Vibroacoustic Modeling of Noise in Magnetic Resonance Imagers

Introduction Magnetic Resonance Imagers (MRI) are unpleasantly loud for medical diagnostic devices. Noise levels in excess of 115 dBA have been observed at a patient's ear during certain scan sequences [1]. These high noise levels cause problems in patient comfort, doctorpatient communications, and occupational noise exposure for healthcare workers. We have developed a vibroacoustic model of a typical MRI, based on a technique known as Statistical Energy Analysis (SEA). The model serves as a guide in understanding the vibroacoustic environment of an imager, for the purpose of quieting existing products and in designing quiet into new products. Experiments have provided data that serve to validate the estimates of subsystem vibration and the final noise level predictions. Description of the Model Statistical Energy Analysis relies on the fact that, under certain circumstances, the flow of vibrational energy between coupled mechanical systems is proportional to the difference in average modal energy between them. This condition occurs at frequencies where there is a sufficient density of vibrational modes that the modes are statistically similar. In a structure with the size and construction of an MRI scanner, this happens at frequencies above about 400 Hz. Since most MR scan sequences are composed of multiple harmonics of 100 Hz or so, with harmonic components extending beyond 5 kHz, the majority of MR vibroacoustic energy falls into this statistically valid range [2]. The high frequency content of the noise, the simple, lightly damped structure, and the structural-acoustic interactions involved make MRI scanners well suited to this modeling technique. Application of SEA principles allows us to reduce a complicated vibration problem to a simpler energy balance problem similar to classical conductive heat transfer. We use a commercially available SEA code known as AutoSEA, which is marketed by Vibroacoustic Sciences, Inc., of San Diego CA, USA. The physical model is depicted schematically in Figure 1.