edge-wise (at each entry of the correlation matrix)Īdditionally, the relationship of the matrix of edge-wise correlations between FC and motion (across participants) to the matrix of Euclidean distances between pairs of regional centroids can be studied. regionally (averaging FC for each node, by calculating the average over rows (or columns) of the correlation matrix) globally (averaging FC over the upper triangular part of the correlation matrix) The mean FD can be used to detect potential residual relationships between functional connectivity (FC, usually defined as the cross-correlation between processed fMRI time-series between pairs of regions) and movement, across subjects. Relationship of Framewise Displacement to Functional Connectivity ![]() ![]() The thresholds to choose are likely to be study-specific - but should be based on the knowledge that even seemingly very small magnitude of motion (< 0.05 mm) can have artefactual impacts on the data ( Byrge and Kennedy, 2018). (2015) excluded participants with mean FD > 0.2 mm, and those with > 20/124 volumes with FD > 0.25 mm. There is no universally accepted threshold for movement, but past studies have excluded participants both on the grounds of high average FD, and of a high proportion of motion-contaminated time points. The exclusion of a small subset of high-motion participants can substantially improve overall data quality, and decrease the strength of the relationship between motion and functional connectivity across subjects, which can be driven by a few high-motion outliers. The values of FD should be inspected - both the mean FD, as well as subject-specific distributions of FD over time. For each participant, a single (scalar) estimate of overall motion, the mean FD, can be calculated by averaging the FD time series. Subsequently, these were used to calculate an overall estimate of motion - the framewise displacement (FD), defined as the sum of the absolute temporal derivatives of the six motion parameters, following conversion of rotational parameters to distances by computing the arc length displacement on the surface of a sphere with radius 50 mm (as in Power et al. During processing, re-alignment of scans is used to estimate 6 motion parameters for each participant (3 translation parameters and 3 rotation parameters). One key quantity used for diagnostic purposed is the framewise displacement (FD), a subject-specific time-series indexing an overall estimate of movement over time. The following steps apply across types of fMRI acquisition (e.g.: single- or multi-echo data) and across processing pipelines. Multiple diagnostic measures can be investigated during quality control of fMRI data, to ensure that the processed data are maximally free of motion-related (or other) artefacts. From left to right: rotation, translation, 1st derivative of rotation and translation Other prominent artifacts, particularly in fMRI, relate to respiration, vasculature and arousal (e.g.: Murphy et al., 2013 Caballero-Gaudes and Reynolds, 2017). The latter method was subsequently extended to provide revised estimates of voxel-wise effective degrees of freedom (df) of the BOLD time series, which due to denoising are lower than the nominal N(df ) = N(time-points), and which affect estimates of edge probability when incorporated into network analysis ( Patel and Bullmore, 2016). Later solutions included censoring of motion-affected frames ( Power et al., 2012), removal of non-BOLD (artefactual) signal using independent component analysis applied to multi-echo fMRI data, based on the dependence of the BOLD signal on echo time (Kundu et al., 2012, 2013), or "despiking" of motion-related non-stationary events from a wavelet decomposition of the signal ( Patel et al., 2014). ![]() Initial motion-correction approaches involved regression of motion parameters and their derivatives from voxel-wise BOLD time series ( Friston et al., 1996), or regression of the average or "global" signal ( Aguirre et al., 1998 for a recent review, see Murphy and Fox, 2017). Crucially, motion can substantially affect estimates of functional connectivity ( Power et al., 2012 Satterthwaite et al., 2012). This was recognized as an issue in fMRI long ago ( Hajnal et al., 1994 Friston et al., 1996).ĭue to the spin excitation history issue already pointed out by Bullmore et al., 1999( PDF), motion has complex effects on the signal - including increases, decreases or complex waveforms, depending on factors such as the timing, duration and trajectory of motion ( Power et al., 2015, Byrge and Kennedy 2018). ![]() Participant in-scanner motion is one of the prominent sources of artefacts in fMRI data.
0 Comments
Leave a Reply. |