Motivation

The following are the broad topics and specific questions which motivated this advanced tutorial. We expect the talks to address some of these issues during the course of the day. Unanswered questions and unresolved issues will be the starting point for the panel discussion at the end of the day.

A. Reconstruction of the diffusion model

• models and representation of diffusion data: physical, biological and mathematical underpinnings for the same
• choosing an "optimal" gradient direction set
• choice of b value in terms of research data and clinical viability
• noise models
• biological basis of diffusion

B. Quality Analysis of acquired data

• what is a feasible quality control protocol for acquired data
• what is the way to compare DWI volumetric data across scanners, across different gradient directions and b-values
• comparison of acquisition across subjects, across time points in a longitudinal study
• protocol for automated QA post-acquisition, prior to analysis.

C. Registration of reconstructed post-processed data

• how to evaluate and compare different registration methods (full-tensor, scalar-based, multi-channel)
• optimal metrics and features (tract-based, GM, WM cues)
• methods for high higher dimensional data
• template-free/group-based registration
• creating an atlas

D. Metrics and statistics on tensors

• what is the metric of choice to measure distance between DWI, tensors
• physical/biological rationale for a representation-metric combination
• how can the metric be used to obtain the sample mean and variance, for smoothing & interpolation
• are these generalizable to higher dimensional data
• how to do population statistics / determine the underlying distribution
• how to identify that distinct ROIs have a uniform statistical distribution
• method(s) adopted to compare (qualitatively and quantitatively) results from different paradigms

E. Tracking, Segmentation and Clustering

• effect of acquisition parameters on tracking
• how to evaluate different tracking methods on the same data set
• segmentation of data into WM/CM/CSF
• clustering of tracts
• physical/biological underpinnings of fiber tracking
• identifying fiber crossings: HARDI & beyond

 

Parent: DMRI08 Revision: r17 - 02 Aug 2008 - 20:44:35 - CarlFredrikWestin

Edit  |  Attach image or document  |  More topic actions  |  Backlinks

---