roiextract.optimize.ctf_optimize#

roiextract.optimize.ctf_optimize(leadfield, template, mask, lambda_, mode='similarity', criteria='rat', threshold=None, initial='auto', tol=0.001, reg=0.001, quantify=False, ch_names=None, name='')#

Derive a spatial filter that optimizes properties of the CTF for the extracted ROI time series

Parameters:

  • leadfield (array_like) – Lead field matrix for dipoles with fixed orientation with shape (channels, voxels).

  • template (array_like) – Template CTF pattern for voxels with shape (voxels,). Only values within the ROI are used.

  • mask (array_like) – Voxel mask of the region of interest with shape (voxels,). Contains ones and zeros for voxels within and outside ROI, respectively.

  • lambda (float) – If 0, only ratio is optimized. If 1, only dot product. Values in between allow tweaking the balance between optimization for dot product or ratio.

  • mode (str) – Optimize a combination of ratio and similarity (mode=”similarity”) or ratio and homogeneity (mode=”homogeneity”).

  • reg (float) – Regularization parameter to ensure that it is possible to calculate the inverse matrices.

  • quantify (bool) – Whether to calculate CTF properties for the optimized spatial filter.

Returns:

  • sf (SpatialFilter) – Spatial filter produced by the optimization.

  • props (dict, only returned if quantify=True) – Dictionary that contains the estimates of CTF-based properties (ratio, similarity and/or homogeneity).

Raises:

ValueError – If provided lambda_ value is out of [0, 1] range.