-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy path3_graph_liver.py
210 lines (190 loc) · 9.17 KB
/
3_graph_liver.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
209
import os
import networkx as nx
import numpy
import numpy as np
from rivuletpy.rivulet import rtrace
import nibabel as nib
from mpl_toolkits.mplot3d import Axes3D
import skfmm
from scipy import ndimage
import pickle as pkl
import shutil
import time
import torch
import matplotlib.pyplot as plt
# _cenertline
def graph_construct(img_file,seg_file,size,edge_dist_thresh,save_dir):
image_name = str(img_file.split('/')[-1].split('.')[0].split('_img')[0])
print(image_name,"start")
image_itk = nib.load(img_file)
image_arr = image_itk.get_fdata().astype(np.float32) # x, y, z
seg_itk = nib.load(seg_file)
seg_arr = seg_itk.get_fdata().astype(np.float32) # x, y, z
# print(seg_arr.shape)
xyz_list = []
max_val = []
max_pos = []
label_idx = []
num_node = 0
num_edge = 0
z_size = image_arr.shape[2]
padding_z = (16 - z_size % 16) % 16
img = np.pad(image_arr, ((padding_z // 2, padding_z - padding_z // 2), (0, 0), (0, 0)), 'constant')
vessel_prob = np.pad(seg_arr, ((padding_z // 2, padding_z - padding_z // 2), (0, 0), (0, 0)), 'constant')
y_size = img.shape[1]
padding_y = (16 - y_size % 16) % 16
img = np.pad(img, ((0, 0), (padding_y // 2, padding_y - padding_y // 2), (0, 0)), 'constant')
vessel_prob = np.pad(vessel_prob, ((0, 0), (padding_y // 2, padding_y - padding_y // 2), (0, 0)), 'constant')
x_size = img.shape[0]
padding_x = (16 - x_size % 16) % 16
img = np.pad(img, ((0, 0), (0, 0), (padding_x // 2, padding_x - padding_x // 2)), 'constant')
vessel_prob = np.pad(vessel_prob, ((0, 0), (0, 0), (padding_x // 2, padding_x - padding_x // 2)), 'constant')
# find local maxima
im_x = img.shape[0]
im_y = img.shape[1]
im_z = img.shape[2]
x_quan = range(0, im_x, int(size / 2))
x_quan = sorted(list(set(x_quan) | set([im_x])))
y_quan = range(0, im_y, size)
y_quan = sorted(list(set(y_quan) | set([im_y])))
z_quan = range(0, im_z, size)
z_quan = sorted(list(set(z_quan) | set([im_z])))
for x_idx in range(len(x_quan) - 1):
for y_idx in range(len(y_quan) - 1):
for z_idx in range(len(z_quan) - 1):
cur_patch = vessel_prob[x_quan[x_idx]:x_quan[x_idx + 1], y_quan[y_idx]:y_quan[y_idx + 1], z_quan[z_idx]:z_quan[z_idx + 1]]
if np.sum(cur_patch) == 0 :
max_val.append(0)
max_pos.append((x_quan[x_idx] + int(cur_patch.shape[0] / 2),
y_quan[y_idx] + int(cur_patch.shape[1] / 2),
z_quan[z_idx] + int(cur_patch.shape[2] / 2)))
label_idx.append(0)
continue
else:
# temp = np.unravel_index(cur_patch.argmax(), cur_patch.shape)
# max_pos.append((x_quan[x_idx] + temp[0], y_quan[y_idx] + temp[1],z_quan[z_idx] + temp[2]))
max_val.append(np.max(cur_patch))
temp = np.zeros(3)
count = 0
for i_0 in range(cur_patch.shape[0]):
for i_1 in range(cur_patch.shape[1]):
for i_2 in range(cur_patch.shape[2]):
if cur_patch[i_0, i_1, i_2] == 1:
temp += np.array([i_0, i_1, i_2])
count += 1
temp = np.around(temp / count).astype(int)
max_pos.append((x_quan[x_idx] + temp[0], y_quan[y_idx] + temp[1], z_quan[z_idx] + temp[2]))
label_idx.append(1)
num_node += 1
graph = nx.Graph()
# add nodes
for node_idx, (node_x, node_y,node_z) in enumerate(max_pos):
graph.add_node(node_idx , x=node_x, y=node_y, z=node_z ,node_label = label_idx[node_idx])
xyz_list.append((node_x,node_y,node_z))
# print(node_x,node_y,node_z)
speed = seg_arr
#边构建的距离阈值
edge_method = 'geo_dist'
node_list = list(graph.nodes)
# print(len(node_list),node_list)
# print(graph.nodes.data())
# print(graph.nodes[n]['x'], graph.nodes[n]['y'], graph.nodes[n]['z'])
for i, n in enumerate(node_list):
if speed[graph.nodes[n]['x'], graph.nodes[n]['y'], graph.nodes[n]['z']] == 0:
continue
neighbor = speed[max(0, graph.nodes[n]['x'] - 1): min(im_x, graph.nodes[n]['x'] + 2),
max(0, graph.nodes[n]['y'] - 1): min(im_x, graph.nodes[n]['y'] + 2),
max(0, graph.nodes[n]['z'] - 1): min(im_x, graph.nodes[n]['z'] + 2)],
if np.mean(neighbor) < 0.1:
continue
if edge_method == 'geo_dist':
# phi = np.ones_like(speed)
# phi[graph.nodes[n]['x'], graph.nodes[n]['y'],graph.nodes[n]['z']] = -1
mask = np.zeros_like(speed, dtype=bool)
mask[graph.nodes[n]['x'], graph.nodes[n]['y'], graph.nodes[n]['z']] = True
dist = ndimage.distance_transform_edt(~mask)
phi = np.where(mask, -dist, dist)
tt = skfmm.travel_time(phi, speed, narrow=edge_dist_thresh) # travel time
for n_comp in node_list[i + 1:]:
geo_dist = tt[graph.nodes[n_comp]['x'], graph.nodes[n_comp]['y'],graph.nodes[n_comp]['z']] # travel time
if geo_dist < edge_dist_thresh:
graph.add_edge(n, n_comp, weight=edge_dist_thresh / (edge_dist_thresh + geo_dist))
