Source code for niftynet.layer.mean_variance_normalisation
# -*- coding: utf-8 -*-
from __future__ import absolute_import, print_function
import numpy as np
import numpy.ma as ma
from niftynet.layer.base_layer import Layer
from niftynet.layer.binary_masking import BinaryMaskingLayer
[docs]class MeanVarNormalisationLayer(Layer):
"""
This class defines image-level normalisation by subtracting
foreground mean intensity value and dividing by standard deviation
"""
def __init__(self, image_name, binary_masking_func=None):
self.image_name = image_name
super(MeanVarNormalisationLayer, self).__init__(name='mean_var_norm')
self.binary_masking_func = None
if binary_masking_func is not None:
assert isinstance(binary_masking_func, BinaryMaskingLayer)
self.binary_masking_func = binary_masking_func
[docs] def layer_op(self, image, mask=None):
if isinstance(image, dict):
image_data = np.asarray(image[self.image_name], dtype=np.float32)
else:
image_data = np.asarray(image, dtype=np.float32)
if isinstance(mask, dict):
image_mask = mask.get(self.image_name, None)
elif mask is not None:
image_mask = mask
elif self.binary_masking_func is not None:
image_mask = self.binary_masking_func(image_data)
else:
# no access to mask, default to the entire image
image_mask = np.ones_like(image_data, dtype=np.bool)
if image_data.ndim == 3:
image_data = whitening_transformation(image_data, image_mask)
if image_data.ndim == 5:
for m in range(image_data.shape[4]):
for t in range(image_data.shape[3]):
image_data[..., t, m] = whitening_transformation(
image_data[..., t, m], image_mask[..., t, m])
if isinstance(image, dict):
image[self.image_name] = image_data
if isinstance(mask, dict):
mask[self.image_name] = image_mask
else:
mask = {self.image_name: image_mask}
return image, mask
else:
return image_data, image_mask
