A novel method of calculating cardiac pulsatility from resting state fMRI data

Measures of cardiac pulsatility were generated from resting-state fMRI data using ICA information and processed HRV traces. This was achieved by performing a correlation between the ICA component and HRV time series to isolate cardiac-related components from non-cardiac-related components. Classified datasets were used to train the FIX noise correction algorithm. FIX was used to classify cardiac components in 4123 rfMRI datasets. Measures of voxelwise cardiac pulsatility were generated be determining the R2 variance explained in the voxel time series by the classified cardiac components. Comparisons with gold standard pulsatility measures derived from the HRV traces showed high correlation values.