def process(folder, censor_date):
""" Post process data """
print(folder)
util.setup(folder)
image_ids, survival_days = util.get_image_id_and_survival_days(
study_id="GBM_survival_time", censor_date_str=censor_date)
result = util.post_calculations(image_ids)
print('Total: ' + str(len(result['all'])) + ' patients')
util.avg_calculation(result['all'],
'tumor',
None,
True,
folder,
save_sum=True)
util.avg_calculation(result['img'], 'volume', None, True, folder)
util.mortality_rate_calculation(result['all'],
'',
survival_days,
True,
folder,
default_value=-1)
image_ids, survival_days = util.get_image_id_and_survival_days(
study_id="GBM_survival_time",
censor_date_str=censor_date,
resection=True)
result = util.post_calculations(image_ids)
print('Resected: ' + str(len(result['all'])) + ' patients')
util.avg_calculation(result['all'],
'tumor_resected',
None,
True,
folder,
save_sum=True)
util.avg_calculation(result['img'], 'volume_resected', None, True, folder)
util.mortality_rate_calculation(result['all'],
'_resected',
survival_days,
True,
folder,
default_value=-1)
survival_groups = [[0, 182], [183, 730], [731, float('Inf')]]
for group in survival_groups:
image_ids, survival_days = util.get_image_id_and_survival_days(
study_id="GBM_survival_time",
censor_date_str=censor_date,
survival_group=group)
result = util.post_calculations(image_ids)
print('Group ' + str(group) + ': ' + str(len(result['all'])) +
' patients')
label = 'tumor_' + str(group[0]) + '-' + str(group[1])
util.avg_calculation(result['all'],
label,
None,
True,
folder,
save_sum=True)
def process(folder, censor_date, survival_groups, pids_to_exclude=None):
""" Post process data """
print(folder)
util.setup(folder)
image_ids, survival_days = util.get_image_id_and_survival_days(study_id="GBM_survival_time", exclude_pid=pids_to_exclude, censor_date_str=censor_date)
result = util.post_calculations(image_ids)
print('Total: ' + str(len(result['all'])) + ' patients')
util.avg_calculation(result['all'], 'tumor', None, True, folder, save_sum=True)
util.mortality_rate_calculation(result['all'], '_all_year', survival_days, True, folder, default_value=-1, max_value=150, per_year=True)
util.avg_calculation(result['img'], 'volume', None, True, folder)
image_ids, survival_days = util.get_image_id_and_survival_days(study_id="GBM_survival_time", exclude_pid=pids_to_exclude, censor_date_str=censor_date,resection=True)
result = util.post_calculations(image_ids)
print('Resected: ' + str(len(result['all'])) + ' patients')
util.avg_calculation(result['all'], 'tumor_resected', None, True, folder, save_sum=True)
util.avg_calculation(result['img'], 'volume_resected', None, True, folder)
util.mortality_rate_calculation(result['all'], '_resected_year', survival_days, True, folder, default_value=-1, max_value=150, per_year=True)
for group in survival_groups:
image_ids, survival_days = util.get_image_id_and_survival_days(study_id="GBM_survival_time", exclude_pid=pids_to_exclude, censor_date_str=censor_date,survival_group=group)
result = util.post_calculations(image_ids)
print('Group ' + str(group) + ': ' + str(len(result['all'])) + ' patients')
label = 'tumor_' + str(group[0]) + '-' + str(group[1])
util.avg_calculation(result['all'], label, None, True, folder, save_sum=True)
def process3(folder):
""" Post process data """
print(folder)
util.setup(folder)
params = ['Mobility', 'Selfcare', 'Activity', 'Pain', 'Anxiety', 'karnofsky', 'Index_value']
image_ids = find_images()
result = util.post_calculations(image_ids)
print(len(result['all']))
# util.avg_calculation(result['all'], 'all_N=112', None, True, folder, save_sum=True)
# util.avg_calculation(result['img'], 'img_N=112', None, True, folder)
for qol_param in params:
if qol_param == "Delta_qol2":
(image_ids_with_qol, qol) = util.get_qol(image_ids, "Delta_qol")
qol = [-1 if _temp <= -0.15 else 0 if _temp < 0.15 else 1 for _temp in qol]
else:
(image_ids_with_qol, qol) = util.get_qol(image_ids, qol_param)
if qol_param not in ["karnofsky", "Delta_kps"]:
qol = [_temp * 100 for _temp in qol]
default_value = -100
print(qol_param)
print(len(qol))
result = util.post_calculations(image_ids_with_qol)
for label in result:
if label == 'img':
continue
print(label)
util.avg_calculation(result[label], label + '_' + qol_param + '_N=112', qol, True, folder, default_value=default_value)
util.median_calculation(result[label], label + '_' + qol_param + '_N=112', qol, True, folder, default_value=default_value)
def process2(folder):
""" Post process data Delta"""
print(folder)
util.setup(folder)
params = ['Delta_qol', 'Delta_qol2', 'Delta_mobility', 'Delta_selfcare', 'Delta_activity', 'Delta_pain', 'Delta_anixety', 'Delta_kps']
image_ids = find_images()
result = util.post_calculations(image_ids)
print(len(result['all']))
util.avg_calculation(result['all'], 'all_N=112', None, True, folder, save_sum=True)
util.avg_calculation(result['img'], 'img_N=112', None, True, folder)
print("\n\n\n\n\n")
for qol_param in params:
if qol_param == "Delta_qol2":
(image_ids_with_qol, qol) = util.get_qol(image_ids, "Delta_qol")
qol = [-1 if _temp <= -0.15 else 0 if _temp < 0.15 else 1 for _temp in qol]
else:
(image_ids_with_qol, qol) = util.get_qol(image_ids, qol_param)
qol = [_temp * 100 for _temp in qol]
default_value = -300
print(qol_param, len(qol))
result = util.post_calculations(image_ids_with_qol)
for label in result:
if label == 'img':
continue
print(label)
# util.avg_calculation(result[label], label + '_' + qol_param, qol, True, folder, default_value=default_value)
util.median_calculation(result[label], label + '_' + qol_param, qol, True, folder, default_value=default_value)
def process(folder):
""" Post process data distribution and baseline"""
print(folder)
util.setup(folder)
