Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • May 13 14:01

    adamltyson on master

    update url (compare)

  • May 13 11:11
    adamltyson commented #182
  • May 13 10:48
    atranvan commented #182
  • May 13 09:33
    adamltyson commented #182
  • May 13 09:33
    adamltyson commented #182
  • May 13 09:25
    atranvan labeled #182
  • May 13 09:25
    atranvan assigned #182
  • May 13 09:25
    atranvan opened #182
  • May 10 13:28
    adamltyson closed #181
  • May 10 13:13
    adamltyson commented #181
  • May 10 11:38
    Marilynb24 commented #181
  • May 07 17:22
    adamltyson commented #181
  • May 07 14:41

    adamltyson on master

    Update README.md (compare)

  • May 07 14:41

    adamltyson on master

    Update README.md (compare)

  • May 07 07:39
    Marilynb24 labeled #181
  • May 07 07:39
    Marilynb24 assigned #181
  • May 07 07:39
    Marilynb24 opened #181
  • May 06 11:25
    adamltyson labeled #180
  • May 06 11:25
    adamltyson assigned #180
  • May 06 11:25
    adamltyson opened #180
cindy12-gao
@cindy12-gao
Capture.JPG
this is the image properties show in the fiji.so i definite the -v 1 3 0.812 orientation air
cindy12-gao
@cindy12-gao
actually, im not sure whether its right or not
1 reply
cindy12-gao
@cindy12-gao
hello,Adam. Ive trained my own model, but I dont know how to apply it in the cell detection. Is the command like this right?--cellfinder -s D:\coronal\C0 -b D:\coronal\C1 -o D:\coronal\output8 -v 1 3 0.812 --orientation air --no-register --trained-model D:\coronal\output7\points\training\model-epoch.47-loss-0.661.h5
Adam Tyson
@adamltyson
Yep that looks about right.
cindy12-gao
@cindy12-gao
do I need change the parameter of cell detection, like --soma-diameter, --threshold and so on?
Adam Tyson
@adamltyson
Not unless you want to.
cindy12-gao
@cindy12-gao
after I used the trained model, the detection accuracy seems have no improve.
and how to check the detected spots is right, which the ring definitely lay on the signal spots? now I use cellfinder_curate to show the detected output. I want a figure like this. how to do it? thank you~
cindy12-gao
@cindy12-gao
image.png
Adam Tyson
@adamltyson
That figure was made with a napari script, not a specific tool in cellfinder. I can share specifically how to do it, but i won’t be back at work until January.
cindy12-gao
@cindy12-gao
all right. so I can try make this figure by matlab. but hope for your script! re-confirm it - if I want to detect the signal of whole brain, I just need the command----cellfinder -s D:\coronal\C0 -b D:\coronal\C1 -o D:\coronal\output9 -v 1 3 0.812 --orientation air --trained-model D:\coronal\output7\points\training\model-epoch.47-loss-0.661.h5 .. does it can detect the cell refer to my own model?
Adam Tyson
@adamltyson
The syntax looks fine, but that model may not work too well. Does it have the lowest loss value of all the saved models?
cindy12-gao
@cindy12-gao
yep, the accuracy value kept on 0.9868
Jan C. Frankowski
@janfpl
Hi Adam,
Is there a way to use brainreg-segment to manually count cells and for visualization/quantification?
Adam Tyson
@adamltyson
There is, but it’s not yet released/advertised. Could you maybe raise a GitHub issue (ideally on the cellfinder repo) outlining what you would like to do, and I’ll see if I can release something that would work for you once I’m back from holiday (next week).
Nastacia Goodwin
@goodwinnastacia
cellfinder.png
hello, I'm working on pip installing cellfinder in a python 3.7 anaconda environment and am running into these errors and was wondering if you have any advice
Adam Tyson
@adamltyson
Are you just running pip install cellfinder or do you have any version downloaded from github etc?
6 replies
Adam Tyson
@adamltyson
FYI - from version 0.4.8, cellfinder now requires CUDA 11.0.
stasys-hub
@stasys-hub
Hello, Cellfinder seems to be an awesome softwearesuite. Thank you for provinding with such awesome software! I am tying to perform a testrun on a dataset we acquired. The Data was processed with Idisco, clearing and light-sheet microscopy. For the testrun i just wanted to see whether cellfinder will run through. Unfirtunately i get an error in the inference step, saying:
ValueError: could not broadcast input array from shape (5113,2114) into shape (5119,2116)
So it seems some dimensions do not match. Do you have a clue why this happens? And if yes, what can I do to fix this.
Many thanks in advance
5 replies
stasys-hub
@stasys-hub
Transformation artefact.png
Hello Adam! We able to fix the dimensions problem using some preprocessing in python. Cellfinder was able run through without any errors.
Now we have encountered another problem. In the screenshot attached you can see a distortion in hippocampus area ('downsampled_channel_0' on the right and 'downsampled_standart_channel_0' on the left). There are also other distortions throughout the brain.
Maybe it's because of one hemisphere missing? We are imaging on a light-sheet microscope, in saggital oientation, right side up (iDISCO+ clearing). It seems the resampling worked fine. And we also see that some cells were detected, mostly only in the right hemisphere. On the screenshot you see the sample on the left, which is weird, but the images are probably mirrored, correct?
We haven't trained the network on our data yet, this is the next step.
Is there a way to make cell finder work with only one hemisphere? And is/will there be official support for one hemisphere?
Thank you in advance for your help!
3 replies
stasys-hub
@stasys-hub
Thank you very much for your detailed answer! This sounds really promising. I hope it will get implemented soon.
Rob Campbell
@raacampbell
I have a question about how many channels are needed. Say we have a rabies tracing experiment with labelled cells in red and starters in green. Is acquiring red and green sufficient or do we also need blue? So far I've been doing the latter.
2 replies
Venus Sherathiya
@venus-sherathiya
Hello, If the images are taken in horizontal plane, cellfinder can still do atlas registration. Right ?
Screen Shot 2021-03-01 at 3.28.32 PM.png
Also, when I am trying to open the brain-reg output folder in Napari viewer for your sample data. I am getting error as shown in the screenshot.
3 replies
wudustan
@wudustan

