[go: up one dir, main page]
More Web Proxy on the site http://driver.im/

CA2475368A1 - Rapid hearing screening and threshold evaluation - Google Patents

Rapid hearing screening and threshold evaluation Download PDF

Info

Publication number
CA2475368A1
CA2475368A1 CA002475368A CA2475368A CA2475368A1 CA 2475368 A1 CA2475368 A1 CA 2475368A1 CA 002475368 A CA002475368 A CA 002475368A CA 2475368 A CA2475368 A CA 2475368A CA 2475368 A1 CA2475368 A1 CA 2475368A1
Authority
CA
Canada
Prior art keywords
data
stimulus
response
stimuli
noise
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Abandoned
Application number
CA002475368A
Other languages
French (fr)
Inventor
Michael Sasha John
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Individual
Original Assignee
Individual
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Individual filed Critical Individual
Publication of CA2475368A1 publication Critical patent/CA2475368A1/en
Abandoned legal-status Critical Current

Links

Classifications

    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00Measuring for diagnostic purposes; Identification of persons
    • A61B5/12Audiometering
    • A61B5/121Audiometering evaluating hearing capacity
    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00Measuring for diagnostic purposes; Identification of persons
    • A61B5/72Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235Details of waveform analysis
    • A61B5/7253Details of waveform analysis characterised by using transforms
    • A61B5/7257Details of waveform analysis characterised by using transforms using Fourier transforms

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Medical Informatics (AREA)
  • Acoustics & Sound (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Otolaryngology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Multimedia (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • General Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

Systems (10) and methods for performing automatic and rapid screening tests that involve acoustically presenting at least one modulated white noise stimulus to at least one ear of a subject with an acoustic transducer (26), recording response data obtained with at least one sensor (28), performing signal analysis on the response data and using the result to statistically evaluate the presence of at least one auditory steady-state response at a processor (12), and providing a pass/fail result to the user via a display (36) to indicate whether the subject has passed or failed the screening test.
responses.

Description

TITLE
RAPID HEARING SCREENING AND THRESHOLD EVALUATION
REFERENCE TO RELATED APPLICATION
This application claims the benefit of my provisional application serial No.
601354,991 filed February 8, 2002, which is hereby incorporated by reference.
FIELD
This patent specification is in the field of auditory assessment and relates to fihe rapid initial evaluation of hearing impairment also known as "universal newborn hearing screening". It describes systems and methods for quickly and objectively performing a screening test by using steady-state noise stimuli and periodic stimuli such as clicks. It further describes systems and. methods for evaluating an individual's hearing abilities and determining auditory thresholds by using ramped stimuli and evaluating the changes that these stimuli evoke in the individual's brain activity at different moments in time.
BACKGROUND
Hearing-impairment affects about three newborn babies per thousand.
These babies must be identified as early as possible so that adequate treatment can be provided while the baby learns to hear and speak, The early detection and treatment of hearing loss helps the hearing-impaired child to communicate effectively, and benefits society since this individual then requires less in the way of support. Governments in Canada, the USA and Europe have therefore instituted programs for universal newborn hearing screening. Infants cannot reliably respond to sound - one cannot ask a baby if it can hear a sound.
Screening must therefore be performed by measuring the response of the ear or the brain to sound. The ear's response can be measured using "otoacoustic emissions" and the brain's response can be tested using the "auditory brainstem response". Both tests have their drawbacks. The otoacoustic emissions are fast, but do not check whether the brain is receiving information from the ear. The transient evoked auditory brainstem responses take longer to record than would be optimal for a screening test.
Currently, the click-evoked auditory brainstem response (ABR) is the standard test for screening and evaluating infant hearing, but this is often not automatic and is not frequency specific. While this test will detect gross hearing loss, it may not detect hearing loss at specific frequencies that are important to the development of speech and language. Tone-evoked ABRs (Stape(Is, 1997) can be used to assess frequency-specific thresholds, but this testing procedure takes too long to be used routinely. The use of otoacoustic emissions (OAEs) is popular for quick screening of normal auditory function, but will only detect peripheral loss and will not detect problems related to central hearing loss.
Furthermore, while OAE methods can detect hearing impairment, these techniques cannot be used to determine the actual extent of the hearing loss.
Frequency specific audiometric techniques that are rapid and accurate are therefore important for detecting auditory thresholds and fitting hearing aids in infants who are unable to provide indication of their hearing abilities. The inventor has previously described novel techniques to record the brain's response to sound using auditory steady-state responses (ASSR). These techniques provide a very fast and accurate assessment of hearing at specific frequencies (John et al, 1998; John et al 2000; PCT/CA 01/00715). The new set of techniques can accomplish rapid screening and hearing evaluation in adults and in infants using methods that are faster and more accurate than those previously described.
. There are two main types of tests which may be implemented by an audiologist: screening tests and, threshold tests. These tests can be carried out for either frequency specific stimuli, or for non-frequency specific stimuli.
Current screening tests are done using non-frequency specific stimuli. The tests indicate whether or not an infant has a minimal acceptable level of hearing and usually provide a simple pass/fail result. In the case where and individual fails a screening test (hearing thresholds are elevated), a threshold test can provide a more full assessment of auditory abilities. tn a threshold test hearing is tested at several higher intensity levels in order to determine the thresholds of a subject.
In one embodiment, a rapid and non-frequency specific hearing screening test is described which uses amplitude modulated noise stimuli and periodic stimuli such as clicks, which are presented at a constant intensity and at sufficiently rapid rates so that steady-state responses can be obtained from the subject's EEG. In alternative embodiment, threshold tests are described in which tamping stimuli are used in order to determine non-frequency specific hearing thresholds.
In alternative embodiment, tamping stimuli are used in order to determine frequency specific hearing thresholds. It is an object to use tamping stimuli to provide frequency specific hearing assessment for at least one stimulus at several intensities. These new techniques can be used to test one ear at a time, or can be used to test both ears simultaneously, and can test several stimuli simultaneously in a single ear.
Two basic techniques are described. The first technique is a screening procedure which uses steady-state responses evoked by stimuli that are relatively non-frequency specific. The prior art includes a series of scientific publications on using single steady-state responses or the MASTER technique (John et at, 2000) enables the assessment of frequency specific hearing.
Because the responses to non-frequency specific stimuli, such as clicks and amplitude modulated noise are larger than responses to frequency specific stimuli, the novel methods described here can be done faster and can be done at lower intensities than the prior art and therefore are appropriate for a rapid screening test. Steady-state evoked potentials evoked by amplitude modulated noise have been investigated for many years (e.g., Rees et al., 1986). The inventor (John et al, 1998), used amplitude modulated noise and showed that the evoked steady-state responses were larger than those evoked by amplitude modulated tonal stimuli. However, until several experiments were done by the inventor, as are reported here, it was not understood that the increased size of these responses, which are also robust at low intensities, led to such a rapid detection of the responses that these stimuli could be used in a screening test.
Normally steady-state responses are recorded to using tonal carriers, which require a relatively long test duration and would not be appropriate for a rapid screening test. Additionally, click stimuli are used with the new methods.
Click stimuli have been used for screening for many years, but were not presented at sufficiently rapid rates to generate steady-state responses. Unlike convention screening tests that use clicks, when clicks are used in order to generate steady-state responses, the clicks must occur at a repetition rate that causes the time between click stimuli to be a sub-multiple of the epoch length. The second technique involves a set of methods for analyzing tamping auditory evoked response which can be evoked by either frequency specific or non-frequency specific tamping stimuli. The tamping technique provides information about response magnitude at several intensity levels. Ramping techniques have been described for responses evoked by amplitude magnitude tamping stimuli presented at rates of 5-60 Hz (Linden et al, 1987). Linden et al concluded that the variability in the estimates of threshold were not accurate enough to be useful as a clinical method. However, unlike the methods disclosed here, Linden et al did not utilize methods for increasing the homogeneity of the collected EEG
response data. Norcia and Tyler (1985) have used the sweep technique in the visual modality to test visual acuity. The tamping techniques described here are novel and offer advantages over the prior art because they use novel methods of data collection and signal processing techniques.
Systems and methods are described to measure hearing quickly and automatically. The methods can be completely objective, which means that subjective responses are not required from the person being tested and subjective interpretations are not required from the clinical personnel conducting the test. The instrument presents acoustic stimuli to a patient, and concurrently records brain activity with the premise that if the brain activity is altered by the sounds, then the individual can hear the sounds being tested. The instrument contains both hardware and software components. However, it is the software components and algorithms that are the main emphasis.
The proposed methods and system for performing rapid hearing testing relies on novel stimuli, signal processing techniques, and statistical methods that allow the testing to occur more rapidly, accurately, and thoroughly than currently available methods. The instrument may be used to test hearing in infants, elderly people, workers claiming compensation for noise-induced hearing loss, and any other individuals who are unable or unwilling to provide reliable behavioral responses during conventional hearing tests. The instrument may be used as a rapid screening test and can also be used to provide threshold information.
SUMMARY
Disclosed is , an apparatus for presenting stimuli while simultaneously acquiring EEG data, and analyzing the EEG data to detect the presence of steady-state auditory evoked potentials (SS-AEP) or ramping auditory evoked potentials (R-AEP) in order to rapidly achieve auditory screening and threshold estimation. The apparatus comprises hardware and software. Methods are disclosed for collecting and evaluating the data in order to evaluate changes that occur in the evoked responses over time. The software further enables displaying the results of ongoing testing and the final results of the test as well as the storage of the raw data and test results for subsequent viewing and/or analysis.
Also included is software that incorporates methods both for noise reduction, such as weighted averaging, and for data rejection/replacement. The software is also adapted to effect signal processing and statistical testing in order to detect the SS-AEPs and R-AEPs that are evoked by acoustic stimuli. The software is also adapted to carry out tests of data quality and consistency in order to determine if the results are meaningful and reliable.
Particular types of stimuli are used to increase the amplitude of the resulting responses or offer other advantages over alternative stimuli. These stimuli include amplitude modulated white noise, band-limited noise which can be either modulated or un-modulated. These stimuli also include ramped acoustic stimuli that are characterized by continuously changing amplitudes. These stimuli also include multiple intensity stimuli in which at least 2 intensities are simultaneously tested, using different modulation/repetition rates for a single type of sfiimulus.
The systems and methods disclosed here can be used for a variety of objective tests such as providing a simple normal/abnormal evaluation for simple screening, or for providing more sensitive estimation of audiometric threshold.
The stimuli and methods can be used for testing the aided and unaided thresholds of a subject.
The systems and methods further comprise a database of normative data, such as phase data, or response amplitude vs. intensity curves, which can be used to help detect SS-AEPs or R-AEPs and to determine whether these are indicative or normal or abnormal hearing. The database contains normative data which are grouped in terms of the subject's characteristics such as age, sex and arousal state over a variety of stimulus characteristics such as type of stimulus, the modulation rates (or repition rates).
Further objects and advantages will be apparent from the following description, taken together with the accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
For a better understanding of the written description and to show more clearly how it may be carried into effect, reference will now be made, by way of example, to the accompanying drawings which show preferred embodiments and in which:
Figure 1 a is a schematic of an embodiment of a suitable apparatus of the present invention;
Figure 1b is a flow diagram illustrating the general objective auditory test methodology;

