Figure 1.
Application of the null-beamformer.
The figure shows the estimated power of the sources in the mid-sagittal (top) and mid-coronal (bottom) view following the use of A) conventional beamformer and B) null-beamformer. The threshold value is 70% of the peak of the power spectrum. As can be seen, the null-beamformer has successfully removed the interference outside of the brain. Please note that in order to best depict the sources of brain activity, the null location is approximate and its actual location is in other anatomical planes (not shown).
Figure 2.
Comparison of the accuracy of using null and conventional beamformers for the localization of known locations of DBS electrodes.
A) We used the null beamformer to localize the DBS electrodes when the stimulator was ON at 130 Hz. The coronal view of the lowest electrodes as localised on the patient’s MRI (red markers) compared with the overlay of contours of the estimated power using the null-beamformer. Two sagittal slices through lower left electrode and sagittal view of the lower right electrode. The fit is especially good on the left side. B) Similar localization using the conventional beamformer method shows a less good fit. In particular the method is unable to localize both electrodes.
Figure 3.
Summary of changes in brain activity between DBS ON and OFF after one week and one year after surgery.
Four main midline brain regions were identified that had increases in power following pain related changes when DBS was OFF: the pre-supplementary motor area (pre-SMA), caudal-ACC (c-ACC), rostral-ACC (r-ACC) and brainstem (periaqueductal grey, PAG). A) The figure shows the mid-sagittal slice with the contours of the reconstructed MEG sources at 10Hz using null-beamformer with DBS OFF. Below each brain slice is shown the corresponding 3D mesh plot of the reconstructed neural activity in the mid-sagittal slice at 10Hz (with the lower electrode location inserted in red). B) Similarly, when DBS is ON, the figure shows the reconstructed MEG sources and the corresponding 3D mesh plot. This shows a decrease in activity in the caudal and rostral ACC with pain relief. C) One year later, after a continuous DBS, a similar pattern of changes in activity emerges when the patient is in pain (DBS OFF). D) Similarly, a decrease in activity in rostral ACC is evident following pain relief with DBS ON. Interestingly, the activity in the caudal ACC appears to be depressed during pain after one year of DBS and thus show a much smaller decrease in activity upon pain relief. This could be suggestive of plastic changes following long term DBS. Please note that the vertical axes (depicting normalised power) are the same for Figures A and B but not for Figures C and D. As can be seen from the figure, the power of the reconstructed sources is independent from the number of rejected channels (see Methods). It is also notable that the full width at half maximum (FWHM) of each peak is also independent from the number of rejected channels (e.g. see the pre-SMA). The number of rejected channels is therefore likely only to have a minor influence on the source space results.
Table 1.
Changes in activity in brain regions one week and one year after surgery.