Hi,
Pablo is right: going up with the echo spacing makes EPI less loud. Not only that: it also lowers the frequency of the emitted EPI sound (which may be perceived less annoying).
Even though EPI implementations vary (i.e. w/r to ramp sampling etc), the latter relation is pretty tight and you can approximate EPI sound frequency by calculating the reciprocal of twice the echo spacing.
However, the former greatly depends on the sound attenuation embedded in your system. The relation of EPI sound levels to r/o frequency or echo spacing, respectively, tends to be quasi-logarithmical: plotting log(1/2xESP) against peak EPI SPL [dB] tends to give a line. Essentially, this reflects the linear increase of sound pressure p(t) with gradient performance g(t).
In my experience, you can save up to 15dB by going from ~0.6ms ESP to 0.9ms on older systems, for example. However, newer scanners have their sound attenuation much improved, and here increasing the ESP may save you less than 10dB, or only 5dB (I've looked at this on 5 different scanners).
But consider: EPI sound on newer scanners is already much lower to start with. Slope and intercept of SPL = f(ESP) differ across systems and primarly depend on the sound attenuation embedded. In fact, the sound attenuation often has more of an impact than the field strength. For example, Siemens has greatly improved their sound attenuation / noise protection measures over the years ("AudioComfort") and even their new 7T is less loud than old 1.5Ts (if you measure with identical EPIs). We are working on a paper to highlight these issues, so just stay online if that is of interest to you.
I doubt, however, that increasing the ESP will be enough to let people fall asleep but it may be worth to try.
One more thing to consider though: if you hit a audio-mechanical resonance frequency or one of its harmonics at a particular ESP, your sound level rises dramatically by 10-25dB (see, for example, a Tomasi and Ernst paper J Magn Reson Imaging. 2003 Jul;18(1):128-30.). The audio-mechanical resonance frequency depends on your system, the bore but also on site-specific issues like the ground, the cabin etc. Thomas Benner from the MGH has developed a nice tool to determine the resonance frequency based on secondary ghosting intensities (very cool but you may want to validate this by SPL-measurements). Siemens now broadly blocks these presumed resonance frequencies of their systems to protect the hearing as well as the system from secondary damage (even though the resonance at yours may somewhat deviate from their center frequency - it should fall into the blocked intervall), but I'm not sure about the other vendors.
Others have developed headphones to cancel out the EPI noise - that may also be an option for you.
HTH-
Andreas
________________________________________
Von: FSL - FMRIB's Software Library [[log in to unmask]] im Auftrag von Pablo Velasco [[log in to unmask]]
Gesendet: Samstag, 19. Dezember 2009 13:54
An: [log in to unmask]
Betreff: Re: [FSL] "silent" EPI
Hi Diederick,
Since the principal source of noise are vibrations produced by the gradients
blipping, you can try decreasing the bandwidth/increasing the echo-spacing.
Your total readout time will go up, so you will have more distortions and
drop-out, so you will have to reach some compromise.
HTH,
-Pablo
On Fri, 18 Dec 2009 18:48:05 +0100, Diederick Stoffers
<[log in to unmask]> wrote:
>Hi all,
>
>A question not relating to software, but perhaps someone would be wiling to
help us out.
>
>We are trying to heave subjects fall asleep during fMRI acq, we would like
to have a more "silent" EPI sequence. There are quite some publications from
groups that have a relatively silent EPI sequence running on a Siemens
machine. We have a 3T GE Signa HD and a 3T Philips Intera available. We have
been playing around with the factory provided silent mode options on both,
but are not particularly happy with the results (high TR/TE/slice thickness,
not much noise reduction). Does anyone have a reasonably performing "silent"
EPI on a GE or Philips system running? Can anyone recommend what settings
to use?
>
>Thanks,
>
>Diederick
|