params = ['Mobility', 'Selfcare', 'Activity', 'Pain', 'Anxiety', 'karnofsky', 'Index_value']
image_ids = find_images_163()
result = util.post_calculations(image_ids)
print(len(result['all']))
util.avg_calculation(result['all'], 'all_N=163', None, True, folder, save_sum=True)
util.avg_calculation(result['img'], 'img_N=163', None, True, folder)
image_ids = do_img_registration_GBM.find_images()
result = util.post_calculations(image_ids)
print(len(result['all']))
util.avg_calculation(result['all'], 'all_N=170', None, True, folder, save_sum=True)
util.avg_calculation(result['img'], 'img_N=170', None, True, folder)
for qol_param in params:
(image_ids_with_qol, qol) = util.get_qol(image_ids, qol_param)
if qol_param not in ["karnofsky", "Delta_kps"]:
qol = [_temp * 100 for _temp in qol]
default_value = -100
print(qol_param)
print(len(qol))
result = util.post_calculations(image_ids_with_qol)
for label in result:
if label == 'img':
continue
print(label)
# util.avg_calculation(result[label], label + '_' + qol_param, qol, True, folder, default_value=default_value)
util.median_calculation(result[label], label + '_' + qol_param, qol, True, folder, default_value=default_value)
def validate(folder):
""" Post process data tumor volume"""
print(folder)
util.setup(folder, 'MolekylareMarkorer')
conn = sqlite3.connect(util.DB_PATH, timeout=120)
conn.text_factory = str
cursor = conn.execute(
'''SELECT pid from MolekylareMarkorer ORDER BY pid''')
brain_mask = nib.load(util.TEMPLATE_MASK).get_data()
max_val = 0
max_pid = -1
for pid in cursor:
pid = pid[0]
_id = conn.execute('''SELECT id from Images where pid = ?''',
(pid, )).fetchone()
if not _id:
print("---No data for ", pid)
continue
_id = _id[0]
_filepath = conn.execute(
"SELECT filepath_reg from Labels where image_id = ?",
(_id, )).fetchone()[0]
tumor_data = nib.load(util.DATA_FOLDER + _filepath).get_data()
union_data = (1 - brain_mask) * tumor_data
print(pid, np.sum(union_data[:]))
if np.sum(union_data[:]) > max_val:
max_val = np.sum(union_data[:])
max_pid = pid
print("---------- ", max_pid, max_val)
def process(folder):
""" Post process data """
print(folder)
util.setup(folder)
for grade in [2, 3, 4]:
image_ids, survival_days = util.get_image_id_and_survival_days(
glioma_grades=[grade])
result = util.post_calculations(image_ids)
print(len(result['all']))
util.avg_calculation(result['all'],
'all_',
None,
True,
folder,
save_sum=True)
util.avg_calculation(result['img'], 'img_', None, True, folder)
for label in result:
if label == 'img':
continue
util.avg_calculation(result[label],
'survival_time_grade_' + str(grade),
survival_days,
True,
folder,
default_value=-1)
def test_selectbox(select_name, select_option, url_param):
'''Changes the selectbox value and ensures that it was applied to the search
Parameters:
select_name: The id of the select element (string)
select_option: The text of the option to select (string)
url_param: The URL query string that should be applied (string)
'''
# Make sure this test is from a clean state
setup("Ensuring that the search applies the parameter {}".format(url_param))
# Change the dropdown option
change_dropdown_selection(select_name, select_option)
# Allow time for search options to update in JS
time.sleep(1)
# Press search
search_button = return_when_visible(By.XPATH, "//button[@type='submit']")
search_button.click()
# Slight delay to see search results screen
time.sleep(1)
# Check if the url query string matches the search options
test_passed(assert_url_contains(url_param))
def process_vlsm(folder, n_permutations):
""" Post process vlsm data """
print(folder)
util.setup(folder)
image_ids = find_images()
params = [
'Index_value', 'karnofsky', 'Mobility', 'Selfcare', 'Activity', 'Pain',
'Anxiety'
]
alternative = [
'less', 'less', 'greater', 'greater', 'greater', 'greater', 'greater'
]
stat_func = [
util.brunner_munzel_test, util.mannwhitneyu_test,
util.mannwhitneyu_test, util.mannwhitneyu_test, util.mannwhitneyu_test,
util.mannwhitneyu_test, util.mannwhitneyu_test
]
for (qol_param, stat_func_i, alternative_i) in zip(params, stat_func,
alternative):
(image_ids_with_qol, qol) = util.get_qol(image_ids, qol_param)
result = util.post_calculations(image_ids_with_qol)
for label in result:
print(label)
if label == 'img':
continue
util.vlsm(result[label],
label + '_' + qol_param,
stat_func_i,
qol,
folder,
n_permutations=n_permutations,
alternative=alternative_i)
def process_labels2(folder):
""" Post process data tumor volume"""
print(folder)
util.setup(folder)
conn = sqlite3.connect(util.DB_PATH, timeout=120)
conn.text_factory = str
cursor = conn.execute('''SELECT pid from Patient where study_id = ?''', ("qol_grade3,4", ))
atlas_path = "/home/dahoiv/disk/Dropbox/Jobb/gbm/Atlas/Hammers/Hammers_mith-n30r95-MaxProbMap-full-MNI152-SPM12.nii.gz"
resample = slicer.registration.brainsresample.BRAINSResample(command=util.BRAINSResample_PATH,
inputVolume=atlas_path,
outputVolume=os.path.abspath(folder +
'Hammers_mith-n30r95-MaxProbMap-full'
'-MNI152-SPM12_resample.nii.gz'),
referenceVolume=os.path.abspath(util.TEMPLATE_VOLUME))
resample.run()
img = nib.load(folder + 'Hammers_mith-n30r95-MaxProbMap-full-MNI152-SPM12_resample.nii.gz')
lobes_brain = img.get_data()
label_defs = util.get_label_defs_hammers_mith()
res_lobes_brain = {}
book = Workbook()
sheet = book.active
sheet.cell(row=1, column=1).value = 'PID'
sheet.cell(row=1, column=2).value = 'Lobe'
# sheet.cell(row=1, column=3).value = 'Center of mass'
k = 2
for pid in cursor:
pid = pid[0]
_id = conn.execute('''SELECT id from Images where pid = ?''', (pid, )).fetchone()
if not _id:
print("---No data for ", pid)
continue
_id = _id[0]
_filepath = conn.execute("SELECT filepath_reg from Labels where image_id = ?",
