Which MONAI model to use for liver segmentation

[gokceay]

Hi,

There are many models that segments liver in MONAI. What I want to ask is which one should I use to get the most accurate and correct liver segmentation? To increase the accuracy and smoothness of the related organ, do I have to retrain the network by resampling the dataset to a lower spacing value? If so, how should I retrain? Is there any specific tutorial about it?

Thanks in advance

[diazandr3s]

Hi @gokceay,

Very good question.
I'd suggest you try the default Segmentation model in the Radiology app: https://github.com/Project-MONAI/MONAILabel/tree/main/sample-apps/radiology#segmentation

It is not only for the Liver segmentation, it actually segments 20+ organs. But it works great and you could tweak it to only show the liver if needed.

To increase the accuracy and smoothness of the related organ, do I have to retrain the network by resampling the dataset to a lower spacing value?

That's correct. Be mindful that this change will require more GPU memory to retrain the model.

Hope this helps,