updates 2016-05 May - July

Owned by Paul Macey (admin)
Last updated: Aug 02, 2016
4 min read
2016-07-30
  • Tweak to new visualization - to avoid warning messages (might work)
2016-07-29
  • Tidied messages for studies with no MRI's (when loading sets)
  • Smoothed display for T1 processing (removed clutter, fixed spelling mistake)
  • Fixed a T1 preprocessing bug
  • Tidied display GUI to disable atlases that do not work
2016-07-28
  • Fixed missing "Sinc" ASL option
  • Tweak for T1, still not ready
  • Fixed Anatomical atlas bug (may be more - testing now)
2016-07-25
  • Fixed DTI indices error from recent change (21st)
2016-07-22
  • Option to view DTI files from single study, access from Stats GUI; only does DTI, but will add others over time

....

2016-07-21
  • More tweaks to FIRST - may display a message to run command from Linux
  • Fixed bug from yesterday's preprocessing fix
  • Fixed free water bug - was using mm/2, should have been using m/2 for water diffusion (need to recalculate FW maps).
2016-07-20
  • Fixed preprocessing issue that could sometimnes re-preprocess the T1's. This was due to the time of parameters being saved being later than the time they were created with manual realignment. 
  • Adjusted physiology RMANOVA, fMRI RMANOVA
2016-07-19
  • Fixed fMRI temporal smooth bug
2016-07-17
  • Tried more checking on FSL FIRST
  • Tweak for physiology
2016-07-14,15
  • Working on T1R, nothing ready yet but some changes
  • Fixed timetrend bugs ("Clear" and improved RMANOVA output for copying to CorelDraw to make figures..
2016-07-11
  • Removed code that will become obsotele in a couple of MATLAB releases.
  • Fixed fMRI timetrend "Plot Combined" error
  • Fixed error for B1 maps in Philips DICOM conversion (via Rania); however, get large number of series, which is probably incorrect.
  • Fixed ASL bug introduced with previous updates.
    • Note: correcting DTI-to-T1 registration issue below corrected some registration errors.

IXI: 386 subjects had DTI scans; 8 had DTI calculation errors (tensors not producing recognisation FA maps). Of the remaining, 20 were not coregistered with T1 originally; fixing rigid-body shift error below resulted in 15/20 resolving the coregistration issue, leaving 5/378 (1.3%) with remaining failure to match T1. See QC details.

2016-07-08
  • !! Fixed bug in match-to-T1 for DTI, and possibly fMRI and ASL. This could lead to poor T1 matching, and hence poor normalization. There was a change in SPM12 from SPM8 in the parameter names for rigid-body registration before unified segmentation; this meant that there was a bug with no rigid body normalization first, so if DTI images were shifted significantly from T1's, the warping to T1 could fail, and lead to poor MNI normalization. This seemed to affect ~5% of subjects in one dataset I tested (IXI)
    • You can (and should) always check for accurate normalization in the "summary" check reg files.
  • Fixed T1/VBM bug - VBM option would overwrite a normalized gray matter file, which causes errors when doing the summary maps.
  • Fixed overwrite option for anatomical - works now for T1 processing.
2016-07-06
  • Added option to match timetrend colors to strings, with settings recorded in database. 
  • Fixed bug with "Setup T1".
2016-07-05
  • Added option in VOI timetrend analysis to select reference for differences; also have option to only show differences.
  • Fixed resize errors with project configuration
  • Added option to convert T1 to FreeSurfer-compatible format (*.nii, no NaN's)

  • Fixed CSPM analysis bug introduced with ASL subtraction option.
2016-07-01
  • CSPM should allow for subtraction type.... untested!
  • ASL GUI now allows option of subtraction type - simple, surround, sinc.
2016-06-15
  • Try to catch FSL FIRST error - weird one where it misses seeing a newly-created file.
2016-06-14
  • Working on changes to bring in "metrics" for survey data, and other one-off measures
2016-05-31
  • Added extra summary files: creates masks of voxels where ALL subjects are non-zero, and where ANY subject is non zero. This can be useful for ASL, where there may be regions where some subjects have missing values. This will also show up if one subject is way off -  the "ALL" will be missing a lot. Below is an example from gray matter in 78 subjects:

  • Fixed ASL stats bug - would always take run stats on first option, even if 2nd or later was selected.
2016-05-27
  • Fixed cspm_im_mean, cspm_im_median to run on 4D nifti files.
  • Fixed ASL bug where it was not masking - masking now works (runs by default).
  • Fixed ASLtbx bug whereby it returned incorrect mean map name
  • Added ASL option to turn pre-smoothing on or off. *** New analyses will show pre-smoothing; old ones will look as if no presmoothing. 
  • Option to print out current SPM clusters - useful for creating table of clusters:

2016-05-02
  • Correct to FIRST running design matrix for no significant results.