Introduction

The nose is the major organ responsible for olfaction and air intake (1). Due to its rich innervation by the autonomic nervous system (2,3), it can play a pivotal role in helping us to understand the complex interactions between the peripheral and central nervous systems. Yet, the characterization of system-wide functional connections of the nose is not possible with standard fMRI techniques because of signal loss resulting from susceptibility effects originating from the air-tissue interfaces. To overcome these challenges, we exploit the resilience of ultrashort and zero echo time MRI techniques to susceptibility effects for conducting fMRI studies. Previously we demonstrated that these methods are capable of providing fMRI contrast primarily mediated by blood flow (4).

Method

Brain images of 5 subjects (age, mean ± SD = 51.4 ± 19.9 years, 2 males) were acquired on a 7T Siemens Magnetom scanner with a 32-channel receive NOVA head coil. Resting-state fMRI was performed with a slab-selective UTE sequence with 1070 radial view, FOV=(192 mm)³, isotropic 2 mm voxels, TR/TE=1.4/0.12 ms, flip angle=2°, 244 volumes, time for single 3D-image 1.5 s (∼6 min acquisition time). For anatomical reference, an MP2RAGE sequence was acquired with 240 slices, TR/TE=5000/2.27 ms, FOV=240×225 mm, voxel size 0.75x0.75x0.8 mm³. Also for anatomical reference, a high-resolution UTE was acquired with 4096 radial views, 24 radial interleaves, FOV=(192 mm)³, isotropic voxels of (0.75 mm)³, TR/TE=3/0.11 ms, flip angle=3.5° (∼5 min acquisition time). Throughout functional scanning, the respiratory belt and pulse plethysmograph were used to record the physiological status of the subjects. One resting state study on an awake mouse was performed at 9.4T with Multi-Band Sweep Imaging with Fourier Transformation (MB-SWIFT)(4). Transceiver surface RF-coil covering nose and brain was used. Acquisition parameters were as follows: 2047 radial views, FOV (30 mm)³, isotropic voxels (0.47 mm)³, TR=0.81 ms, flip angle 2.0°, time for single 3D-image 1.7 s. Resting-state data was collected for 5 minutes. Human fMRI data were processed using Brain Voyager QX including motion correction and coregistration to the anatomical reference. Single-subject independent components analysis (ICA) was performed with fastICA (5) and 30 independent components. For analysis of physiological signals, we used PhysioNet Cardiovascular Signal Toolbox (6) along with custom MATLAB scripts. After synchronization with MRI, pulse signals were smoothed with a Savitzky-Golay FIR filter (order 3, frame size 67 samples), and time between successive pulse wave onsets (PP) was calculated in order to obtain the heart rate variability (HRV) as the time series of the root mean square of successive differences in PP intervals (7). This signal was resampled at the fMRI temporal resolution and convolved with the human hemodynamic response function (from SPM). Pearson’s coefficient was calculated to evaluate the correlation between ICA component time courses and HRV. Respiration signals were resampled to fMRI temporal resolution, and a frequency analysis was performed. Resting-state mouse data were corrected for motion, and ICA components were estimated using FSL-MELODIC.

Results

Figure 1A shows single subjects ICA z-maps of a network encompassing the nasal cavity, which was lateralized and highly reproducible. The same component is shown for one subject overlaid on the anatomical UTE with high anatomical details (Figure 1B). Figure 1C shows another network extending from the nose to CSF around the brainstem and major vessels, also highly reproducible. Figure 2 illustrates the temporal signals of those networks; the nose-only component presents only slow fluctuations, while the other component presents also higher frequency fluctuations. The signal of the nose-only component was highly correlated with the HRV (Figure 3) in all available subjects (subject #1: r=0.48; subject #2: r=0.73; subject #4: r=0.90; subject #5 r=0.68; p<0.001). No overlapping was found between the main frequency component of the respiration (between 0.2 and 0.3 Hz) and the frequency components of the ICA. At least one prominent component, resembling nose network in humans was observed also in the mouse (Figure 4).

Discussion

In the current study, we demonstrate that it is possible to evaluate nose functional activity with ultrashort and zero echo time fMRI. We observed in humans robust and reproducible nose networks, one of which was lateralized on the nostril with open airways, thus reflecting the nasal cycle (8), and was strongly correlated with a proxy of HRV reflecting autonomic activity (9). Results in the mouse were consistent with the observations in humans.

Conclusion

We conclude that ultrashort and zero echo time fMRI provide the opportunity to monitor nose functional networks that have been inaccessible so far with traditional BOLD techniques, opening a new investigational window into the nervous system and its related pathological changes.