Comparing 1 second resolution estimates of the HR from jittered ER designs to actual 1 second resolution measurements.
This image shows power spectra of fMRI data for nine voxels in visual cortex from an event related visual chequerboard stimulus task. The top power spectra are for an event related design with a TR of 3 seconds, and an SOA of 20 seconds. The resulting 1 second difference between the SOA and a integer multiple of the TR allows a 1 second resolution estimate of the haemodynamic response. The bottom spectra are for the same event related design, but with an actual time resolution of 1 second (i.e. a TR of 1 second). Note that the axes are scaled differently in both figures.
This image shows the estimated haemodynamic responses from the same two event related runs as described above. It can be seen that the estimated 1 second resolution HR (top) is similar to the actual 1 second resolution measured HR (bottom), but contains more “noise”, in the form of high frequency oscillations. These oscillations might be due to the effects of low frequency fluctuations in the raw time series.