QualityIndex brick¶
Computes a quality index for each sub-brick (3D volumes for each TR) in a 3D+time dataset using AFNI 3dTqual¶
The output is a 1D time series with the index for each sub-brick.
Mandatory inputs parameters:
- in_file (a string representing an existing file)
Input image.
ex. '/home/username/data/derived_data/reg_func_valid.nii'
Optional inputs with default value parameters:
- autoclip (a boolean, optional, default value is False)
Clip off small voxels. Mutually exclusive with mask.
ex. False
- automask (a boolean, optional, default value is True)
Clip off small voxels. Mutually exclusive with mask
ex. True
- interval (a boolean or an integer, optional, default value is False)
Write out the median + 3.5 MAD of outlier count with each timepoint.
ex. False
- out_prefix (a string, optional, default value is ‘QI’)
Specify the string to be prepended to the filename of the output file.
ex. 'QI_'
- quadrant (a boolean or an integer, optional, default value is False)
Similar to spearman parameter, but using 1 minus the quadrant correlation coefficient as the quality index.
ex. False
- spearman (a boolean or an integer, optional, default value is False)
Quality index is 1 minus the Spearman (rank) correlation coefficient of each sub-brick with the median sub-brick.
ex. False
Optional inputs:
- mask_file (a string representing an existing file, optional)
Mask image. Compute correlation only across masked voxels. Mutually exclusive with automask and autoclip.
ex. '/home/username/data/derived_data/automask_mean_reg_func_valid.nii'
- polort (an integer, optional)
Detrend each voxel timeseries with polynomials. Default value is Undefined (i.e parameter not used)
ex. 3
Outputs parameters:
- out_file (a strings representing a file)
Out file.
ex. '/home/username/data/derived_data/outliers_reg_func_valid.out'
Useful links: