Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

DOI: 10.1148/radiol.2321030955
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Moelker, A.
Right arrow Articles by Pattynama, P. M. T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Moelker, A.
Right arrow Articles by Pattynama, P. M. T.
(Radiology 2004;232:107-113.)
© RSNA, 2004

Special Report

Verbal Communication in MR Environments: Effect of MR System Acoustic Noise on Speech Understanding1

Adriaan Moelker, MD, Ronald A. J. J. Maas, PhD and Peter M. T. Pattynama, MD

1 From the Departments of Radiology (A.M., P.M.T.P.) and Audiophysics (R.A., J.J.M.), Erasmus Medical Center Rotterdam, 50 Dr Molewaterplein, PO Box 1738, 3000 DR Rotterdam, the Netherlands. Received June 18, 2003; revision requested August 13; final revision received November 4; accepted November 20. Address correspondence to A.M. (e-mail: a.moelker@erasmusmc.nl).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
PURPOSE: To assess the masking effect of magnetic resonance (MR)-related acoustic noise and the effect of passive hearing protection on speech understanding.

MATERIALS AND METHODS: Acoustic recordings were made at 1.5 T at patient and operator (interventionalist in the MR suite) locations for relevant pulse sequences. In an audiologic laboratory, speech-to-noise ratios (STNRs) were determined, defined as the difference between the absolute sound pressure levels of MR noise and speech. The recorded noise of the MR sequences was played simultaneously with the recorded sentences at various intensities, and 15 healthy volunteers (seven women, eight men; median age, 27 years) repeated these sentences as accurately as possible. The STNR that corresponded with a 50% correct repetition was used as the measure for speech intelligibility. In addition, the effect of passive hearing protection on speech intelligibility was tested by using an earplug model.

RESULTS: Overall, speech understanding was reduced more at operator than at patient location. Most problematic were fast gradient-recalled-echo train and spiral k-space sequences. As the absolute sound pressure level of these sequences was approximately 100 dB at patient location, the vocal effort needed to attain 50% intelligibility was shouting (>77 dB). At operator location, less effort was required because of the lower sound pressure levels of the MR noise. Fast spoiled gradient-recalled-echo and echo-planar imaging sequences showed relatively favorable results with raised voice at operator location and loud speaking at patient location. The use of hearing protection slightly improved STNR.

CONCLUSION: At 1.5 T, the level of MR noise requires that large vocal effort is used, at the operator and especially at the patient location. Depending on the specific MR sequence used, loud speaking or shouting is needed to achieve adequate bidirectional communication with the patient. The wearing of earplugs improves speech intelligibility.

© RSNA, 2004

Index terms: Magnetic resonance (MR), biological effects • Magnetic resonance (MR), functional imaging • Magnetic resonance (MR), guidance • Magnetic resonance (MR), safety


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
Acoustic noise generated during magnetic resonance (MR) imaging is an unwanted side effect (1) that may, it has been suggested, negatively affect verbal communication between patient and operator (the interventionalist in the MR suite) and between multiple operators (2). From a practical point of view, adequate speech understanding is a prerequisite in a number of circumstances. First, speech intelligibility between operators is essential for MR-guided interventional procedures, particularly in potentially dangerous situations. Second, in functional MR imaging of the auditory brain, both the instruction and presentation of verbal stimuli to subjects require clear speech intelligibility (3). In addition, in audiologic experiments, the subject’s verbal responses to language tasks should be accurately perceived by the operator (4). To our knowledge, the effects of MR acoustic noise on speech intelligibility at functional and interventional MR imaging have not yet been investigated.

Various techniques of acoustic noise reduction have been proposed and implemented in the MR environment with the goal of improving speech understanding (5). These techniques are used in an attempt to reduce MR acoustic noise while minimally affecting or enhancing speech understanding (6). One such technique is passive hearing protection (eg, by means of earplugs or earmuffs), which substantially reduces acoustic noise levels (5). Only a few investigators have emphasized that passive hearing devices might be favorable for speech intelligibility in noisy environments (7,8). This assumption, to our knowledge, also has never been validated in the MR environment.

Thus, the purpose of our study was to assess the masking effect of MR-related acoustic noise and the effect of passive hearing protection on speech understanding.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
Study Design
In audiology, the measure commonly applied for speech intelligibility is the speech reception threshold, which is the speech-to-noise ratio (STNR) that corresponds with a 50% correct response by subjects to speech in noisy conditions (9). The speech reception threshold is a robust and validated measure of speech intelligibility and makes use of a large set of sentence material (9). In this study, the STNR was defined as the difference between the absolute sound pressure levels (the continuous sound pressure level expressed in decibels with an A-weighted scale [10]) of speech and masking MR noise; this difference is typically –5 dB for listeners with normal hearing ability. In other words, a lower STNR (larger negative value) corresponds with better speech understanding. The speech level required for 50% intelligibility can be found by iteratively adjusting the level of speech (which is masked by fixed MR noise) and can be qualitatively classified in terms of normal, raised, loud, or shouting voice (Table 1) (11). The MR imaging sequences tested were chosen on the basis of their potential relevance to vascular, interventional, and functional MR imaging. Speech intelligibility was quantified for MR acoustic noise recorded both at the operator’s location next to the MR system and at the patient’s location. In addition, we assessed the effect of passive hearing protection on speech reception thresholds during MR imaging by using an earplug model.


View this table:
[in this window]
[in a new window]

  		TABLE 1. Overall Sound Level of Speech at Each Level of Vocal Effort

 
Volunteers and Speech Reception Threshold Measurements
Our analysis was based on measurements obtained in 15 native Dutch-speaking volunteers (seven women, eight men; median age, 27 years) without hearing impairment. Although our institutional review board did not require its approval, informed consent was obtained from all participants. The acoustic exposures were well below daily permissible limits (12). Before the experiments were performed, a baseline audiogram (OB822 clinical audiometer; Madsen Electronics, Bloomington, Minn) was obtained in all subjects to exclude hearing impairment (13). In all subjects, the minimum hearing levels on the pure tone audiogram did not exceed 20 dB, at octave frequencies from 125 to 8,000 Hz; thus, hearing levels were normal in all subjects.

As all subjects were native Dutch speakers, the measurements for speech reception threshold were obtained by using Dutch test sentences (14) that had previously been validated for phonetic balance (15,16). These sentences are short and redundant to enhance their intelligibility against distortions or interfering sounds to listeners with normal or impaired hearing (15). None of the subjects had heard or read the sentences before, and each sentence was presented only once to each subject to avoid memory effects. In addition, the listeners participated in a brief speech reception threshold training session to reduce learning effects of speech intelligibility in noise.

In the adaptive procedure of measuring the speech reception threshold (9), prerecorded MR noise was played simultaneously with the test sentences through loudspeakers (Lab-501; Westra Electronic, Wertingen, Germany) in an anechoic environment (Industrial Acoustic Company, Bronx, NY) (15). The listeners were seated perpendicular to a loudspeaker that reproduced the MR noise and facing another loudspeaker that provided the test sentences (13). This arrangement was considered a plausible representation of an interventional MR imaging procedure. After a sentence was presented, the subject responded by repeating it as accurately as possible. The first sentence (from a list of 13) was initially presented at such a low STNR that it was unlikely to be intelligible to the subject. This sentence was then presented repeatedly, at increasing sound levels (in 4-dB steps), until the listener could reproduce the sentence correctly. This method provided a quick convergence to the 50% speech intelligibility threshold. The remaining 12 sentences were presented only once, at a sound level that depended on the subject’s response to the previous sentence (ie, the level was 2 dB higher after an incorrect response and 2 dB lower after correct repetition of the complete sentence). By averaging the STNR values for the last 10 sentences, a 50% sentence intelligibility threshold was obtained and initialization effects were eliminated. Because little deviation of STNR values between sentence lists has been reported previously (9,15), an average of three sentence lists per patient was taken as a speech reception threshold value for that particular condition.

A computer equipped with Matlab (R13; MathWorks, Novi, Mich) performed STNR adjustments with custom-written software and delivered the MR noise and sentences to the active loudspeaker systems. The speech reception threshold measurements were automatically logged and stored for analysis.

MR Equipment and Acoustic Noise Measurement
The acoustic data were obtained with a 1.5-T cardiovascular MR imager (Signa CV/i, GE Medical Systems, Milwaukee, Wis), with LX 8.4 software with gradients of up to 40 mT · m–1 and slew rates of 150 T · m–1 · sec–1, by using an integrated quadrature-driven transceiver and a radiofrequency body coil. The pulse sequences tested were as follows: echo-planar, fast spoiled gradient-recalled-echo (FSPGR), fast gradient-recalled-echo train (FGRET, a hybrid echo-planar and FSPGR sequence), spiral trajectory k-space, and conventional fast spin-echo imaging.

