Phase-coherence classification: A new wavelet-based method to separate local field potentials into local (in)coherent and volume-conducted components

• Published in: Journal of neuroscience methods Vol. 291; pp. 198 - 212
• Main Authors: von Papen, M., Dafsari, H.S., Florin, E., Gerick, F., Timmermann, L., Saur, J.
• Format: Journal Article
• Language: English
• Published: Netherlands Elsevier B.V 01-11-2017
• Subjects: Aged; Algorithms; Apomorphine - pharmacology; Beta Rhythm - drug effects; Beta Rhythm - physiology; Coherence; Cohort Studies; Deep Brain Stimulation; Dopamine Agonists - pharmacology; Female; Humans; Implantable Neurostimulators; Local field potential; Male; Middle Aged; Motor Activity - physiology; Neurons - drug effects; Neurons - physiology; Parkinson Disease - physiopathology; Parkinson Disease - therapy; Parkinson's disease; Phase coupling; Subthalamic Nucleus - drug effects; Subthalamic Nucleus - physiopathology; Synaptic Transmission - drug effects; Synaptic Transmission - physiology; Tremor - physiopathology; Tremor - therapy; Volume conduction; Wavelet Analysis; Local field potential; Wavelet analysis; Parkinson's disease; Phase coupling; Volume conduction; Coherence
• ISSN: 0165-0270; 1872-678X
• DOI: 10.1016/j.jneumeth.2017.08.021

•The novel wavelet-based phase-coherence classification (PCC) is introduced in detail.•Local field potentials are split in time-frequency domain into three signal components.•Spectra of incoherent, coherent and volume conducted components are analyzed separately.•In Parkinson's disease components are differently modulated by medication and movement.•The PCC components may represent activity of physiologically different networks.
Local field potentials (LFP) reflect the integrated electrophysiological activity of large neuron populations and may thus reflect the dynamics of spatially and functionally different networks.
We introduce the wavelet-based phase-coherence classification (PCC), which separates LFP into volume-conducted, local incoherent and local coherent components. It allows to compute power spectral densities for each component associated with local or remote electrophysiological activity.
We use synthetic time series to estimate optimal parameters for the application to LFP from within the subthalamic nucleus of eight Parkinson patients. With PCC we identify multiple local tremor clusters and quantify the relative power of local and volume-conducted components. We analyze the electrophysiological response to an apomorphine injection during rest and hold. Here we show medication-induced significant decrease of incoherent activity in the low beta band and increase of coherent activity in the high beta band. On medication significant movement-induced changes occur in the high beta band of the local coherent signal. It increases during isometric hold tasks and decreases during phasic wrist movement.
The power spectra of local PCC components is compared to bipolar recordings. In contrast to bipolar recordings PCC can distinguish local incoherent and coherent signals. We further compare our results with classification based on the imaginary part of coherency and the weighted phase lag index.
The low and high beta band are more susceptible to medication- and movement-related changes reflected by incoherent and local coherent activity, respectively. PCC components may thus reflect functionally different networks.