Artikel
Automated, individual thalamus segmentation based on non-linear atlas co-registration for deep brain stimulation
Suche in Medline nach
Autoren
Veröffentlicht: | 8. Juni 2016 |
---|
Gliederung
Text
Objective: Current methods for thalamus segmentation based on probabilistic diffusion tensor imaging are computationally intensive, time consuming and limited with regard to their intra- and intersubject reliability. Refined atlases of the histo-anatomy and connectivity are currently not considering the individual anatomy of the patients. Individualizing these atlases might overcome previous limitations.
Method: Preoperative MR (T1, diffusion-weighted images) and postoperative CT images of twelve patients undergoing bilateral thalamic deep brain stimulation for tremor were co-registered and automatically segmented using the FMRIB Software Library (1). Based on the individual thalamic outlines, a twelve degree of freedom linear co-registration of the 3D Morel Atlas (2) was performed first and followed by a subsequent nonlinear adjustment to achieve the best fit on the individual anatomy. The transformation matrices were then applied on the 49 thalamus sub-nuclei and compared to the postoperative electrode contact positions (12x4x2 contacts).
Results: The procedure was automated and achieved the co-registration in each subject. The electrode contacts projected to VLpv(VIM), VM, VLa, VPM, VL and beyond the thalamus border 49,19, 13, 9, 1 and 5 times, respectively.
Conclusions: The introduced procedure might offer an alternative to other approaches for thalamic targeting during DBS therapy. Future studies will need to directly compare this strategy to other segmentation methods.