Objective: To evaluate the effectiveness of the proposed adaptive speech enhancement (ASE) system for the magnetic resonance imaging (MRI) environment to reduce the loud scanning noise without disrupting the communication between patients and MRI operators.
Materials and methods: The developed system employed the idea of differential directional microphones for measuring and distinguishing the speech signals and MRI acoustic noises simultaneously. Two-stage adaptive filters with normalized least mean square algorithms were adopted. Two common MRI scanning sequences, echo planar imaging (EPI) and gradient echo multi-slice (GEMS), were tested using a 4T MRI scanner.
Results: A total of 1.4 and 3.3 dB speech enhancements quantified by the cepstral distance assessment were achieved for the speech signal contaminated with the EPI and GEMS noises, respectively. The speech signal was noticeably recovered, and a clear speech waveform was observed after treated with the ASE system. Furthermore, a non-adaptive post-processing approach [i.e. simply using spectral subtraction (SS) technique] was also adopted to process the abovementioned results. Additional reductions were achieved for the non-coherent MRI acoustic noises.
Conclusion: The results showed that combining the proposed ASE system along with the SS approach has a great potential for treating MRI acoustic noise to guarantee an effective communication from patient to MRI operators.
Keywords: Active noise control; Adaptive speech enhancement; MRI acoustics; Noise; Sound.