(_id, )).fetchone()[0]
if _filepath is None:
print("No filepath for ", pid)
continue
com, com_idx = util.get_center_of_mass(util.DATA_FOLDER + _filepath)
lobe = label_defs.get(lobes_brain[com_idx[0], com_idx[1], com_idx[2]], 'other')
res_lobes_brain[pid] = lobe
sheet.cell(row=k, column=1).value = pid
sheet.cell(row=k, column=2).value = lobe
sheet.cell(row=k, column=4).value = str(com_idx[0]) + " " + str(com_idx[1]) + " " + str(com_idx[2])
k += 1
book.save("brain_lobes_Hammers_mith_n30r95.xlsx")
print(res_lobes_brain, len(res_lobes_brain))
def export_labels(folder, censor_date, survival_groups, pids_to_exclude):
""" Post process data """
print(folder)
util.setup(folder)
image_ids, survival_days = util.get_image_id_and_survival_days(study_id="GBM_survival_time", exclude_pid=pids_to_exclude, censor_date_str=censor_date)
result = util.post_calculations(image_ids)
print('Total: ' + str(len(result['all'])) + ' patients')
util.export_labels_and_survival_groups(result['all'], 'tumor', survival_days, survival_groups, True, folder )
def process_labels(folder):
""" Post process data tumor volume"""
print(folder)
util.setup(folder)
conn = sqlite3.connect(util.DB_PATH, timeout=120)
conn.text_factory = str
cursor = conn.execute('''SELECT pid from Patient where study_id = ?''', ("qol_grade3,4", ))
img = nib.load("/home/dahoiv/disk/data/MolekylareMarkorer/lobes_brain.nii")
lobes_brain = img.get_data()
label_defs = util.get_bigger_label_defs()
label_defs_r_l = util.get_right_left_label_defs()
res_lobes_brain = {}
book = Workbook()
sheet = book.active
sheet.cell(row=1, column=1).value = 'PID'
sheet.cell(row=1, column=2).value = 'Lobe'
sheet.cell(row=1, column=3).value = 'Right/Left'
# sheet.cell(row=1, column=3).value = 'Center of mass'
k = 2
for pid in cursor:
pid = pid[0]
_id = conn.execute('''SELECT id from Images where pid = ?''', (pid, )).fetchone()
if not _id:
print("---No data for ", pid)
continue
_id = _id[0]
_filepath = conn.execute("SELECT filepath_reg from Labels where image_id = ?",
(_id, )).fetchone()[0]
if _filepath is None:
print("No filepath for ", pid)
continue
com, com_idx = util.get_center_of_mass(util.DATA_FOLDER + _filepath)
lobe = label_defs.get(lobes_brain[com_idx[0], com_idx[1], com_idx[2]], 'other')
res_lobes_brain[pid] = lobe
sheet.cell(row=k, column=1).value = pid
sheet.cell(row=k, column=2).value = lobe
sheet.cell(row=k, column=3).value = 'left' if com_idx[0] < 99 else 'right'
rl = label_defs_r_l.get(lobes_brain[com_idx[0], com_idx[1], com_idx[2]], 'other')
if rl != 'unknown' and rl != sheet.cell(row=k, column=3).value:
print("\n\n\n", pid)
# sheet.cell(row=k, column=3).value = str(com[0]) + " " + str(com[1]) + " " + str(com[2])
# sheet.cell(row=k, column=4).value = str(com_idx[0]) + " " + str(com_idx[1]) + " " + str(com_idx[2])
k += 1
book.save("brain_lobes.xlsx")
print(res_lobes_brain, len(res_lobes_brain))
def process4(folder):
""" Post process data tumor volume"""
print(folder)
util.setup(folder)
default_value = 0
label = 'all'
for image_ids in [do_img_registration_GBM.find_images(), find_images(), find_images_163()]:
result = util.post_calculations(image_ids)
(image_ids_with_qol, qol) = util.get_tumor_volume(image_ids)
num = len(result['all'])
print(num)
util.median_calculation(result[label], 'tumor_volume_N=' + str(num), qol, True, folder, default_value=default_value)
def process(folder, glioma_grades):
""" Post process data """
print(folder)
util.setup(folder)
params = [
'Index_value', 'Global_index', 'Mobility', 'Selfcare', 'Activity',
'Pain', 'Anxiety', 'karnofsky'
]
(image_ids, qol) = util.get_image_id_and_qol(None,
exclude_pid,
glioma_grades=glioma_grades)
print(len(image_ids))
result = util.post_calculations(image_ids)
print(len(result['all']))
util.avg_calculation(result['all'],
'all',
None,
True,
folder,
save_sum=True)
util.avg_calculation(result['img'], 'img', None, True, folder)
# (image_ids, qol) = util.get_image_id_and_qol('Index_value', exclude_pid, glioma_grades=glioma_grades)
# result = util.post_calculations(image_ids)
# util.calculate_t_test(result['all'], 1)
for qol_param in params:
(image_ids, qol) = util.get_image_id_and_qol(qol_param, exclude_pid)
if not qol_param == "karnofsky":
qol = [_temp * 100 for _temp in qol]
if not qol_param == "Index_value":
default_value = -100
else:
default_value = 0
print(image_ids)
result = util.post_calculations(image_ids)
for label in result:
print(label)
if label == 'img':
continue
util.avg_calculation(result[label],
label + '_' + qol_param,
qol,
True,
folder,
default_value=default_value)
util.std_calculation(result[label], label + '_' + qol_param, qol,
True, folder)
def process(folder, exclude):
""" Post process data """
print(folder)
util.setup(folder, "meningiomer")
image_ids = find_images(exclude=exclude)
result = util.post_calculations(image_ids)
print(len(result['all']))
util.avg_calculation(result['all'],
'all',
None,
True,
folder,
save_sum=True,
save_pngs=True)
util.avg_calculation(result['img'], 'img', None, True, folder)
def find_bad_registration():
util.setup(folder, "meningiomer")
image_ids = find_images()
template_mask = nib.load(util.TEMPLATE_MASK).get_data()
conn = sqlite3.connect(util.DB_PATH, timeout=120)
k = 0
exclude = []
for _id in image_ids:
_filepath = conn.execute(
"SELECT filepath_reg from Labels where image_id = ?",
(_id, )).fetchone()[0]
com, com_idx = util.get_center_of_mass(util.DATA_FOLDER + _filepath)
if template_mask[com_idx[0], com_idx[1], com_idx[2]] == 0:
print(_filepath)
k += 1
exclude.append(_id)
print(k)
return exclude
def main():
config, device = util.setup()
logger.setLevel(getattr(logging, config['log_level'].upper()))
gnn = import_module('gnn' if config['mlp_arch'] else 'gnn_old')
