compare_nii_images_using_checksum.py

import os
import glob
import hashlib
import argparse


def compute_checksum(file_path):
    with open(file_path, 'rb') as f:
        file_bytes = f.read()
    return hashlib.md5(file_bytes).hexdigest()


if __name__ == '__main__':
    parser = argparse.ArgumentParser(description='Check if image in folder1 is presented in folder2 '
                                                 '(sct-testing-large).')
    parser.add_argument('-folder1', type=str, help='path to folder 1 with NIfTI images')
    parser.add_argument('-folder2', type=str, help='path to folder 2 with NIfTI images (sct-testing-large)')
    args = parser.parse_args()

    if not args.folder1 or not args.folder2:
        parser.print_help()
        exit(1)

    # Get all .nii.gz files without seg.nii.gz suffix in the input folder 1
    files1 = [os.path.abspath(f) for f in glob.glob(args.folder1 + '/*.nii.gz') if f.endswith('.nii.gz') and not f.endswith('seg.nii.gz')]

    # Get all .nii.gz files in the input folder 2. This is a recursive search because sct-testing-large is BIDS
    files2 = [os.path.abspath(f) for f in glob.glob(args.folder2 + '/**/*.nii.gz', recursive=True) if 'anat' in f]

    # Loop across files in folder1
    for file1 in files1:
        print(file1)
        checksum1 = compute_checksum(file1)
        # Loop across files in folder2
        for file2 in files2:
            checksum2 = compute_checksum(file2)
            if checksum1 == checksum2:
                print(f"{file1} and {file2} are identical.")
                break