dipy_correct_biasfield

Synopsis

Correct bias field.

Usage

dipy_correct_biasfield [OPTIONS] input_files

Input Parameters

• input_files

  Path to the input volumes. This path may contain wildcards to process multiple inputs at once.

General Options

• --bval

  Path to the b-value file. (default: None)

• --bvec

  Path to the b-vector file. (default: None)

• --method

  Bias field correction method. Choose from:

  • ‘n4’: DeepN4 bias field correction. See [1] for more details.

  • ‘poly’: Legendre polynomial regression bias correction.

  • ‘bspline’: Cubic B-spline regression bias correction.

  • ‘auto’: Run both poly and bspline, return the one with lower Coefficient of Variation within the brain mask.

  ‘n4’ method is recommended for T1-weighted images. ‘poly’ and ‘bspline’ methods are recommended for diffusion-weighted images. (default: auto)

• --threshold

  Threshold for cleaning the final correction field in DeepN4 method. (default: 0.5)

• --use_cuda

  Use CUDA for DeepN4 bias field correction. (default: False)

• --verbose

  Print verbose output. (default: False)

• --order

  Maximum Legendre polynomial degree (used with method=’poly’). (default: 3)

• --n_control_points

  Control grid size per axis (used with method=’bspline’). (default: 8)

• --pyramid_levels

  Comma-separated downsampling factors for coarse-to-fine pyramid, e.g. ‘4,2,1’ (used with method=’poly’ or ‘bspline’). (default: 4,2,1)

• --n_iter

  Reweighting iterations per pyramid level (poly/bspline methods). (default: 4)

• --lambda_reg

  Ridge regularization strength (poly/bspline methods). (default: 0.001)

• --robust

  Apply Tukey biweight robust reweighting (poly/bspline methods). (default: True)

• --gradient_weighting

  Apply gradient-based edge suppression (poly/bspline methods). (default: True)

• --zero_background

  If True, set the saved bias field to 1.0 outside the brain mask, suppressing extrapolation artifacts in the background (poly/bspline methods). (default: False)

Output Options

• --out_dir

  Output directory. (default: current directory)

• --out_corrected

  Name of the corrected volume to be saved. (default: biasfield_corrected.nii.gz)

• --out_bias_field

  Name of the bias field volume to be saved (poly/bspline methods). (default: bias_field.nii.gz)

References

[1]

Praitayini Kanakaraj, Tianyuan Yao, Leon Y. Cai, Ho Hin Lee, Nancy R. Newlin, Michael E. Kim, Chenyu Gao, Kimberly R. Pechman, Derek Archer, Timothy Hohman, Angela Jefferson, Lori L. Beason-Held, Susan M. Resnick, Eleftherios Garyfallidis, Adam Anderson, Kurt G. Schilling, Bennett A. Landman, Daniel Moyer, The Alzheimer’s Disease Neuroimaging Initiative (ADNI), and The BIOCARD Study Team. DeepN4: Learning N4ITK Bias Field Correction for T1-weighted Images. Neuroinformatics, 22(2):193–205, April 2024. URL: https://doi.org/10.1007/s12021-024-09655-9, doi:10.1007/s12021-024-09655-9.

Garyfallidis, E., M. Brett, B. Amirbekian, A. Rokem, S. Van Der Walt, M. Descoteaux, and I. Nimmo-Smith. Dipy, a library for the analysis of diffusion MRI data. Frontiers in Neuroinformatics, 1-18, 2014.