Noise-canceling Electrostatic Headphones for fMRI

Home - Updates


Electrostatic Transducers

Noise Cancellation



About Me

Detailed understanding of the brain's reaction to sound, language, and music is a critical aspect of many fields of research. However, the current state-of-the-art for brain imaging, fMRI, presents a number of significant obstacles to accurate sound reproduction.

No ferrous metals:

The powerful magnetic field in and around the MRI chamber prohibits the use of any ferrous metals. At the very least, these objects will disrupt the uniform magnetic field necessary for the scan; at worst, objects may be accelerated through the air from surprising distances, causing injury or death.

This rules out any moving-coil dynamic (normal speakers and headphones) and planar-magnetic transducer design. The only options then, are pneumatic (sound is produced normally outside of the MRI chamber and sent through tubes), piezoelectric, and electrostatic transducers. Of those three, only electrostatic transducers are capable of producing high-fidelity sound.

Scanning noise:

Noise inside the chamber during scanning measures about 103dB, with a spectrum between 1kHz and 7kHz (the most sensitive, or "piercing" part of our hearing). Lower-frequency background machine noise registers about 70dB. EEG studies on subjects listening to a voice recording while subjected to a recording of fMRI scanning noise showed significantly less correlated brain activity, compared to subject listening only to the voice recording.

The best insulated ear protectors provide 30dB of reduction at specific frequencies, reducing the noise to a bearable level, but not enough to conduct accurate studies on music or voice cognition.

To reduce noise further, active noise cancellation is required, either by real-time open/closed loop noise cancellation--or, because scanning noise is predictably related to scanning parameters, cancellation by a parametrically generated anti-noise.

Radio-frequency interference/drain:

fMRI scanning is accomplished in part by sending RF energy at 64mHz or 128mHz into the chamber and detecting the far weaker response from the scanned body. Any introduction of outside energy at these frequencies would disrupt the scan, and any leakage or absorption of the scanned body's response reduces the imaging resolution.

This affects signal cabling to the headset. Ideally, very high resistance cables and high-quality RF chokes impervious to energy at 64 and 128mHz would be used.