Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
    tasmiaorni
    @tasmiaorni
    Now I have installed histomicsTK fully and it's running with the code provided in the documentation link.
    But when I am adding the last part i.e. Nuclei segmentation I get this error.
    @manthey Can you please help me in this issue
    image.png
    Deepak Roy Chittajallu
    @cdeepakroy
    Looks like you put np.int64_t there
    can you use np.int32 as in the master version here
    tasmiaorni
    @tasmiaorni
    yes I did
    image.png
    Deepak Roy Chittajallu
    @cdeepakroy
    This is the error you reported previously. Did you try changing all memoryviews (int[:, :] im_fgnd_mask) to explicit declarations like "np.ndarray[np.int32_t, ndim=2] im_fgnd_mask" in https://github.com/DigitalSlideArchive/HistomicsTK/blob/master/histomicstk/segmentation/nuclear/_max_clustering_cython.pyx#L8
    For some reason, memoryviews don't seem to work in your environment
    Here is a brief tutorial on Cython memory views for some background -- https://cython.readthedocs.io/en/latest/src/userguide/memoryviews.html
    tasmiaorni
    @tasmiaorni
    I changed this :
    image.png
    But still showing some error:
    image.png
    Nelson Tsaku
    @Tsakunelson
    Hey @tasmiaorni did you finally get nuclei segmentation with HistomicsTK working?
    AgrawalAman
    @AgrawalAman
    was anyone able to install on a macos/ubuntu? I am getting the cmake build errors specified in the issues
    krushnavadan
    @kvbids_gitlab
    Hi,
    i wants to generate button in openseadragon viewer for actual zoom level , 5x,10x 15x 20x
    is there any opensource code i can find
    darkmich
    @darkmich
    Are there any good places to get started looking at customization? I’d like to have a particular collection require specific metadata and have that be searchable. I’ve seen some oblique references, but nothing that really spells it out (unless I’m just dense!). Thanks!
    David Manthey
    @manthey
    kvbids_gitlab, This HistomicsTK source code shows how to set actual zoom level with the geojs viewer. Openseadragon has widgets that show scale, so I assume there are widgets to do what you want, but we haven't integrated any of them into Girder or HistomicsTK.
    darkmich, generally, custom metadata searches work best if you write an endpoint that does what you want. That way, if your data would benefit from an index you can ensure the index exists, and you can scope the search to whatever you need. There are plugins that add some search -- for instance there is two branches on the digital_slide_archive that do different metadata searches in different ways ...
    When you say you'd like to have a collection require particular metadata, how would you like that enforced? For instance, would you want it to reject adding items if they don't have the correct metadata? Or just not search such items?
    krushnavadan
    @kvbids_gitlab
    Thanks David and darkmich , i will read more and try to implement
    Komal
    @komzy
    Hi there. Do you provide an offline GUI for visualization and annotations with your python tool kit?
    Komal
    @komzy
    I was going through your repository and it seems that you have a web based one. Is there any reason as to why you opted to make a web based UI? and is it possible to make non-webbased python gui for WSI visualization using your toolkit?
    David Manthey
    @manthey
    You can always use a web browser offline to view annotations. Fundamentally, we wanted to be able have multiple users collaborate on creating annotations and viewing images, so the UI has used javascript and html. The restful python library can be used without the UI (via direct python or via REST calls), but since we can use the existing UI online or offline, we haven't had any need to make it another way.
    Komal
    @komzy
    ok thanks for the reply.
    alasla
    @alasla
    Hey. I’m having a bit of trouble wrangling out versions of things for HistomicsTK. I can get girder3 and large_image up and running, but it appears histomicsTK is only for girder 2.5 ? Is that correct?
    David Manthey
    @manthey
    Yes. We probably won't convert HistomicsTK to Girder 3 for another month or so.
    alasla
    @alasla
    thank you @manthey - that is helpful to know.
    @manthey I see you’ve some commits to large_image as well. Do you have a suggestion for a branch or tag I should use to get large_image running with histomicsTK in girder2.5 ?
