Orientation-selective and directional deep brain stimulation in swine assessed by functional MRI at 3T

Abstract

Functional MRI (fMRI) has become an important tool for probing network-level effects of deep brain stimulation (DBS). Previous DBS-fMRI studies have shown that electrical stimulation of the ventrolateral (VL) thalamus can modulate sensorimotor cortices in a frequency and amplitude dependent manner. Here, we investigated, using a swine animal model, how the direction and orientation of the electric field, induced by VL-thalamus DBS, affects activity in the sensorimotor cortex. Adult swine underwent implantation of a novel 16-electrode (4 rows x 4 columns) directional DBS lead in the VL thalamus. A within-subject design was used to compare fMRI responses for (1) directional stimulation consisting of monopolar stimulation in four radial directions around the DBS lead, and (2) orientation-selective stimulation where an electric field dipole was rotated 0°-360° around a quadrangle of electrodes. Functional responses were quantified in the premotor, primary motor, and somatosensory cortices. High frequency electrical stimulation through leads implanted in the VL thalamus induced directional tuning in cortical response patterns to varying degrees depending on DBS lead position. Orientation-selective stimulation showed maximal functional response when the electric field was oriented approximately parallel to the DBS lead, which is consistent with known axonal orientations of the cortico-thalamocortical pathway. These results demonstrate that directional and orientation-selective stimulation paradigms in the VL thalamus can tune network-level modulation patterns in the sensorimotor cortex, which may have translational utility in improving functional outcomes of DBS therapy.

Keywords: Directional DBS; Functional magnetic resonance imaging; High frequency stimulation; Motor cortex; Orientation selective DBS; Somatosensory cortex; Thalamus.

Fig. 1.

Experimental setup. A swine atlas was registered to a pre-operative MRI (a) to determine the location of the VL thalamus, a 16 channel (4 rows × 4 columns) DBS lead (b) was stereotactically implanted in the VL thalamus followed by postoperative anatomical MRI (c).

Fig. 2.

Stimulation was applied in two paradigms through a 16-contact directional DBS lead. (A) Unwrapped version of the lead showing directional stimulation where a pair of cathodes in each column were used to stimulate in each of the 4 directions around the lead. (B) Orientation-selective stimulation consisted of rotating a dipole 360° around a quad of electrodes.

Fig. 3.

fMRI BOLD contrast in the sensorimotor cortex to VL-thalamus DBS. (A) Atlas-based segmentations of the somatosensory, premotor, and primary motor cortices. (B) An example of the BOLD contrast to VL-thalamus stimulation from a single trial. (C) An example mean time series of the response shown in (B). DBS was applied for 20 s followed by 60 s of no stimulation, repeated 3 times.

Fig. 4.

Localization of 16-contact DBS lead implants to the VL thalamus. (VL = ventral lateral, VA = ventral anterior, VP = ventral posterior, R = reticular). Anatomical post-implant UTE scan showing the artifact of the DBS lead is overlaid with a swine atlas. (Félix et al., 1999; Saikali et al., 2010).

Fig. 5.

fMRI negative BOLD contrast is shown in the motor, premotor, and somatosensory cortices (A) when stimulating in four radial directions around the DBS lead. The negative BOLD contrast was quantified by the average t-statistic in the primary motor (blue), premotor (orange), and somatosensory (yellow) cortices using the GLM method (B) and the ICA method (C).

Fig. 6.

fMRI BOLD contrast during orientation-selective stimulation through a quad of electrodes on the medial side of the lead. 0°/180° represents orientation of the electric field parallel to the lead and 90°/270° represents orientation of the electric field perpendicular to the lead. The average t-statistic of significant voxels within each cortical region was quantified for each orientation using the GLM method (A) and the z-statistic using the ICA method (B).