Choosing Wavelet Methods, Filters, and Lengths for Functional Brain Network Construction

Abstract

Wavelet methods are widely used to decompose fMRI, EEG, or MEG signals into time series representing neurophysiological activity in fixed frequency bands. Using these time series, one can estimate frequency-band specific functional connectivity between sensors or regions of interest, and thereby construct functional brain networks that can be examined from a graph theoretic perspective. Despite their common use, however, practical guidelines for the choice of wavelet method, filter, and length have remained largely undelineated. Here, we explicitly explore the effects of wavelet method (MODWT vs. DWT), wavelet filter (Daubechies Extremal Phase, Daubechies Least Asymmetric, and Coiflet families), and wavelet length (2 to 24)—each essential parameters in wavelet-based methods—on the estimated values of graph metrics and in their sensitivity to alterations in psychiatric disease. We observe that the MODWT method produces less variable estimates than the DWT method. We also observe that the length of the wavelet filter chosen has a greater impact on the estimated values of graph metrics than the type of wavelet chosen. Furthermore, wavelet length impacts the sensitivity of the method to detect differences between health and disease and tunes classification accuracy. Collectively, our results suggest that the choice of wavelet method and length significantly alters the reliability and sensitivity of these methods in estimating values of metrics drawn from graph theory. They furthermore demonstrate the importance of reporting the choices utilized in neuroimaging studies and support the utility of exploring wavelet parameters to maximize classification accuracy in the development of biomarkers of psychiatric disease and neurological disorders.

Positive affect, surprise, and fatigue are correlates of network flexibility

“…We repeated this analysis for all scan sessions.

Figure 1Schematic figure illustrating network construction, multi-layer network model, and community detection output. ( A ) Functional connectivity networks were generated by calculating the magnitude squared wavelet coherence for all pairs of regional fMRI BOLD time series 73 . The resulting coherence estimates are arranged in a region-by-region functional connectivity matrix.

…”

“…( A ) Functional connectivity networks were generated by calculating the magnitude squared wavelet coherence for all pairs of regional fMRI BOLD time series 73 . The resulting coherence estimates are arranged in a region-by-region functional connectivity matrix.…”

Multi-scale brain networks

“…Functional connectivity (FC) networks, on the other hand, refer to the strength of the statistical relationship between nodes’ activity over time (Friston, 2011). Usually this statistical relationship is operationalized as a Fisher-transformed correlation coefficient (Zalesky et al, 2012) or a coherence measure (Zhang et al, 2016). Both SC and FC networks are represented with a connectivity matrix, A , whose element A ij is equal to the connection weight between regions i and j .…”

Cohesive network reconfiguration accompanies extended training

“…These scans were divided into 112 regions based on the Harvard-Oxford Atlas, a probabilistic atlas covering cortical and subcortical areas [Desikan et al, 2006]. Connections or edges between nodes represented the pairwise coherence of the average fMRI time series for a pair of brain regions [Bassett et al, 2011;Braun et al, 2015;Zhang et al, 2016].…”