Projects:RegistrationDocumentation:RegEval Anisotropy

Effects of Voxel Anisotropy and Intensity-Inhomogeneity on Image-based 3D Registration

Dominik Meier, Andryi Fedorov, William Wells, C.F. Westin, Ron Kikinis

Summary / Questions

This is a planned experiment for (mostly empirical) data and models on how voxel-anisotropy affects the quality/outcome of automated, intensity-based image registration. Chief questions to address are:

• at what point does voxel anisotropy seriously affect registration performance?
• are there remedies?
• does bias field inhomogeneity in MRI images affect the quality of automated registration?
• What is the relative sensitivity of different cost functions (MI vs. NormCorr)?
• What is the relative sensitivity to DOF?
• Should a bias-field correction be applied beforehand or after? Is bias correction affected by prior registration?
• what is the effect of "differential bias"? for the purpose of subtraction and/or ratio images, is differential bias correction (Lewis et al. NeuroImage 2004) preferable to correcting each bias individually? Determined as residual in validation images of zero or predefined diff. -> moving out of scope a bit, because it is no longer pure registration but includes change detection via subtraction. On the other hand that is the common application and hence clarity on bias sources is needed. We can argue that all intra-subject intra-modality registration is done for the purpose of change detection.

Hypotheses

• registration is particularly sensitive to rotation perpendicular to largest voxel dim. -> need to decouple rotations from translation in variational framework
• resampling to isotropic voxel size before registration does reduce albeit not eliminate the "anisotropy threshold"

Methods

• Test Data Candidates
  • must have 1mm isotropic resolution
  • Brainweb
  • kidney, breast MRI pairs,
  • modalities where anisotropy is common

• Anisotropy Experiment

1. take 1mm iso ref volume and move a known amount
2. filter (1-D avg) & subsample both image grids
3. register & evaluate residual error (evaluate RMS distance: distance of ICC points sent through R1*inv(R2)

• BiasField Experiment:

1. take 1mm iso ref volume and move a known amount
2. apply bias field to both image grids
3. register & evaluate residual error

• Variational Parameters
  • voxel size factors: x 1 , 1.2 , 1.5 , 3 , 5, 10
  • bias field: derive from actual case, then amplify x 1 , 1.2 , 1.5
  • reference motion
  • reg. sampling rate
• to obtain Relevant Reference XForm
  • take 2 real-life scans of different protocols, e.g. FLAIR and T1, and perform BSpline registration, use that as reference + add additional translation & rotation
  • take average of real-life bias fields then amplify x 1 , 1.2 , 1.5

• Evaluation
  • registration error as RMS residual
  • cost function
  • ROC?