Magn Reson Med. 1997 Apr;37(4):631-6.

Limitations of temporal resolution in functional MRI.

Kim SG1, Richter W, Uğurbil K.

Author information

1Department of Radiology, University of Minnesota Medical School, Minneapolis 55455, USA.

Abstract

In fMRI, images can be collected in a very short time; therefore, high temporal resolution is possible in principle. However, the temporal resolution is limited by a blurred intrinsic hemodynamic response and a finite signal-to-noise ratio. To determine the upper limit of temporal resolution in a single area during repeated tasks, motor cortex activity was investigated during visually instructed finger movements. Without averaging, a sequence of four single-finger movements with an execution time of approximately 2 s can be resolved when the delay time between consecutive sequences is at least 3 s. The hemodynamic response time is constant for each subject, but not among different subjects. The temporal resolution can be better when the signal from spatially distinct regions is examined. For a series of experiments involving a visually instructed delayed cued finger movement task with a well-defined, independently determined, variable delay time, time courses in the motor area are distinct from each other in two experiments if the difference in delay time is as little as 2 s. The activation in the visual area due to the presentation of the task serves here as an internal time reference. By comparing a set of fMRI time courses in multiple distinct areas, serial neural processing may be investigated.

Look if this helps you and contact the researchers, they may be here on RG.

It means that it takes 'a long time' to acquire two consecutive brain volumes. Usually, fMRI acquires one volume every 2 or 3 seconds. This is a disadvantage, since we have no signal of what happened between that time interval. It also means that at the end of the experiment we will only have a relatively few number of volumes. For example, if you scan a person for 5 minutes, you will get less than 300 time points. Compare this with other methods, like EEG, that can acquire hundreds of time points per second.

Why does it matter? Many common signal processing techniques require many time points to 'perform well'. So it can be challenging to adapt then to fmri

In the end, it all boils down to the usual weakness of nuclear magnetic resonance which is its limited sensitivity. Low sensitivity can be compensated by long monitoring (data acquisition) times, but that reduces the temporal resolution. It is as simple as that.

For fMRI, the issue is not solely sensitivity. For typical resolutions, using multi-band techniques it is now possible to get whole brain coverage is less than a second. But fMRI is measuring the BOLD response, which is a complicated sampling of the hemodynamic response. The hemodynamic response to neuronal activity takes several seconds to peak. Neuronal events, on the other hand, happen on the millisecond time scale. So the BOLD response is a slow indirect measure of neuronal activity. That said, people have shown consistent differences in the onset of the BOLD response to small delays in neuronal activity, But since the hemodynamic response function varies across the brain, comparing onset times of different locations can be misleading.

The metabolic activity recorded by fMRI takes place after the actual shifts have occurred, so that the temporal resolution of this methods is weak.

fMRI utilizes very fast magnetic resonance images collected serially in order to measure the changes in blood flow associated with cognitive activity. Thus, spatial resolution with fMRI is excellent despite weak temporal resolution as the changes in blood flow do not precisely co-occur with the event. More recently, the development of methods for rapid acquisition of MRI slices has largely been proven to be more sensitive to brain function, with much better resolution. These methods essentially use conventional MRI scanners with software that permits fast image acquisition to detect alterations in blood flow and volume.

I think that account should also be taken of the fact that temporal processes recorded via fMRI are part of longer drawn neuronal sequences of activation initiated in the limbic system and that also terminate back into the limbic system.

The intensity of the temporal signals seem to be related to the intensity of the amygdala triggering response to a wide range of stimuli:

https://doi.org/10.4236/ce.2019.102028

It depends on the research question or context. It may not be too bad.