num_edge += 1
# print('An edge BTWN', 'node', n, '&', n_comp, 'is constructed')
# print('Generate total', num_node, 'nodes, ', num_edge, 'edges.')
#
## The graph to visualize
# node_xyz_list = np.array(xyz_list)
# edge_list = np.array(graph.edges)
# edge_xyz_list = []
# #
# for idx in range(len(edge_list)):
# for node_idx in range(len(node_list)):
# if edge_list[idx][0] == node_list[node_idx] :
# edge_xyz_list.append([tuple(node_xyz_list[edge_list[idx][0]]),tuple(node_xyz_list[edge_list[idx][1]])])
# # Create the 3D figure
# fig = plt.figure()
# ax = fig.add_subplot(111, projection="3d", )
# # Plot the nodes - alpha is scaled by "depth" automatically
# ax.scatter(*node_xyz_list.T, s=5, ec="r")
# # Plot the edges
# for vizedge in np.array(edge_xyz_list):
# ax.plot(*vizedge.T, color="tab:gray")
# def _format_axes(ax):
# """Visualization options for the 3D axes."""
# # Turn gridlines off
# ax.grid(False)
# # Suppress tick labels
# for dim in (ax.xaxis, ax.yaxis, ax.zaxis):
# dim.set_ticks([])
# # Set axes labels
# ax.set_xlabel("x")
# ax.set_ylabel("y")
# ax.set_zlabel("z")
# _format_axes(ax)
# fig.tight_layout()
# plt.show()
# Save the graph as files
file_name = image_name.split('_p')[0]
save_dir = save_dir + '/'+ file_name + '/'
if not os.path.exists(save_dir):
os.makedirs(save_dir)
graph_save_path = os.path.join(save_dir, image_name + '_' + str(size) + '.gpickle')
nx.write_gpickle(graph, graph_save_path, protocol=pkl.HIGHEST_PROTOCOL)
return graph
if __name__ == '__main__':
centerline = '/media/Data/yanxc/Liver_vessel/pre_data/val/'
save_dir = '/media/Data/yanxc/Liver_vessel/pre_data/graph_val'
seg_paths = [os.path.join(centerline, x, y)
for x in os.listdir(centerline)
for y in os.listdir(os.path.join(centerline, x))
if y.endswith('_seg.nii.gz')]
img_paths = [os.path.join(centerline, x, y)
for x in os.listdir(centerline)
for y in os.listdir(os.path.join(centerline, x))
if y.endswith('_img.nii.gz')]
seg_paths.sort()
img_paths.sort()
print(len(seg_paths),seg_paths)
print(len(img_paths), img_paths)
for idx in range(len(img_paths)):
print(img_paths[idx])
time_start = time.time()
graph = graph_construct(img_paths[idx], seg_paths[idx], size=12, edge_dist_thresh=20,save_dir=save_dir)
lens = len(list(graph.nodes))
pos = torch.zeros((lens, 3))
for i in range(lens):
x, y, z = graph.nodes[i]['x'] + 1, graph.nodes[i]['y'] + 1, graph.nodes[i]['z'] + 1 # 加一是因为在三维矩阵中索引是从0开始(0-127)，而在图像中是从1开始(1-128)，为了保证坐标轴对应
pos[i] = torch.tensor([x, y, z], dtype=torch.float)
edge_index = torch.tensor(list(graph.edges), dtype=torch.long)
x = torch.tensor(list(graph.nodes), dtype=torch.float)
target = torch.tensor([graph.nodes[i]['node_label'] for i in range(lens)], dtype=torch.long)
print('节点个数:', len(list(graph.nodes)), ' ', '边个数:', len(list(graph.edges)), ' ')
print('********** Time:', time.time() - time_start, '**********')
# try:
# graph,_ = graph_construct(patch_paths[idx], seg_paths[idx],size=8,edge_dist_thresh=15)
# except ValueError:
# print('跳过了',patch_paths[idx])
# patch_name = patch_paths[idx].split('\\')[-1].split('_c')[0]
# shutil.move(patch_paths[idx], "E:/graph_data/del_data/" + patch_name + '_cenertline.nii.gz')
# continue