FSPGR, FGRET, and spiral k-space sequences are especially suitable for real-time imaging and were therefore considered most relevant to interventional MR imaging (17). Echo-planar imaging is the most extensively employed pulse sequence in functional MR imaging and was therefore not assessed at the operator location (18). The acoustic characteristics for all pulse sequences are given in Table 2. The imaging parameters, in particular the repetition time, were chosen to test pulse sequences under worst-case conditions for clinical imaging (19). The bandwidth was similar for all pulse sequences tested (100–125 kHz), with the exception of the fast spin-echo sequence (62 kHz). For safety reasons, however, all pulse sequences, especially those recorded in the magnet bore, were limited to a maximum sound intensity of approximately 100 dB (A-weighted scale) (20) by omitting pulse sequences with extremely short repetition time or by increasing repetition time slightly (FGRET and spiral k-space).


View this table:
[in this window]
[in a new window]

  		TABLE 2. Parameters and Acoustic Characteristics for Each Pulse Sequence

 
The sound levels of the imaging pulse sequences and their acoustic waveforms were recorded by using a digital sound analyzer (Investigator 2260; Brüel & Kjær, Nærum, Denmark) connected to a computer that was equipped with a sound card (Audio DSP24; Hoontech, Bucheon City, Korea) (21). To quantify the speech intelligibility at the operator position, the acoustic recordings were made with a microphone (model 4189; Brüel & Kjær) placed at a distance of 0.80 m from the MR imager entrance and at a height of 1.70 m, close to the MR imaging table, which is a plausible location for the ear of the interventional radiologist. Also, similar recordings were made with the microphone positioned inside the magnet bore (at the isocenter), deployed for quantifying speech intelligibility at the location of the patient’s ear.

Calibration and Analysis
The digital sound analyzer was calibrated with a sound calibrator (model 4231; Brüel & Kjær) at regular intervals and showed an accuracy of better than 0.1 dB. The sound card setup for recording the acoustic MR noise was calibrated by matching sound card and sound analyzer at all relevant frequencies. The sound pressure level calibration for delivered speech and MR noise in the anechoic room was performed as follows: In the MR suite, the continuous-equivalent sound level and waveform of an FGRET pulse sequence were recorded with the sound analyzer at the operator location. Next, the waveform was replayed in the anechoic room while the sound intensity was adjusted to its original level as measured in the MR environment. The sound pressure levels of all other pulse sequences were registered in the same experimental setting and could be easily compared with this reference level and calibrated accordingly.

Since human hearing is less sensitive to frequencies below 1 kHz, a so-called A-weighted filter was applied to both speech and MR noise waveforms (expressed as decibels with an A-weighted scale) before calculating the STNR (22). Also, a level thresholding method was required for suppression of the silent periods in the speech signals and MR noise (23). This was done by filtering the waveforms with a low-pass filter (47 Hz) to obtain an intensity envelope of speech and MR noise. The resultant waveform was thresholded (14 dB below its root-mean-square value) to eliminate the silent periods, and its sound pressure level was finally used for determining the actual STNR (23).

In the speech reception thresholding procedure, speech was presented at levels ranging from approximately 50 to 90 dB (A-weighted scale), depending on the required speech reception threshold and the absolute sound pressure level of the pulse sequence investigated. In other words, the speech intensity ranged from a normal conversational level to extremely loud shouting. With increasing vocal effort, however, the frequency spectrum of speech changes; the level of frequencies around 2 kHz is relatively increased (11). This was simulated by digitally filtering the sentences in the frequency domain before presentation, in accordance with internationally standardized values (11).

Passive Hearing Protection
In addition to the 12 different experiment conditions already described (Table 2), additional speech reception threshold measurements were obtained in all 15 subjects by using digital filtering of speech and MR noise that represented the wearing of passive hearing protection. Toward this end, the frequency transfer function of earplugs as inserted by untrained users was adapted from a recent publication (24). The frequency transfer function of earplugs is comparable with that of earmuffs, which are more frequently used in functional MR imaging (5). The rationale for using digital filtering rather than real passive hearing protection is the relatively large difference between auditory attenuation with earplugs in subjects who have been taught proper insertion and that in untrained subjects (24). Recall that the computed STNR equals the difference between the sound levels of MR noise and speech (ie, the difference between the sound intensities measured in the MR environment). The presented speech level is the unfiltered speech level produced by, for example, the operator. The STNR, based on sound pressure levels of MR noise and speech after the filtering procedure, represents the performance of the subject’s hearing rather than the required speech level during MR imaging.