Figure 2 shows amplitude spectra after 8 and 16 seconds of recording, two components in the spectra which are at the frequency of modulation of the periodic acoustic stimulus are shown;
Figure 3 shows the amount of time required for responses to reach significance in 20 ears using an exponentially modulated noise carrier stimulus Figure 4a shows a significance series for responses that was above the subject's hearing threshold.
Figure 4b shows a significance series for responses that was below the subject's hearing threshold.
Figure 5 shows a procedure for generating R-AEPs and a spectrogram and obtaining amplitude and phase plots using data from 1 subject Figure 6 shows examples of multiple intensity amplitude modulated noise stimuli that were created using either sinusoidal or exponential envelopes (and well chosen stimulus parameters), or created using clicks of 2 intensities.
DETAILED DESCRIPTION
An apparatus is disclosed for recording steady-state and ramp evoked potentials and a set of methods for using the apparatus to rapidly screen for hearing pathology or to obtain estimates of hearing threshold. The basic hardware and software components of the apparatus will be discussed first.
Methods for screening and then for threshold estimation will then be discussed.
For both these techniques, novel types of acoustic stimuli, data analysis, noise reduction and response detection methods will be discussed. Protocols for objective audiometric testing based on ramp stimuli and R-AEPs will be discussed. A main goal of developing a clinical screening instrument and set of techniques is to enable the testing to be achieved as quickly and accurately as possible. This allows the test to be performed in the brief periods when an infant is sleeping and allows many infants to be tested in a short time. Three fundamental methods of achieving a faster test are by making the signal larger, the noise background smaller, or interpolating from existing data. New procedures to address each of these areas are employed.
Hardware and software components Referring to Figure 1 a, the objective audiometric test apparatus 10 includes a processor 12, a data acquisition system that can be a board 14 having a digital to analog converter (DAC) 16 and an analog to digital converter (ADC) 18, an audiometer 20 having a filter 22 and an amplifier 24, a transducer 26, a sensor 28, a second amplifier 30, a second filter 32, a storage device 34 and a display monitor 36. The processor 12 is suitably programmed with a software program 40 comprising a signal creator module 42, and an analysis module 46 having. a noise reduction module 48 a detection module 50, and a database 52 having a plurality of normative values.
A persona! computer (PC), for example a Pentium 750 running Windows 2000, may provide the processor 12, storage device 34 and display monitor 36.
The software program 40 is run on the PC and the database 52 can be stored in the memory of the personal computer and can communicate with the software program. Alternatively, these components may be efFected on a laptop, a handheld computing device, such as a palmtop, or a dedicated electronics device.
The objective audiometric test apparatus 10 can be used to assess the auditory system of a subject 60 by presenting acoustic stimuli to~the subject 60, while simultaneously amplifying and digitizing the sensed potentials (i.e., EEG
data). The EEG data is then processed and statistically evaluated to determine if this data contains evoked responses (i.e, SS-AEPs or R-AEPs). For example, data processing may show the responses that are statistically significantly different than the background EEG noise levels. The design of the objective audiometric test apparatus 10 follows clear principles concerning the generation of the acoustic stimuli, the acquisition of artifact-free data or use of weighted averaging, the analysis of EEG data in the frequency-domain and the objective detection of the responses (e.g., John et al, 2000, 2001).
The data acquisition board 14 is a commercial data acquisition board (e.g., AT-MIO-16E-4) available from National Instruments, or is a comparable alternative. The data acquisition board 14 allows for the output of data via the DAC 16 as well as the input of data via the ADC 18.
The output from the DAC 16 is sent to a signal conditioner such as an audio amplifier or audiometer 20 which may also be under the control of the processor 12. The audiometer 20 acts to condition the stimulus which is presented to the subject 60 via the filter 22 and the amplifier 24. Rather than using the audiometer 20, functionally similar amplifyinglattenuating and filtering hardware can be incorporated into the audiometric test apparatus 10 to control the intensity and frequency content of the stimulus.
The stimulus is presented to the subject 60 via the transducer 26 which may be at least one free-field speaker, headphone, insert earphone (e.g., ER3A
from Etymotics Research) or bone conduction vibrator. The transducer 26 allows fihe steady-state or tamping stimulus to be presented to the left and/or right ears of the subject 60.
While the stimulus is presented to the subject 60, the EEG is substantially simultaneously sensed using sensors 28 which are typically electrodes. For example, one active electrode placed at the vertex location of the head of the subject 60, one reference electrode placed on the neck of the subject 60. and a ground electrode placed at the clavicle of the subject 60. Other configurations for the electrodes are possible. It is also possible to use more electrodes.
The sensed EEG data is routed to an amplifier 30 which amplifies the sensed EEG data to a level that is appropriate for the input range of the ADC
18.
The amplifier 30 may use a gain of 50,000. The amplified sensed EEG data is then sent to the filter 32 which filters the amplified sensed EEG data such that sampling can be done without aliasing by the ADC 18. The filter 32 may have a band-pass of 1-300 Hz. The ADC 18 digitizes the filtered amplified EEG data at a rate of approximately 1000 Hz, or other rate, provided the filter band-pass is set so that the Nyquist rate is not violated as is well understood by those skilled in the art.
The objective audiometric test apparatus 10 can be embodied in various ways. For example, multiple outputs may also be used (e.g. 8 DACs) to create the acoustic stimuli that are presented to each ear of the subject 60. This would allow some components of the stimulus to easily be dynamically manipulated in real-time independently from the others.
The software program 40, which is based upon an earlier program known as the MASTER (Multiple Auditory Steady-State Response) program, allows a user to select a particular auditory test to perform on the subject 60. The software program 40 comprises a plurality of modules that are not all shown in Figure 1a to prevent cluttering the Figure. The software program 40 controls test signal generation via the signal creator module 42. The software program 40 also allows the operator to select from a variety of pre-defined objective audiometric tests. The signal creator module 42 allows the creation of the time series waveforms that are used as the acoustic stimulus. The software program 40 then controls analog to digital conversion and digital to analog conversion according to the protocol of the auditory test that is being performed.
During an auditory test, the software program 40 analyzes the sensed EEG data via the analysis module 46 which includes the noise reduction module 48 and the response detection module 50. The noise reduction module 48 may employ weighted averaging, time averaging and/or various types of artifact rejection (which will all be described later in more detail). The reduced noise signal is then sent to the detection module 50 to determine whether at least one evoked response is present within the data. The detection module 50 may employ the phase weighted t-test, or other methods which will later be described in more detail. In the case of R-AEPs, the detection module 50 may also be used to determine if a reliable threshold estimate has been detected.

