Frequency-domain active noise control for magnetic resonance imaging acoustic noise

Abstract The purpose of this paper is to propose a frequency-domain active noise control algorithm for acoustic noise reduction in magnetic resonance imaging and experimentally verify its feasibility. The sound pressure level of noise approaches up to 130 dB in 3 T magnetic resonance imaging. Even it’s expected to increase for higher-resolution images in the future magnetic resonance. For this noise, some active methods have been researched, but the reduction performance is not enough to satisfy the safety standard. The noise in 3 T magnetic resonance imaging is measured and analyzed in the time-frequency domain. It has fundamental frequency with its higher harmonics, and the Fourier coefficients of all peaks are slowly varying over time. The variation is small enough, so it can be assumed that the noise is periodic in short time. Based on this assumption, the control signal is determined as the sum of sinusoidal signals designed by using the Fourier coefficients calculated from a fundamental period of noise, then it is applied to the next period. To do this in real-time process, domain transform algorithm, which is time to frequency based on discrete Fourier series, is developed with the same level of computational complexity compared to the conventional active noise control algorithm. Experimental results show approximately 35-dB overall reduction in the frequency range of interest (80–1600 Hz).