Statistical Analysis
Measured values for speech intelligibility at the operator and patient locations (as an average, as well as for each type of filtering) were compared and tested for statistically significant differences by using pulse sequences with imaging parameters that were similar for operator and patient locations (FSPGR, FGRET, spiral k-space, and fast spin-echo with long repetition time) (14). Also, for the filtered and unfiltered measurement conditions, the speech intelligibility (as an average and for each recording location) was tested for statistically significanct differences. Toward this end, Student t tests were performed by using a statistical software package (SPSS version 11.0 for Windows; SPSS, Chicago, Ill). A P value of .05 (with 95% CIs) was considered to denote a statistically significant difference (14).


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
Characteristics of MR Acoustic Noise
The acoustic noise levels in the isocenter of the MR imager were, on average, 10 dB higher than those measured at the entrance of the MR system; this is consistent with previously reported values (25). For all sequences except FGRET and FSPGR with short repetition time, the MR noise had a broadband frequency distribution that peaked between 1 and 3 kHz (Figure). This was a result of the typically low fundamental frequency (equal to the inverse of repetition time) and the associated higher harmonics that were, consequently, closely spaced in the frequency domain. As the fundamental frequency for FGRET and FSPGR with short repetition time was relatively high, these pulse sequences showed distinct frequency components (Figure). Also, the frequency distribution of the pulse sequences differed with respect to the location at which they were recorded; higher frequencies (up to 3 kHz) were recorded at the isocenter of the imager, while lower frequencies (up to 1 kHz) were recorded near the imager. As an example, FGRET noise peaked at 1,158 Hz whenever recorded beside the MR imager, whereas for a similar pulse sequence recorded in the magnet bore, the most intense frequencies were near 2 kHz (Figure, C, D).



View larger version (57K):
[in this window]
[in a new window]
[Download PPT slide]
  		Acoustic noise spectra of FGRET and FSPGR at operator location with short repetition time (A,B) and of FGRET at operator location and patient location with long repetition time (TR) (C,D). In A and B, the harmonics are clearly distinguishable, whereas in C and D, the harmonics are too closely spaced, giving rise to a broadband noise spectrum. Note the peak frequency at approximately 2 kHz in D. Spectra were obtained with a 4,096-point fast Fourier transform and a Hannig filter. SPL = sound pressure level.

 
Passive hearing protection greatly reduced MR noise (Table 3), by 25 dB on average at the operator location, and showed better reduction for imaging sequences recorded at the isocenter (noise reduced by 29 dB). This was evidently a result of the low-pass characteristics of passive filtering in concurrence with higher frequencies generated in the magnet bore (7).


View this table:
[in this window]
[in a new window]

  		TABLE 3. Mean STNR Values Required for 50% Speech Intelligibility

 
Speech Intelligibility at Operator Location
Table 3 shows both the speech reception thresholds and sound pressure levels of the speech. Of the pulse sequences recorded at the operator location, the most problematic were FGRET, with a STNR of –11.8 dB, and spiral k-space sequences, with a STNR of –16.2 dB. FSPGR and fast spin-echo sequences showed more favorable results, with speech reception thresholds ranging from –20.3 to –20.9 dB STNR and –18.5 dB STNR, respectively. As the sound pressure level of FGRET and spiral k-space sequences were up to 97 dB (A-weighted scale), the 50% intelligibility threshold required sound pressure levels that ranged from 72.9 to 81.2 dB (A-weighted scale). In other words, the vocal effort with which one needed to speak to attain 50% intelligibility was loud speaking, which escalated to extremely loud shouting during imaging with worst-case imaging parameters (short repetition time). On the other hand, both the advantageous STNR and low absolute sound pressure levels of fast spin-echo and FSPGR allowed for ambient communication with only mildly raised vocal effort.

Speech Intelligibility at Patient Location
It can be appreciated from Table 3 that the speech reception threshold results determined at patient location were opposite to those found at operator location: FGRET and spiral k-space sequences were most favorable (–23.3 and –19.6 dB STNR, respectively), whereas FSPGR and fast spin-echo sequences substantially reduced speech understanding (–16.9 and –16.6 dB STNR, respectively). In terms of vocal effort, the absolute sound pressure levels ranged from 73.7 to 80.4 dB (A-weighted scale), corresponding to loud speaking and shouting. Despite the intense masking level of echo-planar imaging (100 dB [A-weighted scale]), the sound pressure level of speech required for 50% intelligibility was only 73.7 dB on the A-weighted scale (loud speaking). Note that these pulse sequences did not represent worst-case scenarios, as they were restricted to a maximum sound pressure level for safety reasons. Speech levels would be higher if worst-case imaging parameters were used.

For similar pulse sequences, speech understanding proved significantly better at the isocenter of the imager with regard to the STNR (2.5 dB on average, P < .05; Table 4), but more advantageous at operator location with regard to the absolute sound pressure levels (–7.0 dB on average, P < .05; Table 4).


