for testing and deploying your application
for finding and fixing issues
for empowering human code reviews
import SimpleITK as sitk
from .label_transform import LabelTransform
class Contour(LabelTransform):
r"""Keep only the borders of each connected component in a binary image.
Args:
**kwargs: See :class:`~torchio.transforms.Transform` for additional
keyword arguments.
"""
def __init__(self, **kwargs):
super().__init__(**kwargs)
self.args_names = []
def apply_transform(self, subject):
for image in self.get_images(subject):
assert image.data.ndim == 4 and image.data.shape[0] == 1
sitk_image = image.as_sitk()
contour = sitk.BinaryContour(sitk_image)
tensor, _ = self.sitk_to_nib(contour)
image.set_data(tensor)
return subject