Has anyone here worked with CLARITY tissue cleared samples and Cellfinder?

I'm also looking for support on masking one hemisphere. Taking the 25um mouse atlas as an example, the directory contains the image .tiff files and then an atlas of the structures and a folder of meshes. If I seek to mask the right hemisphere manually, then what should I do with the structure .csv list and the meshes? Do these need to be edited too?

Screenshot 2021-03-05 at 07.02.26.png
Adam Tyson
@adamltyson
I think there are some labs that have used it with CLARITY cleared tissue, but I don’t know how well it worked yet.
Support for a hemisphere is better added into the registration software itself, rather than editing the atlas. This is something I’ve been meaning to work on for a while. If you want to give it a go, I’m happy to help. If you leave a comment on the existing GitHub issue, we can discuss. I’ve got a good idea about how it can be done. See here: brainglobe/brainreg#6
Pasha Davoudian
@davoudian
has anyone had success running cell candidate identification/detection on macOS? I know its not fully supported technically, but thought some people may have tried it out. I'm able to register brain (both my own and test data) just fine, but cell finder seems to be hanging on cell identification with something related to python's multiprocessing code.
41 replies
Rob Campbell
@raacampbell
We are soon going to try cellfinder on SHIELD cleared brains imaged on a mesoSPIM, so will post results here. I'm sure the outcome will be similar to CLARITY and how well it works will surely be influenced by clearing quality. I expect registration to the atlas to be much more challenging, as these samples deform inhomogenously.
Adam Tyson
@adamltyson
Has anyone used cellfinder on an AMD (i.e. not Intel) CPU?
John Tukker
@jjtukker_gitlab
hi Adam, thanks a lot for sharing this great resource! I just installed it on our Linux compute server (openSUSE Xfce v4.14), and I was wondering if you could provide any tips on how to use CellFinder with many CPUs. Is there a way to divide the work up over several nodes? Unfortunately, our GPU (Nvidia RTX2080) is currently not working, after a recent OS version update (not under my control), so I was hoping as an alternative to at least do some parallel work over several CPU nodes (may also useful for others working without GPU but with many cores/nodes).
3 replies
John Tukker
@jjtukker_gitlab
(cellfinder) john@alz1:/alzheimer/AnatomyAnalysis/CellFinder/test_brain/outputfull$ napari
WARNING: X server does not support XInput 2
12:28:29 WARNING X server does not support XInput 2
WARNING: QXcbConnection: XCB error: 1 (BadRequest), sequence: 169, resource id: 213, major code: 130 (Unknown), minor code: 47
12:28:29 WARNING QXcbConnection: XCB error: 1 (BadRequest), sequence: 169, resource id: 213, major code: 130 (Unknown), minor code: 47
WARNING: Application calling GLX 1.3 function "glXCreatePbuffer" when GLX 1.3 is not supported! This is an application bug!
12:28:32 WARNING Could not import pyopencl. Falling back to scipy for CPU affine transforms
Loading cellfinder directory
Loading brainreg directory
allen mouse atlas (res. 10um)
From: http://www.brain-map.org (Wang et al 2020, https://doi.org/10.1016/j.cell.2020.04.007 )
Found additional downsampled image: downsampled_channel_0.tiff, adding to viewer
WARNING: QXcbConnection: XCB error: 3 (BadWindow), sequence: 41139, resource id: 6300687, major code: 40 (TranslateCoords), minor code: 0
16:16:22 WARNING QXcbConnection: XCB error: 3 (BadWindow), sequence: 41139, resource id: 6300687, major code: 40 (TranslateCoords), minor code: 0
8 replies
Hugo McGurran
@hugomcg_twitter

Hey Adam,
I have run into an issue when running cellfinder on a serial 2p/brainsaw dataset, and it will run for ~ 10 minutes and load the images, but it runs into the error:

