not found get_segmentation_connected_components function.

[wangnanv5]

Hello, when i run this project, it's raise a error in file assign_ribs_to_vert.py line 66, do you know how to solve it?
AttributeError: 'NII' object has no attribute 'get_segmentation_connected_components'. Did you mean: 'get_largest_k_segmentation_connected_components'?

Looking forward to your reply.

[Hendrik-code]

Hey
yeah, the underlying Toolbox TPTBox has been updated and this function no longer exists, give me a little time to fix it, should not be a big issue. Thanks for bringing it to my attention!

[wangnanv5]

okay~ If there is any progress in the work, please @ me here😁. Thank you again and your team for making this ribs labels and code. This work can effectively make up for the deficiency of the rib segmentation results of TotalSegmentator.

[Hendrik-code]

This work can effectively make up for the deficiency of the rib segmentation results of TotalSegmentator.

This is exactly one of the reasons why we started on this, because for some reason, no public model can segment the beginning part of the ribs (closest to the vertebrae), even though for a model, this is still trivial to segment.

Yes, will let you know.

[Hendrik-code]

can you retry this? @wangnanv5

[wangnanv5]

can you retry this? @wangnanv5

Sorry, I have been busy with other things during this period. I will test the effect on the private dataset as soon as possible and give you feedback in time😁

[wangnanv5]

can you retry this? @wangnanv5
when i check this related manual data, i found there are many unlabel region, like this,

They are neither cartilage nor bone tumors. Is there a mistake in the labeling?

[Hendrik-code]

Hey @wangnanv5
What do you mean with "related manual data"? Is this a prediction on your data?
In any case, the shown segmentation definitely has errors. But without further info, I cannot know what might have caused this.

[wangnanv5]

hey,
What I described is not very accurate, specifically, the above graph occurs in the data "verse_rib_manual\sub-verse768" , it has also appeared in other data, but not frequently,i will check the data again and report the incorrect data to you

[Hendrik-code]

hey @wangnanv5

okay, that is great to know! Maybe I made a mistake when uploading, because those errors should not be present in the manual folders.
Are you sure this is sub-verse768? Because both my image and segmentation look completely different and correct.

[wangnanv5]

my bad, The id in the above picture is sub-verse502, for other incorrect images, I will check them as soon as possible