View this table:
[in this window]
[in a new window]

  		TABLE 4. Average Differences in STNR and Sound Pressure Level for Parameters Tested with Similar Pulse Sequences

 
Effect of Passive Hearing Protection on Speech Intelligibility
The effect of passive hearing protection on speech understanding with MR noise was assessed by using sound filters that simulated the effect of earplugs. Table 4 shows the improvement in speech intelligibility with hearing protection at operator and patient locations for similar imaging conditions. The average increase was 1.5 dB, which was predominantly attributed to pulse sequences recorded in the imager bore. Although the speech reception threshold did increase considerably at operator location with earplug simulation, this finding was exclusive to worst-case pulse sequences (high sound pressure level, STNR of 4.4 dB). At relatively low sound pressure levels, passive hearing protection had only insignificant adverse effects on intelligibility (–0.2 dB, filtered minus unfiltered STNR averaged for FSPGR, spiral k-space, FGRET, and fast spin-echo sequences). The beneficial effect of passive (low-pass) protection on intelligibility could be partially explained by the different frequency distributions of speech and MR noise. In particular, at the isocenter, the higher frequencies of the MR noise resulted in greater sound attenuation in comparison with that of spoken sentences. Consequently, the computing of the subjective STNRs (ie, as perceived by the listener after filtering) almost completely compensated for the improvement in STNR (–0.2 dB). In contrast, at operator location, earplug filtering proved substantially beneficial with worst-case pulse sequences (subjective improvement of 3.4 dB, filtered minus unfiltered STNR averaged for FSPGR, spiral k-space, and FGRET sequences), but intelligibility deteriorated during pulse sequences with long repetition times (–2.2 dB, filtered minus unfiltered STNR averaged for FSPGR, spiral k-space, and FGRET sequences).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
 
Our measurements show that, depending on the specific MR imaging sequence used, speech understanding is greatly reduced at both operator and patient locations at 1.5 T. Speech levels of up to 80 dB (A-weighted scale) were common with use of the MR sequences most likely to be used for interventional MR imaging (FGRET, spiral k-space, and FSPGR). In other words, 50% speech understanding can be attained on the condition that the voice level is raised to extremely loud or shouting levels. Although speech levels could be as low as 63.1 dB (A-weighted scale), especially with the less demanding pulse sequences, and seem to approach conversational levels (11), the real intelligibility level in clinical practice is expected to be worse and should be placed in proper context.

First, this experiment was designed so that the interfering MR noise was presented toward the listener’s right ear. Such an arrangement improves speech intelligibility by 10 dB compared with that at binaural presentation, during which it is decreased due to head shadow (decrease of 3 dB) and differences in arrival time of MR noise and speech (decrease of 7 dB) (13,26), and so speech levels at the patient location were underestimated. Second, the standard speech reception threshold of 50% is too low for adequate verbal communication (16,27,28). A more relevant measure of communication in a noisy environment is the range of STNRs for which intelligibility is about 80% (27). Fortunately, near the 50% speech reception threshold, a 1-dB increase in STNR results in a 20% higher intelligibility score (9,26,29). This implies that, for satisfying communication, the speech level should be raised by 1–2 dB. Finally, our data represent speech levels as measured at the subject’s ear; in the actual MR environment, however, the spatial distance between speaker and listener would have to be compensated for by further raising of the speaker’s voice.

Speech intelligibility scores differed considerably between the various imaging sequences and recording locations, ranging from –11.8 to –26.3 dB. One tentative explanation is in the dissimilarities of the frequency spectra of these sequences. In conventional audiometry, the test for speech reception threshold is conducted with modified white masking noise that has a frequency spectrum similar to that of normal speech (15). Speech is consequently masked at its greatest extent with a STNR of about –5 dB (15). Analogously, we can appreciate, for example, that FSPGR noise masks speech only at specific frequencies (STNR, –20.3 dB), whereas FGRET has a broadband masking profile that resembles that of speech (STNR, –11.8 dB). In addition, the absolute sound pressure levels of the MR noise might have influenced speech intelligibility. Although it is widely believed that the speech reception threshold depends only on the STNR, whenever the absolute noise level is less than 120 dB (26), some level-dependent effect has recently been described by Van Wijngaarden and Steeneken (30). A theoretical reason was that auditory masking at high levels is different from masking at low levels (less masking) (8,30). Remarkably, this relation was stronger for low-frequency noise (so-called upward spread of masking), in accordance with our finding that this level-dependent effect was present particularly at the operator location (30).

