Limits...
MEG and fMRI Fusion for Non-Linear Estimation of Neural and BOLD Signal Changes.

Plis SM, Calhoun VD, Weisend MP, Eichele T, Lane T - Front Neuroinform (2010)

Bottom Line: Neural activity estimates were derived using a dynamic Bayesian network with continuous real valued parameters by means of a sequential Monte Carlo technique.The highly non-linear model of the BOLD response poses a difficult inference problem for neural activity estimation; computational requirements are also high due to the time and space complexity.We show that joint analysis of the data improves the system's behavior by stabilizing the differential equations system and by requiring fewer computational resources.

View Article: PubMed Central - PubMed

Affiliation: The Mind Research Network Albuquerque, NM, USA.

ABSTRACT
The combined analysis of magnetoencephalography (MEG)/electroencephalography and functional magnetic resonance imaging (fMRI) measurements can lead to improvement in the description of the dynamical and spatial properties of brain activity. In this paper we empirically demonstrate this improvement using simulated and recorded task related MEG and fMRI activity. Neural activity estimates were derived using a dynamic Bayesian network with continuous real valued parameters by means of a sequential Monte Carlo technique. In synthetic data, we show that MEG and fMRI fusion improves estimation of the indirectly observed neural activity and smooths tracking of the blood oxygenation level dependent (BOLD) response. In recordings of task related neural activity the combination of MEG and fMRI produces a result with greater signal-to-noise ratio, that confirms the expectation arising from the nature of the experiment. The highly non-linear model of the BOLD response poses a difficult inference problem for neural activity estimation; computational requirements are also high due to the time and space complexity. We show that joint analysis of the data improves the system's behavior by stabilizing the differential equations system and by requiring fewer computational resources.

No MeSH data available.


Blood oxygenation level dependent (top) and neural activity (bottom) estimation plots. Each plot displays the averaged BOLD response (red circles) plotted with the corresponding signal estimate produced by our Bayesian sensor fusion model (blue squares). Since true neural activity is not known for the real dataset only the estimate is displayed in the bottom plots. Horizontal axes give time (seconds), while vertical axes are arbitrary units for signal response. The BOLD response is perfectly tracked by fMRI (A) as well as by the joint analysis (C). However, MEG completely fails to track it. fMRI-only analysis fails to track neural response (B). In both cases this is due to varying parameters of the HFM, required since the true parameters are unknown. This lead to underdetermined problem in case of fMRI, creating random results while maintaining perfect fit, and also allowed the MEG-only case to drift off completely, since it was not constrained by the BOLD response. The joint analysis is closely tracking the BOLD response and provides neural activity estimation consistent with expectation from previous knowledge about the experiment (C).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2991230&req=5

Figure 14: Blood oxygenation level dependent (top) and neural activity (bottom) estimation plots. Each plot displays the averaged BOLD response (red circles) plotted with the corresponding signal estimate produced by our Bayesian sensor fusion model (blue squares). Since true neural activity is not known for the real dataset only the estimate is displayed in the bottom plots. Horizontal axes give time (seconds), while vertical axes are arbitrary units for signal response. The BOLD response is perfectly tracked by fMRI (A) as well as by the joint analysis (C). However, MEG completely fails to track it. fMRI-only analysis fails to track neural response (B). In both cases this is due to varying parameters of the HFM, required since the true parameters are unknown. This lead to underdetermined problem in case of fMRI, creating random results while maintaining perfect fit, and also allowed the MEG-only case to drift off completely, since it was not constrained by the BOLD response. The joint analysis is closely tracking the BOLD response and provides neural activity estimation consistent with expectation from previous knowledge about the experiment (C).

Mentions: The results of application of our method to the real data are displayed in Figure 14. They are presented in the same manner as the simulation results of Figure 5. The gray strip signifies the interval at which the flashing checkerboard stimulus was presented. fMRI-only estimation in Figure 14A is able to track the data (the BOLD response) almost perfectly. However, it produces a noisy neural activity estimate. Although it does increase together with the data similarly as observed in simulations. Note, that when estimation of neural activity is performed together with estimating the system's parameters we are dealing with an underdetermined system, i.e., many different solutions fit the objective perfectly and we are left with inverse problem (Riera et al., 2004). We attribute poor performance of the fMRI-only estimation to this inverse problem. MEG-only estimation is quite opposite: it estimates the neural activity consistently with the stimulus presentation (Figure 14B) and fails at tracking the data. This happens since the MEG-only estimation is unconstrained by the fMRI data and, obviously, HFM parameters freely drift away from the solution region. Results of the joint analysis of MEG and fMRI (Figure 14C) support our findings in the simulated experiments: fMRI data is traced exactly and neural activity is estimated as expected. Figure 14 displays neural activity as the absolute value of the estimate, the way it inputs the HFM.


