-
Notifications
You must be signed in to change notification settings - Fork 1
/
predict.py
320 lines (267 loc) · 13.6 KB
/
predict.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
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
import paths
import csv
import os
import logging
import argparse
import SimpleITK as sitk
import numpy as np
import matplotlib.pyplot as plt
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import RepeatedStratifiedKFold, LeaveOneOut
from sklearn.feature_selection import VarianceThreshold
from sklearn.feature_selection import RFECV
from sklearn.feature_selection import SelectKBest, f_regression, mutual_info_regression
from sklearn import svm
from sklearn.calibration import CalibratedClassifierCV
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import mean_absolute_error, accuracy_score
from sklearn.ensemble import RandomForestRegressor, RandomForestClassifier
from utils import ReadImage, find_list, threshold_connectivity_matrix, weight_conversion, get_lesion_weights, get_train_dataset
from utils import extract_gt_mRS, extract_volumetric_features, extract_tractographic_features, extract_spatial_features
from utils import extract_volumetric_spatial_features, extract_morphological_features, extract_modified_volumetric_spatial_features
from utils import extract_new_tractographic_features, extract_tract_features
from xgboost import XGBRegressor
# =========================================== setup logs ====================================== #
log = os.path.join(os.getcwd(), 'log.txt')
fmt = '%(asctime)s %(message)s'
logging.basicConfig(level=logging.INFO, format=fmt, filename=log)
console = logging.StreamHandler()
console.setLevel(logging.INFO)
console.setFormatter(logging.Formatter(fmt))
logging.getLogger('').addHandler(console)
# ======================================= Parameters Setting ================================== #
use_original_feature = False
rfecv_feature_selection = True
# ======================================== Select Feature ==================================== #
#feature_type = 'volumetric'
#feature_type = 'spatial'
#feature_type = 'morphological'
# lesion in brain parcellation regions and the remaining region
#feature_type = 'volumetric_spatial'
# lesion in brain parcellation regions
#feature_type = 'original_volumetric_spatial'
# modified volumetric and spatial feature
#feature_type = 'modified_volumetric_spatial'
#atlas_name = 'HarvardOxfordSub'
#atlas_name = 'HarvardOxfordCort'
#atlas_name = 'aal'
#atlas_name = 'JHU-WhiteMatter-labels-1mm'
#atlas_name = 'MNI'
#atlas_name = 'OASIS_TRT_20'
#feature_type = 'oskar'
#feature_type = 'num_fibers'
#feature_type = 'new_fib_pass'
#feature_type = 'new_fib_end'
# original
#weight_type = 'original'
# modified
#weight_type = 'modified'
# one
#weight_type = 'one'
#aal_regions = 116
#feature_type = 'tract_dsi_pass'
#feature_type = 'tract_nrm_pass'
#feature_type = 'tract_bin_pass'
#feature_type = 'tract_dsi_end'
#feature_type = 'tract_nrm_end'
feature_type = 'tract_bin_end'
# original
#weight_type = 'original'
# modified
#weight_type = 'modified'
# one
weight_type = 'one'
aal_regions = 116
# =================================== Groundtruth of mRS scores ================================ #
logging.info('Extracting mRS scores...')
mRS_gt = extract_gt_mRS()
# ======================================== Feature Extraction ================================= #
# ======================================== Voluemtric Feature ================================= #
if 'volume' in feature_type and 'spatial' not in feature_type:
logging.info('Extracting volumetric features...')
features, features_list = extract_volumetric_features()
if 'spatial' in feature_type and 'volume' not in feature_type:
logging.info('Extracting spatial features...')
features, features_list = extract_spatial_features()
if 'morpho' in feature_type:
logging.info('Extracting morphological features...')
features, features_list = extract_morphological_features()
if 'volume' in feature_type and 'spatial' in feature_type:
logging.info('Extracting volumetric and spatial features...')
logging.info(atlas_name)
if 'ori' not in feature_type and 'mod' not in feature_type:
logging.info('Included the lesion outside brain parcellation regions...')
features, features_list = extract_volumetric_spatial_features(atlas_name)
if 'ori' in feature_type:
logging.info('Removing the lesion outside brain parcellation regions...')
features, features_list = extract_volumetric_spatial_features(atlas_name)
features = features[:,1:]
del features_list[0]
if 'mod' in feature_type:
logging.info('Using modified version of volumetric and spatial features...')
features, features_list = extract_modified_volumetric_spatial_features(atlas_name)
if 'tract' in feature_type:
logging.info('Extracting tractographic features...')
