Jin Kyu Gahm1,2, Nicholas Wisniewski3,
William S. Klug4, Alan Garfinkel3,5, Daniel B. Ennis1,6
1Department of Radiological
Sciences, University of California, Los Angeles, CA, United States; 2Department
of Computer Science, University of California, Los Angeles, CA, United
States; 3Department of Medicine, University of California, Los
Angeles, CA, United States; 4Department of Mechanical &
Aerospace Engineering, University of California, Los Angeles, CA; 5Department
of Physiological Science, University of California, Los Angeles, CA, United
States; 6Biomedical Engineering Interdepartmental Program,
University of California, Los Angeles, CA, United States
DT-MRI interpolation is the process of estimating diffusion tensors at arbitrary points in space from regularly sampled tensor data. Tensor interpolation is important for tensor-based fiber tractography, registration, volume rendering, and computational model building. In this work, we use bootstrap statistical methods to compare four different DT-MRI interpolation methods accuracies for recovering the tensor shape (invariants) and orientation of unknown tensors from known tensor data. By using a cardiac DT-MRI dataset, we show the statistical bootstrapping results for the paired comparisons, and present recommendations for the selection of the DT-MRI interpolation method.