Limits...
Noise-assisted instantaneous coherence analysis of brain connectivity.

Hu M, Liang H - Comput Intell Neurosci (2012)

Bottom Line: Characterizing brain connectivity between neural signals is key to understanding brain function.In our method, fully data-driven MEMD, together with Hilbert transform, is first employed to provide time-frequency power spectra for neural data.Furthermore, a statistical randomization procedure is designed to cancel out the effect of the added noise.

View Article: PubMed Central - PubMed

Affiliation: School of Biomedical Engineering, Science & Health Systems, Drexel University, 3141 Chestnut Street, Philadelphia, PA 19104, USA.

ABSTRACT
Characterizing brain connectivity between neural signals is key to understanding brain function. Current measures such as coherence heavily rely on Fourier or wavelet transform, which inevitably assume the signal stationarity and place severe limits on its time-frequency resolution. Here we addressed these issues by introducing a noise-assisted instantaneous coherence (NAIC) measure based on multivariate mode empirical decomposition (MEMD) coupled with Hilbert transform to achieve high-resolution time frequency representation of neural coherence. In our method, fully data-driven MEMD, together with Hilbert transform, is first employed to provide time-frequency power spectra for neural data. Such power spectra are typically sparse and of high resolution, that is, there usually exist many zero values, which result in numerical problems for directly computing coherence. Hence, we propose to add random noise onto the spectra, making coherence calculation feasible. Furthermore, a statistical randomization procedure is designed to cancel out the effect of the added noise. Computer simulations are first performed to verify the effectiveness of NAIC. Local field potentials collected from visual cortex of macaque monkey while performing a generalized flash suppression task are then used to demonstrate the usefulness of our NAIC method to provide highresolution time-frequency coherence measure for connectivity analysis of neural data.

Show MeSH

Related in: MedlinePlus

Significance test of difference between two perceptual conditions revealed by the NAIC (a) and the wavelet-based method (b). General agreement of two methods is evident, yet the NAIC is able to detect statistically significant difference of perceptual suppression occurring as early as 400 msec after surrounding onset. Level lines are depicted at P < 0.05 (red) and P < 0.01 (blue), respectively.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3368308&req=5

fig11: Significance test of difference between two perceptual conditions revealed by the NAIC (a) and the wavelet-based method (b). General agreement of two methods is evident, yet the NAIC is able to detect statistically significant difference of perceptual suppression occurring as early as 400 msec after surrounding onset. Level lines are depicted at P < 0.05 (red) and P < 0.01 (blue), respectively.

Mentions: As described in Method part, the MEMD was first performed on multichannel multitrial LFP data to produce the IMFs, followed by Hilbert transform to obtain the analytic matrix of data. A random noise matrix with noise variance of 10−4 was then added to the analytic matrix of data to facilitate the calculation of coherence. The high-resolution time-frequency coherence spectrum was finally obtained by applying the proposed statistical randomization procedure in which the noise variance was set to 10−4. Figures 8(a) and 8(b) showed the grand average of the NAIC spectra in the Visible and Invisible conditions, respectively. From this figure, we can see clearly that the 10 Hz coherence initially appeared in both conditions for about 200 msec after the surrounding onset. We then observed a slightly shift of oscillatory frequency to 10–20 Hz with reduced coherence, yet the Visible condition exhibited greater coherence than the Invisible condition. As comparisons, we applied Fourier- and wavelet-based coherence methods to the same neural data, with results shown in Figures 9 and 10, respectively. Based on these figures, we can see that Fourier- and wavelet-based methods exhibited similar coherence patterns but with poor time-frequency resolution. Furthermore, we compared the NAIC spectra between Invisible and Visible conditions to reveal how neural connectivity reflected perceptual suppression. We initially performed point-wise significance test by applying t-test to every time-frequency index between two conditions. As shown in Figure 11(a), significant perceptual suppression effect was evident in about 400 msec after the surrounding onset between 10 and 20 Hz, in which Visible condition showed significantly larger coherence than Invisible condition (P < 0.05, uncorrected). To deal with multiple-comparison problem, for which several methods have been proposed [21, 25, 26], we adopted a clustered-based nonparametric method [21] and found that the significant difference observed between two conditions still survived (P < 0.05). For comparison, we repeated the same statistical procedure to the wavelet-based coherence between two conditions. The resulting significant difference at both P < 0.05 and P < 0.01 is shown in Figure 11(b). Both NAIC and wavelet methods show general agreement about the concentration of significant difference in frequency. However, the NAIC is more sensitive in revealing significant difference of perceptual suppression that occurred as early as 400 msec after surrounding onset. These results together suggest that neural coherence reflects perceptual suppression, and significantly reduced coherence in Invisible condition may be associated with the reduction of brain connectivity.


