P.M. Drysdale

Comparing hemodynamic models with DCM

The classical model of blood oxygen level-dependent (BOLD) responses by Buxton et al. [Buxton, R.B., Wong, E.C., Frank, L.R., 1998. Dynamics of blood flow and oxygenation changes during brain activation: the Balloon model. Magn. Reson. Med. 39, 855–864] has been very important in providing a biophysically plausible framework for explaining different aspects of hemodynamic responses. It also plays an important role in the hemodynamic forward model for dynamic causal modeling (DCM) of fMRI data. A recent study by Obata et al. [Obata, T., Liu, T.T., Miller, K.L., Luh, W.M., Wong, E.C., Frank, L.R., Buxton, R.B., 2004. Discrepancies between BOLD and flow dynamics in primary and supplementary motor areas: application of the Balloon model to the interpretation of BOLD transients. NeuroImage 21, 144–153] linearized the BOLD signal equation and suggested a revised form for the model coefficients. In this paper, we show that the classical and revised models are special cases of a generalized model. The BOLD signal equation of this generalized model can be reduced to that of the classical Buxton model by simplifying the coefficients or can be linearized to give the Obata model. Given the importance of hemodynamic models for investigating BOLD responses and analyses of effective connectivity with DCM, the question arises which formulation is the best model for empirically measured BOLD responses. In this article, we address this question by embedding different variants of the BOLD signal equation in a well-established DCM of functional interactions among visual areas. This allows us to compare the ensuing models using Bayesian model selection. Our model comparison approach had a factorial structure, comparing eight different hemodynamic models based on (i) classical vs. revised forms for the coefficients, (ii) linear vs. non-linear output equations, and (iii) fixed vs. free parameters, ε, for region-specific ratios of intra- and extravascular signals. Using fMRI data from a group of twelve subjects, we demonstrate that the best model is a non-linear model with a revised form for the coefficients, in which ε is treated as a free parameter.

Mean-field modeling of the basal ganglia-thalamocortical system. II: Dynamics of parkinsonian oscillations

Neuronal correlates of Parkinsons disease (PD) include a shift to lower frequencies in the electroencephalogram (EEG) and enhanced synchronized oscillations at 3-7 and 7-30 Hz in the basal ganglia, thalamus, and cortex. This study describes the dynamics of a recent physiologically based mean-field model of the basal ganglia-thalamocortical system, and shows how it accounts for many key electrophysiological correlates of PD. Its detailed functional connectivity comprises partially segregated direct and indirect pathways through two populations of striatal neurons, a hyperdirect pathway involving a corticosubthalamic projection, thalamostriatal feedback, and local inhibition in striatum and external pallidum (GPe). In a companion paper, realistic steady-state firing rates were obtained for the healthy state, and after dopamine loss modeled by weaker direct and stronger indirect pathways, reduced intrapallidal inhibition, lower firing thresholds of the GPe and subthalamic nucleus (STN), a stronger projection from striatum to GPe, and weaker cortical interactions. Here it is shown that oscillations around 5 and 20 Hz can arise with a strong indirect pathway, which also causes increased synchronization throughout the basal ganglia. Furthermore, increased theta power with progressive nigrostriatal degeneration is correlated with reduced alpha power and peak frequency, in agreement with empirical results. Unlike the hyperdirect pathway, the indirect pathway sustains oscillations with phase relationships that coincide with those found experimentally. Alterations in the responses of basal ganglia to transient stimuli accord with experimental observations. Reduced cortical gains due to both nigrostriatal and mesocortical dopamine loss lead to slower changes in cortical activity and may be related to bradykinesia. Finally, increased EEG power found in some studies may be partly explained by a lower effective GPe firing threshold, reduced GPe-GPe inhibition, and/or weaker intracortical connections in parkinsonian patients. Strict separation of the direct and indirect pathways is not necessary to obtain these results.

BOLD responses to stimuli: dependence on frequency, stimulus form, amplitude, and repetition rate.

