Header

6. STEP 2: PREPARING AN ANATOMICAL IMAGE FOR EPI DISTORTION CORRECTION REORIENTATION OF FINAL IMAGE

If your data has an associated T2w (fat suppressed- highly recommended) dataset, EPI correction can be successfully performed using an anatomical image based registration in DIFFPREP. 

***Please note: This correction is as good as the quality of the diffusion and the structural data provided***

T1W image can be used in the absence of T2W image, although the high contrast of tissue to skull may result in suboptimal correction. Changing the contrast from T1w to T2w can be performed and the resulting image maybe be used in correction.

***Note: T1 to T2 converted image needs to be used with caution in performing EPI distortion correction. Unless masked appropriately, several areas with soft tissue in regions near brain stem, tissue skull interfaces, may get included in converted T2. Please check the imitation T2 and the resulting correction results from DIFFPREP/DRBUDDI. Using reference image should ensure correction and not add more artifacts***

Note1: if T1w or T2w image is unavailable, we recommend  turning the EPI correction option off (explained in upcoming section). In addition, one may choose not to perform EPI correction in the absence of inadequate reference image to avoid introducing more artifacts. 

We have adapted a practice of standardizing the diffusion data,in a study, by performing AC-PC alignment on the associated structural data. By identifying certain landmarks (Anterior commissure, Posterior Commissure) on the subject anatomical, in post processing, we can try to reduce the variability that may arise from subjects being scanned in various head positions. We will describe three procedures to align a structural data in standard space, with advice to choose the method best suited for your purpose and also paying close attention to the resulting output.

6.1. Option1: AUTO ACPC DETECT

‘AUTO ACPC detect’ software aligns T1w contrast images to its default T1 template that comes with the package. Since T2w image contrast is **required for EPI distortion correction in TORTOISE, especially DRBUDDI, you may rigidly align the T2w image to T1W image that has undergone auto acpc alignment.

Please note If you are only performing DIFFPREP processing and will not be continuing with DRBUDDI, you may provide the non oriented T2w image for the -s tag and provide an auto acpc aligned T1w image as a reorientation template. This will put your final reoriented image from DIFFPREP in ACPC aligned space.

The auto acpc detect software can be downloaded from the following link: http://www.nitrc.org/projects/art/

6.2. Option2: AFNI AXIALIZATION

Please make sure you have AFNI downloaded to use this script. AFNI can be downloaded from : https://afni.nimh.nih.gov/pub/dist/doc/htmldoc/background_install/download_links.html Please refer: https://sscc.nimh.nih.gov/pub/dist/doc/htmldoc/tutorials/fatcat_prep/Prepreprocessing_I.html

  1. Note:

    Please note that axialization is NOT ac-pc alignment script. While, the output images may end up in AC-PC alignment depending on if the match between reference and input image. We highly recommend checking the .ax.nii output image to check if the AC-PC points are in the same axial plane.
  2. Note:

    The reference image should be close in intensity and structure to the population of subjects, that are being registered to. Occasionally, the intensity differences, large differences in structure or size, can cause certain misalignment to be introduced. Please refer to the fat_profat_proc_axialize_anat command  in AFNI for additional tips on axializing. 
  3. Note:

    Please consider consulting the AFNI message boards for questions regarding axialization problems.

6.3. Option3: LANDMARKING AC-PC POINTS

In option 3 you can generate an AC-PC aligned structural image. AFNI or MIPAV have capabilities in generating landmark based AC-PC aligned images. Both are user dependent in identifying the landmarks. 

AFNI landmarks

Please refer to the following link for more information on the method: https://msu.edu/~zhuda/fmri_class/labs/lab3/afni08_talairach.pdf

MIPAV landmarks

If using MIPAV, please download it from the following link: (https://mipav.cit.nih.gov/) The following link provides details about the ac-pc alignment procedure: https://mipav.cit.nih.gov/pubwiki/index.php/Select_Algorithms_and_Brain_Tools_for_Talairach_Transform

  1. Note:

    We perform ac-pc alignment in native space and do not transform the images to tailarach co-ordinates. After you pick the 5 points, please hit apply and then save the resulting image.
  2. Note:

    MIPAV settings: Please make sure the saved file post AC-PC alignement is not flipped left right with respect to the DWI b0. The saved AC-PC file will have a set dimension and resolution (1mm isotropic). For a study, consistently use the same version of TORTOISE/AFNI/MIPAV to avoid any discrepancy. 
  3. Note:

    The output file saved in MIPAV has an intensity that is smoother than the original image. Whereas, AFNI generated output files, both from axialization or landmark based AC-PC alignment, maintain the original quality of the structural data provided. 
PreviousNext