logging.info('Feature type: %s' %feature_type)
logging.info('Weight type: %s' %weight_type)
W_dsi_pass, W_nrm_pass, W_bin_pass, W_dsi_end, W_nrm_end, W_bin_end, tract_list = extract_tractographic_features(weight_type, aal_regions)
if 'dsi' in feature_type and 'pass' in feature_type:
features, features_list = W_dsi_pass, tract_list
elif 'nrm' in feature_type and 'pass' in feature_type:
features, features_list = W_nrm_pass, tract_list
elif 'bin' in feature_type and 'pass' in feature_type:
features, features_list = W_bin_pass, tract_list
elif 'dsi' in feature_type and 'end' in feature_type:
features, features_list = W_dsi_end, tract_list
elif 'nrm' in feature_type and 'end' in feature_type:
features, features_list = W_nrm_end, tract_list
else:
features, features_list = W_bin_end, tract_list
if 'oskar' in feature_type:
logging.info('Extracting Oskar features...')
features = np.load('./features/oskar_features.npy')
features_list = list(range(1662))
if 'num_fibers' in feature_type:
logging.info('Extracting number of tracts...')
features, features_list = extract_tract_features()
if 'new_fib' in feature_type:
logging.info('Extracting new tractographic features...')
logging.info('Feature type: %s' %feature_type)
logging.info('Weight type: %s' %weight_type)
W_pass_histogram_features, W_end_histogram_features, tractographic_list = extract_new_tractographic_features(weight_type, aal_regions)
if 'pass' in feature_type:
features, features_list = W_pass_histogram_features, tractographic_list
else:
features, features_list = W_end_histogram_features, tractographic_list
# =============================== Save Original Features ======================================= #
print(features[0,:])
logging.info('Saving Features...')
if not os.path.exists('./features/'):
os.mkdir('./features/')
#np.save('./features/volumetric_features.npy', features)
#np.save('./features/spatial_features.npy', features)
#np.save('./features/morphological_features.npy', features)
#np.save('./features/ori_aal_features.npy', features)
#np.save('./features/ori_tract_nrm_end_aal_features.npy', features)
#np.save('./features/mod_tract_bin_end_aal_features.npy', features)
# ============================== Feature Normalization ========================================= #
logging.info('Normalizing features...')
scaler = StandardScaler()
normalized_features = scaler.fit_transform(features)
# =========================== Remove features with low variance ================================= #
logging.info('Removing features with low variance...')
sel = VarianceThreshold(0.85*(1-0.85))
selected_normalized_features = sel.fit_transform(normalized_features)
selected_features_list = [name for idx, name in enumerate(features_list) if sel.get_support()[idx]]
# =========================== Select which feature to use =========================================== #
if use_original_feature:
# ====================================== Original Features ====================================== #
logging.info('Using original features...')
X, X_list = normalized_features, features_list
elif rfecv_feature_selection:
# ======================================= RFECV Features ======================================== #
logging.info('Using RFECV features...')
# Cross Validation Model
loo = LeaveOneOut()
estimator_rfecv = RandomForestRegressor(n_estimators=300, max_depth=3, random_state=1989, n_jobs=-1)
rfecv = RFECV(estimator_rfecv, step=1, cv=loo, scoring='neg_mean_absolute_error', n_jobs=-1)
rfecv_selected_normalized_features = rfecv.fit_transform(selected_normalized_features, mRS_gt)
rfecv_selected_features_list = [name for idx, name in enumerate(selected_features_list) if rfecv.get_support()[idx]]
X, X_list = rfecv_selected_normalized_features, rfecv_selected_features_list
else:
# ======================================= Features with high variance =========================== #
logging.info('Using features with high variance...')
X, X_list = selected_normalized_features, selected_features_list
# =============================================== Start Prediction ========================================= #
logging.info('Predicting...')
