gistfile1.txt

#!/usr/bin/env python

import pandas as pd
import numpy as np
from argparse import ArgumentParser
import sys
import os

def parse_outliers(outliers, volumes):
    if not outliers:
        return []
    
    # initialize array of volumes
    arr = np.zeros(volumes)
    arr[outliers[0]] = 1
    
    return [arr] + parse_outliers(outliers[1:], volumes)

def save_outliers(outliers):
    # save as dataframe
    df = pd.DataFrame()
    for i, outlier in enumerate(outliers):
        df[str(i)] = pd.Series(outlier).astype(int)
    
    return df


def get_parser():
    docstr = """Generate outlier files compatible with SPM from fmriprep outputs"""
    parser = ArgumentParser(description="Generate outlier files compatible with SPM")
    parser.add_argument('bids_dir', action='store', help='the root folder of a BIDS dataset')
    parser.add_argument('-f', '--fmriprep', action='store', help='fmriprep directory')
    parser.add_argument('-o', '--outdir', action='store', help='outdir')
    parser.add_argument('-c', '--cutoff', action='store', default=0.2, help='FD cutoff')
    parser.add_argument('--overwrite', action='store_true', help='regenerate files')
    return parser
    
    
def main(argv=None):
    
    parser = get_parser()
    args = parser.parse_args(argv)
    
    if args.fmriprep:
        from glob import glob
        path = os.path.join(args.fmriprep, 'sub-*', '**', '*_confounds.tsv')
        confounds = [x for x in glob(path)]
    else:
        from bids.grabbids import BIDSLayout
        layout = BIDSLayout(args.bids_dir)
        confounds = layout.get(type='confounds', return_type='file')

    if not confounds:
        print("No confounds found")
        sys.exit(1)
    
    if not args.cutoff:
        print("FD cutoff missing")
        sys.exit(1)
        
    outdir = args.outdir if args.outdir else os.path.join(args.bids_dir, 
                                                          'derivatives', 
                                                          'outliers') 

    for confound in confounds:
        outname = os.path.join(outdir, os.path.basename(confound.split('_bold')[0] + '_fd.txt'))
        if os.path.exists(outname) and not args.overwrite:
            print(outname, 'already exists, to regenerate run with --overwrite option')
            continue

        df = pd.read_csv(confound, sep='\t')
        volumes = len(df)
        outliers = df[df['FramewiseDisplacement'].gt(args.cutoff)].index.tolist()
        outlier_arr = parse_outliers(outliers, volumes)
        if not outlier_arr:
            continue
        outdf = save_outliers(outlier_arr)

        outname = os.path.join(outdir, os.path.basename(confound.split('_bold')[0] + '_fd.txt'))
        if not os.path.exists(outdir):
            os.makedirs(outdir)
        outdf.to_csv(outname, index=False, header=False, sep="\t")

if __name__ == '__main__':
    main()