Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
    janfpl
    @janfpl
    amap.PNG
    Adam Tyson
    @adamltyson
    I didn’t mean measure it, if you view it overlaid on top of the downsampled.nii file, you can see if it’s scaled appropriately.
    janfpl
    @janfpl
    So this would show the boundaries before the freeform transformation? I'm not familiar with how to work with .nii files, is there a way to do this with cellfinder/amap?
    Adam Tyson
    @adamltyson
    This will show the registered data itself (ie the atlas template image warped to your data). You should be able to open the nii image in FIJI or in Napari with the right plugins (this will have been installed with amap).
    If you open the downsampled image, then overlay the affine registered image, you should see an image that’s roughly aligned with your data.
    The free form registered image will then be the properly (or not in your case) aligned image.
    If the affine image is of the right size, and roughly aligned with your brain, then the problem probably lies in the free form step.
    I’d like get this fixed though, so I’m happy to have a quick chat on zoom, or if you can send me some data I’ll take a look.
    janfpl
    @janfpl
    nii overlay.PNG
    The affine_registered_atlas_brain.nii (magenta) overlays fairly well over the downsampled image (green).
    I'll send you a link to the raw data privately. Thanks for your help!
    Adam Tyson
    @adamltyson
    Great, I'll take a look. That affine registration looks pretty good, so hopefully we can tweak the freeform step.
    Chouchi974
    @Chouchi974
    Hi !
    I have a problem regarding the training of the network.
    When I use the curation command, this line appears just after my command :
    WARNING: Cannot mix incompatible Qt library (version 0x50c03) with this library (version 0x50c09)
    Do you have any idea how to fix it ?
    Thanks for your help !
    6 replies
    Adam Tyson
    @adamltyson
    brainreg.png
    Hi @janfpl, I had a quick look at your data, and it seems like some of the registration problems might be due to the data itself (damage to the front of the brain, what looks like stitching artefacts, and some bright signal outside of the brain). I'll have a play around to see if I can improve it, but interestingly, using the 25um resolution atlas in brainreg (https://github.com/brainglobe/brainreg) seems to look a bit better than the image you sent (see above).
    It's still not great though.
    If you want to try out brainreg, this is the command I used:
    brainreg /path/to/data /path/to/output -x 3.6498 -y 3.6498 -z 15.3767 --orientation lsp --atlas allen_mouse_25um
    janfpl
    @janfpl
    Awesome! I’ll try it out and let you know how it works.
    adam-romanski
    @adam-romanski

    Hi!
    I am having a problem with registering a brain for a current project. Right before the Debug process starts to run I receive this error:

    Traceback (most recent call last):
    File "c:\users\administrator\miniconda3\envs\cellfinder\lib\concurrent\futures\process.py", line 239, in _process_worker
    r = call_item.fn(call_item.args, *call_item.kwargs)
    File "c:\users\administrator\miniconda3\envs\cellfinder\lib\site-packages\brainio\brainio.py", line 465, in load_from_paths_sequence
    return volume
    UnboundLocalError: local variable 'volume' referenced before assignment

    This error only occurs with this one brain. All other brains I register succeed. Thank you for your help!

    Adam Tyson
    @adamltyson
    @adam-romanski is there anything different about this brain? Number of planes, naming?
    adam-romanski
    @adam-romanski
    @adamltyson All of the brains I register have a variable number of planes. I have not encountered this issue with brains that have a higher or lower number of planes. All of our brains are named in the same file format (#_background, #_cells, such as 165_background)
    Adam Tyson
    @adamltyson
    Could you try moving one of the images out of the directory, and also trying to set —n-free-cpus to a differrnt number? There may be some weird issue with the nunber of planes and the number of cpu cores for loading the data in parallel.
    If it still doesnt work, if you can share the data with me, ill take a look.
    adam-romanski
    @adam-romanski
    @adamltyson Setting --n-free-cpus to a different number seemed to work. Thank you for your help!
    Chouchi974
    @Chouchi974
    HI Adam. I just have a question regarding the generation of the training data. there are a lot of cells that are considerate non-cells by the classification, so I think I have to retrain the network properly. But do I have to generate these training data for the whole brain to generate the Yaml files ? Or I only can do the correction for some plane ? Thanks for your help
    Adam Tyson
    @adamltyson
    Hi @Chouchi974, I've answered your question over at https://gitter.im/cellfinder/celldetection, so that others can find the answer in the right place.