Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Mainak Jas
    @jasmainak
    @sappelhoff sorry for being late to the party. I was asleep :) trans is documented here in the glossary: https://martinos.org/mne/stable/glossary.html. Yes, agree with Eric. The trans gives you mri <-> head whereas hpi is needed for meg <-> head. The head moves during the recording but you can also think of it another way, that the MEG moves and the head is fixed :p So no need to recompute mri <-> head again
    Stefan Appelhoff
    @sappelhoff
    thanks Eric and Mainak!
    what does dev_head_t stand for?

    Wednesday, 2019-06-26

    Mainak Jas
    @jasmainak
    dev_head_t stands for "device to head transform"
    Stefan Appelhoff
    @sappelhoff
    Can I locally build the MNE-Python docs for just a single file?
    Mainak Jas
    @jasmainak
    Yes use $ PATTERN=‘blah.py’ make html-dev_pattern
    The environment variable PATTERN might be different. Check the makefile :)
    Stefan Appelhoff
    @sappelhoff
    cool, thanks
    Stefan Appelhoff
    @sappelhoff
    what is _get_trans actually doing?
    I am currently trying to get the MRI landmarks using trans and raw.info['dig'], and Alex hinted at _get_trans
    in that function there is a dangerous amount of "probably" and "usually" in the comments :-D
    Stefan Appelhoff
    @sappelhoff
    and what does fro_to_t mean?
    Mainak Jas
    @jasmainak
    from to to :)
    the "probably" and "usually" is in almost any coordinate transform code :P
    Stefan Appelhoff
    @sappelhoff
    from to to what? --> and why would you not spell it out then? from_to_to
    it's also a bit weird IMO that you input a trans ... and in return you get a from_to_t (which is actually a trans now) ... and a string variable called trans
    Mainak Jas
    @jasmainak
    the coordinate frame that the trans takes points "from" and transforms "to"
    Stefan Appelhoff
    @sappelhoff
    and that string variable does not hold any transformation, just info from where the from_to_t came from?
    Mainak Jas
    @jasmainak
    it's just a function to homogenize the different ways trans can be handled in MNE
    I agree the code is a bit confusing though
    Stefan Appelhoff
    @sappelhoff
    yes, I think even private functions should have docstrings ;-)
    Eric Larson
    @larsoner
    from_to_t is short for "from" to "to" transform
    no idea why we used fro instead of from, except that fro is often used in the phrase "to and fro"
    Stefan Appelhoff
    @sappelhoff
    and why t instead of to ?

    I think I am slowly making progress ... this progress could be significantly sped up if you took some time looking at my writeup: https://github.com/mne-tools/mne-bids/issues/210#issuecomment-506403864

    @larsoner @jasmainak

    I tried to be as precise as possible to show you my current understanding ... that way you can correct me more easily I hope :)

    Thursday, 2019-06-27

    Mainak Jas
    @jasmainak

    and why t instead of to ?

    https://www.youtube.com/watch?v=TJ8KIzkCAto :)

    I don't think there is a good reason ...
    Eric Larson
    @larsoner
    the t stands for transform not to
    like dev_head_t is dev device _ to head head _t transform
    Stefan Appelhoff
    @sappelhoff

    and why t instead of to ?

    https://www.youtube.com/watch?v=TJ8KIzkCAto :)

    :laughing:

    Stefan Appelhoff
    @sappelhoff

    @jasmainak if you have some time, please make an initial review of mne-tools/mne-bids#211

    I also need feedback for the problem I encountered: https://github.com/mne-tools/mne-bids/pull/211#issuecomment-506707030

    After that, I can add tests, and a function that gets trans using fit_matched_points

    Alexandre Gramfort
    @agramfort
    @sappelhoff how did you test your process is correct?
    do you ahve some data with fiducial points in MRI coord from which you could derive a trans file/object?
    do you know about the vox2ras transform?
    you'll need this to put the surface RAS coord to vox space
    Stefan Appelhoff
    @sappelhoff

    @sappelhoff how did you test your process is correct?

    no tests in place (yet) ... currently just by eyeballing and seeing that it's quite close with a small error

    do you know about the vox2ras transform?

    I don't ... but it sounds intriguing ... perhaps I don't have to make another PR to BIDS-specification after all :) Can you point me to it?

    Alexandre Gramfort
    @agramfort
    @jasmainak you know about this right ?
    It must be used in 3D orthoviewer
    Grep for vox2ras or ras2vox in mne source code
    Mainak Jas
    @jasmainak
    gosh, I always winged coordinate systems :-) @sappelhoff see my comment on the PR
    Alexandre Gramfort
    @agramfort
    Nice blog post
    Alexandre Gramfort
    @agramfort
    @sappelhoff does the link above help?
    Stefan Appelhoff
    @sappelhoff
    starting to look at it now!
    I think the issue is that BIDS wants the coords in voxels