Context Image Annotation

Context Image Annotation  

  By: soldeepml on Jan. 9, 2025, 8:13 a.m.

Dear Organizers!

Trusting this message finds you well. I want to know if the context images also have their annotation provided, it does not seem to be mentioned in the data section.

 Last edited by: soldeepml on Jan. 9, 2025, 8:14 a.m., edited 1 time in total.

Re: Context Image Annotation  

  By: mschuiveling on Jan. 9, 2025, 11:11 a.m.

Dear Soldeepml,

There are not context image annotations. You can, however, always create them yourself from the centered image to generate more data if needed.

Good luck and thank you for joining the challenge!

Kind regards, Mark

Re: Context Image Annotation  

  By: soldeepml on Jan. 10, 2025, 2:04 a.m.

Thank you for your reply, Mark! Can you kindly help me understand this, if the context image is cropped to 1024 x 1024, won't that be the same image as the ROI? Also, does the area outside the ROI on the context image not contain any of the cells annotated on the ROI?

Re: Context Image Annotation  

  By: mschuiveling on Jan. 10, 2025, 11:06 a.m.

Hi Soldeepm,

I understand the confusion and I think you do understand how the data works. To be sure I'll show it with an example:

The annotations are for 1024 x 1024 image which is the center of the context image. It is possible with an offset to display the annotations in a context image. Annotations for the full context image, though, are not present. In the context image the same nuclei types are present (with exceptions because some primary melanoma for example have glands in their primary image which consists out of different tissue and nuclei types not represented in the dataset). What I meant in my previous message is that it is possible to infer from the ROI on the context ROI to generate more data.

I hope this helps and if you have more questions please let me know!

Kind regards, Mark

Re: Context Image Annotation  

  By: soldeepml on Jan. 10, 2025, 3:41 p.m.

Thank you for the clarity