MEG and fMRI Fusion for Non-Linear Estimation of Neural and BOLD Signal Changes.

Plis SM, Calhoun VD, Weisend MP, Eichele T, Lane T - Front Neuroinform (2010)

Blood oxygenation level dependent (top) and neural activity (bottom) estimation plots. Each plot displays the averaged BOLD response (red circles) plotted with the corresponding signal estimate produced by our Bayesian sensor fusion model (blue squares). Since true neural activity is not known for the real dataset only the estimate is displayed in the bottom plots. Horizontal axes give time (seconds), while vertical axes are arbitrary units for signal response. The BOLD response is perfectly tracked by fMRI (A) as well as by the joint analysis (C). However, MEG completely fails to track it. fMRI-only analysis fails to track neural response (B). In both cases this is due to varying parameters of the HFM, required since the true parameters are unknown. This lead to underdetermined problem in case of fMRI, creating random results while maintaining perfect fit, and also allowed the MEG-only case to drift off completely, since it was not constrained by the BOLD response. The joint analysis is closely tracking the BOLD response and provides neural activity estimation consistent with expectation from previous knowledge about the experiment (C).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2991230&req=5

Figure 14: Blood oxygenation level dependent (top) and neural activity (bottom) estimation plots. Each plot displays the averaged BOLD response (red circles) plotted with the corresponding signal estimate produced by our Bayesian sensor fusion model (blue squares). Since true neural activity is not known for the real dataset only the estimate is displayed in the bottom plots. Horizontal axes give time (seconds), while vertical axes are arbitrary units for signal response. The BOLD response is perfectly tracked by fMRI (A) as well as by the joint analysis (C). However, MEG completely fails to track it. fMRI-only analysis fails to track neural response (B). In both cases this is due to varying parameters of the HFM, required since the true parameters are unknown. This lead to underdetermined problem in case of fMRI, creating random results while maintaining perfect fit, and also allowed the MEG-only case to drift off completely, since it was not constrained by the BOLD response. The joint analysis is closely tracking the BOLD response and provides neural activity estimation consistent with expectation from previous knowledge about the experiment (C).
Mentions: The results of application of our method to the real data are displayed in Figure 14. They are presented in the same manner as the simulation results of Figure 5. The gray strip signifies the interval at which the flashing checkerboard stimulus was presented. fMRI-only estimation in Figure 14A is able to track the data (the BOLD response) almost perfectly. However, it produces a noisy neural activity estimate. Although it does increase together with the data similarly as observed in simulations. Note, that when estimation of neural activity is performed together with estimating the system's parameters we are dealing with an underdetermined system, i.e., many different solutions fit the objective perfectly and we are left with inverse problem (Riera et al., 2004). We attribute poor performance of the fMRI-only estimation to this inverse problem. MEG-only estimation is quite opposite: it estimates the neural activity consistently with the stimulus presentation (Figure 14B) and fails at tracking the data. This happens since the MEG-only estimation is unconstrained by the fMRI data and, obviously, HFM parameters freely drift away from the solution region. Results of the joint analysis of MEG and fMRI (Figure 14C) support our findings in the simulated experiments: fMRI data is traced exactly and neural activity is estimated as expected. Figure 14 displays neural activity as the absolute value of the estimate, the way it inputs the HFM.

Bottom Line: Neural activity estimates were derived using a dynamic Bayesian network with continuous real valued parameters by means of a sequential Monte Carlo technique.The highly non-linear model of the BOLD response poses a difficult inference problem for neural activity estimation; computational requirements are also high due to the time and space complexity.We show that joint analysis of the data improves the system's behavior by stabilizing the differential equations system and by requiring fewer computational resources.

View Article: PubMed Central - PubMed

Affiliation: The Mind Research Network Albuquerque, NM, USA.

ABSTRACT
The combined analysis of magnetoencephalography (MEG)/electroencephalography and functional magnetic resonance imaging (fMRI) measurements can lead to improvement in the description of the dynamical and spatial properties of brain activity. In this paper we empirically demonstrate this improvement using simulated and recorded task related MEG and fMRI activity. Neural activity estimates were derived using a dynamic Bayesian network with continuous real valued parameters by means of a sequential Monte Carlo technique. In synthetic data, we show that MEG and fMRI fusion improves estimation of the indirectly observed neural activity and smooths tracking of the blood oxygenation level dependent (BOLD) response. In recordings of task related neural activity the combination of MEG and fMRI produces a result with greater signal-to-noise ratio, that confirms the expectation arising from the nature of the experiment. The highly non-linear model of the BOLD response poses a difficult inference problem for neural activity estimation; computational requirements are also high due to the time and space complexity. We show that joint analysis of the data improves the system's behavior by stabilizing the differential equations system and by requiring fewer computational resources.

No MeSH data available.