Neuronal activity has been directly measured with electroencephalography (EEG) and magnetoencephalography (MEG), which have limited spatial resolution. Functional MRI has been used for mapping neuronal activity in higher spatial resolution, but has limited temporal resolution and limited spatial specificity to neuronal activity because of its dependence on blood hemodynamics rather than direct neuronal activity. Many researchers have tried to directly detect the neuronal activity using MRI, such as direct detection of phase signal change induced by oscillating magnetic field changes, ultra-low field MRI which lowered Larmor frequency using a particular detector, and a detection method based on spin-locking mechanism, et al. Most of the techniques showed successful results in phantom experiment, but the detectability of neuronal current in vivo is still controversial over a decade.
In this study, we suggested a novel method which can detect neuronal oscillation directly without synchronization. By combining dynamic multi-phase acquisition and Fourier analysis, we could detect oscillating magnetic field with a magnitude of ~1nT. The proposed approach was robust to random changes in oscillating magnetic fields such as multiple frequency components and random on/off intervals. Multiple refocusing radiofrequency (RF) pulses were use to increase its sensitivity when detecting particular frequency component. The approach with $180^\circ$ refocusing RF pulses increased both sensitivity and specificity for weak oscillating magnetic field. And we also suggested a method to identify the oscillation frequency by combining two dataset with different times to repeat (TR).