Helper Module for Deep Learning.

Common functions to normalize intensities. Code: https://github.com/jcreinhold/intensity-normalization

pynet.preprocessing.intensity.kde_normalize(arr, mask=None, modality='T1w', norm_value=1)[source]¶

Use kernel density estimation to find the peak of the white matter in the histogram of a skull-stripped image. Then normalize intensitites to a normalization value.

Parameters

arr: array

the input data.

mask: array, default None

the brain mask.

modality str, default ‘T1w’

the modality (T1w, T2w, FLAIR, MD, last, largest, first).

norm_value: float, default 1

the new intensity value for the detected WM peak.

Returns

normalized: array

the normalized input data.

pynet.preprocessing.intensity.rescale(arr, mask=None, percentiles=(0, 100), dynamic=(0, 1))[source]¶

Performs a rescale of the image intensities to a certain range.

Parameters

arr: array

the input data.

mask: array, default None

the brain mask.

percentiles: 2-uplet, default (0, 100)

percentile values of the input image that will be mapped. This parameter can be used for contrast stretching.

dynamic: 2-uplet, default (0, 1)

the intensities range of the rescaled data.

Returns

rescaled: array

the rescaled input data.

pynet.preprocessing.intensity.zscore_normalize(arr, mask=None)[source]¶

Performs a batch Z-score normalization.

Parameters

arr: array

the input data.

mask: array, default None

the brain mask.

Returns

normalized: array

the normalized input data.