Real-time Motion Detection Algorithms for Spherical Navigator Echoes in Functional MRI
Kraft R, Ari N
Wake Forest University
Spherical navigator echoes (SNAV) can be used to track head position throughout the functional MRI. However, current motion detection algorithms for SNAV require significant computation times and hence limit the feasibility of SNAV technique for real-time functional MRI applications. We developed a real-time SNAV technique, optimized to have a motion detection accuracy of 0.3mm and timing performance of 2300ms for a TR of 2500ms. We developed k-space thresholded Least Squares Conjugate Gradient (kt-LSCG) algorithm for translation detection, and multi-scale DownHill Simplex (ms-DHS) algorithm with normalized Least Squares cost function for rotation detection.