    David Manthey
    @manthey
    large_image master is still the Girder 2.x branch (and works with HistomicsTK). There is a large_image girder-3 branch. I expect that in the near future (a week or two), that the girder-3 branch will become the new master and there will be a long-standing grider 2.x branch for maintenance.
    alasla
    @alasla
    do you have any idea what the versions/branches/tags of things should be (girder/large_image/histomicsTK) ?
    girder-large-image-1.0.0-dev52 from pypi requires girder3,
    Large_image master on git is 0.3.0, but doesn’t seem to satisfy HistomicsTK which requires things like 'large-image-source-openslide’, which aren’t the way 0.3.0 are organized.
    David Manthey
    @manthey
    HistomicsTK has instructions on installation: https://digitalslidearchive.github.io/HistomicsTK/installation.html . The toolkit is pip installable. The Girder 2.x plugin is not. The easiest way to install it would be via Docker or Vagrant.
    zzzghm
    @zzzghm
    Hey, I had a web-viewer based on OpenSeadragon, and I would like to combine the viewer with HistomicsTK and girder to better analysis and manage the whole slide images. I am wondering where I can find the lower level code that called the OpenSeadragon as a function that I can modify, or where I can put that as a plugin, in that case, I can start from there to combine our viewer with Girder. Please see the demo as https://zzzghm.github.io Thank you!
    Pete Lawson
    @geekypete
    I am attempting to upload an Aperio Imagescope annotation .xml file, associate it with a large_image file, and display the image, with imported annotation, in the HistomicsTK viewer. Is this possible? I know there is a POST /item/{id}/aperio API command to import an item (.xml) as an Aperio annotation and associate it with an image_id but when I go to view the annotations on HistomicsTK after doing so, nothing is present.
    David Manthey
    @manthey
    Looking at the source (in the digital_slide_archive plugin), when you import an aperio annotation, it adds information to the Girder item document under the aperio tag.
    This is searchable on specific items via the GET /item/{id}/aperio endpoint, or across items via mongo queries. It isn't exposed on the HistomicsTK interface.
    Pete Lawson
    @geekypete
    @manthey Thank you, that clarifies things. I ended up writing a function to convert Aperio xml annotations to a compliant JSON annotation which does the trick.
    Pete Lawson
    @geekypete
    Is there a good tutorial for extending the functionality of the HistomicsTK web-viewer? I understand we need to use the slicer-cli-web plugin as a template for building in CLI analysis functionality to Girder and linking that to the web interface, but I don't really understand where we are dropping in our custom Python code, and how we build the modified slicer-cli-web plugin into a Docker image, nor how we are supposed to load new plugins. An end-to-end tutorial would be very helpful to more naive users (such as myself) that would like to extend the functionality of the HistomicsTK web interface for their research.
    David Manthey
    @manthey
    There is a sample on making a CLI analysis plugin here: https://github.com/cdeepakroy/slicer_cli_web_plugin .
    @cooperlab Didn't we also have another document explaining how to do this?
    Pete Lawson
    @geekypete
    @manthey Thank you that was helpful, its still a little opaque how to build a new plugin from scratch, but this is a good start. I see that the example is in Python2, is that a requirement for plugins to HistomicsTK and or can I build them out in Python3?
    David Manthey
    @manthey
    You can use Python 3. We test with 2.7, 3.5, 3.6, and 3.7.
    Yu Qin
    @demiqin_gitlab
    Can you tell me What's the best way to add new analyses function of HistomicsTK plugin and update via Docker? And why the test_data folder is empty when we export HISTOMICS_TESTDATA_FOLDER?
    David Manthey
    @manthey
    There is an example to create a new docker analysis plugin here: https://github.com/cdeepakroy/slicer_cli_web_plugin . There is some information about installing them here: https://digitalslidearchive.github.io/HistomicsTK/installation.html
    The test data is downloaded when you run the cmake tests. If you are using the deploy scripts, this should happen via the deploy/test.sh script.
    Pete Lawson
    @geekypete
    @manthey Following up to your response above, what is the appropriate way to deploy a new Docker image once it has been created, and refresh HistomicsTK/Girder so that it may be used? Particularly with the Docker based HistomicsTK deployment. It seems that existing Docker containers are pulled from Dockerhub, but what about ones we build locally? I see that their is an API endpoint for loading a custom Docker image, but where should that Docker image be built and live? In the case of the Docker deployed HistomicsTK is that within the Docker container that contains the running instance of HistomicsTK?