if config['method'] == 'reinforce':
model = gnn.ReinforcePolicy
elif config['method'] == 'reinforce_multi':
model = gnn.ReinforcePolicy
elif config['method'] == 'pg':
model = gnn.PGPolicy
elif config['method'] == 'a2c':
model = gnn.A2CPolicy
if config['model_path']:
logger.info('Loading model parameters from {}'.format(
config['model_path']))
policy = torch.load(config['model_path']).to(device)
if config['load_with_noise']:
with torch.no_grad():
# for p in policy.parameters():
# p.add_(torch.randn(p.size()) * 0.02)
for p in policy.policy_readout.parameters():
p.add_(torch.randn(p.size()) * 0.1)
else:
if config['mlp_arch']:
policy = model(3, config['gnn_hidden_size'],
config['readout_hidden_size'], config['mlp_arch'],
config['gnn_iter'], config['gnn_async']).to(device)
else:
policy = model(3, config['gnn_hidden_size'],
config['readout_hidden_size']).to(device)
optimizer = getattr(optim, config['optimizer'])(policy.parameters(),
lr=config['lr'])
scheduler = optim.lr_scheduler.MultiStepLR(
optimizer,
milestones=config['lr_milestones'],
gamma=config['lr_decay'])
ls = LocalSearch(policy, device, config)
train_sets, eval_set = load_data(config['data_path'], config['train_sets'],
config['eval_set'],
config['data_shuffle'])
for i in range(1, config['cycles'] + 1):
logger.info(f'Cycle: {i}')
for train_set in train_sets:
logger.info('Train set: {}'.format(train_set['name']))
train(ls, optimizer, scheduler, (train_set, eval_set), config)
def process_vlsm(folder, glioma_grades):
""" Post process vlsm data """
print(folder)
util.setup(folder)
params = [
'Index_value', 'Mobility', 'Selfcare', 'Activity', 'Pain', 'Anxiety',
'karnofsky'
]
stat_func = [util.brunner_munzel_test, [util.mannwhitneyu_test] * 6]
for (qol_param, stat_func_i) in zip(params, stat_func):
(image_ids, qol) = util.get_image_id_and_qol(qol_param, exclude_pid)
print(image_ids)
result = util.post_calculations(image_ids)
for label in result:
print(label)
if label == 'img':
continue
util.vlsm(result[label],
label + '_' + qol_param,
stat_func_i,
qol,
folder,
n_permutations=100)
def main():
config, device = util.setup()
logger.setLevel(getattr(logging, config['log_level'].upper()))
train_data, dev_data = load_data(config['data_path'], config['no_dev'])
if dev_data:
dev_batches = list(
create_batches(dev_data, config['eval_batch_size'], device))
gnn = GraphClassifier(2, config['gnn_hidden_size'],
config['readout_hidden_size']).to(device)
optimizer = optim.Adam(gnn.parameters(), lr=config['lr'])
criterion = nn.BCELoss()
batch_count = 0
best_dev_acc = 0
for epoch in range(1, config['epochs'] + 1):
train_loss = train_correct = sample_count = 0
for data in create_batches(train_data, config['batch_size'], device):
gnn.train()
z = gnn(data)
loss = criterion(z, data.y)
optimizer.zero_grad()
loss.backward()
optimizer.step()
batch_count += 1
sample_count += data.y.shape[0]
train_loss += loss.item() * data.y.shape[0]
train_correct += pred_correct(z, data.y)
if batch_count % config['report_interval'] == 0:
log(epoch, batch_count, train_loss / sample_count,
train_correct / sample_count)
train_loss = train_correct = sample_count = 0
if not config[
'no_dev'] and batch_count % config['eval_interval'] == 0:
gnn.eval()
dev_loss, dev_acc = eval(gnn, criterion, dev_batches)
logger.info('(Dev) Loss: {:.4f}, Acc: {:.4f}'.format(
dev_loss, dev_acc))
if dev_acc > best_dev_acc:
best_dev_acc = dev_acc
torch.save(gnn, join(config['dir'], 'model.pth'))
torch.save(gnn, join(config['dir'], 'model_final.pth'))
def evaluate(path_to_config, path_to_model):
"""
Evaluate the network on test data using the model stored in `path_to_model`.
Args:
path_to_config (str): Path to configuration file
path_to_model (str): Path to the saved model
Returns:
object: the paths of the session
"""
config, paths, session_id = setup(path_to_config, 1)
assert isinstance(config, ExperimentConfig)
logger = logging.getLogger("%s.main" % config.name)
logger.info("Evaluating network on test data")
network = Network(config, paths, session_id, 0)
network.build()
network.evaluate(DATA_TYPE_TEST, model_path=path_to_model)
return paths
# -*- coding: utf-8 -*-
"""
Created on Mon Sep 19 15:32:12 2016
@author: dahoiv
"""
import os
import sqlite3
import util
if __name__ == "__main__":
os.nice(19)
util.setup("temp/", "")
all_filepaths = ['brainSegmentation.db']
conn = sqlite3.connect(util.DB_PATH)
conn.text_factory = str
cursor = conn.execute('''SELECT filepath, transform, filepath_reg from Images''')
for (filepath, transform, filepath_reg) in cursor:
all_filepaths.extend(util.ensure_list(filepath))
all_filepaths.extend(util.ensure_list(filepath_reg))
if transform is None:
continue
for _transform in transform.split(","):
all_filepaths.append(_transform.strip())
cursor = conn.execute('''SELECT filepath, filepath_reg from Labels''')
import datetime
import sqlite3
import glob
import util
from img_data import img_data
from image_registration import move_vol
# pylint: disable= invalid-name
if __name__ == "__main__": # if 'unity' in hostname or 'compute' in hostname:
# new_segmentations_folder = '/Volumes/Neuro/Segmentations/oppdaterte_filer/'
new_segmentations_folder = '/media/leb/data/oppdaterte_filer/'
temp_folder = "ADD_SEGMENTATIONS_" + "{:%Y%m%d_%H%M}".format(
datetime.datetime.now()) + "/"
util.setup(temp_folder, 'glioma')
conn = sqlite3.connect(util.DB_PATH)
cursor = conn.cursor()
modified_patients = [
subfolder_name
for subfolder_name in os.listdir(new_segmentations_folder) if
os.path.isdir(os.path.join(new_segmentations_folder, subfolder_name))
]
for pid in modified_patients:
#print('PATIENT ' + pid)
cursor.execute("SELECT id FROM Images WHERE pid = ?", (int(pid), ))
image_id = cursor.fetchone()[0]
print(image_id)
from laplacian_loss import laplacian_loss
