Hi Martin,

When I try installing the latest oxasl version into the fslpython environment, I get the following error:
ModuleNotFoundError: No module named 'Cython'

Installing the Cython module does not help (cython in /Library/Python/2.7/site-packages (0.29.23))

This happens both on my own computer and on the FMRIB jalapeño software. Would you have any tips on how to get around this issue?

Many thanks

From Winok::

ERROR: 0-dimensional array given. Array must be at least two-dimensional
Traceback (most recent call last):
  File "/home/fs0/winokl/Programmes/oxasl/oxasl/oxford_asl.py", line 160, in main
    oxasl(wsp)
  File "/home/fs0/winokl/Programmes/oxasl/oxasl/oxford_asl.py", line 180, in oxasl
    oxasl_preproc(wsp)
  File "/home/fs0/winokl/Programmes/oxasl/oxasl/oxford_asl.py", line 249, in oxasl_preproc
    calib.calculate_m0(wsp)
  File "/home/fs0/winokl/Programmes/oxasl/oxasl/calib.py", line 66, in calculate_m0
    wsp.calibration.m0 = get_m0_refregion(wsp)
  File "/home/fs0/winokl/Programmes/oxasl/oxasl/calib.py", line 440, in get_m0_refregion
    get_tissrefmask(wsp)
  File "/home/fs0/winokl/Programmes/oxasl/oxasl/calib.py", line 617, in get_tissrefmask
    reg.reg_struc2std(wsp)
  File "/home/fs0/winokl/Programmes/oxasl/oxasl/reg.py", line 234, in reg_struc2std
    wsp.reg.std2struc = np.linalg.inv(wsp.reg.struc2std)
  File "/opt/fmrib/fsltmp/fsl_3a805695/fslpython/envs/fslpython/lib/python3.6/site-packages/numpy/linalg/linalg.py", line 522, in inv
    _assertRankAtLeast2(a)
  File "/opt/fmrib/fsltmp/fsl_3a805695/fslpython/envs/fslpython/lib/python3.6/site-packages/numpy/linalg/linalg.py", line 201, in _assertRankAtLeast2
    'at least two-dimensional' % a.ndim)
numpy.linalg.linalg.LinAlgError: 0-dimensional array given. Array must be at least two-dimensional

While the output seems pretty close the regenerated mask is significantly different. Need to compare OXASL with Flora's updated oxford_asl code to see what is causing this. To my eye the OXASL mask looks better but this may not be the case!

• OXASL
  

• oxford_asl
  

np.int etc are now deprecated. Definitely crashing as of np 1.24.2.
https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations

See below image from data set which had a handful of nan values in the output. It displayed ok in fsleyes but the report looks bad.

Need to ignore or remove nan values from data sets when doing reporting (and maybe zero them out when outputting as well?)

Currently just have --debug. Would like:

• --log-fslcmd - log the command line for each FSL command
• --log-fslout - log the standard output from each FSL command

--debug would still exist and set full output logging and full data output.

It is not obvious if partial volume correction is valid for VE data since we are doing modelling on a data set where only part of the brain contains a perfusion signal.

We should run some tests on this and see if it makes sense. If it doesn't then the option should generate a warning/error if it is used with a VE data set.

This is related to the difficulty in obtaining a PWI prior to vessel decoding. The PWI is required for EPI_REG.

However since the encoding cycles typically include a 'control all' and 'label all' image, it should be possible to implement this for VE data.

From Tom. Not completely clear what this is about but to investigate:

"Somewhat confusingly, oxasl with pvec outputs arrival time maps anyway with the default aslrest values (0.7 and 1.0 respectively). Had me pretty concerned for a while trying to work out why it was ignoring my commands, but it seems to simply chuck them out by default. The logfiles for basil do at least show the right numbers."

Hi all,

I was using the oxasl with FSL version 6.0.1 and it was running well. Last week I updated my FSL to 6.0.4 and when using oxasl (or oxford_asl), the Step 2 of BASIL analysis is taking too much time, so much that I cant finish (one hour for 1%). Here is the print of my try. I am using Ubuntu 18.04.

Did it happen with anybody else? What should I do to solve it?

I already tried to re-install my FSL as well as the oxasl.

The structural space mask seems to be transformed but not binarized:

Generating HTML report
usage: oxasl [OPTIONS] SOURCEDIR OUTPUTDIR [FILENAMES...]
oxasl: error: cannot find files ['/Users/tokell/Data/7T_ASL_Tests/2019-06-13- 
Subject_MPPCASL_with_multi_PLD/MPPCASL_HighRes_oxasl_phase_prior_prec0.01_output/report’]

Reproduced twice on Mac. Does not happen on Linux.

Hi Martin,

I'm trying to run a multiphase data on oxasl. In the documentation (https://oxasl.readthedocs.io/en/latest/example_mp.html) there is the option for --phases=

However, when I am trying to run it, I'm finding this error: oxasl: error: no such option: --phases

I also couldn't find the option --phases when looking at the oxasl help. Is this option not available anymore?

Best,
Andre

e.g. --infert1

The most common ones should be supported as CL options. Also, there should be provision for a file containing extra options so we don't have to add every possible model option to the list of valid CL arguments.

Report generation failed due to not having sphinx and rst2pdf. Given this runs by default, should these not be listed as requirements?

Before doing model fitting, oxasl reorders and performs label-control subtraction on the ASL data. This means the output modelfit is unlikely to 'match' the input ASL data.

We should reorder the modelfit data before output and also provide a subtracted version of the input data for direct comparison.

This option is not work for roi extraction from processed asl data.
I checked codes in oxford_asl and region_analysis. It looks like not enabled and involved this option to oxford_asl.
Do you have any plan to include this code in oxasl or do you have any solution for this?

Thanks,
Huijin

Should automatically save the stdout log to the output directory. This is pretty easy for the user to do anyway by redirection but helpful to do it especially as there is useful info in there (e.g. M0 values from calibration).