Disturbances in speech understanding are critical in potentially dangerous situations such as the MR environment. It is evident that clear bidirectional speech understanding between cooperating operators is essential during interventional procedures, and low sound levels are required. Most MR systems have a pause capability, but this halts the image acquisition. Interventional procedures do not continuously require full system capabilities; this allows for intervals with less demanding gradient pulses during imaging. To this end, we recently developed a tool, to be located in the MR room, that hooks up with the imager interface and remotely lowers the gradient performance (and acoustic noise level) without ceasing imaging (31). Furthermore, verbal communication is critical in emergency situations (situations related to the MR procedure or MR hardware), and the exchange of information such as the type and extent of emergency situation must be adequate. MR-related acoustic noise may, therefore, pose an occupational hazard, particularly to interventionalists. An emergency button is generally present, but it should be used with care, especially in superconductive systems, because it initiates quenching. In the United States, the Occupational Safety and Health Administration of the Department of Labor has advised that industrial environments with high ambient noise levels must be equipped with a dedicated voice alarm system (32). The power of such a system should be at least 15 dB above the speech reception threshold, but a range of 15–25 dB is more desirable. This equals a speech level of approximately 105 dB in the MR environment.

Echo-planar imaging is the most widely used pulse sequence in functional MR imaging (18). Although the lowest speech intelligibility scores were noted for echo-planar imaging (–29.6 dB), the intense sound levels generated during the imaging process may interfere with functional MR imaging examinations (33). This is problematic in studies of the auditory system, specifically in studies in which speech perception is being investigated (4). Furthermore, results in some reports have demonstrated that the MR noise impairs the mapping of brain functions by evoking undesirable blood oxygen level–dependent, or BOLD, signals (34,35). Fortunately, unilateral communication is sufficient during most functional MR imaging examinations that include an auditory component, because responses by subjects are often not required or are expressed through nonverbal communication (eg, by pressing a button) (4). This allows for the application of abundant acoustic noise insulators (eg, earplugs, earmuffs, cushions) with concurrent delivery of speech signals either via the audio system with compensation for passive attenuation (5) or via pneumatic headsets with integrated probe tubes (ie, audio capability). The latter reduce MR noise levels without compromising speech levels.

A limitation of our study might be in the magnetic field strength of the MR system used. Many interventional procedures are currently performed at less than 1.0 T, and the absolute speech levels measured with our 1.5-T MR system may not be directly applicable to systems with lower field strengths (36). Also, there is a growing trend toward use of MR systems with field strengths greater than 1.5 T for functional MR imaging. Such systems provide better homogeneity and stability of the main magnetic field, higher signal-to-noise ratios and spatial resolution, and faster imaging (37). As a relative measure, however, the STNR values are likely to be similar at field strengths lower and higher than 1.5 T, because speech intelligibility is only slightly dependent on the sound pressure level of the masking MR imager noise. Absolute speech levels can therefore be calculated by taking the absolute sound levels of the particular MR system into account. In addition, conclusions about absolute sound levels may be derived from cautious extrapolation of MR sound levels based on the linear relationship between magnetic field strength and sound pressure level (38).

In contrast to what is generally believed (1), we found a positive and distinct gain in speech intelligibility with use of simulated earplugs for MR noise. This effect was negligible or even negative at low MR noise levels. On the other hand, a typical attenuation of –27 dB with passive hearing protection is known to largely reduce the MR-related risks of hearing damage (by approximately 10,000 times) (25,39). We deem this risk reduction more relevant than the minor adverse effect on speech intelligibility and, therefore, recommend the usage of earplugs.

A final consideration is the possibility of combined use of passive and active noise reduction, in which additional sounds interfere with the MR noise according to the superposition principle (6). From the noise reduction perspective, this is beneficial, as active noise reduction is best suited for low-frequency noise (<1 kHz) (8), whereas passive devices progressively attenuate noise at frequencies above 1 kHz (5). The effect of combining active noise reduction with passive aids for speech understanding is unclear, but evidence suggests a small (10%) improvement in intelligibility for subjects with normal hearing (8). A more viable concept, not currently applied in MR imaging, may be in the capture of the voice with a highly directional microphone located close to the speaker’s mouth, the amplification of the sound, and the projection of the sound through headphones or a loudspeaker. Such a setup would be expected to result in a better STNR and, potentially, better speech understanding.