File "/home/hugo/miniconda3/envs/cellfinder/lib/python3.8/site-packages/tifffile/tifffile.py", line 2904, in init
raise TiffFileError('not a TIFF file')
tifffile.tifffile.TiffFileError: not a TIFF file

Is this something you have ever ran across? Would appreciate any input,
Cheers

6 replies
Hugo McGurran
@hugomcg_twitter

Hi Adam,
I have come across a problem during the 'cell positioning' step of cellfinder:

INFO - MainProcess main.py:207 - Analysing cell positions
Traceback (most recent call last):
File "/home/hugo/miniconda3/envs/cellfinder2/bin/cellfinder", line 8, in <module>
sys.exit(main())
File "/home/hugo/miniconda3/envs/cellfinder2/lib/python3.8/site-packages/cellfinder/main.py", line 88, in main
run_all(args, what_to_run, atlas)
File "/home/hugo/miniconda3/envs/cellfinder2/lib/python3.8/site-packages/cellfinder/main.py", line 208, in run_all
analyse.run(args, atlas, downsampled_space)
File "/home/hugo/miniconda3/envs/cellfinder2/lib/python3.8/site-packages/cellfinder/analyse/analyse.py", line 233, in run
run_analysis(
File "/home/hugo/miniconda3/envs/cellfinder2/lib/python3.8/site-packages/cellfinder/analyse/analyse.py", line 277, in run_analysis
transformed_cells = transform_points_to_atlas_space(
File "/home/hugo/miniconda3/envs/cellfinder2/lib/python3.8/site-packages/cellfinder/analyse/analyse.py", line 178, in transform_points_to_atlas_space
downsampled_points = transform_points_to_downsampled_space(
File "/home/hugo/miniconda3/envs/cellfinder2/lib/python3.8/site-packages/cellfinder/analyse/analyse.py", line 162, in transform_points_to_downsampled_space
points = source_space.map_points_to(target_space, points)
File "/home/hugo/miniconda3/envs/cellfinder2/lib/python3.8/site-packages/bg_space/core.py", line 25, in decorated
return method(spaceconv_instance, space_description, args, *kwargs)
File "/home/hugo/miniconda3/envs/cellfinder2/lib/python3.8/site-packages/bg_space/core.py", line 355, in map_points_to
pts_to_transform = np.insert(pts, 3, np.ones(pts.shape[0]), axis=1)
File "<__array_function__ internals>", line 5, in insert
File "/home/hugo/miniconda3/envs/cellfinder2/lib/python3.8/site-packages/numpy/lib/function_base.py", line 4558, in insert
raise IndexError(
IndexError: index 3 is out of bounds for axis 1 with size 0

It appears to manage with all other steps of cellfinder, have you come across this index issue before?
Cheers

1 reply
Hugo McGurran
@hugomcg_twitter
4 replies
cindy12-gao
@cindy12-gao
Hi Adam,
error log.JPG
Hi Adam, when I used the cellfinder0.4.15, the cell detection still has some trouble
cindy12-gao
@cindy12-gao
and i didn`t have the training model, so the classification trained model is default
1 reply
Guido Meijer
@guidomeijer
Hi all, I'm a new user to cellfinder, thanks so much for developing and maintaining this software! I experienced an issue and solved it so just wanted to let you folks know about it: when pip installing cellfinder and using the napari plugin to curate cells I got an error (TypeError: init() got an unexpected keyword argument 'extract') when clicking 'Save training data'. This was solved by upgrading cellfinder-core from 0.1.1 to 0.2.1. It might be nice to have 0.2.1 installed automatically when pip installing cellfinder.
3 replies
Marilynb24
@Marilynb24
Hi all, I'm also new user to cellfinder, thanks for this software. I've just integrated a lab team who used to work on cellfinder 0.4.3 and I’m trying to install the latest version (0.4.15) on my computer. I don’t have any error message during installation, but when I run the cellfinder command (cellfinder -s test_brain/ch00 -b test_brain/ch01 -o test_brain/output -v 5 2 2 --orientation psl --atlas allen_mouse_10um –start-plane 1000 –end-plane 1500) on your test data and trying to visualize the results on napari, it seems there are some troubles with the cell detection. In fact, I can’t see any cells, or just few markers which don’t correspond to real cells; real cells aren’t recognized. Did I miss some update to run cellfinder with the latest version or does someone know where the issue comes from and how can I fix it? Thanks
5 replies
cindy12-gao
@cindy12-gao
hi Adam, after registered the brain data, there is a profile 'registered atlas' which is 32bit in the register folder. i opened it by Fiji and found that each brain area with different intensity. Do you have a list or a table showing the relationship between the intensity and brain area?
1 reply
cindy12-gao
@cindy12-gao
Thank you! I found that , the number of ID means the intensity value on the atlas map?
1 reply
Adam Tyson
@adamltyson
Hi all. BrainGlobe has joined https://forum.image.sc as a community partner. Going forward, please ask any BrainGlobe questions over there, tagging brainglobe and the tool you're asking about, e.g. cellfinder. This Gitter channel will no longer be monitored. Thanks!