-
Notifications
You must be signed in to change notification settings - Fork 40
/
Copy patht_el_9.py
234 lines (170 loc) · 8.42 KB
/
t_el_9.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
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
import numpy as np
import data_transforms
import data_iterators
import pathfinder
import lasagne as nn
from collections import namedtuple
from functools import partial
import lasagne.layers.dnn as dnn
import lasagne
import theano.tensor as T
import utils
restart_from_save = None #"/home/eavsteen/dsb3/storage/metadata/dsb3//models/eavsteen/t_el_4-20170323-010448.pkl"
rng = np.random.RandomState(33)
# transformations
p_transform = {'patch_size': (48, 48, 48),
'mm_patch_size': (48, 48, 48),
'pixel_spacing': (1., 1., 1.)
}
p_transform_augment = {
'translation_range_z': [-3, 3],
'translation_range_y': [-3, 3],
'translation_range_x': [-3, 3],
'rotation_range_z': [-180, 180],
'rotation_range_y': [-180, 180],
'rotation_range_x': [-180, 180]
}
# based om malignancy chances found here: http://www.radiologyassistant.nl/en/p460f9fcd50637/solitary-pulmonary-nodule-benign-versus-malignant.html
# confusion_weights = np.array([[0, 0.1, 1, 15, 75, 100],
# [0.1, 0, 0.9, 14.9, 74.9, 99.9],
# [1, 0.9, 0, 14, 74, 99],
# [15, 14.9, 14, 0, 60, 85],
# [75, 74.9, 74, 60, 0, 25],
# [100, 99.9, 99, 85, 25, 0]])
malignancy_ground_truth = np.array([0, .001, .01, .15, .75, 1.])
# data preparation function
def data_prep_function(data, patch_center, pixel_spacing, luna_origin, p_transform,
p_transform_augment, world_coord_system, **kwargs):
x, patch_annotation_tf = data_transforms.transform_patch3d(data=data,
luna_annotations=None,
patch_center=patch_center,
p_transform=p_transform,
p_transform_augment=p_transform_augment,
pixel_spacing=pixel_spacing,
luna_origin=luna_origin,
world_coord_system=world_coord_system)
x = data_transforms.hu2normHU(x)
return x
data_prep_function_train = partial(data_prep_function, p_transform_augment=p_transform_augment,
p_transform=p_transform, world_coord_system=True)
data_prep_function_valid = partial(data_prep_function, p_transform_augment=None,
p_transform=p_transform, world_coord_system=True)
# data iterators
batch_size = 16
nbatches_chunk = 1
chunk_size = batch_size * nbatches_chunk
train_valid_ids = utils.load_pkl(pathfinder.LUNA_VALIDATION_SPLIT_PATH)
train_pids, valid_pids = train_valid_ids['train'], train_valid_ids['valid']
bin_borders = [4,8,20,30,100]
train_data_iterator = data_iterators.CandidatesLunaSizeBinDataGenerator(data_path=pathfinder.LUNA_DATA_PATH,
batch_size=chunk_size,
transform_params=p_transform,
data_prep_fun=data_prep_function_train,
rng=rng,
patient_ids=train_valid_ids['train'],
full_batch=True, random=True, infinite=True,
positive_proportion=0.5)
valid_data_iterator = data_iterators.CandidatesLunaSizeBinValidDataGenerator(data_path=pathfinder.LUNA_DATA_PATH,
transform_params=p_transform,
data_prep_fun=data_prep_function_valid,
patient_ids=train_valid_ids['valid'])
no_final_units = 1 + len(train_data_iterator.bin_borders)
assert(len(train_data_iterator.bin_borders) == len(valid_data_iterator.bin_borders))
nchunks_per_epoch = train_data_iterator.nsamples / chunk_size
max_nchunks = nchunks_per_epoch * 100
validate_every = int(5. * nchunks_per_epoch)
save_every = int(5. * nchunks_per_epoch)
learning_rate_schedule = {
0: 4e-4,
int(max_nchunks * 0.5): 1e-4,
int(max_nchunks * 0.6): 5e-5,
