Coil PCA

Coil compression using principal component analysis.

mr_utils.coils.coil_combine.coil_pca.coil_pca(coil_ims, coil_dim=-1, n_components=4, give_explained_var=False, real_imag=True, debug_level=30)[source]

Reduce the dimensionality of the coil dimension using PCA.

Parameters:
  • coil_ims (array_like) – Coil images.
  • coil_dim (int, optional) – Coil axis, default is last axis.
  • n_components (int, optional) – How many principal components to keep.
  • give_explained_var (bool, optional) – Return explained variance for real,imag decomposition
  • real_imag (bool, optional) – Perform PCA on real/imag parts separately or mag/phase.
  • debug_level (logging_level, optional) – Verbosity level to set logging module.
Returns:

  • coil_ims_pca (array_like) – Compressed coil images representing n_components principal components.
  • expl_var (array_like, optional) – complex valued 1D vector representing explained variance. Is returned if give_explained_var=True
mr_utils.coils.coil_combine.coil_pca.python_pca(X, n_components=False)[source]

Python implementation of principal component analysis.

To verify I know what sklearn’s PCA is doing.

Parameters:
  • X (array_like) – Matrix to perform PCA on.
  • n_components (int, optional) – Number of components to keep.
Returns:

P – n_component principal components of X.
Return type:

array_like