Acoustic, psychophysical, and neuroimaging measurements of the effectiveness of active cancellation during auditory functional magnetic resonance imaging
The average power spectrum of the scanner noise generated during functional brain imaging (EPI) using a Philips Intera MR scanner. The black line (noncancelled) indicates the spectrum of the noise recorded under normal operation and shows a peak of power at and lesser peaks at 150 and . The gray line (cancelled) indicates the spectrum when the ANC is operative. The power around the dominant peak is reduced by . The inset (upper right) shows the time-domain waveform of a single burst of EPI noise, a sequence of 18 discrete pings corresponding to a series of 18 consecutive slices through the brain. Each ping lasts about , and the total duration of the burst is about .
Examples of the time course of acoustic noise reduction and its relationship to detection thresholds for a target. Diamonds and squares denote the rms level at the left and right ears, respectively. The first scan corresponds to the recording of the noise reference and indicates the noncancelled level of the EPI noise. (A) Example taken from listener 4 in experiment 1. The level of the EPI noise rapidly reduces over the four training scans (gray infills) as the ANC system optimises the adaptive filter. In each pair of subsequent trials, the level in one interval is higher than in the other because it contains both the masker and the target signal. The black dots denote the actual presentation level of the target, the first two practice trials being at SPL. The target level tracks the interval that represents the combined level of the noise and target reasonably well. (B) Example taken from listener 6 in experiment 3. The level of the EPI noise rapidly reduces over the two training scans (gray infills), but the amount of noise reduction is generally less than for experiment 1 and the masker attenuation in the right ear can be seen to deteriorate by over over the course of the adaptive run.
Envelopes of the recordings of three noise maskers taken from one adaptive track of listener 4. The top panel shows one of the non-cancelled maskers at the beginning of the run, and the other two panels show two separate cancelled maskers from later in the run. The envelopes were calculated using the Hilbert transform, using edited waveforms of duration that were approximately centered on the scan. The waveforms were initially filtered to a wide band, centered on (see Sec. II B ).
Same/different frequency discrimination performance for experiment 2 measured using . Mean performance is shown by the height of the bar, whilst individual scores are represented by the ellipse (listener 1), the square (listener 2), and the circle (listener 4).
Sound-evoked brain activity for experiment 2. Each row displays the results for an individual listener overlaid onto a slice that is oriented through auditory cortical regions on the upper surface of the superior temporal gyrus. Each column displays the pattern of significant brain activity, in white, for one of the four major contrasts (voxel-level threshold, ).
The extent of sound-evoked auditory activity (voxel-level threshold, ) within two auditory cortical regions. The primary auditory region contains a total of 2172 voxels across both hemispheres. The (posterior) nonprimary region contains a total of 7982 voxels. Extent of activation is plotted separately for the different sound levels and for the different ANC conditions, with standard error bars.
Individual sound levels measured in dB SPL at the listeners’ ears in experiment 1. Values in parentheses represent the standard error of the measurement. The three different masker levels were computed from the DAT recording of the error microphone using a wide filter centered on . The masker levels reported for the cancelled condition were measured for trials at the reversal points in the adaptive run. The individual threshold levels of the signal are provided for comparison with the masker levels in the better ear.
Masked perceptual thresholds in the cancelled and noncancelled conditions for experiments 1 and 3. Mean levels are reported in dB SPL, with standard errors in parentheses.
The physical levels of the stimuli and the resulting sensation levels, for each listener who participated in experiment 2. The sensation levels for the conditions are determined from the individual cancelled and noncancelled thresholds reported in experiment 1 (Table II ).
Mean sound levels measured in dB SPL at the listeners’ ears across different frequencies of interest in experiment 3. The masker levels reported for the cancelled condition correspond to those trials at the reversal points in the adaptive run. The mean threshold level of each target signal is provided for comparison with the masker levels in the better ear. Values in parentheses represent the standard error of the measurement.
Article metrics loading...
Full text loading...