Figure 1
From: Using carpet plots to analyze transit times of low frequency oscillations in resting state fMRI
Demonstration of carpet plot reconstruction and slope identification. (a) Carpet plot reconstruction for DSC-MRI and for rs-fMRI. Carpet plots for DSC-MRI scans were constructed by reordering voxels in descending order based on time to peak (TTP). Carpet plots for rs-fMRI were reconstructed by reordering voxels based on the delay time derived from the cross-correlation between the sLFO signal for a given voxel and the averaged global sLFO across all voxels in the brain. MRI data processing and image visualization was assisted by use of the FMRIB Software Library (FSL, v5.0 for DSC and rs-fMRI data, v6.0 for the CO2-MRI data, https://fsl.fmrib.ox.ac.uk/fsl/fslwiki)24. (b) Slope-detecting algorithm developed for computation of carpet plot edge slopes and transit times. The original image (i) is smoothed with a 2D blurring filter (ii), followed by a horizontal derivative filter (iii). For every row of (iii), the maximum derivative location is located and denoted by a red point. These points are used as data to compute a linear regression line or edge, shown in (iv), from which the slope (steepness of the edge) and the transit time (time difference between the bottom and top of an identified edge indicating the time taken for the corresponding fluctuation to traverse all voxels in the carpet plot) can be calculated.