   FOOTNOTES
 
Abbreviations: FGRET = fast gradient-recalled-echo train, FSPGR = fast spoiled gradient-recalled echo, STNR = speech-to-noise ratio

Author contributions: Guarantors of integrity of entire study, A.M., R.A.J.J.M., P.M.T.P.; study concepts and design, A.M., R.A.J.J.M., P.M.T.P.; literature research, A.M., R.A.J.J.M.; experimental studies, A.M.; data acquisition, A.M.; data analysis/interpretation, A.M., R.A.J.J.M.; statistical analysis, A.M.; manuscript preparation, definition of intellectual content, editing, revision/review, and final version approval, A.M., R.A.J.J.M., P.M.T.P.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 

  1. McJury M, Shellock FG. Auditory noise associated with MR procedures: a review. J Magn Reson Imaging 2000; 12:37-45.[CrossRef][Medline]
  2. Edelstein WA, Hedeen RA, Mallozzi RP, El-Hamamsy SA, Ackermann RA, Havens TJ. Making MRI quieter. Magn Reson Imaging 2002; 20:155-163.[CrossRef][Medline]
  3. Shah NJ, Steinhoff S, Mirzazade S, et al. The effect of sequence repeat time on auditory cortex stimulation during phonetic discrimination. Neuroimage 2000; 12:100-108.[CrossRef][Medline]
  4. Mohr CM, King WM, Freeman AJ, Briggs RW, Leonard CM. Influence of speech stimuli intensity on the activation of auditory cortex investigated with functional magnetic resonance imaging. J Acoust Soc Am 1999; 105:2738-2745.[CrossRef][Medline]
  5. Ravicz ME, Melcher JR. Isolating the auditory system from acoustic noise during functional magnetic resonance imaging: examination of noise conduction through the ear canal, head, and body. J Acoust Soc Am 2001; 109:216-231.[CrossRef][Medline]
  6. Chen CK, Chiueh TD, Chen JH. Active cancellation system of acoustic noise in MR imaging. IEEE Trans Biomed Eng 1999; 46:186-191.[CrossRef][Medline]
  7. Abel SM, Spencer DL. Speech understanding in noise with earplugs and muffs in combination. Appl Acoust 1999; 57:61-68.[CrossRef]
  8. Abel SM, Spencer DL. Active noise reduction versus conventional hearing protection: relative benefits for normal-hearing and impaired listeners. Scand Audiol 1997; 26:155-167.[Medline]
  9. Plomp R, Mimpen AM. Improving the reliability of testing the speech reception threshold for sentences. Audiology 1979; 18:43-52.[Medline]
  10. American National Standard S1.13–1995. Measurement of sound pressure levels in air Melville, NY: Acoustical Society of America, 1995.
  11. American National Standard S3.79–1993. Methods for the calculation of the speech intelligibility index Melville, NY: Acoustical Society of America, 1993.
  12. National Institute for Occupational Safety and Health (NIOSH). Criteria for a recommended standard: occupational noise exposure—revised criteria, 1998 Publication no. 98–126. Washington, DC: U.S. Department of Health and Human Services, 1998.
  13. Festen JM, Plomp R. Speech-reception threshold in noise with one and two hearing aids. J Acoust Soc Am 1986; 79:465-471.[CrossRef][Medline]
  14. Van Wijngaarden SJ, Steeneken HJ, Houtgast T. Quantifying the intelligibility of speech in noise for non-native listeners. J Acoust Soc Am 2002; 111:1906-1916.[CrossRef][Medline]
  15. Versfeld NJ, Daalder L, Festen JM, Houtgast T. Method for the selection of sentence materials for efficient measurement of the speech reception threshold. J Acoust Soc Am 2000; 107:1671-1684.[CrossRef][Medline]
  16. Van Wijngaarden SJ, Steeneken HJ, Houtgast T. Quantifying the intelligibility of speech in noise for non-native talkers. J Acoust Soc Am 2002; 112:3004-3013.[CrossRef][Medline]
  17. Moelker A, Vogel MW, Pattynama PM. Efficacy of passive acoustic screening: implications for the design of imager and MR-suite. J Magn Reson Imaging 2003; 17:270-275.[CrossRef][Medline]
  18. Parrish T. Functional MR imaging. Magn Reson Imaging Clin N Am 1999; 7:765-782.[Medline]
  19. Shellock FG, Morisoli SM, Ziarati M. Measurement of acoustic noise during MR imaging: evaluation of six "worst-case" pulse sequences. Radiology 1994; 191:91-93.[Abstract/Free Full Text]
  20. Guidance for the submission of premarket notifications for magnetic resonance diagnostic devices Washington, DC: U. S. Department of Health and Human Services, Food and Drug Administration, Center for Devices and Radiological Health, 1998.
