Hi,
we have brain images stored as DICOM and as NIFTI files. For some time we used to import the NIFTI files to freesurfer for segmentation. I was curious if it makes a difference if we would import the DICOM files directly, and was surprised that it actually did. Here are the commands we used:
# Import DICOM files recon-all -s 3003PC_dcm -i ../../Daten/DICOM/3003PC/M_MPRAGE_sag/IM-0010-0001.dcm
# Convert DICOM to NIFTI and import NIFTI file mri_convert ../../Daten/DICOM/3003PC/M_MPRAGE_sag/IM-0010-0001.dcm 3003PC.nii.gz recon-all -s 3003PC_nii -i 3003PC.nii.gz
# Segmentation of DICOM file import recon-all -s 3003PC_dcm -all
# Segmentation of NIFTI file import recon-all -s 3003PC_nii -all
Here are the first 3 volumes from aseg.stats for 3003PC_dcm (DICOM import)
1 4 5612 5611.8 Left-Lateral-Ventricle 27.6355 12.8465 8.0000 93.0000 85.0000 2 5 452 452.2 Left-Inf-Lat-Vent 42.1100 14.9377 7.0000 86.0000 79.0000 3 7 13765 13765.1 Left-Cerebellum-White-Matter 87.7808 6.6008 36.0000 109.0000 73.0000
and here the the first 3 volumes for 3003PC_nii (NIFTI import)
1 4 5582 5582.2 Left-Lateral-Ventricle 27.4507 12.7212 8.0000 93.0000 85.0000 2 5 421 421.1 Left-Inf-Lat-Vent 40.8032 14.6554 7.0000 83.0000 76.0000 3 7 13899 13898.6 Left-Cerebellum-White-Matter 87.6707 6.7539 36.0000 109.0000 73.0000
The differences are small but I think there should be no differences at all. Has someone an explanation for this differences and which file format is better?
Greetings Peter
P. S.: Our system: freesurfer-Linux-centos6_x86_64-stable-pub-v5.3.0 on Ubuntu 14.04