As in the aforementioned title,
**Is there an automatic heart segmentation tool for segmenting the important heart structures:
Left/Right (Ventricle/Atrium) , Superior/inferior vena cava/pulmonary artery etc
Since some CT scans shows plague and some structures are damaged, so it is hard to manually select precisely each structure for different patient CT scans.
WIll there be a tool in the future, or is there already one(Like Nvidia for tumor segmentation) in the Slicer already?
I’m not aware of any freely available model for it, but you can train your own. For heart segmentation on good-quality contrasted CT of healthy people you probably don’t need a lot of training data - maybe a few dozens of CT scans (you need more if you need a robust solution for various abonrmalities). You can do the same what @rbumm has done recently for lungs - used semi-automatic LungCTAnalyzer extension (internally uses Grow from seeds effect) to generate training data for MONAILabel.
Most of the CT based segmentation paper cite the paper as this:
Tautz L, Neugebauer M, Hüllebrand M, et al. Extraction of open-state mitral valve geometry from CT volumes[J]. International Journal of Computer Assisted Radiology and Surgery, 2018, 13(11): 1741-1754.
But there still exist so much work to reproduce it.
Automatic segmentation may need landmark labels mannully.