SliceTiming for GE Diffusion

[mr-jaemin]

Although GE EPI diffusion "epi2" does not store slice timing information in DICOM header, the slice timing information can be calculated using this method/code used in fMRI;

The provided C source code slicetime.cpp demonstrates one method. Here, the DICOM header is read to determine the multi-band factor , number of slices per volume , sampling rate , sequential or interleaved order , ascending or descending order , software release (versions prior to 27.0 R3 used different mulit-band slice order) and group delay .

The diffusion EPI "epi2" is a special case of this approach:

• Interleaved=true
• is27r3=false (i.e. if #slices/MB is an even integer, add extra MB excitation so that #slices/MB = odd)
• group delay=0

Exception:
Three configurations (Premier, UHP, 7T) uses diffusion gradient cycling for thermal management. There are different cycling modes in product PSD:

• All-TR (default): one diffusion volume collected over all-TR
  -- CV12 (0019,10B3) UserData12 = 1
• 2-TR: one diffusion volume collected over 2-TR
  -- CV12 (0019,10B3) UserData12 = 2
• 3-TR: one diffusion volume collected over 3-TR
  -- CV12 (0019,10B3) UserData12 = 3
• Special OFF: one diffusion volume collected over 1-TR
  -- For 4 popular multi-site studies: ABCD, ADNI3-Adv, HCP-lifespan, UK BioBank
  -- CV15 (0019,10B6) UserData15 = 0.72

I would propose the following approach to generate SliceTiming for GE Diffusion:

if `(0008,1090)` ManufacturersModelName == Premier or UHP or 7T
   if CV15 `(0019,10B6)` == 0.72
     Cycling Special OFF
   else if CV12 `(0019,10B3)` == 2
     Cycling 2TR
   else if CV12 `(0019,10B3)` == 3
     Cycling 2TR
   else
      Cycling ALL-TR
else
     Cycling OFF

To accommodate for/support the following scenarios:

• If user manually turns off the diffusion gradient cycling (for Premier, UHP, 7T) by changing research CV (or custom PSD), DICOM header doesn't store this information.
• Also, user-custom PSD may use CV12 for other purpose beside controlling cycling mode.
  I would propose the (undocumented) program option(--diffCyclingModeGE) to specify/override cycling mode

I will submit a pull requests for your review.

[neurolabusc]

Thanks for your PR

[mr-jaemin]

v1.0.20220915+

[mharms]

@mr-jaemin: Just wanted to mention that this PR is reporting values for the NumberOfDiffusionT2GE parameter for the "Calibration" scans, and that plus the NumberOfDiffusionDirectionGE variable for the research/ABCD/mprage_promo sequence (under DV26). So, the fields queried for those two variables are apparently not unique to the diffusion sequence (and presumably have a different interpretation in those other sequences).

[mr-jaemin]

@mharms Thanks for catching and reporting this issue. You are right: the NumberOfDiffusionT2GE is from UserData23 (0019,10DF) and the NumberOfDiffusionDirectionGE is from UserData24 (0019,10E0). The UserData have application specific interpretation. So, I will submit PR to limit them to GE diffusion only.

[mr-jaemin]

@mharms Could you please try the fix, now in the development branch?

[mharms]

The minor issue that I raised here has been resolved (on the particular data I was testing) with #645.