The software program 40 can display on fihe display monitor 36, the real-time test results and the incoming EEG data in bofih the time and/or frequency domain, and can save the test results and recorded EEG data on the sfiorage device 34, and also allows the test results to be printed by a printer (not shown).
The software program 40 also communicates with the database 52 which comprises a plurality of normative data values. The database 52 contains normative data from sample populations of subjects relating to a variety of parameters for screening or threshold testing. For instance, the database 52 includes normative mean and variance values for measures such as phase data, which can be used during a phase-biasing t-test. The database 52 can also include normative data for slopes of amplitude data for different types of stimuli and different modulation/repetition rates, and or other normative data.
The software program 40 implements a graphical user interface which includes a series of interactive screens. These interactive screens allow a user to control the software program 40, perform a desired auditory test, create customized stimuli, analyze and summarize test results, and enter demographic data about a patient.
The software program 40 permits the user to define the acoustic test signals. In addition to the amplitude and frequency modulated signals previously described in PCT/CA 01/00715, and the options which were available for modification, (e.g., modulation rate, carrier frequency), the software program allows for the creation of un-modulated or modulated noise which may be band-limified, and allows for the creation of ramp stimuli by allowing the shapes of the ramp functions to be defined. Ramp functions for creating ramp stimuli, are defined by parameters such as range of the tamping function and duration of tamping function (which define the rate of change over time), as well as the tamping function shape (e.g., linear, logarithmic, smooth or stepped). Both symmetrical and non-symmetrical tamping functions may be used. Additionally, the upward or downward portion of a tamping function may consist of more than one slope. For example, in a dual slope function the slope of the ramp is less while the intensity of the stimulus is lower and changes to a steeper slope when the ramping function traverses the higher intensity range. If dual slope functions are used, then two threshold estimations may be computed separately, using the R-AEP responses evoked by the two separate portions of the ramp.
Once the stimulus parameters are chosen, the signal creator 42 automatically adjusts them in order to ensure that certain rules are followed.
For instance, it ensures that an integer number of cycles of the modulation signal fit in the output buffer of the DAC 16 and the input buffer of the ADC 18. In the case of transient stimuli, the repetition rate is chosen so that duration of the stimulus combined with the post-stimulus period is an integer sub-multiple of the duration of input/output buffer (i.e., of the epoch length). This is important to avoid spectral spreading in the generated acoustic stimulus as well as to avoid spectral spreading in the sensed EEG data which are digitized by the ADC 18.
The signal creator 42 may also be used to present test signals to the subject with constant peak-to-peak amplitudes or constant RMS amplitudes, whereby the amplitude of the envelope of the test signal is increased to compensate for the modulation depth.
The signal creator 42 can also generate stimuli consisting of tones, broad-band noise, high-pass noise, low-pass noise, or band-pass noise, all of which can be either modulated or un-modulated. In the case of noise, the signal creator 42 may allow the user fo adjust the band-pass and band-stop characteristics of the noise including the roll-off of the transition region that is between the band-pass and band-stop regions.
The software program 40 permits the user to define the rate of the ADC
18, the rate of the DAC 16 (which must be a multiple of the A/D rate) and the epoch duration (i.e. the size of the input buffer contained in the ADC 18).
The user may also define an artifact rejection technique and associated parameters, calibration coefficients, phase adjustment coefficients, and whether on-line computations are made upon weighted or un-weighted (i.e. raw) data. The artifact rejection level may be based one or more criteria. For example, criteria may include an absolute threshold value, the average amplitude of the high frequency range of the sensed EEG data, or the characteristics of data already collected for that subject.
Prior to running the test, the software program 40 permits the user to view the stimuli that will be presented to the subject 60. During a test, the user may view the sensed EEG data for the current epoch that is being sampled. The user can also view the amplitude spectra of the average sweep (a sweep is a concatenation of epochs and the average sweep is the result from averaging a plurality of sweeps). When the spectra of the average sweep is displayed, the frequencies of the SS-AEPs or R-AEPs in the EEG data are highlighted for easy comparison with background EEG activity (i.e. background noise). The software also allows the user to view both the numerical and graphical results of statistical analyses that are conducted on the EEG data to detect the presence of at least one evoked response to a stimulus.
The software program 40 enables the user to choose different methods of viewing, storing, combining and analyzing data sets, which either the averaged data or the responses from a single subject or from a plurality of subjects.
The data sets may be combined so that each sweep that goes into a final average is weighted by the amount of data from which it is created or by the number of separate data sets combined. The data sets may also be subtracted, in order to enable the user to calculate, for example, derived-band responses.
The software program 40 has options for collecting and displaying data.
For example, as is commonly incorporated into clinical audiometric devices, the parameters for several clinical protocols can be stored in several parameter files to enable several tests to be run automatically, for example, each with different stimulus intensities or different stimuli. The results for tests incorporating different stimuli and different stimulus intensity levels can be displayed in several Test Summary screens where all of the audiometric test results of the subject 60 are presented, for example, in traditional audiogram format.

