wholeBody_ct_segmentation model for DICOM input file

[ytl0623]

Hi, I was used wholeBody_ct_segmentation model to test custom clinical image.

But the format file is NIFITI (.nii), how can I inference the model with DICOM file format.

Is there any method to prove it.

Thanks in advance.

[diazandr3s]

Hi @ytl0623,

This is a good question.
I guess there are two options here:

• Use monaibundle app in MONAI Label and a DICOM Web server (Slicer, Orthanc) to run batch inference
• Convert DICOM images to NIfTI/NRRD and run batch inference as it is shown in the tutorial

Hope this helps,

[ArthurRomansini]

If you're using the model via a python script you can just pass the dicom directory to the preprocess like that:

CTDicomFolder = "./data/yourdicomfolder"
configPath = "./models/yourmodelhere/configs/inference.json"

config = ConfigParser()
config.read_config(configPath)
preprocessing = config.get_parsed_content("preprocessing")
data = preprocessing({'image': CTDicomFolder})