fit_harmonics_grids_regularize

Spherical harmonic fitting with Legendre polynomial regularization.

fit_harmonics_grids_regularize <grid1> <mask1> [...] <output.par> [options]

DESCRIPTION

fit_harmonics_grids_regularize fits spherical harmonic coefficients to deformation grids with an additional Legendre polynomial regularization term. The regularization penalizes high-order coefficients to produce smoother distortion models and reduce overfitting, particularly when the input data is noisy or sparsely sampled.

The regularization strength is controlled by the --lambda parameter. Higher values produce smoother fits at the expense of fitting accuracy, while lower values allow the model to follow the data more closely.

OPTIONS

--verbose

Print progress information during processing.

--clobber

Overwrite the output file if it already exists.

--order n

Maximum order of the spherical harmonic expansion.

--skip n

Skip every n-th voxel during fitting to reduce computation time.

--limit val

Exclude voxels with displacement magnitude above this threshold.

--par file

Initialize fitting from an existing parameter file.

--lambda val

Regularization weight. Higher values produce smoother fits (default: 0).

EXAMPLES

# Regularized fit with default lambda
fit_harmonics_grids_regularize grid.mnc mask.mnc output.par --order 10

# Strong regularization for noisy data
fit_harmonics_grids_regularize grid.mnc mask.mnc output.par \
    --order 12 --lambda 0.1

# Joint regularized fit across multiple grids
fit_harmonics_grids_regularize grid1.mnc mask1.mnc grid2.mnc mask2.mnc \
    output.par --order 10 --lambda 0.01 --skip 2

AUTHOR

Vladimir S. Fonov - McConnell Brain Imaging Centre, Montreal Neurological Institute.

COPYRIGHTS

Copyright © 2009-2024 by Vladimir S. Fonov

SEE ALSO

fit_harmonics_grids , fit_harmonics_grids_diff , param2grid