1], parameters=parameters), points_chunk)
pool.close()
pool.join()
pool.terminate()
except:
pool.terminate()
t1 = time()
print(">>>>> Elapsed time : " + str(np.round(t1-t0, decimals=3)))
return skeletons
def crop_cell(skeletons, points_list):
# Crop skeletons
bound_list = points_list[0]
print(bound_list.shape)
print(len(skeletons))
for i in range(len(skeletons)):
skeleton = skeletons[i]
if skeleton.nodes.shape[0] != 0:
bound = np.zeros([2,3])
bound[0,:] = bound_list[i,0,:] + 50
bound[1,:] = bound_list[i,1,:] - 50
skeleton = crop_skeleton(skeleton, bound)
skeletons[i] = skeleton
return skeletons
def skeletonize_file(points_file, output_file, n_core=2, soma=0, soma_coord=np.array([])):
print('Loading points...')
p = np.load(points_file)
if soma:
p_wsoma = np.copy(p)
print('Removing soma...')
p = remove_soma(p_wsoma,soma_coord)
points_list = chunk_points(p,512,256)
skeletons = skeletonize_cell_dist(points_list, [10,10], n_core)
skeletons = crop_cell(skeletons, points_list)
print('Merging chunks...')
skeleton = merge_cell(skeletons, points_list)
skeleton = trim_skeleton(skeleton, p)
if soma:
skeleton = connect_soma(skeleton, soma_coord, p)
save_skeleton(skeleton, output_file)