search

abbieneininger / sarcApp

A python-based quantification app for sarcomeres and generally stripey objects
MIT License
0 stars 1 forks source link

auto-clicking next in sarcApp.py + myomesin stain #1

Open ruifanp opened 1 year ago

ruifanp commented 1 year ago

Thank you for this wonderful tool. I work in heart disease research doing imaging analysis, so sarcApp is super interesting in being able to quantify the sarcomeres.

I have 2 questions:

  1. sarcApp.py, runs 1 image at a time, and then prompts me to click next before proceeding to the next image. This is cumbersome if I have several hundred images I want to analyze. Is there a way to run all the images in the folder at once?

  2. My data has the stains myomesin and dapi. If I specify myomesin and dapi channels and run, I get a correct-looking binary image, but then the output myomesin_cellResults0.csv almost completely empty/nan, and myomesinImage0.jpg is a blank white image. There is also no output in the dapi output folder. If I choose actinin and dapi however, but use my myomesin images for actinin, then it runs fine and outputs the expected results but it's the wrong stain. The actininImage0.jpg looks like this. Is this correct? I was expecting a binary image with the myofibrils annotated.

image

Is it possible to run with the myomesin and dapi stains only?

abbieneininger commented 1 year ago

Hello! I'm glad you've been finding the tool useful and apologize for the issues, hopefully we can solve them asap.

For the first issue, you can run all of the images in the folder at once if you input data (csv files) instead of images, but I understand that you if you don't need to individually look at each of the images it will be cumbersome to have to click next every time. I will make a new version as a branch to this repository that does not show any images to click through and link it to you in a comment with the new version tomorrow.

For the next issue, let me ask a few questions first so I can best understand what is going wrong: are you inputting images alone and using the internal binarization method for both myomesin and dapi? Do your images have metadata that have the microns/pixel value stored? The actinin image looks incorrect and should show stripes at each Z-Line with colors denoting individual myofibrils. Let me look into why no results are saving from the DAPI images in the meantime!

Thank you for reaching out, and please continue to raise issues as you find them!

ruifanp commented 1 year ago

Thank you for the response! Once I can get the images to analyze correctly I will try inputting a csv so that all the images can run at once.

Yes, I am inputting the images alone. They are in the .tiff format but there is no additional metadata. I did not run the images through ImageJ or anything like that. My actual images are of the myomesin and DAPI stain, and there are 2 scenarios in which I'm running:

  1. Specify myomesin and DAPI in sarcApp. There is no output for DAPI, and the myomesin_cellResult.csv output is mostly nan. The corresponding myomesinImage.jpg is a blank outline. There is also no myomesin_mfResults.csv output as no myofibrils are identified.
image image
  1. Set actinin in sarcApp while using my myomesin images. Set DAPI using actual DAPI images. Now there is a result for DAPI and actinin, though results seem strange.
image image image

I think it's not super surprising if it doesn't work when I input myomesin images as actinin. Case 1 is more confusing that there is no output when I am inputting the actual correct stains.

My images are 1080 by 1080 pixels at 40x/0.25 microns per pixel, and have around 100 cells present.