External Email - Use Caution

Dear experts, I have been using the mri_normalize in FSv7.3.2 to standardize the intensity of FLAIR images , and I've observed that increasing the sigma parameter up to 25 is effective in improving the quality of the normalized images for noisy datasets. mri_normalize -sigma 25 -g 1 FLAIR_skullstripped.nii FLAIR_skullstripped_norm.mgz

However, I have questions about the n argument, which is set to 2 by default. For noisy images, white matter intensity after normalization is around 140-150. When using n=10, I noticed that the white matter intensity decreased from 140-150 to around 110-120, resulting in better quality normalized images. 1) I'm wondering if there is a cut-off value for the number of iterations, or if a higher number of iterations always yields better results. 2) Also, if the intensity of the white matter after normalization is not around 110, but instead around 140 or 150, does that indicate that the normalization was not effective and that we need to increase the number of iterations?

Thank you for your assistance. John