Figure 1 b illustrates the general steps undertaken by the objective audiometric test apparatus 10. The objective audiometric test apparatus 10 first generates a test signal in step M1 which is appropriate in testing an aspect of the auditory system of the subject 60. The test signal may comprise a wide variety of signals including amplitude modulated noise, click train stimuli, ramped signals, and the like. The next step M2 is to transduce the test signal to create a stimulus and present this stimulus while simultaneously recording the EEG data M3. The presentation of the stimulus and the acquisition of the EEG data must be synchronized to accurately represent signals of interest. The next step M4 consists of analyzing the recorded EEG data to determine whether there are any responses present in the EEG data. This step may also include estimating what auditory threshold may be based upon these responses. This step will typically involve performing a noise reduction method on the EEG data and then applying a detection method to the noise reduced data. The next step M5 may be to report test results. The steps outlined in Figure 1 b may be part of a larger audiometric test that will involve iteratively performing each of the steps several times and at difFerent intensities. These particular audiometric tests and the steps which are involved are discussed in more detail' below.
SS-AEP/R-AEP Detection The EEG data that is sensed during the presentation of rapidly presented click stimuli, steady-state stimuli, or ramp stimuli may contain several superimposed responses (as in the case of binaural or multiple-stimulus testing) as well as background noise. Accordingly, it is difficult to distinguish the SS-AEP/R-AEP responses in the time domain. However, if the EEG data is converted into the frequency domain, using a Fast Fourier Transform (FFT) afor example, the amplitude and phase of each evoked response can be measured at the specific frequency of each modulation/repetition rate in the stimulus.
The SNRs of the SS-AEPs and R-AEPs are very small compared to the background EEG. Accordingly, a sufficient amount of EEG data needs to be collected to increase the SNR of the response data to useful levels.
Conventional approaches to increase the SNR of the evoked response data include artifact rejection and time averaging. These conventional approaches are implemented by the analysis module 46 since these techniques are still fairly popular with clinicians and research scientists in the field of audiometry.
As previously mentioned, epochs of EEG data are acquired during SS-AEP testing. Artifacts may contaminate. the data and introduce large noise spikes that are due to non-cerebral potentials such as movement of facial muscles or the like. Accordingly, artifact rejection involves analyzing each epoch to determine if the epoch contains data points that are higher than a specified threshold level (e.g., 80 ~.V). Artifact rejection is useful in removing spurious noise components to noise reduction techniques such as time averaging to be more effective. The noise reduction module 48 is adapted to effect artifact rejection on the epochs which are recorded. If an epoch is rejected, the next epoch that does not exceed the artifact rejection threshold is concatenated to the last acceptable epoch. This concatenation procedure does not cause discontinuities in the data because the steady-state or ramped stimuli which evoke the steady-state or ramped responses are constructed so that each epoch contains an integer number of periods of the evoked steady-state responses.
Time averaging comprises concatenating epochs to form sweeps. A
plurality of sweeps are then averaged in time to yield an average sweep. Time averaging reduces the level of background noise activity that are not time-locked to the stimuli. After the average sweep is obtained, it is converted into the frequency domain via the FFT. In this case, the sweep duration is an issue since increasing the sweep duration distributes the background noise power across more FFT bins without affecting the amplitude of the SS-AEP response which is confined to a single FFT bin since the SS-AEP response occurs at a single frequency and the noise is broadband. Thus increasing the duration of the sweep increases the frequency-resolution of the FFT. The specific frequencies available from the FFT are integer multiples of the resolution of the FFT which is 1/(Nt), where N is the number of data points and t is the sampling rate. One possible implementation uses a sampling rate of 1000 Hz, an epoch length of 1.024 points and sweeps that are 16 epochs long (16,384 points). Accordingly, the resulting frequency resolution is 0.61 Hz (1/(16*1.024*.001)) and the frequency region in the FFT spans DC (0 Hz) to 500 Hz. Alternatively, sweeps may also be 8 epochs long or 12 epochs long.
The detection module 50 may provide a noise estimate which is derived from neighboring frequencies in the amplitude spectrum (i.e. FFT) at which no steady-state response occurs. If there were no steady-state response in the recorded data then the power at the modulation frequency, where the response should occur, would be within the range of the noise power at the neighboring frequencies. An F-ratio may then be used to estimate the probability that the amplitude at the modulation frequency in the resulting FFT is not statistically different from the noise estimate. When this probability is less than 0.05 (p<0.05), the SS-AEP may be considered significantly different from noise, and the subject 60 is considered to have heard the steady-state stimulus. A more stringent criteria of p<0.01 can also be chosen. The objective audiometric test apparatus can provides an F-Ratio where each SS-AEP response in the amplitude spectrum associated with a frequency of modulation is compared to the FFT data in 60 noise bins above and 60 noise bins below the FFT bin that contains the SS-AEP. In this case, this ratio is evaluated as an F-statistic with 2 and 240 degrees of freedom.
The objective audiometric test apparatus 10 further comprises the noise reduction module 48 which may be adapted to employ artifact rejection in which epochs are rejected based a specific criteria such as amount high frequency (e.g., 70-200 Hz) activity. The noise reduction module 48 may further employ other types of weighted averaging as previously described by the inventor (John et al., 2001, PCT/CA 01/00715). As will be described later, artifact rejection and noise weighting are more complicated for R-AEPs because the intensity of the stimulus is constantly changing. Accordingly, if an epoch is rejected, the subsequent epoch can not take its place because this epoch contains R-AEPs elicited by acoustic stimuli with a different range of intensities.
Referring now to the detection module 50, a phase weighted t-test may be used to detect the presence of SS-AEP responses in the recorded EEG data (Picton et al 2001, PCT/CA 01/00715). The phase weighted t-test employs data biasing to detect the SS-AEP response based on a priori knowledge about the SS-AEP response. As described in the prior art, several approaches can be used to define the expected phase. First and foremost, the database 52 may contain normative expected phase values. Other methods are also described for choosing correct reference phase values.
The detection module 50 may be further adapted to perform other statistical methods for detection, such as the MRC method. The use of an expected phase angle has been incorporated as a variant of the Rayleigh test for circular uniformity (RC) termed the modified Rayleigh test (MRC). The RC
method can be made more statistically powerful if an expected phase angle is known. Probabilities for these two types of tests are computed using critical values available in standard statistical reference materials (e.g. Zar, 1999).
The objective audiometric test apparatus 10, contains software that allows the signal creator 42 which can create considerably more types of acoustic stimuli than has been described in the prior art. The software not only an create steady-state tonal stimuli, but also is adapted to construct a variety of test signals which can be used to evoke the SS-AEPs and R-AEPs. These test signals include amplitude modulated noise, transient stimuli such as click trains, and ramp stimuli. The signal creator 42 can also generate signals consisting of high-pass, low-pass, or band-pass noise, all of which can be either modulated or un-modulated. The signal creator can also generate a train of rarefaction, condensation, or alternating polarity clicks.
Several novel methods will be described which offer advantages over the current art. The first method is for rapid screening. The second method is for obtaining threshold estimates and utilizes ramping stimuli. This method is used for obtaining threshold data and may be used both with relatively non-frequency specific stimuli or with frequency specific stimuli and provides more information about a subjects audiogram. The third method is a screening test that also may provide some information about threshold.
Method 1: Rapid Screening Stimuli In one preferred embodiment, a rapid screening test is accomplished by recording the auditory steady-state response to at least one amplitude modulated white noise stimulus which is presented to at least one ear of a subject. The inventor has shown that this stimulus evokes a significantly larger steady-state response than the modulated tonal stimuli normally used in frequency specific auditory tests. In our initial experiments using amplitude modulated noise stimuli, the response was about 1.5 times the size of the response to a pure tone that was presented at the same intensity (John et al., 1998). In the experimental results of Figure 2, data are provided that this increase in amplitude enables the extremely rapid detection times required in a screening technique. The efficiency of the detection procedures depends on square of the signal-to-noise ratio, and increasing the signal/response amplitude by 2 will therefore increase the speed of the test by 4. Repetition/modulation rates should be sufficient to create steady-state responses so that these can be evaluated in the frequency domain.
In adults this should be above approximately 30 Hz and usually less than approximately 300 Hz, while in infants this should be above approximately 70 Hz.
Figure 2 shows data from one adult subject and demonstrates the feasibility of a screening protocol. The responses to the left ear and right ear stimulus and to the right stimulus appear 85 and 95 Hz, respectively, which were the rates at which the noise stimuli were modulated. The stimuli were presented at an intensity of 50 dB SPL, which is about 25 dB above normal hearing threshold. The responses were assessed after 8 seconds and after 16 seconds.

