search

ENHANCE-PET / MOOSE

MOOSE (Multi-organ objective segmentation) a data-centric AI solution that generates multilabel organ segmentations to facilitate systemic TB whole-person research.The pipeline is based on nn-UNet and has the capability to segment 120 unique tissue classes from a whole-body 18F-FDG PET/CT image.
https://enhance.pet
GNU General Public License v3.0
196 stars 28 forks source link

MRI model is for rats but not for humans. #94

Closed LinLiMax1 closed 10 months ago

LinLiMax1 commented 10 months ago

Describe the bug I tried the CT models and it worked very well, but the MRI model preclin_mr_all didn't generate good results. Update: I missed the information in the document. The MRI model is for rats but I used it in humans. Thank you for your clarification!

Screenshots

image
LalithShiyam commented 10 months ago

Hi @LinLiMax1

I saw the screenshot and it looks like a clinical-MR of the heart. I can't view the actual NIFTI image as I am on vacation.
We don't have models for clinical MRs. The model that you used is for "preclinical MR", that's why it starts with "preclin_".

Refer this:

`Update AVAILABLE_MODELS List ✏️

Include your model's unique identifier to the AVAILABLE_MODELS list in the resources.py file. The model name should follow a specific syntax: 'clin' or 'preclin' (indicating Clinical or Preclinical), modality tag (like 'ct', 'pt', 'mr'), and then the tissue of interest.`

Kindly confirm if your data is clinical or preclinical, so that we can help you accordingly. If you tried to use a preclinical model on a clinical datasets. Of course your "meaningless" results makes sense.

Cheers, Lalith

LalithShiyam commented 10 months ago

Hi @LinLiMax1, Preclinical means it is a model for segmenting organs from rats/mice. Clinical = humans.

Simply put, you used a segmentation model created for segmenting mice and rats mr organs to segment human MR.

That will not work.

Cheers, Lalith

LinLiMax1 commented 10 months ago

Thank makes sense now.....Thank you so much......