Averaging

Implementation of several image averaging algorithms.

Available averaging methods are:

  • mean, compute the mean value for each voxel;

  • robust-mean, compute the trimmed mean keeping only a fraction of values for each voxel [Truncated_mean];

  • median, compute the median value for each voxel;

  • minimum, keep the minimum value for each voxel;

  • maximum, keep the maximum value for each voxel;

  • quantile, retain only the given quantile_value for each voxel [Quantile];

  • sum, compute the sum for each voxel;

  • var, compute the variance for each voxel;

  • stddev, compute the standard deviation for each voxel.

timagetk.algorithms.averaging.images_averaging(images, method=None, masks=None, **kwargs)[source]

Images averaging algorithm.

Parameters:

  • images (list(SpatialImage)) – List of SpatialImage to average.

  • method ({"mean", "robust-mean", "median", "minimum", "maximum", "quantile", "sum", "var", "stddev"}, optional) – Image averaging method to use, “mean” by default.

  • masks (list of SpatialImage, optional) – If any, a list of masks to apply to images before averaging.

  • window (list) – Window size for processing, default is [1, 1(, 1)].

  • quantile_value (float) – Quantile of the retained value, should be in the [0, 1] range. See notes for detailled explanations.

  • lts_fraction (float in [0, 1]) – Fraction of points to be kept for the computation of the robust mean (trimmed estimation). Should be in the [0, 1] range.

  • params (str, optional) – CLI parameter string used by vt.mean_images method.

Returns:

timagetk.SpatialImage – Average image and its metadata.

Raises:

ValueError – If the image returned by vt.mean_images is None.

See also

timagetk.third_party.vt_parser.AVERAGING_METHODS, timagetk.third_party.vt_parser.DEF_AVG_METHOD, timagetk.third_party.vt_parser.averaging_kwargs, timagetk.third_party.vt_parser.general_kwargs, timagetk.third_party.vt_parser.parallel_kwargs

Notes

All images should have the same physical properties: shape & voxelsize.

Explanations for quantile_value parameters:

  • 0: minimum value, equivalent to “min” method

  • 0.5: median value, equivalent to “median” method

  • 1: maximum value, equivalent to “max” method

References

Examples

>>> from timagetk.algorithms.averaging import images_averaging
>>> from timagetk.synthetic_data.wall_image import example_layered_sphere_wall_image
>>> image = example_layered_sphere_wall_image()
>>> list_images = [image, image, image]
>>> mean_image = images_averaging(list_images, method='mean')
>>> rob_mean_image = images_averaging(list_images, method='robust-mean', lts_fraction=0.8)
>>> q_mean_image = images_averaging(list_images, method='quantile', quantile_value=0.8)