CATNAP : Technical Notes

(Coregistration, Adjustment, and Tensor-solving, a Nicely Automated Program)

BUG: Eigenvalue Scaling (April 18, 2007)

An astute user noticed that the eigenvalues that CATNAP produced seemed to be a scalar multiple off of those produced by other programs. Indeed, this is the case. Due to a problem with our revision control system after our validation testing, the b-value was unintentionally reset to 700 s/mm² regardless of the b-value in the par file. This impacts all public releases of CATNAP. A new, fixed version (3.13) is available. Please upgrade.
We anticipate that this mistake will impact users to differing degrees:
If you used the dtiproc option:

If you use a b-value of 700 s/mm²_:

Your data has been processed correctly.

If you use a different b-value:

Your anistropy measures (FA, RA, sRA, VR, iVR), colormap (CMAP), orientations (VEC), and image contrasts (MB0, MDW) are correct.
The following files are systematically incorrect: (Suppose b is your b-value in s/mm²)

The eigenvalues (EVAL1, EVAL2, EVAL3) and mean diffusivity (ADC) should be multiplied by 700/b.
The tensor coefficients (DXX, DXY, DXZ, DYY, DYZ, DZZ) should be multiplied by 700/b.

Additionally, prior to release 3.13, the AFNI processing options did not compensate for b-value as per the notation on the AFNI message board. With release 3.13, we have added a correction step so that the eigenvalues are in physical units and are consistent with dtiproc. These changes will impact all users of the AFNI options:
If you used either AFNI option:

Your anistropy measures (FA, RA, sRA, VR, iVR), colormap (CMAP), orientations (VEC), and image contrasts (MB0, MDW) are correct.
The following files are systematically incorrect: (Suppose b is your b-value in s/mm²)

The eigenvalues (EVAL1, EVAL2, EVAL3) and mean diffusivity (ADC) should be multiplied by 1/b.
The tensor coefficients (DXX, DXY, DXZ, DYY, DYZ, DZZ) should be multiplied by 1/b.

We sincerely appologize for any inconvenience. To possiblely alleviate future problems, we decided to post the JAVA source files for dtiproc and the contrast computations. Previously, these packages were available upon request. This is a service for advanced useres, and use of these tools outside of CATNAP cannot be supported. However, comments, improvements, and optimizations are welcome and appreciated.

PAR File Format (April 11, 2007)

If your DTI data was acquired on a Philips MR Scanner with new software (release 1.5, 1.7, 2.0 or 2.1), the PAR file version may be V4.1. You can verify this by looking at the top right hand corner of the file for the tag: Research image export tool V4.1. The other known versions are V3 or V4.
The V4.1 par files list the diffusion weighting directions. HOWEVER, these directions are not in pixel space. Specifically, in the case of gradient overplus = YES tables, the directions need to be rotated and corrected for slice angulation. This is automatically done with CATNAP.
CATNAP DOES NOT read in the directions from the PAR file because this information is not available for all PAR files (V3 or V4). Instead, it uses the directions documented in the Philips source code.

“Sort Images” Export Option (April 11, 2007)

1) If you are using a user defined gradient table without the “OVERRIDE” option, then do NOT specify the 0,0,0 (b0’s) in the diffusion weighting directions text file. The location of the b0’s within the file will be automatically determined based on the PAR version and “sort images” option.

2) If you are using the “OVERRIDE” option, then you must specify the order to b0’s and diffusion weighted images in the diffusion weighting directions text file. Indicate b0’s with a direction of 0,0,0.

3) CATNAP is using the following logic to place the b0’s:

Version sort_images order b=0 position in REC file

V3 yes volume first volume

V3 no NOT SUPPORTED

V4 yes volume first volume

V4 no slice first sub volume for slices

V4.1 yes volume last or second last volume

V4.1 no slice first sub volume for slices

4) IMPACT on gradient tables for DTIStudio:

% This code writes out gradient tables for DTIStudio. The gradient table % provided will reflect the position of the b=0 volume in the REC file (it % will be first, last or second last depending on the table above). For % the case of V4.1 sort_images = y, the b=0 volume is last. For the case % of V4.1 sort_images = 'n', the b=0 volume must be listed first to work % with DTIStudio and image sequence "slice by slice" must be selected

Last Updated: Tuesday May 27, 2008

THIS PAGE HAS BEEN RETIRED. PLEASE SEE MY NEW WEBSITE.

http://masi.vuse.vanderbilt.edu