A quantitative theory is developed for the relationship between stimulus and the resulting blood oxygen level-dependent (BOLD) functional MRI signal. The relationship of stimuli to neuronal activity during evoked responses is inferred from recent physiology-based quantitative modeling of evoked response potentials (ERPs). A hemodynamic model is then used to calculate the BOLD response to neuronal activity having the form of an impulse, a sinusoid, or an ERP-like damped sinusoid. Using the resulting equations, the BOLD response is analyzed for different forms, frequencies, and amplitudes of stimuli, in contrast with previous research, which has mostly concentrated on sustained stimuli. The BOLD frequency response is found to be closely linear in the parameter ranges of interest, with the form of a low-pass filter with a weak resonance at approximately 0.07 Hz. An improved BOLD impulse response is systematically obtained which includes initial dip and post-stimulus undershoot for some parameter ranges. It is found that the BOLD response depends strongly on the precise temporal course of the evoked neuronal activity, not just its peak value or typical amplitude. Indeed, for short stimuli, the linear BOLD response is closely proportional to the time-integrated activity change evoked by the stimulus, regardless of amplitude. It is concluded that there can be widely differing proportionalities between BOLD and peak activity, that this is the likely reason for the low level of correspondence seen experimentally between ERP sources and BOLD measurements and that non-BOLD measurements, such as ERPs, can be used to correct for this effect to obtain improved activity estimates. Finally, stimulus sequences that optimize the signal-to-noise ratio in event-related BOLD fMRI (efMRI) experiments are derived using the hemodynamic transfer function.

Hemodynamic traveling waves in human visual cortex.

Functional MRI (fMRI) experiments rely on precise characterization of the blood oxygen level dependent (BOLD) signal. As the spatial resolution of fMRI reaches the sub-millimeter range, the need for quantitative modelling of spatiotemporal properties of this hemodynamic signal has become pressing. Here, we find that a detailed physiologically-based model of spatiotemporal BOLD responses predicts traveling waves with velocities and spatial ranges in empirically observable ranges. Two measurable parameters, related to physiology, characterize these waves: wave velocity and damping rate. To test these predictions, high-resolution fMRI data are acquired from subjects viewing discrete visual stimuli. Predictions and experiment show strong agreement, in particular confirming BOLD waves propagating for at least 5–10 mm across the cortical surface at speeds of 2–12 mm s-1. These observations enable fundamentally new approaches to fMRI analysis, crucial for fMRI data acquired at high spatial resolution.

Spatiotemporal hemodynamic response functions derived from physiology

Probing neural activity with functional magnetic resonance imaging (fMRI) relies upon understanding the hemodynamic response to changes in neural activity. Although existing studies have extensively characterized the temporal hemodynamic response, less is understood about the spatial and spatiotemporal hemodynamic responses. This study systematically characterizes the spatiotemporal response by deriving the hemodynamic response due to a short localized neural drive, i.e., the spatiotemporal hemodynamic response function (stHRF) from a physiological model of hemodynamics based on a poroelastic model of cortical tissue. In this study, the models boundary conditions are clarified and a resulting nonlinear hemodynamic wave equation is derived. From this wave equation, damped linear hemodynamic waves are predicted from the stHRF. The main features of these waves depend on two physiological parameters: wave propagation speed, which depends on mean cortical stiffness, and damping which depends on effective viscosity. Some of these predictions were applied and validated in a companion study (Aquino et al., 2012). The advantages of having such a theory for the stHRF include improving the interpretation of spatiotemporal dynamics in fMRI data; improving estimates of neural activity with fMRI spatiotemporal deconvolution; and enabling wave interactions between hemodynamic waves to be predicted and exploited to improve the signal to noise ratio of fMRI.

Mean field theory of the coherent to random-phase state transition in three-wave interactions

The crossover of three-wave interactions from the coherent monochromatic limit to the wide bandwidth random-phase limit is investigated as the bandwidth of the waves is varied in a system exhibiting nonlinear three-wave oscillations. A recently observed sudden transition between the coherent and incoherent interaction is confirmed. As the bandwidth is increased from the monochromatic limit, it is found that the coherence of the interaction decreases slowly. At the transition point of the interaction the coherence then falls abruptly and nonlinear oscillations cease. An analytic mean-field approach is used to model the transition. Below the transition point, each frequency component of the wave spectra oscillates at its own individual frequency about the aggregate quasicoherent oscillation of the wave as a whole. It is when the frequency component at the spectral edge cannot sustain such oscillations that the system switches to random phase behavior. The analytic model’s estimate of the transition point and other interaction properties agrees semiquantitatively with numerical solutions of the full equations.