The graphs plot a portion of the amplitude spectrum near the frequencies at which the responses appear. Each response (arrowheads) can be identified at the frequency of stimulus-modulation. The response is present if the amplitude of the response is statistically larger than the background noise levels. In this subject the responses were statistically significantly different from noise at seconds (open arrowheads) and were highly significant and visibly larger than background noise (filled arrowheads) by 16 seconds.
Figure 3 shows the distribution of the times required before the responses in 10 subjects (20 ears) were significantly difFerent from the background noise.
Seventy-five percent of the responses were significant by one minute, and all were significant before 3 minutes, demonstrating the promise of the procedure as a very rapid screening test. These results have been replicated on other subjects and other types of stimuli have been tested as well. In a recent study (John et al., 2003) the inventor demonstrated that while the average response amplitude for an amplitude modulated broadband noise (BBN) stimuli (band-pass 1-Hz to 8 kHz) was 77 nV, the amplitude was larger for a when the modulated noise did not contain lower frequencies. Using a modulated high-pass noise (HPN) stimulus (2kHz to 8kHz) the amplitude of the steady-state response increased to 89 nV. Further, the Inventor demonstrated that a steady-state response can be amplified by simultaneously presenting sound at a lower frequency. When the HPN stimulus was presented simultaneously with an amplitude modulated low-pass noise (LPN) stimulus the response to the HPN
stimulus increased to 94 nV. This type of enhanced HPN stimulus is called an EHPN stimulus. Accordingly, BBN, HPN, and EHPN stimuli can all work well for screening. Variations of these stimuli are possible. For example, a different range for the frequencies of the HPN (e.g., 1.5 kHz to 7kHz) may be used, and the LPN stimulus can be modulated or un-modulated.
While amplitude modulated noise stimuli have been used to study the auditory system for years, they have not been considered as potentially useful for a screening test. The inventor studied modulated noise in subjects more than five years ago. However, until the inventor conducted a recent study (John et al, 2003) it was not understood that the response to noise stimuli were of sufficient magnitude to cause the SS-AEPs to become significant within the very quick amount of time needed for a screening test. (e.g., within 3 minutes) for all the subjects tested. Although the stimuli used in this study were slightly higher in intensity than would be used in a screening test, the data suggest that presenting these stimuli at slightly lower intensities would still enable a screening test to be clinically useful. The average amount of time for most of the responses to become significant was between 30 and 60 seconds, with the maximum time being no longer than 2 minutes, and no subjects with normal hearing failed to produce a significant response.
Rather than using steady-state stimuli, transient stimuli presented at rapid rates can also produce SS-AEPs. However, in order for the Fourier analysis to work accurately, the transient stimuli should occur at intervals that were equal, to integer sub-multiples of the DA and AD buffers. When both ears are tested at the same time, the stimuli in one ear also should occur at a different rate than in the other ear. An example of how this can be done is as follows. The number of points in the DA buffer are be made equal to the product of the integer numbers of cycles of the two stimuli within a single epoch multiplied by a power of 2 (giving approximately 32,000 data points). A further proviso that the AD
bufFer is exactly 1/32 of the DA buffer is ensured by choosing the two rates so that the final number of DA-buffer points is divisible by 32. For example, the 2 modulation rates can be 90 and 96 cycles per epoch which will result in a product of 8640.
This value is then multiplied by 4 to give 34,560 points. This result is then divided by 32 in order to obtain the number of points (1080) that were in each AD
buffer.
Because the AID rate was set at 1000 Hz and the A/D buffer was 1080, the epoch duration was 1080 ms and the actual frequencies for the two stimuli were 83.33 Hz [i.e., 90*(1000/1080))] and 88.89 Hz, respectively. Both the A/D rate of 1000 Hz and the D/A rate of 32,000 Hz were acceptable since these are both integer submultiples of the clock used by the data acquisition board (e.g., a MHz clock).