opt = get_opt()
print_opt()
# hyper parameters
n_epochs = opt.epochs
batch_size = opt.batch_size
z_dim = opt.z_dim
x_dim = opt.x_dim
sample_size = opt.sample_size
lr = opt.lr
log_step = 100
sample_step = 1000
sample_path, model_path = setup()
image_path = os.path.join(os.getcwd(), 'CelebA', '128_crop')
train_loader = get_loader(image_path=image_path,
image_size=opt.x_dim,
batch_size=opt.batch_size,
num_workers=2)
image_path = os.path.join(os.getcwd(), 'CelebA', '128_crop')
img_size = opt.x_dim
# choose generator type laplacian generator or common dcgan generator
generator = Generator2()
util.generate_image(img.pre_processed_filepath, image_registration.TEMPLATE_VOLUME)
def test_be(moving_image_ids, reg):
start_time = datetime.datetime.now()
for moving_image_id in moving_datasets_ids:
_test_be(moving_image_id, reg)
bet_time = datetime.datetime.now() - start_time
print("\n\n\n\n -- Run time BET: ")
print(bet_time/len(moving_datasets_ids))
# pylint: disable= invalid-name
if __name__ == "__main__":
os.nice(19)
util.setup("GBM_test/", "GBM")
moving_datasets_ids = []
reg = ants.Registration()
# reg.inputs.args = "--verbose 1"
reg.inputs.collapse_output_transforms = True
reg.inputs.moving_image = image_registration.TEMPLATE_VOLUME
reg.inputs.num_threads = 8
reg.inputs.initial_moving_transform_com = True
reg.inputs.transforms = ['Rigid', 'Affine']
reg.inputs.metric = ['MI', 'MI']
reg.inputs.radius_or_number_of_bins = [32, 32]
reg.inputs.metric_weight = [1, 1]
reg.inputs.convergence_window_size = [5, 5]
def process(folder, pids_to_exclude=()):
""" Post process data """
print(folder)
util.setup(folder, 'MolekylareMarkorer')
conn = sqlite3.connect(util.DB_PATH, timeout=120)
conn.text_factory = str
cursor = conn.execute(
'''SELECT pid from MolekylareMarkorer ORDER BY pid''')
image_ids = []
image_ids_1 = []
image_ids_2 = []
image_ids_3 = []
tag_data_1 = []
tag_data_2 = []
tag_data_3 = []
img = nib.load("/media/leb/data/Atlas/lobes_brain.nii")
lobes_brain = img.get_data()
label_defs = util.get_label_defs()
res_right_left_brain = {}
res_lobes_brain = {}
patients = '\nPID MM\n----------------\n'
for pid in cursor:
pid = pid[0]
if pid in pids_to_exclude:
continue
_id = conn.execute('''SELECT id from Images where pid = ?''',
(pid, )).fetchone()
if not _id:
print("---No data for ", pid)
continue
_id = _id[0]
_mm = conn.execute(
"SELECT Subgroup from MolekylareMarkorer where pid = ?",
(pid, )).fetchone()[0]
if _mm is None:
print("No mm data for ", pid)
patients += str(pid) + ': ?\n'
continue
_desc = conn.execute(
"SELECT comments from MolekylareMarkorer where pid = ?",
(pid, )).fetchone()[0]
if _desc is None:
_desc = ""
_filepath = conn.execute(
"SELECT filepath_reg from Labels where image_id = ?",
(_id, )).fetchone()[0]
if _filepath is None:
print("No filepath for ", pid)
continue
com, com_idx = util.get_center_of_mass(util.DATA_FOLDER + _filepath)
val = {}
val['Name'] = str(pid) + "_" + str(_mm)
val['PositionGlobal'] = str(com[0]) + "," + str(com[1]) + "," + str(
com[2])
val['desc'] = str(_desc)
lobe = label_defs[lobes_brain[com_idx[0], com_idx[1], com_idx[2]]]
right_left = 'left' if com_idx[0] < 99 else 'right'
res_lobes_brain[lobe] = res_lobes_brain.get(lobe, [0, 0, 0])
res_right_left_brain[right_left] = res_right_left_brain.get(
right_left, [0, 0, 0])
print(right_left, lobe)
if _mm == 1:
res_lobes_brain[lobe][0] += 1
res_right_left_brain[right_left][0] += 1
image_ids_1.extend([_id])
elif _mm == 2:
res_lobes_brain[lobe][1] += 1
res_right_left_brain[right_left][1] += 1
image_ids_2.extend([_id])
elif _mm == 3:
res_lobes_brain[lobe][2] += 1
res_right_left_brain[right_left][2] += 1
image_ids_3.extend([_id])
image_ids.extend([_id])
print(pid, _mm)
patients += str(pid) + ': ' + str(_mm) + '\n'
if _mm == 1:
tag_data_1.append(val)
elif _mm == 2:
tag_data_2.append(val)
elif _mm == 3:
tag_data_3.append(val)
print(format_dict(res_lobes_brain))
lobes_brain_file = open(folder + "lobes_brain.txt", 'w')
lobes_brain_file.write(format_dict(res_lobes_brain))
lobes_brain_file.close()
lobes_brain_file = open(folder + "lobes_brain.txt", 'a')
lobes_brain_file.write(format_dict(res_right_left_brain))
lobes_brain_file.write(patients)
lobes_brain_file.close()
print(len(image_ids))
result = util.post_calculations(image_ids_1)
util.avg_calculation(result['all'],
'mm_1',
None,
True,
folder,
save_sum=True)
result = util.post_calculations(image_ids_2)
util.avg_calculation(result['all'],
'mm_2',
None,
True,
folder,
save_sum=True)
result = util.post_calculations(image_ids_3)
util.avg_calculation(result['all'],
'mm_3',
None,
True,
folder,
save_sum=True)
result = util.post_calculations(image_ids)
util.avg_calculation(result['all'],
'mm_1_2_3',
None,
True,
folder,
save_sum=True)
return
tag_data = {
"tag_data_1": tag_data_1,
"tag_data_2": tag_data_2,
"tag_data_3": tag_data_3
}
pickle.dump(tag_data, open("tag_data.pickle", "wb"))
cursor.close()
conn.close()
util.write_fcsv("mm_1", folder, tag_data_1, '1 0 0', 13)
util.write_fcsv("mm_2", folder, tag_data_2, '0 1 0', 5)
util.write_fcsv("mm_3", folder, tag_data_3, '0 0 1', 6)
result = util.post_calculations(image_ids)
util.avg_calculation(result['all'],
'all',
None,
True,
folder,
save_sum=True)
util.avg_calculation(result['img'], 'img', None, True, folder)
def process_labels(folder, pids_to_exclude=()):
""" Post process data tumor volume"""
print(folder)
util.setup(folder, 'MolekylareMarkorer')
conn = sqlite3.connect(util.DB_PATH, timeout=120)
conn.text_factory = str
cursor = conn.execute(
'''SELECT pid from MolekylareMarkorer ORDER BY pid''')
atlas_path = util.ATLAS_FOLDER_PATH + 'Hammers/Hammers_mith-n30r95-MaxProbMap-full-MNI152-SPM12.nii.gz'
atlas_resampled_path = folder + 'Hammers_mith-n30r95-MaxProbMap-full-MNI152-SPM12_resample.nii.gz'
