Description

Current BASIL implementation does not allow for Look-Locker processing.

Tasks

• Check automatic label-control order

• Calculate PWI correctly for look-locker order (check if programmed already).

• Fix BASIL ATT quantification settings are wrong #1301

• Fix import of second sessions Correctly convert session 2 in DCM2BIDS #1318

• Fix Philips Look-Locker import Fix import for Philips Look-Locker #1366

• Rebase this branch once all the above is merged.

• Run BIDS validator
  One error, couple of warnings in rawdata.

• Check if M0 is input in BASIL quantification

• Check if PWI4D without averaging works better

• BASIL M0 quantification xASL_quant_M0 for look-locker (asl_calib) https://asl-docs.readthedocs.io/en/latest/aslcalib.html

  • Create M0, based on WM, using asl_calib (FSL)
  • Create .json for that as well
  • Put in nice function
  • Keep code, but use control image as M0 instead of constructed one
    - [ ] Check GM and WM ratio in signal per PLD

• Check where to put the extraction of all control images in the ASL module for Look-Locker/ASL_calib use.
  - [x] cite in the code

• Check motion correction
  Currently skipped

• Check what to skip in xASL_processM0
  - 1. register M0 -> can be skipped for the datasets where a new M0 is created via asl_calib -> look for folder and skip?
  - 2. Quantify M0 -> skip as asl_calib already does that.
  - 3. Masking & smoothing -> not sure?

• UseControlAsM0 automatically added in ASL4D, but should not in this case.

• Comment from Jeroen 1: in the new Basil (oxford_asl in the new fsl 6.0.6.4) there is a -fa option to presumably correct for the Look Look-Locker readout (small flip angle than 90 deg). Currently, we correct the multiPLD label/control images and M0 image signals beforehand. The correction factor per PLD is then based on the PLD value and the flip angle used - inspired by Gunther et al MRM 2001 (https://onlinelibrary.wiley.com/doi/10.1002/mrm.1284) and Petr et al MRM 2013 (https://onlinelibrary.wiley.com/doi/10.1002/mrm.24601). This works remarkebly well, as validated against a single (simple) PLD scan CBF map.

  • Discussed with Jan, correction factors is: PWIcor = PWI / cos ^(n-1) * (flip angle) with n is PLD number according to formula 5 of https://onlinelibrary.wiley.com/doi/10.1002/mrm.1284.
  • cite in the code

• the oxasl code for the Look Look-Locker data - as I also need the option there to set the --aifleadscale (set to very small) otherwise we get a 'funny' looking slice in the arterial arrival time output. According to Martin Craig this apparently has to do with making the model for arterial bolus, normally a boxcar function, a bit less discontinuous at the edges, and help to go smoothly through the inference process for ATT.

• Check single-PLD vs Look-Locker on a composite dataset (Jan will put it to flavors).

• Look-Locker import at the moment creates wrong number of averages
  280 acquired pairs, what should be 560/8/2=35

• Printing stats error

  'Index in position 1 exceeds array bounds (must not exceed 132).

  Error in xASL_stat_PrintStats>xASL_stat_PrintStats_FillStatCellArray (line 390)
  statCell{rowNum,iCell} = xASL_num2str(x.S.DAT(iSubjectSession_DAT, iPrint));

How to test

Optional: insert description about how to test the code changes here

Release notes

Required: summarize the changes for the release notes here