Towards artefact-free awake rat brain fMRI/EEG studies with multiband-SWIFT fMRI
Jaakko Paasonen1, Hanne Laakso1, Tiina Pirttimäki1, Petteri Stenroos1, Lauri Lehto2, Michael Garwood2, Shalom Michaeli2, Djaudat Idiyatullin2, Silvia Mangia2, and Olli Gröhn1

1A.I.V. Institute for Molecular Sciences, University of Eastern Finland, Kuopio, Finland, 2Center for Magnetic Resonance Research, University of Minnesota, Minneapolis, MN, United States


EPI sequence is the most common choice for fMRI investigations. EPI, however, is loud for awake studies, prone to susceptibility-induced artefacts, and sensitive to motion. Therefore, we investigated whether the unique features of MB-SWIFT would make it a potential alternative for EPI to overcome these issues. Our results suggest that MB-SWIFT is dramatically more silent compared to EPI, body movement does not affect the voxels in brain, gradient-switching artefacts are minimal in electrophysiological data, and good quality resting-state fMRI data can be obtained from awake rats. We conclude that MB-SWIFT is an advantageous alternative to EPI in demanding experimental applications.


Rodent fMRI studies exploiting animal models can provide invaluable information of normal and abnormal brain function. When fMRI is combined with optical and/or electrical recording and stimulation techniques, multiple scales of brain network activity can be assessed and manipulated. However, simultaneous use of several sensitive measurement techniques is not straightforward and can cause artefacts, such as susceptibility-induced distortions and magnetic field gradient switching artefacts. Rodent fMRI studies are mostly performed with echo planar imaging (EPI), which is one of the loudest pulse sequences and is sensitive to movement and magnetic susceptibility differences caused by surgical manipulations of skull and/or implantation of measuring or stimulation devices.

Another caveat for animal fMRI studies is anesthesia, which is known to alter brain function 1. Several awake rodent fMRI approaches have been introduced 2, but they are still influenced by the loud noise of the scanner causing stress to the animals and movement artefacts. Recently, we introduced multiband (MB) -SWIFT fMRI in the context of deep brain stimulation of rat and were able to acquire fMRI responses comparable to those obtained with spin-echo (SE) EPI with minimal susceptibility artefact from tungsten electrode implanted into ventral posteromedial nucleus 3. Therefore, the aim of this work was to further test how unique features of MB-SWIFT, namely, large bandwidth and small steps in gradient switching, can be exploited to reduce susceptibility-related artefacts, audible noise, sensitivity to movement, and gradient switching artefacts in simultaneous electrophysiological recordings.


Eight male rats (Sprague-Dawley or Wistar, total n=8) were used in experiments. The rats were anesthetized with either urethane (1.25 g/kg), isoflurane (1.3-2.0%), or combination of medetomidine (0.025mg/kg) and isoflurane (0.25-0.5%).

All MRI data was acquired in 9.4T/31 cm bore magnet interfaced to Agilent console using surface transmit-receive RF-coil. MB-SWIFT data were acquired with TR = 0.97 ms, 2000 spokes per volume, temporal resolution of 2 s, bandwidth = 192 kHz, matrix size = 643, FOV = 3.5x3.5x6.4 cm3 or 4.8x4.8x 4.8 cm3, and flip angle = 6°. SE-EPI data were acquired with following parameters: TR = 2 s, TE = 30/35 ms, matrix size = 64x32/64, FOV = 4.0x4.0 cm2 or 4.8x4.8 cm2, and 8 slices with 1/1.5 mm thickness.

Acoustic noise level inside the magnet bore was measured by using Audio-Technica MT830R omnidirectional condenser microphone, attached to Focusrite Scarlett 2i2 audio interface.

One isoflurane-anesthetized rat was used to compare the sensitivity of MB-SWIFT and SE-EPI fMRI to body movement in a “marionette test“, where controlled body movement was induced by pulling threads lifting the animal body inside the bore during the scan. The experiment consisted of 60s baseline and 30s controlled movement, repeated three times with both fMRI sequences.

Five rats underwent a four-day training period for awake imaging, and were scanned for resting-state fMRI on the fifth day as described earlier 2.

