医学图像预处理----Padding

class Padding(object):
	"""
	Add padding to the image if size is smaller than patch size

	Args:
		output_size (tuple or int): Desired output size. If int, a cubic volume is formed
	"""

	def __init__(self, output_size):
		self.name = 'Padding'

		assert isinstance(output_size, (int, tuple, list))
		if isinstance(output_size, int):
			self.output_size = (output_size, output_size, output_size)
		else:
			assert len(output_size) == 2
			self.output_size = output_size

		assert all(i > 0 for i in list(self.output_size))

	def __call__(self,sample):
		image, label = sample['image'], sample['label']

		size_old = image[0].GetSize()

		if (size_old[0] >= self.output_size[0]) and (size_old[1] >= self.output_size[1]):
			return sample
		else:
			output_size = list(self.output_size)
			if size_old[0] > self.output_size[0]:
				output_size[0] = size_old[0]
			if size_old[1] > self.output_size[1]:
				output_size[1] = size_old[1]

			output_size = tuple(output_size)

			for image_channel in range(len(image)):
				resampler = sitk.ResampleImageFilter()
				resampler.SetOutputSpacing(image[image_channel].GetSpacing())
				resampler.SetSize(output_size)

				# resample on image
				resampler.SetInterpolator(sitk.sitkLinear)#线性插值
				resampler.SetOutputOrigin(image[image_channel].GetOrigin())
				resampler.SetOutputDirection(image[image_channel].GetDirection())
				image[image_channel] = resampler.Execute(image[image_channel])

			# resample on label
			resampler = sitk.ResampleImageFilter()
			resampler.SetOutputSpacing(image[image_channel].GetSpacing())
			resampler.SetSize(output_size)
			resampler.SetInterpolator(sitk.sitkNearestNeighbor)#最近邻插值
			resampler.SetOutputOrigin(label.GetOrigin())
			resampler.SetOutputDirection(label.GetDirection())

			label = resampler.Execute(label)

			return {'image': image, 'label': label}