Mapping human brain networks with cortico-cortical evoked potentials

Abstract

The cerebral cortex forms a sheet of neurons organized into a network of interconnected modules that is highly expanded in humans and presumably enables our most refined sensory and cognitive abilities. The links of this network form a fundamental aspect of its organization, and a great deal of research is focusing on understanding how information flows within and between different regions. However, an often-overlooked element of this connectivity regards a causal, hierarchical structure of regions, whereby certain nodes of the cortical network may exert greater influence over the others. While this is difficult to ascertain non-invasively, patients undergoing invasive electrode monitoring for epilepsy provide a unique window into this aspect of cortical organization. In this review, we highlight the potential for cortico-cortical evoked potential (CCEP) mapping to directly measure neuronal propagation across large-scale brain networks with spatio-temporal resolution that is superior to traditional neuroimaging methods. We first introduce effective connectivity and discuss the mechanisms underlying CCEP generation. Next, we highlight how CCEP mapping has begun to provide insight into the neural basis of non-invasive imaging signals. Finally, we present a novel approach to perturbing and measuring brain network function during cognitive processing. The direct measurement of CCEPs in response to electrical stimulation represents a potentially powerful clinical and basic science tool for probing the large-scale networks of the human cerebral cortex.

Keywords: cortico-cortical evoked potential; effective connectivity; electrocorticography; graph theory; stimulation.

Figure 1.

Interventional techniques for measuring effective connectivity. (a) Microstimulation: stimulation and measurement of neural activity can be performed within the same cortical microcolumn. (b) Cortico-cortical evoked potentials: current is injected across electrodes placed on the cortical surface, and the strength and latency of propagating electrical activity is measured at distant sites. (c) Transcranial magnetic stimulation: generation of a large magnetic field outside the skull induces an electrical current inside the skull. Neural activity can be monitored with scalp EEG or functional MRI. Adapted with permission from [30,51,52].

Figure 2.

CCEP mapping and the comparison to anatomical and functional connectivity. (a) Components of the CCEP include the early N1 and late N2. (b) Spatial and temporal distribution of CCEPs. Green and grey coloured electrodes represent significant and non-significant CCEPs, respectively. Bipolar stimulation is applied between the adjacent electrodes (dotted white lines). Examples of CCEP waveforms are shown at several significant (black) and non-significant (grey) regions. (c) Comparison of structural and effective connectivity. The number of white matter tracts measured with DTI are positively correlated with the strength of the CCEP's N1 component and negatively correlated with its latency. Black circles denote the stimulating electrodes. The N1 response is represented by latency (colour of electrode) and amplitude (size of electrode). Electrodes without notable N1 responses are shown in white. All DTI pathways passing through the stimulation site are shown in green. Adapted with permission from [61]. (d) Comparison of functional and effective connectivity. Regions exhibiting strong N1 and N2 CCEP responses demonstrate correlations as measured by fMRI at rest. CCEP responses to stimulation of the white electrodes are depicted as significant (green) and non-significant (grey) circles. The BOLD correlation map with reference to the seed region at the stimulation site is represented by a heat map plotted on the pial surface. Results are from one representative patient. RSFC, resting-state functional connectivity. Adapted with permission from [52].

Figure 3.

Proposed mechanism of CCEP generation. (a) Generation of firing rate changes in pyramidal cells at the stimulation site. The stimulation protocol is shown below. Pyramidal cells are the principal cells of long-range transmission of electrical activity. Electrical current injected at the cortical surface propagates to local pyramidal cells via direct dendritic activity (blue arrows), adjacent interneurons (green arrows) or white matter traversing the stimulated region (black arrows). Solid arrows represent the region of the neuron that is first modulated by stimulation, while dotted arrows denote the direction of propagation within the neuron. (b) Electrical activity is transmitted to distant pyramidal neurons through direct and subcortical pathways. An example of the evoked potential at the target site is shown below. (c) Multi-unit response to electrical stimulation. Red and blue colours denote increases and decreases in multi-unit activity, respectively. Cortical layers are estimated on the left of the multi-unit colour plot. Curve below depicts a representative recording from the deeper layers. Results are from one representative patient. Adapted from [73]. pyr, pyramidal cell; int, interneuron; wm, white matter.

Figure 4.

CCEP mapping probes the directionality of complex brain networks. (a) Illustration of indegree and outdegree measures derived from CCEP mapping. (b) Weighted and binary connectivity matrices for one patient. Each row represents the strength of the CCEP at each electrode following the stimulation of one set of electrodes. Black regions in the binary matrix represent significant CCEPs. (c) Measures of network centrality and information flow expressed as z-scores and plotted on the cortical surface of one patient. Note the high outdegree, centrality and netflow in sensorimotor cortex with high indegree in the temporal lobe. Adapted with permission from [95].

Figure 5.

Task-based CCEPs can examine network reorganization during cognitive processing. (a) Electrical stimulation is applied to the FFA at baseline (during rest) and 100 ms following the presentation of visual stimuli (during task). Visual stimuli decrease the amplitude of the N2 but not the N1 of the CCEP. (b) A face discrimination task was performed while either the FFA or PPA was electrically stimulated. Electrical stimulation was applied 200 ms prior to or 100 or 500 ms following the visual stimuli. Stimulation of the FFA 100 ms after visual stimuli onset increased the reaction time during a face discrimination task when compared with PPA stimulation. Results are from one representative patient.