-
Notifications
You must be signed in to change notification settings - Fork 0
/
dataset_hn2.py
180 lines (141 loc) · 5.96 KB
/
dataset_hn2.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
import os
import nibabel as nib
import numpy as np
import random
from torch.utils.data import Dataset
def normalize(img):
norm_img = np.clip(img, -1024, 1024) / 1024
return norm_img
class HNDataset(Dataset):
def __init__(self, img_dir, shift, transform=None):
self.image_dir = img_dir
self.transform = transform
self.shift = shift
self.images = os.listdir(img_dir)
def __len__(self):
return len(self.images)
def get_locations(self, gt, x_lim, y_lim, z_lim):
random_loc = random.choice(np.argwhere(gt == 1))
x = random_loc[0] + np.random.randint(-x_lim, x_lim)
y = random_loc[1] + np.random.randint(-y_lim, y_lim)
z = random_loc[2] + np.random.randint(-z_lim, z_lim)
return x, y, z
def augment_locations(self, patch_shape, dims, x, y, z):
x_min = x - (patch_shape[0] // 2)
x_max = x + (patch_shape[0] // 2)
y_min = y - (patch_shape[1] // 2)
y_max = y + (patch_shape[1] // 2)
z_min = z - (patch_shape[2] // 2)
z_max = z + (patch_shape[2] // 2)
sampling_loc = [x_min, x_max, y_min, y_max, z_min, z_max]
if sampling_loc[0] < 0:
sampling_loc[0] = 0
sampling_loc[1] = sampling_loc[0] + patch_shape[0]
if sampling_loc[1] > dims[0]:
sampling_loc[1] = dims[0]
sampling_loc[0] = sampling_loc[1] - patch_shape[0]
if sampling_loc[2] < 0:
sampling_loc[2] = 0
sampling_loc[3] = sampling_loc[2] + patch_shape[1]
if sampling_loc[3] > dims[1]:
sampling_loc[3] = dims[1]
sampling_loc[2] = sampling_loc[3] - patch_shape[1]
if sampling_loc[4] < 0:
sampling_loc[4] = 0
sampling_loc[5] = sampling_loc[4] + patch_shape[2]
if sampling_loc[5] > dims[2]:
sampling_loc[5] = dims[2]
sampling_loc[4] = sampling_loc[5] - patch_shape[2]
return sampling_loc
def sample_patches(self, ct, gt, sampling_loc):
ct_patch = ct[sampling_loc[0]:sampling_loc[1],
sampling_loc[2]:sampling_loc[3],
sampling_loc[4]:sampling_loc[5]]
gt_patch = gt[sampling_loc[0]:sampling_loc[1],
sampling_loc[2]:sampling_loc[3],
sampling_loc[4]:sampling_loc[5]]
return ct_patch, gt_patch
def __getitem__(self, index):
patient = os.path.join(self.image_dir, self.images[index])
ct_path = os.path.join(patient, "CT/CT.nii.gz")
gt_path = os.path.join(patient, "GT/GT.nii.gz")
ct = nib.load(ct_path).get_fdata()
gt = nib.load(gt_path).get_fdata()
gt[gt > 0] = 1
ct = normalize(ct)
dims = np.shape(ct)
patch_shape = [144, 144, 144]
x, y, z = self.get_locations(gt,
x_lim=self.shift,
y_lim=self.shift,
z_lim=self.shift)
sampling_loc = self.augment_locations(patch_shape, dims, x, y, z)
ct_patch, gt_patch = self.sample_patches(ct, gt, sampling_loc)
sample = dict()
sample['input'] = ct_patch
sample['target'] = gt_patch
if self.transform is not None:
sample = self.transform(sample)
return sample
class HNDatasetVal(Dataset):
def __init__(self, img_dir, transform=None):
self.image_dir = img_dir
self.transform = transform
self.images = os.listdir(img_dir)
def __len__(self):
return len(self.images)
def augment_locations(self, patch_shape, dims, x, y, z):
x_min = x - (patch_shape[0] // 2)
x_max = x + (patch_shape[0] // 2)
y_min = y - (patch_shape[1] // 2)
y_max = y + (patch_shape[1] // 2)
z_min = z - (patch_shape[2] // 2)
z_max = z + (patch_shape[2] // 2)
sampling_loc = [x_min, x_max, y_min, y_max, z_min, z_max]
if sampling_loc[0] < 0:
sampling_loc[0] = 0
sampling_loc[1] = sampling_loc[0] + patch_shape[0]
if sampling_loc[1] > dims[0]:
sampling_loc[1] = dims[0]
sampling_loc[0] = sampling_loc[1] - patch_shape[0]
if sampling_loc[2] < 0:
sampling_loc[2] = 0
sampling_loc[3] = sampling_loc[2] + patch_shape[1]
if sampling_loc[3] > dims[1]:
sampling_loc[3] = dims[1]
sampling_loc[2] = sampling_loc[3] - patch_shape[1]
if sampling_loc[4] < 0:
sampling_loc[4] = 0
sampling_loc[5] = sampling_loc[4] + patch_shape[2]
if sampling_loc[5] > dims[2]:
sampling_loc[5] = dims[2]
sampling_loc[4] = sampling_loc[5] - patch_shape[2]
return sampling_loc
def sample_patches(self, ct, gt, sampling_loc):
ct_patch = ct[sampling_loc[0]:sampling_loc[1],
sampling_loc[2]:sampling_loc[3],
sampling_loc[4]:sampling_loc[5]]
gt_patch = gt[sampling_loc[0]:sampling_loc[1],
sampling_loc[2]:sampling_loc[3],
sampling_loc[4]:sampling_loc[5]]
return ct_patch, gt_patch
def __getitem__(self, index):
patient = os.path.join(self.image_dir, self.images[index])
ct_path = os.path.join(patient, "CT/CT.nii.gz")
gt_path = os.path.join(patient, "GT/GT.nii.gz")
ct = nib.load(ct_path).get_fdata()
gt = nib.load(gt_path).get_fdata()
gt[gt > 0] = 1
ct = normalize(ct)
dims = np.shape(ct)
patch_shape = [144, 144, 144]
pos = np.argwhere((gt == 1))
x, y, z = random.choice(pos)
sampling_loc = self.augment_locations(patch_shape, dims, x, y, z)
ct_patch, gt_patch = self.sample_patches(ct, gt, sampling_loc)
sample = dict()
sample['input'] = ct_patch
sample['target'] = gt_patch
if self.transform is not None:
sample = self.transform(sample)
return sample