The inventor recently tested several types of stimuli, which were adjusted to have approximately the same intensity relative to a subject's behavioral threshold (i.e. nHL). The first stimulus was the BBN stimulus. The next two stimuli were condensation clicks (CC) and rarefaction clicks (RC), lasting 125 ps.
The remaining stimuli were 1 ms bursts with instantaneous rise and fall times.
These bursts contained BBN or a tone (e.g. 1400 Hz). The average response amplitude for the BBN was 90 nV, and increased to 129 and 137 for the CC and RC stimuli, respectively. The burst-BBN and burst-tone stimuli produced response of 126 and 149 nV, respectively. All of these transient stimuli can be used in the rapid hearing screening method.
When used as a screening test, the stimuli are presented at a single level and the subject receives a pass/fail result depending upon whether the responses statistically determined to be present (e.g., the amplitudes of the responses are larger than the background noise estimate, or the phase of the responses is determined to be statistically stable). Alternatively, the amplitude modulated noise stimuli can be presented at 2 or more levels. The lowest level at which a response occurs for a particular ear is the threshold for that subject in that ear. The test may be repeated 2 or 3 times in order to ensure the reliability of the results. Additionally, the overall amount of background noise will afFect the SNR level, and accordingly, response detection, a limit can be placed on the amount of background noise that is acceptable.
If there is too much noise in the recorded data, a subject may not show significant responses even though hearing is normal. One method of determining the acceptable amount of background noise is to use normative data, whereby the noise estimate must be below some value in order for the test to be regarded as valid. Another method of determining the acceptable amount of background noise is to use an early sample of the response data (i.e., "a self-norm") to determine what the noise level should be by the end of the recording.
Only if the noise level reaches either one or both of these criteria is the test deemed to be acceptable. When records have too much noise the audiologist is warned that the test results are not valid. For example, the device can provide a warning signal, such as, "Too much noise to perform test accurately".
If the noise happens to occur at the same frequency as the response being measured, then the instrument may indicate that the subject can hear a stimulus, even though this is not the case. By computing the SNR for the epochs that are collected, it is possible to determine if one epoch has a much larger SNR
than the others. This would probably not occur physiologically. Accordingly, by using a homogeneity criteria based upon the epochs that are collected it is possible to dynamically remove epochs that do not meet a criterion (e.g., being within 3 standard deviations of the mean SNR level for the amplitude of a frequency bin compared to the amplitude of a noise estimate) as the test progresses.
Determining if the response is present or not may be approached in several manners. One method is to determine the average amount of time needed for the responses of a typical subject to reach significance. The testing can then be limited to this time period and at the end of the test, the response is evaluated. Alternatively, the responses can be evaluated sequentially, after each data sweep is collected, in order to produce a "significance series". A data sweep may be 1.024 seconds in length, or may be longer or shorter. Sequential response testing may lead to a shorter test time, but may lead to an increase in the number of false positives because as the number of statistical tests carried out increases the chance of finding significant results increases.
Figure 4a shows sequential response testing of a significance series in a subject presented with 30 dB SPL white noise stimuli that were about 5 dB
above the subject's behavioral threshold. Figure 4b shows sequential response testing in a subject presented with 20 dB SPL white noise stimuli that were about 5 dB
below the subject's behavioral threshold. The data were generated by collecting 220 sweeps of 4-seconds each and evaluating the data to determine if a response was present at the 0.05 significance level after each sweep was collected. Although the data in 4a reach significance quickly (within 21 epochs or 84 seconds), the data in 4b also become significant for a limited period towards the end of the recording. If testing was simply halted when the response reached significance then the data in 4b would yield a "false positive" result, suggesting that the subject heard the modulated signal, when in fact this was not true.
Below each figure is a series of numbers. The upper row contains points where the response transitioned from being non-significant to significant at the 0.05 level. The lower row contains points where the response transitioned from being significant to non-significant at the 0.05 level. For the stimulus that was heard, the average data made from iteratively adding the current epoch to past epochs indicates that from 8 to 12, 14 to 15, 17 to 19, and then from 20 onward the response reach significance. In figure 4B, the data become significant early in the recording period several times for only 1 epoch, but then there are two sections where the significance is longer, such as from 163 to 171, which is a span of epochs.
This problem can be countered using several techniques termed "statistical conditional criteria" (SCC). One technique, termed the "absolute count", requires an SCC in which a specific number of sweeps must reach significance before the response is considered significant. In other words, the response must remain significant for a specified number of tests which are performed, for example, upon the average sweep after each sweep is collected.
Using an absolute count SCC whereby 10 points must be significant would result in the data of 4b being correctly classified as a "response absent" result (i.e., "Fail"). Another SCC technique, termed the "sequential count", requires that a specific number of consecutive sweeps must reach significance before the response is considered significant. Using a sequential count criteria whereby consecutive points must be significant would result in the data of 4b being classified as a response absent result. Another technique, termed the "relative count", requires that the ratio of the number of sweeps that reach significance divided by the number of sweeps that did not reach significance must be above some value before the response is considered significant. Using a relative count criteria whereby 80% of the total number of points must be significant would result in the data of 6b being classified as a response absent result. The relative count criteria technique can also require that a minimum number, i.e., 30, sweeps must be collected prior to the statistic being used. In figures 4a and 4b, when using a significance series and SCC, the 0.05 critical value becomes more of a descriptive device than a statistic. Accordingly the critical statistic can be decreased to 0.01. By collecting data on a large normal population, the values for absolute count, absolute consecutive count, and relative count can be computed using critical values of 0.05, 0.01, or other critical values so that these SCC yield the desired levels of false positives and false negatives. For example, Bonferroni corrected values may be used as is often done with repeated measure statistics. The number of false positives can be measured by evaluating frequency bins, which do not correspond to a modulation/repetition rate of a stimulus being tested, and computing how may are statistically present (e.g., counting how many bins are significantly larger than the noise estimate). The number of false negatives can be measured by evaluating frequency bins, correspond to a modulation/repetition rate of a stimulus being tested, and computing how many are statistically absent although they should be present (e.g., counting how many bins fail to be significantly . larger than the noise estimate). The appropriate values for various SCC can also be determined using Monte-Carlo simulations.
