Meeting Banner
Abstract #4428

Slice-by-slice regularized registration for spinal cord MRI: SliceReg

J. Cohen-Adad 1,2 , S. Lvy 1 , and B. Avants 3

1 Institute of Biomedical Engineering, Polytechnique Montreal, Montreal, QC, Canada, 2 Functional Neuroimaging Unit, CRIUGM, Universit de Montral, Montreal, QC, Canada, 3 PENN Image Computing & Science Lab, Dept of Radiology, UPENN, Philadelphia, PA, United States

Within-subject registration of spinal cord data is a difficult problem because the articulated nature of the spine can produce non-rigid deformations. To address this, researchers have introduced slice-by-slice registration using only translations within the axial plane. Although more accurate than volume-based transformations, this approach lacks robustness because each slice is treated independently from the others. Here we introduce a novel method (SliceReg) that estimates slice-by-slice transformations with polynomial regularization along the spinal cord axis. SliceReg has been validated in 25 datasets and shows higher accuracy than volume-based transformations and benefit from regularization in comparison with slice-by-slice techniques.

This abstract and the presentation materials are available to members only; a login is required.

Join Here