Towards a model-based integration of co-registered electroencephalography/functional magnetic resonance imaging data with realistic neural population meshes

Abstract

Brain activity can be measured with several non-invasive neuroimaging modalities, but each modality has inherent limitations with respect to resolution, contrast and interpretability. It is hoped that multimodal integration will address these limitations by using the complementary features of already available data. However, purely statistical integration can prove problematic owing to the disparate signal sources. As an alternative, we propose here an advanced neural population model implemented on an anatomically sound cortical mesh with freely adjustable connectivity, which features proper signal expression through a realistic head model for the electroencephalogram (EEG), as well as a haemodynamic model for functional magnetic resonance imaging based on blood oxygen level dependent contrast (fMRI BOLD). It hence allows simultaneous and realistic predictions of EEG and fMRI BOLD from the same underlying model of neural activity. As proof of principle, we investigate here the influence on simulated brain activity of strengthening visual connectivity. In the future we plan to fit multimodal data with this neural population model. This promises novel, model-based insights into the brain's activity in sleep, rest and task conditions.

Figure 1.

(a) Neural population model surface (blue) between the Civet interfaces of grey matter with white matter (yellow) and cerebrospinal fluid (orange). (b) Skull and scalp boundaries (blue) from intensity profiles (black) along outward vectors (green). (c) Visual connectivity used in this study. ‘Regional map’ areas are indicated by colours on an average cortical surface. FEF, frontal eye field; VACd and VACv, dorsal and ventral anterior visual cortex.

Figure 2.

(a) Before (top) and after (bottom) pruning. (b) The bold edge between p₁ and p₂ is replaced by a vertex between p₀ and p₃. (c) If edge 1–2 is removed, triangles 2–3–4 and 1–4–3 collapse. The shaded region is replaced by the 1–2–3 triangle; or (d) a lifted one. (e) Cortical surface before (left) and after (right) pruning to a minimum edge length of 2.5 mm.

Figure 3.

(a) Head model as extracted from structural magnetic resonance imaging. (b) For one vertex, the two kinds of long-range connectivity are illustrated. (c) The neural activity model with excitatory (black) and inhibitory (white) populations and connections.

Figure 4.

(a) Direct connections , , and within the cortical volume are concatenated to the shortest route . (b) Examples of shortest routes so determined.

Figure 5.

(a) Projection vectors for vertex s_i and voxel v_j, and resulting weights from high (red) to low (blue). (b) Vertex to voxel projection of fMRI BOLD in horizontal section.

Figure 6.

Dependence of the power spectral density of the ECoG on the strength of specific connectivity . The power is shown in dB relative to the largest overall value.

Figure 7.

Predictions of (a) scalp EEG, (b) ECoG and (c) fMRI BOLD for different strengths of specific connectivity: 0% (top), 30%, 60%, 85% and 90% (bottom). See also the video in the electronic supplementary material.