def load_3d(self):
"""
dataset_mode: HGG/LGG/ALL
return list(dict[patient_id][modality] = filename.nii.gz)
"""
print("Data Folder: ", self.basedir)
#modalities = ['flair', 't1ce', 't1.', 't2']
modalities = ['ct']
if self.mode == 'test':
imgs = [self.basedir]
patient_ids = ['test' for x in imgs]
else:
imgs = glob.glob(self.basedir + "/*")
patient_ids = [x.split("/")[-1] for x in imgs]
ret = []
print("Preprocessing Data ...")
for idx, file_name in tqdm(enumerate(imgs), total=len(imgs)):
data = {}
data['image_data'] = {}
data['file_name'] = file_name
data['id'] = patient_ids[idx]
# read modality
mod = glob.glob(file_name + "/*.nii*")
#assert len(mod) >= 4, '{}'.format(file_name) # 4mod +1gt
for m in mod:
_m = m.split("/")[-1].split("_")[0]
data['image_data']['ct'] = m
if 'gt' in data:
if True:
#if not config.NO_CACHE and not 'training' in self.basedir:
data['gt'] = glob.glob(
os.path.join(self.basedir, "mask_nii/") +
file_name.split('/')[-1] + "*.nii*")[0]
data['preprocessed'] = crop_brain_region(
data['image_data'], data['gt'])
del data['image_data']
del data['gt']
else:
data['preprocessed'] = crop_brain_region(data['image_data'],
None,
with_gt=False)
del data['image_data']
ret.append(data)
return ret
def load_3d(self):
"""
dataset_mode: HGG/LGG/ALL
return list(dict[patient_id][modality] = filename.nii.gz)
"""
print("Data Folder: ", self.basedir)
modalities = ['flair', 't1ce', 't1.', 't2']
if 'training' in self.basedir:
img_HGG = glob.glob(self.basedir + "/GBM/*")
img_LGG = glob.glob(self.basedir + "/LGG/*")
imgs = img_HGG + img_LGG
else:
imgs = glob.glob(self.basedir + "/*")
imgs = [x for x in imgs if 'survival_evaluation.csv' not in x]
patient_ids = [x.split("/")[-1] for x in imgs]
ret = []
print("Preprocessing Data ...")
for idx, file_name in tqdm(enumerate(imgs), total=len(imgs)):
data = {}
data['image_data'] = {}
data['file_name'] = file_name
data['id'] = patient_ids[idx]
# read modality
mod = glob.glob(file_name + "/*.nii*")
assert len(mod) >= 4, '{}'.format(file_name) # 4mod +1gt
for m in mod:
if 'seg' in m:
data['gt'] = m
else:
_m = m.split("/")[-1].split(".")[0].split("_")[-1]
data['image_data'][_m] = m
if 'gt' in data:
if not config.NO_CACHE and not 'training' in self.basedir:
data['preprocessed'] = crop_brain_region(
data['image_data'], data['gt'])
del data['image_data']
del data['gt']
else:
data['preprocessed'] = crop_brain_region(data['image_data'],
None,
with_gt=False)
del data['image_data']
ret.append(data)
return ret
def load_5fold(self):
with open(config.CROSS_VALIDATION_PATH, 'rb') as f:
data = pickle.load(f)
imgs = data["fold{}".format(config.FOLD)][self.mode]
patient_ids = [x.split("/")[-1] for x in imgs]
ret = []
for idx, file_name in enumerate(imgs):
data = {}
data['image_data'] = {}
data['file_name'] = file_name
data['id'] = patient_ids[idx]
mod = glob.glob(file_name + "/*.nii*")
assert len(mod) >= 4
for m in mod:
if 'seg' in m:
data['gt'] = m
else:
_m = m.split("/")[-1].split(".")[0].split("_")[-1]
data['image_data'][_m] = m
if 'gt' in data:
data['preprocessed'] = crop_brain_region(
data['image_data'], data['gt'])
del data['image_data']
del data['gt']
ret.append(data)
return ret
def load_5fold(self):
# with open(config.CROSS_VALIDATION_PATH, 'rb') as f:
# data = pickle.load(f)
# imgs = data["fold{}".format(config.FOLD)][self.mode]
# patient_ids = [x.split("/")[-1] for x in imgs]
ret = []
patient_ids=[257,258]
imgs=['/home/guan/Documents/codes/3DUnet-Tensorflow-Brats18-master/data/dataset/BRATS2018/val/HGG/BraTS20_Training_257',
'/home/guan/Documents/codes/3DUnet-Tensorflow-Brats18-master/data/dataset/BRATS2018/val/HGG/BraTS20_Training_258']
for idx, file_name in enumerate(imgs):
data = {}
data['image_data'] = {}
data['file_name'] = file_name
data['id'] = patient_ids[idx]
# read modality
mod = glob.glob(file_name+"/*.nii*")
assert len(mod) >= 4 # 4mod +1gt
for m in mod:
if 'seg' in m:
data['gt'] = m
else:
_m = m.split("/")[-1].split(".")[0].split("_")[-1]
data['image_data'][_m] = m
if 'gt' in data:
data['preprocessed'] = crop_brain_region(data['image_data'], data['gt'])
del data['image_data']
del data['gt']
ret.append(data)
return ret