mosamatic-cli 1.5.10

A CLI tool for body composition analysis from DICOM CT scans

mosamatic-cli

Command-line tool for running processing tasks on medical images

To-do

• NumPy to NIFTI conversion

• Create PNG images from DICOM files

• Slice selection using either Total Segmentator or MOOSE

  What you can do is first create spinal segmentations of a CT scan. Then you can use that data to pick a segmentation, e.g., the L3 vertebra and select the middle slice. In terms of commands/tasks, this would mean a command/task:

  • SegmentAnatomyTS: with option to output a specific segmentation object
  • SegmentAnatomyMOOSE: with option to output a specific segmentation object
  • SelectSlice: Given a CT scan and a segmentation object, select the middle or other slice

• CT/MRI registration using ANTsPy

• Total Segmentator liver segmentation CT + MRI (Dixon H2O)

• IMAT analysis

  Discuss with Leroy how to detect fat inside muscle (either as SAT or black pixels). If pixels inside muscle are black, what does that mean? Can I overlay the mask on top of the image and check the pixel gray values and then determine (using the Alberta threshold range) whether these pixels are fat or not?

• Calculate muscle PDFF maps from Dixon MRI after registering with CT

Commands

• numpy2nifti
• createpngsfromdicomfiles
• selectvertebralslice --engine=ts,moose --vertebra=l3,t4 --position=all,top,middle,bottom
• registerl3 --modalities=ct/dixon,ct/t1,ct/t2
• segmentanatomy --engine=ts,moose --masks=all,spine,l3,t4,liver,vessels
• segmentmusclefatl3[tensorflow] --imat=true,false
• calculatepdffmap2d --inphase=/path/to/image --outphase=/path/to/image --water=/path/to/image --mask=/path/to/mask (muscle and fat)
• calculatepdffmap3d --inphase=/path/to/series --outphase=/path/to/series --water=/path/to/series --mask=/path/to/mask (liver)

Next action

• Implement slice selection with Total Segmentator