Yogesh Rathi1,2, James Malcolm, 23,
Sylvain Bouix1, C-F Westin4, Martha E. Shenton1,5
1Psychiatry, Harvard Medical School,
Boston, MA, United States; 2Georgia Institute of Technology,
Atlanta, GA, United States; 3Brigham and Women's Hospital, United
States; 4Radiology, Harvard Medical School, Boston, MA, United
States; 5VA Clinical Neuroscience Division, Boston, MA, United
States
Existing
methods perform model estimation independently at each voxel and tractography
is performed in the next step. We use a nonlinear Kalman filter for
simultaneous model estimation and tractography. The method not only provides
an estimate of the model parameters, but also a confidence in the estimation
in terms of the covariance matrix. We utilize measures derived from this
covariance matrix to detect false positives in the tracts generated.