Two rats were used to compare the magnitude of gradient switching artefacts in simultaneously obtained electrophysiological data between MB-SWIFT and SE-EPI sequences. Electroencephalography (EEG) or local field potentials (LFP) were recorded at 2.5-5 kHz sampling rate with MRI-compatible BrainAmp system from electrodes implanted in either hippocampus (LFP) or on top of somatosensory cortex (EEG).


The measured peak acoustic noise levels were 91 dB and 127 dB for the MB-SWIFT and SE-EPI fMRI sequences, respectively, which means that the peak sound pressure is 63 times lower in MB-SWIFT. In terms of psychoacoustics, the experienced peak loudness is roughly 12 times lower in MB-SWIFT.

Controlled body movement-induced signal variations ranged from -5.5% to 8.9% of baseline in EPI time series, while they were hardly detectable (from -1.0% to 0.9%) in MB-SWIFT data (Figure 1).

The initial analysis of awake fMRI data indicates that several anatomically well-localized independent components, suggesting functional connectivity, can be obtained with MB-SWIFT from conscious animals (Figure 2).

The combined fMRI and EEG/LFP measurements indicate that MB-SWIFT induces hardly detectable gradient switching artefacts to electrophysiological data while compared to SE-EPI (Figure 3). The electrophysiological data measured during MB-SWIFT can be further improved by applying a simple low-pass filter, while SE-EPI-induced artefacts remain clearly visible in the electrophysiological data after similar filtering.

Discussion and Conclusion

Our results suggest that MB-SWIFT has unique features making it a valuable tool for experimental fMRI studies combining different modalities. Furthermore, insensitivity to movement artefacts and relative quietness of the pulse sequence makes it optimal approach for awake fMRI studies. The origin of the fMRI contrast of this almost zero-echo-time sequence is likely related to in-flow of blood 3, which may make the approach sensitive to RF-coil geometry and warrants future characterization.


This work was supported by the NIH grant U01-NS103569-01, the Center for Magnetic Resonance Research NIH grants P30-NS07640 and P41-EB015894, and Jane & Aatos Erkko Foundation.


1. Paasonen J, Stenroos P, Salo RA, et al. Functional connectivity under six anesthesia protocols and the awake condition in rat brain. Neuroimage. 2018;172:9-20.

2. Stenroos P, Paasonen J, Salo RA, et al. Awake Rat Brain Functional Magnetic Resonance Imaging Using Standard Radio Frequency Coils and a 3D Printed Restraint Kit. Front Neurosci. 2018;12:548.

3. Lehto LJ, Idiyatullin D, Zhang J, et al. MB-SWIFT functional MRI during deep brain stimulation in rats. Neuroimage. 2017;159:443-448.


Figure 1. The effect of controlled body movement in “marionette test” on fMRI data acquired with either spin-echo EPI or multiband-SWIFT in head-fixed isoflurane-anesthetized rat. The top row illustrates the voxels affected significantly by the body movement, analyzed with block design general linear model (false discovery rate corrected). The bottom row shows the fMRI time series during baseline and controlled body movement (black bar). Individual trials are shown with colors and average across trials with black. Time serie signals are obtained from somatosensory cortex.

Figure 2. Representative independent components obtained with multiband-SWIFT fMRI from individual awake rats during resting-state scans. The components were obtained with FSL MELODIC independent component analysis toolbox. A, anterior medial cortical component; B, anterior intermediate cortical component; C, lateral anterior cortical component; D, posterior lateral cortical component; E, prefrontal/cingulate cortical component; F, orbital/insular cortical component.

Figure 3. The effect of multiband-SWIFT fMRI sequence (A) and spin-echo EPI (B) on simultaneous local field potential measurements, and representative electroencephalography (EEG) data without (C, top) and with burst-suppression states (C, bottom) acquired during MB-SWIFT fMRI scan. Black signals are raw signals, while for red signals low-pass filter with 100Hz cut-off has been applied. The data in A and B was acquired from hippocampus in urethane-anesthetized rat. The data in C was obtained with EEG electrodes located on top of the cortex in isoflurane-medetomidine anesthetized rat.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)