import collections
import os
import csv
import copy
import glob
import SimpleITK as sitk
from radiomics import featureextractor
from Utility.EcLog import eclog
class RadiomicsFeatureExtractor:
def __init__(self, radiomics_parameter_file, has_label=True, ignore_tolerence=False, ignore_diagnostic=True):
self.feature_values = []
self.case_list = []
self.feature_name_list = []
self.extractor = featureextractor.RadiomicsFeatureExtractor(radiomics_parameter_file)
self.error_list = []
self._file_name = os.path.split(__file__)
self.__has_label = has_label
self.__is_ignore_tolerence = ignore_tolerence
self.__is_ignore_diagnostic = ignore_diagnostic
self.error_list = []
def _RemoveKeyFromPathList(self, path_list, key):
new_path_list = []
for p in path_list:
if key not in str(p):
new_path_list.append(p)
return new_path_list
def __GetFeatureValuesEachModality(self, data_path, roi_path, key_name):
if self.__is_ignore_tolerence:
image = sitk.ReadImage(data_path)
roi_image = sitk.ReadImage(roi_path)
roi_image.CopyInformation(image)
result = self.extractor.execute(image, roi_image)
else:
result = self.extractor.execute(data_path, roi_path)
feature_names = []
feature_values = []
for feature_name, feature_value in zip(list(result.keys()), list(result.values())):
if self.__is_ignore_diagnostic and 'diagnostics' in feature_name:
continue
feature_names.append(key_name + '_' + feature_name)
feature_values.append(feature_value)
return feature_names, feature_values
def __GetFeatureValues(self, case_folder, key_name_list, show_key_name_list, roi_key):
feature_dict = {}
if isinstance(roi_key, str):
roi_key = [roi_key]
roi_key_path = '*'
for one_roi_key in roi_key:
roi_key_path += '{}*'.format(one_roi_key)
roi_candidate = glob.glob(os.path.join(case_folder, roi_key_path))
roi_candidate = [one for one in roi_candidate if 'nii' in one]
if len(roi_candidate) != 1:
eclog(self._file_name).LogError('Check the ROI file path of case: ' + case_folder)
return None
roi_file_path = roi_candidate[0]
quality_feature_path = os.path.join(case_folder, 'QualityFeature.csv')
if os.path.exists(quality_feature_path):
with open(quality_feature_path, 'r') as csvfile:
reader = csv.reader(csvfile, delimiter=',', quotechar='"', quoting=csv.QUOTE_MINIMAL)
for row in reader:
feature_dict['Quality_' + row[0]] = row[1]
for sequence_key, show_key in zip(key_name_list, show_key_name_list):
if isinstance(sequence_key, str):
sequence_key = [sequence_key]
sequence_key_path = '*'
for one_sequence_key in sequence_key:
sequence_key_path += '{}*'.format(one_sequence_key)
sequence_candidate = glob.glob(os.path.join(case_folder, sequence_key_path))
for one_roi_key in roi_key:
sequence_candidate = self._RemoveKeyFromPathList(sequence_candidate, one_roi_key)
if len(sequence_candidate) != 1:
eclog(self._file_name).LogError('Check the data file path of case: ' + sequence_key_path)
return None
sequence_file_path = sequence_candidate[0]
feature_names_each_modality, feature_values_each_modality = \
self.__GetFeatureValuesEachModality(sequence_file_path, roi_file_path, show_key)
for feature_name, feature_value in zip(feature_names_each_modality, feature_values_each_modality):
if feature_name in self.feature_name_list:
feature_dict[feature_name] = feature_value
else:
print('Check the feature name in the feature name list')
feature_dict = collections.OrderedDict(sorted(feature_dict.items()))
feature_values = list(feature_dict.values())
if self.__has_label:
label_value = 0
label_path = os.path.join(case_folder, 'label.csv')
if os.path.exists(label_path):
with open(label_path, 'r') as csvfile:
reader = csv.reader(csvfile, delimiter=',', quotechar='"', quoting=csv.QUOTE_MINIMAL)
for row in reader:
label_value = row[0]
feature_values.insert(0, label_value)
else:
print('No label file!: ', label_path)
return feature_values
def __MergeCase(self, case_name, feature_values):
if case_name in self.case_list:
print('The case exists!')