resample = brainsresample.BRAINSResample(
command=util.BRAINSResample_PATH,
inputVolume=atlas_path,
outputVolume=os.path.abspath(atlas_resampled_path),
referenceVolume=os.path.abspath(util.TEMPLATE_VOLUME))
resample.run()
img = nib.load(atlas_resampled_path)
lobes_brain = img.get_data()
label_defs = util.get_label_defs_hammers_mith()
res_lobes_brain = {}
coordinates_svz = util.get_label_coordinates(util.ATLAS_FOLDER_PATH +
'SubventricularZone2.nii.gz')
surface_dg = util.get_surface(util.ATLAS_FOLDER_PATH +
'DentateGyrus.nii.gz')
book = Workbook()
sheet = book.active
sheet.cell(row=1, column=1).value = 'PID'
sheet.cell(row=1, column=2).value = 'MM'
sheet.cell(row=1, column=3).value = 'Lobe, center of tumor'
sheet.cell(row=1,
column=4).value = 'Distance from SVZ to center of tumor (mm)'
sheet.cell(row=1,
column=5).value = 'Distance from SVZ to border of tumor (mm)'
sheet.cell(row=1,
column=6).value = 'Distance from DG to center of tumor (mm)'
sheet.cell(row=1,
column=7).value = 'Distance from DG to border of tumor (mm)'
sheet.cell(row=1, column=8).value = 'Tumor volume (mm^3)'
i = 8
label_defs_to_column = {}
for key in label_defs:
i += 1
sheet.cell(row=1, column=i).value = label_defs[key]
label_defs_to_column[key] = i
k = 2
for pid in cursor:
pid = pid[0]
if pid in pids_to_exclude:
continue
_id = conn.execute('''SELECT id from Images where pid = ?''',
(pid, )).fetchone()
if not _id:
print("---No data for ", pid)
continue
_id = _id[0]
_filepath = conn.execute(
"SELECT filepath_reg from Labels where image_id = ?",
(_id, )).fetchone()[0]
if _filepath is None:
print("No filepath for ", pid)
continue
com, com_idx = util.get_center_of_mass(util.DATA_FOLDER + _filepath)
surface = util.get_surface(util.DATA_FOLDER + _filepath)
print(pid, com_idx)
dist_from_svz_to_com = distance.cdist(coordinates_svz, [com],
'euclidean').min()
dist_from_svz_to_border = distance.cdist(coordinates_svz,
surface['point_cloud'],
'euclidean').min()
dist_from_dg_to_com = util.get_min_distance(surface_dg, [com])
dist_from_dg_to_border = util.get_min_distance(surface_dg,
surface['point_cloud'])
lobe = label_defs.get(lobes_brain[com_idx[0], com_idx[1], com_idx[2]],
'other')
res_lobes_brain[pid] = lobe
img = nib.load(util.DATA_FOLDER + _filepath)
tumor_data = img.get_data()
voxel_size = img.header.get_zooms()
voxel_volume = np.prod(voxel_size[0:3])
n_voxels = (tumor_data > 0).sum()
tumor_volume = n_voxels * voxel_volume
union_data = lobes_brain * tumor_data
union_data = union_data.flatten()
lobe_overlap = ''
for column in range(1, 1 + max(label_defs_to_column.values())):
sheet.cell(row=k, column=column).value = 0
for _lobe in np.unique(union_data):
column = label_defs_to_column.get(_lobe)
if column is None:
continue
sheet.cell(row=k, column=column).value = 1
lobe_overlap += label_defs.get(_lobe, '') + ', '
_mm = conn.execute(
"SELECT Subgroup from MolekylareMarkorer where pid = ?",
(pid, )).fetchone()[0]
sheet.cell(row=k, column=1).value = pid
sheet.cell(row=k, column=2).value = str(_mm)
sheet.cell(row=k, column=3).value = lobe
sheet.cell(row=k, column=4).value = round(dist_from_svz_to_com, 2)
sheet.cell(row=k, column=5).value = round(dist_from_svz_to_border, 2)
sheet.cell(row=k, column=6).value = round(dist_from_dg_to_com, 2)
sheet.cell(row=k, column=7).value = round(dist_from_dg_to_border, 2)
sheet.cell(row=k, column=8).value = round(tumor_volume, 1)
k += 1
book.save(folder + "brain_lobes_Hammers_mith_n30r95.xlsx")
print(res_lobes_brain, len(res_lobes_brain))
def setUp(self):
super(WechatTest, self).setUp()
u.setup(d)
def setUp(self):
super(VideoTest, self).setUp()
u.setup(d)
def setUp(self):
super(AngrybirdTest, self).setUp()
u.setup(d)
def setUp(self):
super(BaiduTest, self).setUp()
u.setup(d)
def setUp(self):
super(ContactTest, self).setUp()
u.setup(d)
if __name__ == "__main__":
# if False:
# import do_img_registration_LGG_POST as do_img_registration
# util.setup("LGG_POST_RES/", "LGG")
# elif False:
# import do_img_registration_LGG_PRE as do_img_registration
# util.setup("LGG_PRE_RES/", "LGG")
# elif False:
# import do_img_registration_GBM as do_img_registration
# util.setup("GBM_RES2/", "GBM")
params = ['Index_value', 'Global_index', 'Mobility', 'Selfcare', 'Activity', 'Pain', 'Anxiety']
util.mkdir_p("LGG_GBM_RES")
FOLDER = "LGG_GBM_RES/" # "LGG_GBM_RES/GBM"
util.setup(FOLDER)
(image_ids, qol) = util.get_image_id_and_qol('Index_value')
print(image_ids, len(image_ids))
result = util.post_calculations(image_ids)
util.calculate_t_test(result['all'], 0.85)
for qol_param in params:
(image_ids, qol) = util.get_image_id_and_qol(qol_param)
print(image_ids)
result = util.post_calculations(image_ids)
for label in result:
print(label)
if label == 'img':
continue
util.avg_calculation(result[label], label + '_' + qol_param, qol, True, FOLDER)
def find_images():
""" Find images for registration """
conn = sqlite3.connect(util.DB_PATH)
conn.text_factory = str
cursor = conn.execute('''SELECT pid from Patient''')
ids = []
for row in cursor:
cursor2 = conn.execute('''SELECT id from Images where pid = ? AND diag_pre_post = ?''',
(row[0], "pre"))
for _id in cursor2:
ids.append(_id[0])
cursor2.close()
cursor.close()
conn.close()
return ids
# pylint: disable= invalid-name
if __name__ == "__main__":
os.nice(19)
util.setup("LGG_PRE/")
pre_images = find_images()
data_transforms = image_registration.get_transforms(pre_images, image_registration.SYN)
# image_registration.save_transform_to_database(data_transforms)
'limit': 100,
'sr_detail': True
}
usernames = util.get_usernames()
for user in usernames:
print user
continueWithUser = True
while continueWithUser:
r = reddit.api_get('user/' + user.strip() + '/submitted', params)
