def load_metadata(self, index=0):
format = NiftiFormat(self._filename)
metadata = {}
# Get the number of dimensions in the image from the extent
nb_dimensions = len(format.extent)
for i in range(max(len(format.extent) - 3, 0)):
if format.extent[-1] == 1:
nb_dimensions -= 1
else:
break
# Get spacing and origin for those dimensions. Reverse to keep the order
# of the numpy array
metadata["spacing"] = format.asDict()["pixdim"][1:nb_dimensions + 1]
metadata["spacing"].reverse()
metadata["spacing"] = numpy.asarray(metadata["spacing"])
if format.asDict()["scl_slope"] != 0:
metadata["slope"] = format.asDict()["scl_slope"]
metadata["shift"] = format.asDict()["scl_inter"]
metadata["header"] = format.asDict()
metadata["annotations"] = ObservableList()
#########################
# Diffusion information #
#########################
# Load gradient direction file
base_name = os.path.splitext(self._filename)[0]
if base_name.endswith(".nii"):
base_name = os.path.splitext(base_name)[0]
gradient_candidates = [
base_name + ".bvecs", # /foo/bar/image.bvecs
base_name + ".bvec", # /foo/bar/image.bvec
os.path.join(os.path.dirname(self._filename),
"bvecs"), # /foo/bar/bvecs
os.path.join(os.path.dirname(self._filename),
"bvec") # /foo/bar/bvec
]
gradient_file = None
for candidate in gradient_candidates:
if os.path.isfile(candidate):
gradient_file = candidate
break
# Load b-values file
bvalue_candidates = [
base_name + ".bvals", # /foo/bar/image.bvals
base_name + ".bval", # /foo/bar/image.bval
os.path.join(os.path.dirname(self._filename),
"bval"), # /foo/bar/bvals
os.path.join(os.path.dirname(self._filename),
"bvals") # /foo/bar/bval
]
bvalue_file = None
for candidate in bvalue_candidates:
if os.path.isfile(candidate):
bvalue_file = candidate
break
if None not in [gradient_file, bvalue_file]:
gradients = numpy.loadtxt(gradient_file, dtype=numpy.single)
bvalues = numpy.loadtxt(bvalue_file, dtype=numpy.single)
gradients = gradients.T
mr_diffusion_sequence = []
for index, gradient in enumerate(gradients):
dataset = medipy.io.dicom.DataSet()
dataset.diffusion_directionality = "DIRECTIONAL"
dataset.diffusion_bvalue = bvalues[index]
gradient_dataset = medipy.io.dicom.DataSet()
gradient_dataset.diffusion_gradient_orientation = gradient
dataset.diffusion_gradient_direction_sequence = [
gradient_dataset
]
mr_diffusion_sequence.append(dataset)
metadata["mr_diffusion_sequence"] = mr_diffusion_sequence
return metadata
