Meeting Minutes
February 16, 2016
1. Attendees:
- Tom Vercauteren
- Sebastiano Ferraris
- Marc Modat
2. Discussed Topics:
- Draft for WBIR 16, new template presentation.
- Results presentation, and comparison with existing literature.
- Validation strategies.
3. Next Steps
- Using Neuromorphometrics as database instead of Miriad or Adni used so far.
- Implement series method to compute the exponential map. Run preliminary comparisons for this method to validate the results obtained in Bossa Zacur Olmos comparison.
- Search for a justification of the dummy ground truth (bronze standard?) toward non-linear examples with an analytical ground truth: homographies or non linear examples.
- Test scalar associativity over the same SVF (loop in the one parameter subgroup of transformation):
exp((a+b+c)u) = exp(au) exp(bu) exp(cu) = Id for a, b, c scalars, a+b+c = 0
both with synthetic data and real data. - Perform segmentation propagation accuracy test.
- Compute inverse consistency error, with synthetic data and SVF obtained from real cases while performing segentation propagation accuracy test.
- Perform Jacobian determinant to test the smoothness of the exponential methods. Comparison between SVF obtained with Scaling and squaring and SVF obtained with AEI method for real data set (compute it also for synthetic data?).
- Drafting new results and add new corrections to the current draft.
PhD Meeting Minute — Sebastiano Ferraris. — 2016-02-16