if r != None:
for thing in r['data']['children']:
try:
user_comments.insert_one(thing)
except pymongo.errors.DuplicateKeyError:
continue
if r['data']['after'] is None:
continueWithUser = False
break
else:
params['after'] = r['data']['after']
else:
continueWithUser = False
if __name__ == '__main__':
util.setup()
get_reddit_user_content()
import util as util
import math
import time
#start_time = time.time()
# import/parse location and connection file contents
util.setup()
#Receive the input from the user for the starting and ending cities
temp = input("What is the starting city? ")
while(util.verifyInput(temp)!=True):
print("\nPlease input a valid city")
temp = input("What is the starting city? ")
start = temp
temp = input("\nWhat is the ending city? ")
while(util.verifyInput(temp)!=True):
print("\nPlease input a valid city")
temp = input("What is the ending city? ")
end = temp
skip = []
temp = input("\nAre there any cities that you would not want to visit? Please enter \"Help\" if you would like to see a list of cities. Or refer to the \"locations.txt\" file.\n")
while(temp == "Help"):
for i in util.locations:
if(i != "END"):
print(i)
temp = input("What city would you like to skip?\n")
while(temp!= ""):
if(temp in util.locations):
def setUp(self):
super(GalleryTest, self).setUp()
u.setup(d)
util.debug_print('removing from excludes file the line: ' + remove_line)
update_excludes = util.updateFile('excludes', remove_line, addLine = False)
# remove that name from slaves file
util.debug_print('removing from slaves file the line: ' + str(remove_line))
update_slaves = util.updateFile('slaves', remove_line, addLine = False)
# get vmid from slaveName
vmid = util.get_vm_id_by_name(slaveName)
# NOW deestroy vm
util.debug_print('Now we will be trying to destroy the machine with ID: ' + str(vmid))
result = api.destroyVirtualMachine({'id': vmid})
util.debug_print('waiting for the destroyed machine to be finished being destroyed')
waitResult = util.waitForAsync(result.get('jobid'))
# since we destroyed the vm, we can remove from master's /etc/hosts file
hosts = util.get_file_content(config.DEFAULT_DESTINATION_HOSTS_FILENAME)
checker = re.compile('.*' + slaveName + '\n')
to_be_removed_hosts_line = [line for line in hosts if checker.match(line) is not None]
util.debug_print('remove line:' + str(to_be_removed_hosts_line) + ' from /etc/hosts file')
util.updateFile('hosts', to_be_removed_hosts_line[0], addLine = False)
util.debug_print('Done destroying VM.')
return True
# basic global stuff
api, pp = util.setup()
_img = img_data(img_id, db_path, util.TEMP_FOLDER_PATH)
_img.load_db_transforms()
print(_img.transform)
if _img.transform is None:
continue
_img.processed_filepath = image_registration.move_vol(_img.img_filepath, _img.get_transforms())
_img.image_id = ny_img_id
data_transforms.append(_img)
image_registration.save_transform_to_database(data_transforms)
if __name__ == "__main__":
os.nice(19)
util.setup("temp_convert/", "LGG")
util.mkdir_p(util.TEMP_FOLDER_PATH)
util.DATA_FOLDER = "/mnt/dokumneter/data/database/"
if True:
db_path = "/home/dahoiv/disk/data/database3/LGG/"
util.DATA_FOLDER = util.DATA_FOLDER + "LGG" + "/"
util.DB_PATH = util.DATA_FOLDER + "brainSegmentation.db"
convert_table_inv = ConvertDataToDB.get_convert_table('/home/dahoiv/disk/data/Segmentations/NY_PID_LGG segmentert.xlsx')
convert_table = {v: k for k, v in convert_table_inv.items()}
print(convert_table)
print(util.DB_PATH)
conn = sqlite3.connect(util.DB_PATH)
conn.text_factory = str
conn.close()
import datetime
start_time = datetime.datetime.now()
pre_img = img_data(pre_image_id, util.DATA_FOLDER, util.TEMP_FOLDER_PATH)
post_img = img_data(moving_image_id, util.DATA_FOLDER, util.TEMP_FOLDER_PATH)
pre_img = image_registration.pre_process(pre_img, False)
post_img = image_registration.pre_process(post_img, False)
img = image_registration.registration(post_img, pre_img.pre_processed_filepath,
image_registration.RIGID)
print("\n\n\n\n -- Total run time: ")
print(datetime.datetime.now() - start_time)
img.fixed_image = pre_image_id
return img
# pylint: disable= invalid-name
if __name__ == "__main__":
os.nice(19)
util.setup("LGG_POST/")
post_images = find_images()
data_transforms = image_registration.get_transforms(post_images,
process_dataset_func=process_dataset,
save_to_db=True)
# image_registration.save_transform_to_database(data_transforms)
def setUp(self):
super(MusicTest, self).setUp()
u.setup(d)
def setUp(self):
super(MessageTest, self).setUp()
u.setup(d)
cursor = conn.execute('''SELECT pid from Patient''')
ids = []
for row in cursor:
pid = row[0]
if pid not in pids_with_qol:
continue
cursor2 = conn.execute('''SELECT id from Images where pid = ? AND diag_pre_post = ?''',
(pid, "pre"))
for _id in cursor2:
ids.append(_id[0])
cursor2.close()
cursor.close()
conn.close()
return ids
# pylint: disable= invalid-name
if __name__ == "__main__":
os.nice(17)
util.setup("GBM_LGG_TEMP_" + "{:%m_%d_%Y}_BE2".format(datetime.datetime.now()) + "/")
moving_datasets_ids = find_images()
print(moving_datasets_ids, len(moving_datasets_ids))
data_transforms = image_registration.get_transforms(moving_datasets_ids,
image_registration.RIGID,
save_to_db=True)
# image_registration.save_transform_to_database(data_transforms)
help=('A boolean indicating whether running this sample '
'will make changes. No changes will occur if this '
'is set True.'))
def main(doubleclick_bid_manager, body):
# Construct the request.
request = doubleclick_bid_manager.lineitems().uploadlineitems(body=body)
response = request.execute()
if 'uploadStatus' in response and 'errors' in response['uploadStatus']:
for error in response['uploadStatus']['errors']:
print(error)
else:
print('Upload Successful.')