y = mRS_gt
# Cross Validation Model
loo = LeaveOneOut()
accuracy = np.zeros((37,1), dtype=np.float32)
y_pred_label = np.zeros((37,1), dtype=np.float32)
y_abs_error = np.zeros((37,1), dtype=np.float32)
##y_pred_proba = np.zeros((37,5), dtype=np.float32)
subject_feature_importances = np.zeros((37,len(X_list)), dtype=np.float32)
idx = 0
for train_index, test_index in loo.split(X):
X_train, X_test = X[train_index], X[test_index]
y_train, y_test = y[train_index], y[test_index]
# RF Regression
estimator_pred = RandomForestRegressor(n_estimators=300, max_depth=3, random_state=1989, n_jobs=-1)
estimator_pred.fit(X_train, y_train)
subject_importance = estimator_pred.feature_importances_
subject_feature_importances[idx,:] = subject_importance
y_pred_label[idx] = np.round(estimator_pred.predict(X_test))
##y_pred_proba[idx, :] = estimator_pred.predict_proba(X_test)
accuracy[idx] = accuracy_score(np.round(estimator_pred.predict(X_test)), y_test)
y_abs_error[idx] = np.absolute(y_pred_label[idx]-y_test)
idx += 1
logging.info(feature_type+" features with RF Regressor - Accuracy: %0.4f , MAE: %0.4f (+/- %0.4f)" %(np.mean(accuracy), np.mean(y_abs_error), np.std(y_abs_error)))
# =========================== Save Predicted Label ============================== #
logging.info('Saving Predicted Labels...')
if not os.path.exists('./predicted_labels/'):
os.mkdir('./predicted_labels/')
#np.save('./predicted_labels/rfecv_volumetric_pred_loo.npy', y_pred_label)
#np.save('./predicted_labels/rfecv_spatial_pred_loo.npy', y_pred_label)
#np.save('./predicted_labels/rfecv_morphological_pred_loo.npy', y_pred_label)
#np.save('./predicted_labels/morphological_pred_loo.npy', y_pred_label)
#np.save('./predicted_labels/rfecv_ori_aal_pred_loo.npy', y_pred_label)
#np.save('./predicted_labels/ori_aal_pred_loo.npy', y_pred_label)
#np.save('./predicted_labels/rfecv_ori_tract_nrm_end_aal_pred_loo.npy', y_pred_label)
#np.save('./predicted_labels/ori_tract_nrm_end_aal_pred_loo.npy', y_pred_label)
#np.save('./predicted_labels/oskar_pred_loo.npy', y_pred_label)
#np.save('./predicted_labels/mod_tract_bin_end_aal_pred_loo.npy', y_pred_label)
# =========================== Feature Importance ================================ #
importances = np.round(np.mean(subject_feature_importances, axis=0),decimals=2)
feature_importances = [(feature, round(importance, 2)) for feature, importance in zip(X_list, importances)]
# Sort the feature importances by most important first
feature_importances = sorted(feature_importances, key = lambda x: x[1], reverse = True)
# Print out the feature and importances
[logging.info('Variable: {:30} Importance: {}'.format(*pair)) for pair in feature_importances]
logging.info(feature_type+" features with RF Regressor - Accuracy: %0.2f , MAE: %0.2f (+/- %0.2f)" %(np.mean(accuracy), np.mean(y_abs_error), np.std(y_abs_error)))
#x_range = np.arange(volumetric_spatial_features.shape[1])
#lesion_histogram = volumetric_spatial_features.sum(axis=0)
#fig, ax = plt.subplots()
#plt.bar(x_range, lesion_histogram)
#plt.show()
'''# AAL top 7 features
regions_importances = [subject_feature_importances[:,88],
subject_feature_importances[:,6],
subject_feature_importances[:,17],
subject_feature_importances[:,72],
subject_feature_importances[:,64],
subject_feature_importances[:,15],
subject_feature_importances[:,13]]
fig, ax = plt.subplots(1, 1)
ax.boxplot(regions_importances, showmeans=True)
ax.set_title('Region Importance', fontsize = 60)
plt.ylabel('Importance', fontsize = 40)
region_names = ['LITG', 'LMFG', 'RRO', 'LP', 'LAG', 'ORIFG', 'TRIFG']
plt.xticks(np.arange(1,8), region_names, fontsize = 20)
plt.yticks(fontsize = 20)
plt.show()'''
'''# AAL selected features
regions_importances = [subject_feature_importances[:,6],
subject_feature_importances[:,3],
subject_feature_importances[:,0],
subject_feature_importances[:,2],
subject_feature_importances[:,1],
subject_feature_importances[:,4],
subject_feature_importances[:,5],
subject_feature_importances[:,7]]
fig, ax = plt.subplots(1, 1)
ax.boxplot(regions_importances, showmeans=True)
ax.set_title('Region Importance', fontsize = 60)
#plt.xlabel('Region', fontsize = 20)
plt.ylabel('Importance', fontsize = 40)
region_names = ['LITG',
'RRO',
'LMFG',
'ORIFG',
'TRIFG',
'LAG',
'LP',
'RITG']
plt.xticks(np.arange(1,9), region_names, fontsize = 30)
plt.yticks(fontsize = 20)
plt.show()'''