Ultrafast complex-valued 4D fMRI reveals sleep-induced brain respiratory pulsation changes in both magnitude and phase signals

Abstract

Physiological brain pulsations play a critical role in sleep physiology, but their underlying mechanisms remain poorly understood. To study these pulsations more deeply, we employed ultrafast magnetic resonance encephalography (MREG) to capture complex-valued 4D fMRI brain data at a critical 10 Hz sampling rate in healthy volunteers during wakefulness and sleep. We compared the phase and magnitude components of the MREG signal, as the phase component is known to be particularly sensitive to subtle flow and susceptibility changes, offering insights beyond magnitude-only analysis. This approach enabled whole-brain mapping of the amplitudes of all three physiological pulsations - very low frequency (VLF), cardiac, and respiratory - using an extended amplitude of low frequency fluctuation (ALFF) method. We identified significant increases in respiratory amplitudes during sleep compared to wakefulness in both phase and magnitude signals, while the VLF and cardiac phase amplitudes did not show significant differences. Phase respiration map showed increase especially in default mode network regions, while additional patterns were observed in the cerebellum, ventricles, cerebral aqueduct, and subarachnoid cisterns. In contrast, the magnitude maps showed increased amplitudes more widespread across the cerebrum. These findings highlight the complementary nature of phase and magnitude data in fMRI and suggest that combining these signals provides a more comprehensive understanding of brain physiological dynamics during sleep than conventional magnitude-only analyses.