-
Notifications
You must be signed in to change notification settings - Fork 57
/
lib.py
176 lines (151 loc) · 4.96 KB
/
lib.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
import os
import numpy as np
import tensorflow as tf
import nibabel as nib
#### Helpers for file IOs
def _read_lists(fid):
"""
Read all kinds of lists from text file to python lists
"""
if not os.path.isfile(fid):
return None
with open(fid,'r') as fd:
_list = fd.readlines()
my_list = []
for _item in _list:
if len(_item) < 3:
_list.remove(_item)
my_list.append(_item.split('\n')[0])
return my_list
def _save(sess, model_path, global_step):
"""
Saves the current session to a checkpoint
"""
saver = tf.train.Saver()
save_path = saver.save(sess, model_path, global_step = global_step)
return save_path
def _save_nii_prediction(gth, comp_pred, ref_fid, out_folder, out_bname, debug = False):
"""
save prediction, sample and gth to nii file given a reference
"""
# first write prediction
ref_obj = read_nii_object(ref_fid)
ref_affine = ref_obj.get_affine()
out_bname = out_bname.split(".")[0] + ".nii.gz"
write_nii(comp_pred, out_bname, out_folder, affine = ref_affine)
# then write sample
_local_gth = gth.copy()
_local_gth[_local_gth > self.num_cls - 1] = 0
out_label_bname = "gth_" + out_bname
write_nii(_local_gth, out_label_bname, out_folder, affine = ref_affine)
def write_nii(array_data, filename, path = "", affine = None):
"""write np array into nii file"""
if affine is None:
print("No information about the global coordinate system")
affine = np.diag([1,1,1,1])
#pdb.set_trace()
#TODO: to check if it works
# array_data = np.int16(array_data)
array_img = nib.Nifti1Image(array_data, affine)
save_fid = os.path.join(path,filename)
try:
array_img.to_filename(save_fid)
print("Nii object %s has been saved!"%save_fid)
except:
raise Exception("file %s cannot be saved!"%save_fid)
return save_fid
def read_nii_image(input_fid):
"""read the nii image data into numpy array"""
img = nib.load(input_fid)
return img.get_data()
def read_nii_object(input_fid):
""" directly read the nii object """
#pdb.set_trace()
return nib.load(input_fid)
#### Helpers for evaluations
def _label_decomp(num_cls, label_vol):
"""
decompose label for softmax classifier
original labels are batchsize * W * H * 1, with label values 0,1,2,3...
this function decompse it to one hot, e.g.: 0,0,0,1,0,0 in channel dimension
numpy version of tf.one_hot
"""
_batch_shape = list(label_vol.shape)
_vol = np.zeros(_batch_shape)
_vol[label_vol == 0] = 1
_vol = _vol[..., np.newaxis]
for i in range(num_cls):
if i == 0:
continue
_n_slice = np.zeros(label_vol.shape)
_n_slice[label_vol == i] = 1
_vol = np.concatenate( (_vol, _n_slice[..., np.newaxis]), axis = 3 )
return np.float32(_vol)
def _dice_eval(compact_pred, labels, n_class):
"""
calculate standard dice for evaluation, here uses the class prediction, not the probability
"""
dice_arr = []
dice = 0
eps = 1e-7
pred = tf.one_hot(compact_pred, depth = n_class, axis = -1)
for i in range(n_class):
inse = tf.reduce_sum(pred[:, :, :, i] * labels[:, :, :, i])
union = tf.reduce_sum(pred[:, :, :, i]) + tf.reduce_sum(labels[:, :, :, i])
dice = dice + 2.0 * inse / (union + eps)
dice_arr.append(2.0 * inse / (union + eps))
return 1.0 * dice / n_class, dice_arr
def _inverse_lookup(my_dict, _value):
for key, dic_value in list(my_dict.items()):
if dic_value == _value:
return key
return None
def _jaccard(conf_matrix):
"""
calculate jaccard similarity from confusion_matrix
"""
num_cls = conf_matrix.shape[0]
jac = np.zeros(num_cls)
for ii in range(num_cls):
pp = np.sum(conf_matrix[:,ii])
gp = np.sum(conf_matrix[ii,:])
hit = conf_matrix[ii,ii]
if (pp + gp -hit) == 0:
jac[ii] = 0
else:
jac[ii] = hit * 1.0 / (pp + gp - hit )
return jac
def _dice(conf_matrix):
"""
calculate dice coefficient from confusion_matrix
"""
num_cls = conf_matrix.shape[0]
dic = np.zeros(num_cls)
for ii in range(num_cls):
pp = np.sum(conf_matrix[:,ii])
gp = np.sum(conf_matrix[ii,:])
hit = conf_matrix[ii,ii]
if (pp + gp) == 0:
dic[ii] = 0
else:
dic[ii] = 2.0 * hit / (pp + gp)
return dic
def _indicator_eval(cm):
"""
Decompose confusion matrix and get statistics
"""
contour_map = { # a map used for mapping label value to its name, used for output
"bg": 0,
"la_myo": 1,
"la_blood": 2,
"lv_blood": 3,
"aa": 4
}
dice = _dice(cm)
jaccard = _jaccard(cm)
print(cm)
for organ, ind in list(contour_map.items()):
print(( "organ: %s"%organ ))
print(( "dice: %s"%(dice[int(ind)] ) ))
print(( "jaccard: %s"%(jaccard[int(ind)] ) ))
return dice, jaccard