return False
else:
if isinstance(feature_values, list) and len(feature_values) == len(self.feature_values[0]):
self.case_list.append(case_name)
self.feature_values.append(feature_values)
return True
else:
print('Not extract valid features')
return False
def __InitialFeatureValues(self, case_folder, key_name_list, show_key_name_list, roi_key):
feature_dict = {}
if isinstance(roi_key, str):
roi_key = [roi_key]
roi_key_path = '*'
for one_roi_key in roi_key:
roi_key_path += '{}*'.format(one_roi_key)
roi_candidate = glob.glob(os.path.join(case_folder, roi_key_path))
roi_candidate = [one for one in roi_candidate if 'nii' in one]
if len(roi_candidate) != 1:
eclog(self._file_name).LogError('Check the ROI file path of case: ' + case_folder)
return None
roi_file_path = roi_candidate[0]
# Add quality feature
quality_feature_path = os.path.join(case_folder, 'QualityFeature.csv')
if os.path.exists(quality_feature_path):
with open(quality_feature_path, 'r') as csvfile:
reader = csv.reader(csvfile, delimiter=',', quotechar='"', quoting=csv.QUOTE_MINIMAL)
for row in reader:
feature_dict['Quality_' + row[0]] = row[1]
# Add Radiomics features
for sequence_key, show_key in zip(key_name_list, show_key_name_list):
if isinstance(sequence_key, str):
sequence_key = [sequence_key]
sequence_key_path = '*'
for one_sequence_key in sequence_key:
sequence_key_path += '{}*'.format(one_sequence_key)
sequence_candidate = glob.glob(os.path.join(case_folder, sequence_key_path))
for one_roi_key in roi_key:
sequence_candidate = self._RemoveKeyFromPathList(sequence_candidate, one_roi_key)
if len(sequence_candidate) != 1:
eclog(self._file_name).LogError('Check the data file path of case: ' + sequence_key_path)
return None
sequence_file_path = sequence_candidate[0]
feature_names_each_modality, feature_values_each_modality = self.__GetFeatureValuesEachModality(
sequence_file_path, roi_file_path, show_key)
for feature_name, feature_value in zip(feature_names_each_modality, feature_values_each_modality):
feature_dict[feature_name] = feature_value
feature_dict = collections.OrderedDict(sorted(feature_dict.items()))
feature_names = list(feature_dict.keys())
feature_values = list(feature_dict.values())
# Add Label
if self.__has_label:
label_value = 0
label_path = os.path.join(case_folder, 'label.csv')
if os.path.exists(label_path):
with open(label_path, 'r') as csvfile:
reader = csv.reader(csvfile, delimiter=',', quotechar='"', quoting=csv.QUOTE_MINIMAL)
for row in reader:
label_value = row[0]
feature_names.insert(0, 'label')
feature_values.insert(0, label_value)
else:
print('No label file!: ', label_path)
self.feature_name_list = feature_names
self.feature_values.append(feature_values)
self.__TempSave('temp.csv')
def __TempSave(self, store_path):
with open(store_path, 'w', newline='') as csvfile:
writer = csv.writer(csvfile, delimiter=',', quotechar='"', quoting=csv.QUOTE_MINIMAL)
for feature_name, feature_value in zip(self.feature_name_list, self.feature_values[0]):
writer.writerow([feature_name, feature_value])
def __IterateCase(self, root_folder, key_name_list, roi_key, show_key_name_list, store_path=''):
case_name_list = os.listdir(root_folder)
case_name_list.sort()
for case_name in case_name_list:
case_path = os.path.join(root_folder, case_name)
if not os.path.isdir(case_path):
continue
print(case_name)
try:
if self.feature_name_list != [] and self.feature_values != []:
feature_values = self.__GetFeatureValues(case_path, key_name_list, show_key_name_list, roi_key)
if not self.__MergeCase(case_name, feature_values):
self.error_list.append(case_name)
else:
self.__InitialFeatureValues(case_path, key_name_list, show_key_name_list, roi_key)
self.case_list.append(case_name)
except Exception as e:
content = 'In FeatureExtractor, {} extracted failed: '.format(case_name)
eclog(self._file_name).LogError('{}{}'.format(content, str(e)))
self.error_list.append(case_name)
print('{} \n{}'.format(content, e.__str__()))
raise e
if store_path:
self.Save(store_path)
with open(os.path.join(store_path + '_error.csv'), 'w', newline='') as csvfile:
writer = csv.writer(csvfile)
for error_case in self.error_list:
writer.writerow([error_case])
def Save(self, store_path):
header = copy.deepcopy(self.feature_name_list)
header.insert(0, 'CaseID')
with open(store_path, 'w', newline='') as csvfile:
writer = csv.writer(csvfile, delimiter=',', quotechar='"', quoting=csv.QUOTE_MINIMAL)
writer.writerow(header)
for case_name, feature_value in zip(self.case_list, self.feature_values):
row = list(map(str, feature_value))
row.insert(0, case_name)
writer.writerow(row)
def Read(self, file_path):
self.feature_values = []
self.case_list = []
self.feature_name_list = []
with open(file_path, 'r', newline='') as csvfile:
reader = csv.reader(csvfile, delimiter=',', quotechar='"', quoting=csv.QUOTE_MINIMAL)
for row in reader:
if row[0] == '':
self.feature_name_list = row[1:]
else:
self.case_list.append(row[0])
self.feature_values.append(row[1:])
def Execute(self, root_folder, key_name_list, roi_key, store_path, show_key_name_list=[]):
if len(show_key_name_list) == 0:
show_key_name_list = key_name_list
assert(len(show_key_name_list) == len(key_name_list))
if not store_path.endswith('.csv'):
print('The store path should be a CSV format')
else:
self.__IterateCase(root_folder, key_name_list, roi_key, store_path=store_path, show_key_name_list=show_key_name_list)
if __name__ == '__main__':
pass
