Monitoring brain neuroactivity during deep brain stimulation
What is Deep brain stimulation?
Deep brain stimulation (DBS) has been established as a method to control symptoms in a number of neurological disorders. DBS patients have electrodes planted deep in the brain, to which an alternating electric voltage is applied. It has proved very successful in the treatment movement disorders such as Parkinson's disease and dystonia (http://www.parkinsons.org.uk/about_parkinsons/treating_parkinsons/surgery/deep_brain_stimulation.aspx) and is being applied in a range of other diseases.
This is multidisciplinary work, with surgeons in the Oxford John Radcliffe Hospital, the Department of Psychiatry and the Oxford Human Brain Activity Centre, where there is a MEG imaging facility. Our particular interests are tremor in Parkinson's Disease and DBS in the treatment of chronic pain. Through these collaboration we are working to improve stimulation treatment and, associated with that, to get a better understanding of the way the brain is operating, and the effect of stimulation, in these conditions. The work is funded by the EPSRC/Wellcome Centre of Excellence.
1. Recognising abnormal states; adaptive stimulation
At present, stimulation is continuous and this gives poor battery usage and can cause habituation. The goal is to make the next generation of stimulators adaptable to particular patients through demand driven stimulation - altering the pattern and duration of stimulation to the brain's own signals. Tremor in Parkinson's disease (PD) is clearly detectable in the beta wave brain activity (as well of course from external instrumentation such as accelerometers), but its onset is very rapid. We have developed methods using autorgressive models, coherence measures and Hidden Markov modelling to detect the change of state associated with onset of tremor from signals received from implanted electrodes in the subthalmic nucelus (which is a prime target for stimulation in PD patients). More generic work in this area has developed HMM models to identify state transitions between different regions to identify brain networks using MEG imaging. This work is currently been demonstrated on resting state data but will soon be applied to tremor patients.
Chronic pain patients
Associated work is looking at the possibility of methods to use the local field potentials (the signals detected at the DBS sites using the electrodes to receive) to provide quantitive information on the extent of pain. Quantitative methods to relate the power in the characteristic bursts of activity to the patient's own perception of pain, using machine learning techniques are showing early promise.
In collaboration with the Nuffield Department of Clinical Neurosciences we have demonstrated proof of principle of a non-invasive method to reduce tremor, using trasncranial alternating current stimulation (TACs). This work is in its early stages (published in Current Biology, 2013).
2. MEG (Magneto encephalography) imaging and application to chronic pain patients
MEG is the only technology suitable for functional imaging for DBS patients as they have metal implanted in the skull so fMRI cannot be used. MEG uses a set of very sensitive magnetic sensors placed around the head to detect the magnetic fields associated with the neuronal activity. In Oxford (OHBA) we have a Elekta Neuromag facility.
Once the signal have been acquired, an image is formed using a technique known as beamforming which uses them to reconstruct the sources within the brain, a techqnue known as beamforming. The signals acquired are typically with low signal to noise ratio, non-Gaussian distribution and correlated. Beamforming is therefore challenging. It is particularly difficult for DBS patients because artefacts arise from the coil of wire left beneath the burr hole (through which wires are taken to the battery.
We have developed a set of beamformers to improve the resolution of the images to enable the sources to be seen more clearly. Some examples are shown below.
The null beamformer
The left hand images show brain activity using the limear constraint minimum variance (LCMV) beamfomer. In the lower lefthand images an artefact can be seen at the burr hole, where the activity appears to be outside the skull. In the images on the right the null beamformer has been used to correct this artefact.
The null beamformer has been used to investigate the effect of DBS for chornic pain patients. The figures below show source activity in the PAG and the ACC regions in the brain when stimulation is applied. Activity was less without stimulation, when the patient reported pain. Work published recently in PLOSOne, also showed longitudinal changes in the brain over a year.
Researchers: Dr Hamid Mohseni, Dr John Stuart Brittain
DPhil student: Adam Baker
MSc students: Greg Bara, Yashil Handa, Su Zhang
4YP students: Ioana Nica