int(max_nchunks * 0.7): 2.5e-5,
int(max_nchunks * 0.8): 1.25e-5,
int(max_nchunks * 0.9): 0.625e-5
}
# model
conv3d = partial(dnn.Conv3DDNNLayer,
filter_size=3,
pad='same',
W=nn.init.Orthogonal(),
b=nn.init.Constant(0.01),
nonlinearity=nn.nonlinearities.very_leaky_rectify)
max_pool3d = partial(dnn.MaxPool3DDNNLayer,
pool_size=2)
drop = lasagne.layers.DropoutLayer
bn = lasagne.layers.batch_norm
dense = partial(lasagne.layers.DenseLayer,
W=lasagne.init.Orthogonal('relu'),
b=lasagne.init.Constant(0.0),
nonlinearity=lasagne.nonlinearities.rectify)
def inrn_v2(lin):
n_base_filter = 32
l1 = conv3d(lin, n_base_filter, filter_size=1)
l2 = conv3d(lin, n_base_filter, filter_size=1)
l2 = conv3d(l2, n_base_filter, filter_size=3)
l3 = conv3d(lin, n_base_filter, filter_size=1)
l3 = conv3d(l3, n_base_filter, filter_size=3)
l3 = conv3d(l3, n_base_filter, filter_size=3)
l = lasagne.layers.ConcatLayer([l1, l2, l3])
l = conv3d(l, lin.output_shape[1], filter_size=1)
l = lasagne.layers.ElemwiseSumLayer([l, lin])
l = lasagne.layers.NonlinearityLayer(l, nonlinearity=lasagne.nonlinearities.rectify)
return l
def inrn_v2_red(lin):
# We want to reduce our total volume /4
den = 16
nom2 = 4
nom3 = 5
nom4 = 7
ins = lin.output_shape[1]
l1 = max_pool3d(lin)
l2 = conv3d(lin, ins // den * nom2, filter_size=3, stride=2)
l3 = conv3d(lin, ins // den * nom2, filter_size=1)
l3 = conv3d(l3, ins // den * nom3, filter_size=3, stride=2)
l4 = conv3d(lin, ins // den * nom2, filter_size=1)
l4 = conv3d(l4, ins // den * nom3, filter_size=3)
l4 = conv3d(l4, ins // den * nom4, filter_size=3, stride=2)
l = lasagne.layers.ConcatLayer([l1, l2, l3, l4])
return l
def feat_red(lin):
# We want to reduce the feature maps by a factor of 2
ins = lin.output_shape[1]
l = conv3d(lin, ins // 2, filter_size=1)
return l
def build_model():
l_in = nn.layers.InputLayer((None, 1,) + p_transform['patch_size'])
l_target = nn.layers.InputLayer((None, 1))
l = conv3d(l_in, 64)
l = inrn_v2_red(l)
l = inrn_v2(l)
l = inrn_v2_red(l)
l = inrn_v2(l)
l = inrn_v2_red(l)
l = inrn_v2_red(l)
l = dense(drop(l), 512)
l_out = nn.layers.DenseLayer(l, num_units=no_final_units,
W=lasagne.init.Orthogonal(),
b=lasagne.init.Constant(0.1),
nonlinearity=nn.nonlinearities.softmax)
return namedtuple('Model', ['l_in', 'l_out', 'l_target'])(l_in, l_out, l_target)
def build_objective(model, deterministic=False, epsilon=1e-12):
predictions = nn.layers.get_output(model.l_out, deterministic=deterministic)
targets = T.cast(T.flatten(nn.layers.get_output(model.l_target)), 'int32')
#malignancy_broadcastable
p_ground_truth_malignancy = malignancy_ground_truth[targets]
p_pred_malignancy = predictions * malignancy_ground_truth
p_pred_malignancy = T.clip(p_pred_malignancy, epsilon, 1.-epsilon)
loss = - p_ground_truth_malignancy * T.log(p_pred_malignancy) \
- (1.-p_ground_truth_malignancy) * T.log(1.-p_pred_malignancy)
return T.mean(loss)
def build_objective2(model, deterministic=False, epsilon=1e-12):
predictions = nn.layers.get_output(model.l_out, deterministic=deterministic)
targets = T.flatten(nn.layers.get_output(model.l_target))
targets = T.clip(targets, 0, 1)
p_no_nodule = predictions[:,0]
p_nodule = np.float32(1.)-p_no_nodule
bce = T.nnet.binary_crossentropy(p_nodule, targets)
return T.mean(bce)
def build_updates(train_loss, model, learning_rate):
updates = nn.updates.adam(train_loss, nn.layers.get_all_params(model.l_out, trainable=True), learning_rate)
return updates