Brain-wide mapping of resting-state networks in mice using high-frame rate functional ultrasound

Abstract

Functional ultrasound (fUS) imaging is a method for visualizing deep brain activity based on cerebral blood volume changes coupled with neural activity, while functional MRI (fMRI) relies on the blood-oxygenation-level-dependent signal coupled with neural activity. Low-frequency fluctuations (LFF) of fMRI signals during resting-state can be measured by resting-state fMRI (rsfMRI), which allows functional imaging of the whole brain, and the distributions of resting-state network (RSN) can then be estimated from these fluctuations using independent component analysis (ICA). This procedure provides an important method for studying cognitive and psychophysiological diseases affecting specific brain networks. The distributions of RSNs in the brain-wide area has been reported primarily by rsfMRI. RSNs using rsfMRI are generally computed from the time-course of fMRI signals for more than 5 min. However, a recent dynamic functional connectivity study revealed that RSNs are still not perfectly stable even after 10 min. Importantly, fUS has a higher temporal resolution and stronger correlation with neural activity compared with fMRI. Therefore, we hypothesized that fUS applied during the resting-state for a shorter than 5 min would provide similar RSNs compared to fMRI. High temporal resolution rsfUS data were acquired at 10 Hz in awake mice. The quality of the default mode network (DMN), a well-known RSN, was evaluated using signal-noise separation (SNS) applied to different measurement durations of rsfUS. The results showed that the SNS did not change when the measurement duration was increased to more than 210 s. Next, we measured short-duration rsfUS multi-slice measurements in the brain-wide area. The results showed that rsfUS with the short duration succeeded in detecting RSNs distributed in the brain-wide area consistent with RSNs detected by 11.7-T MRI under awake conditions (medial prefrontal cortex and cingulate cortex in the anterior DMN, retrosplenial cortex and visual cortex in the posterior DMN, somatosensory and motor cortexes in the lateral cortical network, thalamus, dorsal hippocampus, and medial cerebellum), confirming the reliability of the RSNs detected by rsfUS. However, bilateral RSNs located in the secondary somatosensory cortex, ventral hippocampus, auditory cortex, and lateral cerebellum extracted from rsfUS were different from the unilateral RSNs extracted from rsfMRI. These findings indicate the potential of rsfUS as a method for analyzing functional brain networks and should encourage future research to elucidate functional brain networks and their relationships with disease model mice.