  21. Counter SA, Olofsson A, Borg E, Bjelke B, Haggstrom A, Grahn HF. Analysis of magnetic resonance imaging acoustic noise generated by a 4.7 T experimental system. Acta Otolaryngol 2000; 120:739-743.[CrossRef][Medline]
  22. American National Standard S3.44–1996. Determination of occupational noise exposure and estimation of noise-induced hearing impairment Melville, NY: Acoustical Society of America, 1996.
  23. Steeneken HJ, Houtgast T. Basics of the STI measuring method. In: Van Wijngaarden SJ, eds. Past, present and future of the Speech Transmission Index. Soesterberg, the Netherlands: TNO Human Factors, 2002; 13-43.
  24. Toivonen M, Paakkonen R, Savolainen S, Lehtomaki K. Noise attenuation and proper insertion of earplugs into ear canals. Ann Occup Hyg 2002; 46:527-530.[Abstract/Free Full Text]
  25. Moelker A, Maas RA, Lethimonnier F, Pattynama PM. Interventional MR imaging at 1.5 T: quantification of sound exposure. Radiology 2002; 224:889-895.[Abstract/Free Full Text]
  26. Plomp R. A signal-to-noise ratio model for the speech-reception threshold of the hearing impaired. J Speech Hear Res 1986; 29:146-154.
  27. Brand T, Kollmeier B. Efficient adaptive procedures for threshold and concurrent slope estimates for psychophysics and speech intelligibility tests. J Acoust Soc Am 2002; 111:2801-2810.[CrossRef][Medline]
  28. Saunders GH, Cienkowski KM. A test to measure subjective and objective speech intelligibility. J Am Acad Audiol 2002; 13:38-49.[Medline]
  29. Plomp R, Mimpen AM. Speech-reception threshold for sentences as a function of age and noise level. J Acoust Soc Am 1979; 66:1333-1342.[CrossRef][Medline]
  30. Van Wijngaarden SJ, Steeneken HJ. Objective prediction of speech intelligibility at high ambient noise levels using the speech transmission index. Presented at the Sixth European Conference on Speech Communication and Technology (Eurospeech 1999), Budapest, Hungary, September 5–9 1999.
  31. Moelker A, Vogel MW, Pattynama PM. Real-time modulation of acoustic gradient noise in interventional MRI. Concepts Magn Reson 2004; 20:34-39.[CrossRef]
  32. Occupational Safety and Health Administration (OSHA) CFR-1910.165, Federal Regulations. Employee alarm systems (Occupational Safety and Health Standards) Washington, DC: U.S. Department of Labor, Occupational Safety and Health Administration, 1981.
  33. Amaro E, Jr, Williams SC, Shergill SS, et al. Acoustic noise and functional magnetic resonance imaging: current strategies and future prospects. J Magn Reson Imaging 2002; 16:497-510.[CrossRef][Medline]
  34. Bandettini PA, Jesmanowicz A, Van Kylen J, Birn RM, Hyde JS. Functional MRI of brain activation induced by scanner acoustic noise. Magn Reson Med 1998; 39:410-416.[Medline]
  35. Cho ZH, Chung SC, Lim DW, Wong EK. Effects of the acoustic noise of the gradient systems on fMRI: a study on auditory, motor, and visual cortices. Magn Reson Med 1998; 39:331-335.[Medline]
  36. Bartels LW, Bakker CJ. Endovascular interventional magnetic resonance imaging. Phys Med Biol 2003; 48:R37-R64.[CrossRef][Medline]
  37. Campeau NG, Huston J, 3rd, Bernstein MA, Lin C, Gibbs GF. Magnetic resonance angiography at 3.0 Tesla: initial clinical experience. Top Magn Reson Imaging 2001; 12:183-204.[CrossRef][Medline]
  38. Moelker A, Wielopolski PA, Pattynama PM. Relationship between magnetic field strength and magnetic-resonance-related acoustic noise levels. MAGMA 2003; 16:52-55.
  39. Foster JR, Hall DA, Summerfield AQ, Palmer AR, Bowtell RW. Sound-level measurements and calculations of safe noise dosage during EPI at 3 T. J Magn Reson Imaging 2000; 12:157-163.[CrossRef][Medline]



This article has been cited by other articles:


Home page
 CMAJHome page
W. E. Hodgetts and R. Liu
Can hockey playoffs harm your hearing?
Can. Med. Assoc. J., December 5, 2006; 175(12): 1541 - 1542.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Moelker, A.
Right arrow Articles by Pattynama, P. M. T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Moelker, A.
Right arrow Articles by Pattynama, P. M. T.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
RADIOLOGY RADIOGRAPHICS RSNA JOURNALS ONLINE