Noise-assisted instantaneous coherence analysis of brain connectivity.

Hu M, Liang H - Comput Intell Neurosci (2012)

Significance test of difference between two perceptual conditions revealed by the NAIC (a) and the wavelet-based method (b). General agreement of two methods is evident, yet the NAIC is able to detect statistically significant difference of perceptual suppression occurring as early as 400 msec after surrounding onset. Level lines are depicted at P < 0.05 (red) and P < 0.01 (blue), respectively.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3368308&req=5

fig11: Significance test of difference between two perceptual conditions revealed by the NAIC (a) and the wavelet-based method (b). General agreement of two methods is evident, yet the NAIC is able to detect statistically significant difference of perceptual suppression occurring as early as 400 msec after surrounding onset. Level lines are depicted at P < 0.05 (red) and P < 0.01 (blue), respectively.
Mentions: As described in Method part, the MEMD was first performed on multichannel multitrial LFP data to produce the IMFs, followed by Hilbert transform to obtain the analytic matrix of data. A random noise matrix with noise variance of 10−4 was then added to the analytic matrix of data to facilitate the calculation of coherence. The high-resolution time-frequency coherence spectrum was finally obtained by applying the proposed statistical randomization procedure in which the noise variance was set to 10−4. Figures 8(a) and 8(b) showed the grand average of the NAIC spectra in the Visible and Invisible conditions, respectively. From this figure, we can see clearly that the 10 Hz coherence initially appeared in both conditions for about 200 msec after the surrounding onset. We then observed a slightly shift of oscillatory frequency to 10–20 Hz with reduced coherence, yet the Visible condition exhibited greater coherence than the Invisible condition. As comparisons, we applied Fourier- and wavelet-based coherence methods to the same neural data, with results shown in Figures 9 and 10, respectively. Based on these figures, we can see that Fourier- and wavelet-based methods exhibited similar coherence patterns but with poor time-frequency resolution. Furthermore, we compared the NAIC spectra between Invisible and Visible conditions to reveal how neural connectivity reflected perceptual suppression. We initially performed point-wise significance test by applying t-test to every time-frequency index between two conditions. As shown in Figure 11(a), significant perceptual suppression effect was evident in about 400 msec after the surrounding onset between 10 and 20 Hz, in which Visible condition showed significantly larger coherence than Invisible condition (P < 0.05, uncorrected). To deal with multiple-comparison problem, for which several methods have been proposed [21, 25, 26], we adopted a clustered-based nonparametric method [21] and found that the significant difference observed between two conditions still survived (P < 0.05). For comparison, we repeated the same statistical procedure to the wavelet-based coherence between two conditions. The resulting significant difference at both P < 0.05 and P < 0.01 is shown in Figure 11(b). Both NAIC and wavelet methods show general agreement about the concentration of significant difference in frequency. However, the NAIC is more sensitive in revealing significant difference of perceptual suppression that occurred as early as 400 msec after surrounding onset. These results together suggest that neural coherence reflects perceptual suppression, and significantly reduced coherence in Invisible condition may be associated with the reduction of brain connectivity.

Bottom Line: Characterizing brain connectivity between neural signals is key to understanding brain function.In our method, fully data-driven MEMD, together with Hilbert transform, is first employed to provide time-frequency power spectra for neural data.Furthermore, a statistical randomization procedure is designed to cancel out the effect of the added noise.

View Article: PubMed Central - PubMed

Affiliation: School of Biomedical Engineering, Science & Health Systems, Drexel University, 3141 Chestnut Street, Philadelphia, PA 19104, USA.

ABSTRACT
Characterizing brain connectivity between neural signals is key to understanding brain function. Current measures such as coherence heavily rely on Fourier or wavelet transform, which inevitably assume the signal stationarity and place severe limits on its time-frequency resolution. Here we addressed these issues by introducing a noise-assisted instantaneous coherence (NAIC) measure based on multivariate mode empirical decomposition (MEMD) coupled with Hilbert transform to achieve high-resolution time frequency representation of neural coherence. In our method, fully data-driven MEMD, together with Hilbert transform, is first employed to provide time-frequency power spectra for neural data. Such power spectra are typically sparse and of high resolution, that is, there usually exist many zero values, which result in numerical problems for directly computing coherence. Hence, we propose to add random noise onto the spectra, making coherence calculation feasible. Furthermore, a statistical randomization procedure is designed to cancel out the effect of the added noise. Computer simulations are first performed to verify the effectiveness of NAIC. Local field potentials collected from visual cortex of macaque monkey while performing a generalized flash suppression task are then used to demonstrate the usefulness of our NAIC method to provide highresolution time-frequency coherence measure for connectivity analysis of neural data.

Show MeSH
Related in: MedlinePlus