def img_to_h5(img_dir, h5_output_name):
""" Saves all slices from an image to a h5 file. """
img = load_dicom_series(img_dir)
with h5py.File(h5_output_name, 'w') as f:
f.create_dataset(H5_DATASET_NAME, data=img)
print("Saved")
def dicom_to_h5(root_dir, h5):
sub_dirs = [x[0] for x in os.walk(root_dir)] # Gather all subdirectories in 'root_dir'
for directory in sub_dirs:
# print(directory)
file_list = glob.glob(directory + '/*.dcm') # Look for .dcm files
if not file_list: # If we find a dir with a .dcm series, process it
continue
dcm_filename = file_list[0] # Checking just one .dcm file is sufficient
img = SimpleITK.ReadImage(dcm_filename) # Read single .dcm file to obtain metadata
# Extract some metadata that we want to keep
patient_id = img.GetMetaData('0010|0020').strip()
patient_age = img.GetMetaData('0010|1010').strip()
series_number = int(img.GetMetaData('0020|0011').strip())
series_description = img.GetMetaData('0008|103e').strip()
data_path = patient_id + '/' + series_description
print('Converting: {}'.format(data_path))
# If we find a DICOM series that already exists, we check if the series number is higher. If so, remove
# series that is already present and add this one.
create = False
if data_path in h5:
if h5[data_path]['pixel_array'].attrs.get('SeriesNr') < series_number:
del h5[data_path]
print('New series has higher series number, so adding.')
create = True
else:
print('New series has lower series number, so not adding.')
else:
create = True
if create:
group = h5.create_group(data_path)
pixeldata = group.create_dataset('pixel_array', data=load_dicom_series(directory))
pixeldata.attrs.create('Age', patient_age)
pixeldata.attrs.create('SeriesNr', series_number)
def dicom_to_h5(root_dir, h5):
sub_dirs = [x[0] for x in os.walk(root_dir)] # Gather all subdirectories in 'root_dir'
ages = {}
for directory in sub_dirs:
# print(directory)
if directory.split(os.sep)[-1] == 'Ktrans': #if the file is from Ktrans, look for the .mhd
Im_Ktrans = True
file_list = glob.glob(directory + '/*.mhd')
else:
Im_Ktrans = False
file_list = glob.glob(directory + '/*.dcm') # Look for .dcm files
if not file_list: # If we find a dir with a .dcm or a .mhd series, process it
continue
series_filename = file_list[0] # Checking just one .dcm file is sufficient
img = sitk.ReadImage(series_filename) # Read single .dcm file to obtain metadata
# Extract some metadata that we want to keep
if Im_Ktrans:
patient_id = series_filename.split(os.sep)[-3]
patient_age = ages[patient_id]
series_number = 0
series_description = 'Ktrans'
else:
patient_id = img.GetMetaData('0010|0020').strip()
patient_age = img.GetMetaData('0010|1010').strip()
series_number = int(img.GetMetaData('0020|0011').strip())
series_description = img.GetMetaData('0008|103e').strip()
# Add the age info in the dictionary for Ktrans
ages[patient_id] = patient_age
# Combine series description and series number to create a unique identifier for this DICOM series.
# Should be unique for each patient. Only exception is ProstateX-0025, hence the try: except: approach.
data_path = patient_id + '/' + series_description #+ '_' + series_number
print('Converting: {}'.format(data_path))
# If we find a DICOM series that already exists, we check if the series number is higher. If so, remove
# series that is already present and add this one.
create = False
if data_path in h5:
if h5[data_path]['pixel_array'].attrs.get('SeriesNr') < series_number:
del h5[data_path]
print('New series has higher series number, so adding.')
create = True
else:
print('New series has lower series number, so not adding.')
else:
create = True
if create:
#print(patient_id)
group = h5.create_group(data_path)
if Im_Ktrans:
data = sitk.GetArrayFromImage(img)
pixeldata = group.create_dataset('pixel_array', data=data)
else:
pixeldata = group.create_dataset('pixel_array', data=load_dicom_series(directory))
pixeldata.attrs.create('Age', patient_age)
pixeldata.attrs.create('SeriesNr', series_number)