search

choosehappy / HistoQC

HistoQC is an open-source quality control tool for digital pathology slides
BSD 3-Clause Clear License
263 stars 105 forks source link

Only left portion of slide HistoQC'd? #249

Closed aday00 closed 1 year ago

aday00 commented 1 year ago

For us, the final HistoQC images only show the leftmost half or third of the WSI pyramid. The WSI data are completely intact and can be dumped via openslide-write-png. Is there a HistoQC bounding box that is less than the entire slide?

nanli-emory commented 1 year ago

Hi @aday00, Could you help me narrow down the issue by answering some questions?

  1. Could you share your config.ini file?
  2. Do you set enable_bounding_box parameter as true on the BaseImage.BaseImage module?
  3. How many WSI images do you run on HistoQC? Do all WSI output images/masks have the same issues?

  4. Can you share or point to one of the WSI images you use?

    Thanks

aday00 commented 1 year ago

Hi @nanli-emory,

Thanks for your fast follow-up and guidance!

  1. As far as I can tell, this happens if we use config_v2.1.ini or a slightly modified ini. I operate in an environment where it's not really possible to move files out.
  2. BaseImage.BaseImage only has image_work_size=1.25x, in_memory_compression=True, confirm_base_mag:False, and mask_statistics=relative2mask. There is no enable_bounding_box in the ini file.
  3. I've tried this on both private and TCGA slides. Seems reproducible. I'll continue tests, particularly on TCGA, because that's more straightforward to share privately.
  4. Is it possible to exchange email addresses via a direct message, or may I email one of the authors (e.g. Andrew or Anant)? We have a collaboration with Anant.

Thanks

nanli-emory commented 1 year ago

Hi @aday00, your comment is very helpful. I will test on my local and see if I can reproduce the issue. This is my email nan.li@emory.edu. Thanks.

choosehappy commented 1 year ago

can you share a screenshot of the result from the TCGA case? does it look like this? https://github.com/choosehappy/HistoQC/issues/215

aday00 commented 1 year ago

No, it doesn't look like that. It's just truncated a bit. Maybe we can have a screenshare sometime? Will follow up via email.

aday00 commented 1 year ago

Nan Li got this working for me, needed to update to https://github.com/choosehappy/HistoQC/commit/a99916c2ceaa61c7ae5d75600f10d3fb553041fc

The code we were running was from https://github.com/choosehappy/HistoQC/commit/1df6420ea830088f8578e53faf7ffab3b3bf2194

Many thanks Nan!