Todo from Previous Meeting:

  • Read thoroughly this paper that approaches davis: link

  • Finish itk wrapping

  • Create figures showing failure modes of

    • no pixel statistics control

    • mean pixel statistics control

    • rolling pixel statistics control

  • FMAPICON try more powerful network?

itk vector saga

itkVectorImagePython is distinct from itkImage[Vector]. This is suffering.

ICON itk wrapping

Approach that doesn't work:

manually create a deformation field that goes straight betweeen images of different spacings, orientations

Approach that is working:

Use linear transforms to move into and out of the space where the neural network defines Phi


Back to weekly reports

Back to Reports