Estimation of head motion in anesthesia – artifact or real data?

Dear all,

I am interested in head motion under anesthesia. As we all know, subjects exhibit variability in head motion when lying awake in the scanner.
Under anesthesia, all subjects of a dataset that I am working on show the same kind of small but ongoing, rapid, and “spike-like” head motion.

Since we can only estimate head motion in fMRI, my question is if the time-series of the motion parameters approximately reflect real head motion, or if this “white noise like head motion” is rather an artifact by AFNI’s head motion estimation (when there is probably little or almost no head motion in anesthesia)?

I attach an paradigmatic image of AFNIs QC for one subject below.


Hi, Philipp-

We have seen similar rapid variations in other cases of breathing apparatuses used in FMRI.

You should open the initial EPI in the GUI, and flip through the volumes (e.g., open a graph window and hit the right arrow key over it to surf through the data). Does the subject appear to be moving?

My guess is that these oscillating, estimated “micromovements”—I wouldn’t call them white noise, because they appear too regular—are likely related to the breathing apparatus, perhaps affecting the magnetic field slightly or related to breathing movements. Unless the head is fixed with a styrofoam cap or something, the subject can still move, even though anaesthetized (e.g., their chest motion will follow breathing), and their breathing pattern can be different than normal. But again, my guess would be, based on the rapid frequency, that it is related to the machine a bit.



thanks for the information and thoughts. There is almost no head motion between the single TRs in the raw data. I therefore assume that your thoughts are going in the right direction, namely that the recorded motion could stem from breathing.

Just adding on to Paul’s comment, we did find the ventilator and related equipment can affect the BOLD signal.

Breathing induces phase encoding directional artifact motion.

Raj, D., Anderson, A.W., Gore, J.C., 2001. Respiratory effects in human functional magnetic resonance imaging due to bulk susceptibility changes. Phys Med Biol 46, 3331-3340.

If your acquisition is A to P PE, each slice has the different amount of A to P directional motion, depending on the breathing cycle, even without the actual head motion. Since EPI volume image stacks up each different amount of A to P motion slices, each volume has different combination of A to P motion slices, from which 3dvolreg estimates the artificial volume motion. Postmortem or ex-vivo brain phantom would generate the actual motion free data. If TR is coherent to the breathing cycle under anesthesia, the slice wise A to P motion has less variation across volumes and the estimated 3d rigid motion is minimized.

Hope it helps

1 Like