A Multiscale “Working Brain” Model

By modeling salient features of the corticothalamic system over multiple spatial and temporal scales, physiologically based neural field theory has yielded numerous successful predictions that interrelate stimuli, neural activity, and measurements. Likewise, physiologically based neural mass theories of the brainstem-hypothalamus sleep-wake switch and associated systems have recently been developed and shown to quantitatively reproduce a wide variety of arousal-state phenomena. In both cases, model parameters have been independently constrained, and each model has integrated multiple phenomena and measurements into a single unified framework, thereby validating the modeling approach and enabling these features to be interrelated and interpreted in terms of underlying physiology and anatomy. Here, a first integration of the corticothalamic and arousal-state models is carried out by incorporating a simple model of their couplings: upward via the neuromodulatory effects of the ascending arousal system, and downward via the gating of light inputs by higher-level behavior. The resulting “working brain” system has a neural-mass-like limit, governed by delay differential equations that enable it to respond correctly to light-dark cycles, sleep deprivation, jetlag, and pharmacological inputs, while driving the corticothalamic system into parameter regions where it reproduces associated electroencephalograms, evoked response potentials, and other phenomena, whose properties are further elucidated by retaining the appropriate neural field equations. Overall, the combined model provides a simple, highly flexible framework for quantitatively modeling a variety of mesoscale to macroscale brain phenomena, ranging from normal behaviors to highly nonlinear dynamics such as found in seizures, and for examining interactions between such phenomena. these findings are illustrated with representative examples. Fitting of the model to data can be used to infer brain states and underlying parameters.

NFTsim: Theory and Simulation of Multiscale Neural Field Dynamics

A user ready, portable, documented software package, NFTsim, is presented to facilitate numerical simulations of a wide range of brain systems using continuum neural field modeling. NFTsim enables users to simulate key aspects of brain activity at multiple scales. At the microscopic scale, it incorporates characteristics of local interactions between cells, neurotransmitter effects, synaptodendritic delays and feedbacks. At the mesoscopic scale, it incorporates information about medium to large scale axonal ranges of fibers, which are essential to model dissipative wave transmission and to produce synchronous oscillations and associated cross-correlation patterns as observed in local field potential recordings of active tissue. At the scale of the whole brain, NFTsim allows for the inclusion of long range pathways, such as thalamocortical projections, when generating macroscopic activity fields. The multiscale nature of the neural activity produced by NFTsim has the potential to enable the modeling of resulting quantities measurable via various neuroimaging techniques. In this work, we give a comprehensive description of the design and implementation of the software. Due to its modularity and flexibility, NFTsim enables the systematic study of an unlimited number of neural systems with multiple neural populations under a unified framework and allows for direct comparison with analytic and experimental predictions. The code is written in C++ and bundled with Matlab routines for a rapid quantitative analysis and visualization of the outputs. The output of NFTsim is stored in plain text file enabling users to select from a broad range of tools for offline analysis. This software enables a wide and convenient use of powerful physiologically-based neural field approaches to brain modeling. NFTsim is distributed under the Apache 2.0 license.

Spatiotemporal dynamics of pattern formation in the primary visual cortex and hallucinations

The existence of visual hallucinations with prominent temporal oscillations is well documented in conditions such as Charles Bonnett Syndrome. To explore these phenomena, a continuum model of cortical activity that includes additional physiological features of axonal propagation and synapto-dendritic time constants, is used to study the generation of hallucinations featuring both temporal and spatial oscillations. A detailed comparison of the physiological features of this model with those of two others used previously in the modeling of hallucinations is made, and differences, particularly regarding temporal dynamics, relevant to pattern formation are analyzed. Linear analysis and numerical calculation are then employed to examine the pattern forming behavior of this new model for two different forms of spatiotemporal coupling between neurons. Numerical calculations reveal an oscillating mode whose frequency depends on synaptic, dendritic, and axonal time constants not previously simultaneously included in such analyses. Its properties are qualitatively consistent with descriptions of a number of physiological disorders and conditions with temporal dynamics, but the analysis implies that corticothalamic effects will need to be incorporated to treat the consequences quantitatively.

Spatiotemporally varying visual hallucinations: I. Corticothalamic theory

The thalamus is introduced to a recent model of the visual cortex to examine its effect on pattern formation in general and the generation of temporally oscillating patterns in particular. By successively adding more physiological details to a basic corticothalamic model, it is determined which features are responsible for which effects. In particular, with the addition of a thalamic population, several changes occur in the spatiotemporal power spectrum: power increases at resonances of the corticothalamic loop, while the loop acts as a spatiotemporal low-pass filter, and synaptic and dendritic dynamics temporally low-pass filter the activity more generally. Investigation of the effect of altering parameters and gains reveals new parameter regimes where activity that corresponds to hallucinations is induced by both spatially homogeneous and inhomogeneous temporally oscillating modes. This suggests that the thalamus and corticothalamic loops are essential components of a model of oscillating visual hallucinations.