if __name__ == '__main__':
args = util.get_arguments(sys.argv, __doc__, parents=[parser])
file_path = args.file_path
if not os.path.isabs(file_path):
file_path = os.path.expanduser(file_path)
with open(file_path, 'rb') as handle:
line_items = handle.read().decode('utf-8')
BODY = {'dryRun': args.dry_run, 'lineItems': line_items}
main(util.setup(args), BODY)
from sys import argv
from util import setup, init, select, test
def show_help():
print("""Usage:
trainer.py [cmd]
cmd: setup or test""")
if __name__ == '__main__':
init()
if len(argv) >= 2:
cmd = argv[1]
if cmd == 'setup':
setup()
elif cmd == 'select':
select(int(argv[2]))
elif cmd == 'test':
test()
# elif cmd == 'add':
# repo = argv[2]
else:
show_help()
else:
show_help()
def find_images():
""" Find images for registration """
conn = sqlite3.connect(util.DB_PATH)
conn.text_factory = str
cursor = conn.execute('''SELECT pid from Patient''')
ids = []
for row in cursor:
cursor2 = conn.execute('''SELECT id from Images where pid = ?''', (row[0], ))
for _id in cursor2:
ids.append(_id[0])
cursor2.close()
cursor.close()
conn.close()
print(ids)
return ids
# pylint: disable= invalid-name
if __name__ == "__main__":
os.nice(19)
util.setup("GBM/")
moving_datasets_ids = find_images()
data_transforms = image_registration.get_transforms(moving_datasets_ids, image_registration.SYN)
# image_registration.save_transform_to_database(data_transforms)
update_slaves = util.updateFile('slaves', remove_line, addLine=False)
# get vmid from slaveName
vmid = util.get_vm_id_by_name(slaveName)
# NOW deestroy vm
util.debug_print('Now we will be trying to destroy the machine with ID: ' +
str(vmid))
result = api.destroyVirtualMachine({'id': vmid})
util.debug_print(
'waiting for the destroyed machine to be finished being destroyed')
waitResult = util.waitForAsync(result.get('jobid'))
# since we destroyed the vm, we can remove from master's /etc/hosts file
hosts = util.get_file_content(config.DEFAULT_DESTINATION_HOSTS_FILENAME)
checker = re.compile('.*' + slaveName + '\n')
to_be_removed_hosts_line = [
line for line in hosts if checker.match(line) is not None
]
util.debug_print('remove line:' + str(to_be_removed_hosts_line) +
' from /etc/hosts file')
util.updateFile('hosts', to_be_removed_hosts_line[0], addLine=False)
util.debug_print('Done destroying VM.')
return True
# basic global stuff
api, pp = util.setup()
def setUp(self):
super(PhoneTest, self).setUp()
u.setup(d)
def setUp(self):
super(CameraTest, self).setUp()
u.setup(d)
def train(path_to_config, verbose=False):
"""
Train the network for multiple runs (can be specified in the configuration file with the "num_runs" option).
After finishing the training for a run, the network is evaluated on the development data and the result is stored.
After finishing all runs, the results are averaged.
Args:
path_to_config (str): Path to the configuration file.
verbose (bool): Whether or not to display additional logging information
"""
config, paths, session_id = setup(path_to_config)
assert isinstance(config, ExperimentConfig)
logger = logging.getLogger("%s.train" % config.name)
results = []
# Average results over `config.num_runs` runs
for i in xrange(config.num_runs):
logger.info("*" * 80)
logger.info("* %d. run for experiment %s", (i + 1), config.name)
logger.info("*" * 80)
network = Network(config, paths, session_id, i)
network.build()
num_actual_epochs, stopped_early = network.train(
verbose=verbose, log_results_on_dev=True)
run_results = network.evaluate(data_type=DATA_TYPE_DEV)
results.append(run_results)
for task_name, result_list in run_results.items():
assert isinstance(task_name, str)
assert isinstance(result_list, ResultList)
# Write a CSV file per task because each task may have different evaluation metrics
csv_out_path = os.path.join(
paths["session_out"],
"session_results.task_%s.csv" % task_name)
network.log_result_list_csv(
task_name, result_list, csv_out_path, {
"# planned epochs": config.epochs,
"# actual epochs": num_actual_epochs,
"stopped early?": stopped_early,
"run": i + 1
})
logger.info("*" * 80)
logger.info("")
# Reset tensorflow variables
tf.reset_default_graph()
logger.info("")
logger.info("Results after %d runs:", config.num_runs)
timestamp = time.strftime("%Y-%m-%d %H:%M")
for task in config.tasks:
task_name = task.name
task_results = [result[task_name] for result in results]
# Write a CSV file per task because each task may have different evaluation metrics
csv_file_path = os.path.join(paths["experiment_out"],
"results.task_%s.csv" % task_name)
logger.info(" - Task %s", task_name)
headers = ["timestamp", "session_id", "num_runs", "task_name"]
values = [timestamp, session_id, config.num_runs, task_name]
for metric in set(config.eval_metrics + task.eval_metrics):
metric_values_sum = 0
for result in task_results:
metric_values_sum += result.compute_metric_by_name(metric)
logger.info(" - Average %s at task %s is %.3f", metric.title(),
task_name, metric_values_sum / float(config.num_runs))
headers += ["AVG:%s" % metric.title()]
values += [metric_values_sum / float(config.num_runs)]
append_to_csv(csv_file_path, headers=headers, values=values)
import os
import sys
import image_registration
import util
# pylint: disable= invalid-name
if __name__ == "__main__": # if 'unity' in hostname or 'compute' in hostname:
HOSTNAME = os.uname()[1]
if 'unity' in HOSTNAME or 'compute' in HOSTNAME:
path = "/work/danieli/GBM_survival/"
else:
os.nice(19)
path = "GBM_" + "{:%m_%d_%Y}".format(datetime.datetime.now()) + "/"
util.setup(path)
#image_ids, survival_days = util.get_image_id_and_survival_days(study_id="GBM_survival_time", registration_date_upper_lim="2018-10-29")
#image_ids = [10]
#image_ids = [10, 19, 35, 371, 71, 83,98, 103, 106, 116, 231, 392, 458]
#image_ids = [10, 19, 71, 83, 98, 103, 106, 116, 231, 392, 458]
#image_ids = range(454,465)
#image_ids = [413, 386, 398, 406, 416, 420, 419, 392, 412, 408, 405, 407]
image_ids = [75, 465, 125, 126, 183, 377, 220]
#image_ids = [413]
util.LOGGER.info(str(image_ids) + " " + str(len(image_ids)))
image_registration.get_transforms(
image_ids,
#reg_type=image_registration.RIGID,
#save_to_db=True,
def setUp(self):
super(WeiboTest, self).setUp()
u.setup(d)
def setUp(self):
super(BrowserTest, self).setUp()
u.setup(d)
