Dear all,

I am a beginners for the brain image analysis. Recently, I have some problems when using the dicom2nii.exe:

1. I am going to using the freesurfer to segment the hippo. Each of the T1 Dicom images has many frames, so I try to use the dicom2nii.exe to convert the dicom-format to nifti-format. I got three types of the output: 'o***.nii', 'co***.nii', '***.nii'. I know that the 'o' is stand for reorient and 'c' is for the crop. But I am a little confused that which one would be better for the further analysis in freesurfer and why ?

2. Due to I am not familiar with the code, could any one tell me the specific steps during the cropping ? Is the cropping carried based on the template or only cut the non-brain tissues ? The cropped output sizes of the different subjects are different, so perhaps it is cropped by cutting the non-brain tissues, right?

Can I suggest you upgrade to 2019 versions of MRIcron or MRIcroGL? These will include the modern dcm2niix rather than the older dcm2nii. While I hope my old tools are robust and mature, the vendors are constantly updating their DICOM images (e.g. moving to enhanced DICOM). You can either use dcm2niix from the command line, or use the MRIcron/MRIcroGL "Import" menu for a graphical interface. dcm2niix will always rotate 3D images to the nearest orthogonal rotation to a NIfTI identity matrix (hence, no 'o' images), but does provide the option to crop 3D images. Cropping attempts to remove excess neck from T1-weigthed MRI scans of the brain. Excess neck can disrupt some tools such as SPM8 and FSL's BET. The newer normalization in SPM12 appears very robust against this. If you are only interested in the brain, the cropped images will require less disk space, lead to more robust processing, and process faster. On the other hand, if you are interested in other body parts, or if the neck is of interest, you will want to preserve it.

Hi， 

Thanks for your kindly response, I will follow your recommends to upgrade the version of the MRIcron. So do you mean all the three outputs(co**.nii, o***.nii, ***.nii) are ok for the further analysis in freesurfer ? And the cropping is to remove excess neck of the T1 Image not based on a Standard template ?

Best wishes

Assuming FreeSurfer reads the NIfTI SForm/QForm then the "img.nii" and "omg.nii" are identical (they are just stored to disk differently). The cropped image may segment better with FreeSurfer, but it really depends on the contrast of your T1 scan (e.g. fat suppression) and the amount of excess neck. In my limited experience, the segmentation of SPM12 is very robust, while the segmentation of FreeSurfer can be outstanding but depends very much on the quality of your T1 images. If you are just acquiring data now, I would suggest using the MGH multi-echo MP-RAGE sequence. The meninges and pial surfaces have different signal decay across the echoes, and FreeSurfer can leverage this information. Please look at the FreeSurfer support page for help with that tool.

Hi，
Thanks for the recommend， I have checked the code crop.pas  at https://github.com/neurolabusc/MRIcron. However, in the crop.pas, it seems the range of the cropping is defined by the parameters: lL, lR, lA, lP, lD, lV. Am I right in understanding ？And I can't find the meanings of these parameters, could you please tell me how these parameters be gotten ?

Best wishes

You are looking at the deprecated code for dcm2nii, development ceased on that tool in 2015 (this is why the final versions of dcm2nii urged users to upgrade to dcm2niix). dcm2nii has been superseded by dcm2niix. Vendors are continually updating their use of DICOM, so I would strongly discourage users from using dcm2nii anymore (with the exception of archival work with proprietary formats that predate DICOM). The cropping code for dcm2niix is the function nii_saveCrop(). The code finds the brightest slice in the head-foot direction, which typically corresponds roughly to the center of the brain (ventricular CSF for T2, anterior commissure for T1). It then looks for most superior slice that exceeds 9% of this value. It then crops any slices that are more than 169mm inferior of that slice.