Method 2: Ramping Stimuli Techniques for Screening and Threshold Tests.
Stimuli In an alternative embodiment, a method is used to achieve rapid estimate of a subject's hearing threshold, whereby tamping evoked potentials are evoked by a rarriping intensity stimulus. Various functions may serve as the tamping envelope. Because intensity is measured upon a log scale, using decibel (dB) units, a ramp which has a linear growth function when plotted in dB units is appropriate. Because any large jump which occurs in the stimulus may be startling to a subject, it may be preferable to use a symmetrical tamping function which consists of a first half that increases in intensity, followed by a second half that decreases in intensity. The last part of the stimulus is presented at an intensity which is equal to that of the first part of a subsequent stimulus.
Symmetrical tamping functions thereby avoid the large changes in stimulus intensity that would exist if a simple continuous increase or decrease of intensity were used. However, either symmetrical or a-symmetrical ramp functions can be used.
A tamping stimulus can be created digitally by multiplying a tamping function with a base signal. A base signal can be conventional types of modulated, steady-state, or periodic stimuli. Repetition/modulation rates of the base signal should be sufficient to create oscillatory tamping responses that can be evaluated in the frequency domain. In adults this should be above approximately 30 Hz and usually less than approximately 300 Hz, while in infants this should be above approximately 70 Hz. The base signal, such as an amplitude modulated BBN stimulus, should range between -1 and 1. In a programming language called LabVIEW, the broadband noise can be created with the subroutine called "uniform white noise.vi", which is then modulated at a particular frequency. After the base signal is created it is multiplied with the tamping function in order to create a tamping stimulus. The tamping function preferably ranges between 1 and zero. When the instantaneous value of the tamping function is 1, the tamping stimulus will therefore have its maximum intensity. The maximum intensity is equal to 1, with the decrease in intensity being determined by the equations:
C=R/(N*20) It=10~(-t*C), Where C is a constant which is equivalent to the intensity step, R is the desired range of intensity in dB, N is the number.of points in the ramp function, It is the SPL level at a given time-point, "t" is the current time-point.
Accordingly, if R is set to 10 and the base signal has an .amplitude which will produce an intensity of 50 dB SPL then the tamped stimulus will decrease from 50 tp.40 dB
SPL. An upward ramp can be generated by taking the original ramp function and reversing the order of the points. By setting N to '/Z of the desired stimulus, and then reversing the original function and adding it this to the original function a ramp stimulus of N length can be created which contains both an upward and , downward going ramp (i.e., a symmetrical ramp). Other types of ramping functions can also be used, including non-linear, and multiple slope functions.
For example, ramp can contain two slopes, the first slope is relatively shallow and lasts for 80% of an upward ramp, while the remaining 20% of the upward ramp has a steeper slope. This may be useful since responses at lower intensities have a lower SNR than ramping responses to higher intensities. A
ramping stimulus can also be created by analog means whereby the intensity of a programmable audio amplifier dynamically adjusted according to the ramping function.
General Method An example of performing a ramped intensity test is shown in figure 5, and is reflective of what the inventor has used in some initial studies with the method.
The first row of the figure shows the intensity of the ramp stimulus, which increased from 20 to 50 db SPL during the first 1h of the data sweep (8.192 seconds) and then decreased for the second half. This is a symmetrical ramp stimulus. As can be seen in the second row, the recorded EEG data were stored in 16 epochs, that lasted 1.024 seconds each. The contiguous epochs were - concatenated into 16.384 sec. sweeps which were averaged together in the time domain (each column is averaged together). In this example, 16 sweeps (i.e., 256 epochs) were collected, causing the recording to last about 4 minutes. The third row, shows the resulting 16.384 sec, averaged EEG waveform, which will contain responses that occurred over the entire range of the ramping stimulus.
This waveform may be analyzed in several manners in order to obtain the time x frequency information. For example, a spectrogram can be created by using a moving window of, for example, 1024 points, with an overlap of 1000 points (i.e., the window is iteratively shifted through the data by 24 points), the entire 16,384 point waveform can be analyzed in 682 (16384/24) separate FFT's which will yield a spectrogram. Such a spectrogram is shown in row 5. The responses to stimuli presented at 85 Hz (left ear) and 95 Hz (right ear) can be seen as 2 horizontal lines towards the top of the spectrogram which are distinctly larger than the background noise (the larger the amplitude, the lighter the color on the plot). A fuzzy line also appears at 60 Hz due to line noise in the building.
By extracting the row equivalent to the frequency of the modulated stimulus (or repetition rate if transient stimuli are used) one obtains the amplitude of the response over time. The amplitude and phase plots of the responses to stimuli presented to the left ear (85 Hz) can be seen, as a function of time, in rows 6 and 7 of figure 5. As the intensity of the ramp is increased, the amplitude of the response increases and the phase values stabilize rather than being random. By extracting the row equivalents of neighboring frequencies to the frequency of modulation of a tamping stimulus or the repetition rate, if a transient stimulus was the base signal, an estimate of the background noise (over time) can be obtained in order to carry out an F test. The current procedure 10 rows above and below the row corresponding to the frequencies of modulation are used in the noise estimate, and the response is compared to the noise using and F test with 2 and 19 degrees of freedom. In generating the noise estimate rows corresponding to other frequencies of modulation are skipped. In the amplitude plot shown in row 6, both the amplitude of the R-AEP over time and the average level of background noise (across the entire averaged sweep) are shown.
In order to estimate threshold the information in the amplitude plot, phase plot, or a mixture of the two types of information can be used as is well known in the art. The estimation of threshold can be done upon the raw data, or obtained using regression. If threshold is estimated from the raw data then threshold could be defined as the point at which the amplitude of the evoked response is not significantly different than the noise estimate. Since the intensity of the tamped stimulus is known at every point in time, it is simple to calculate what the intensity of the signal was when the signal failed to be significant. For example, if the signal starts to be significantly different than noise at 4 seconds, then the threshold is about 35 dB SPL (the slope of the intensity function in figure 5 is 30 dB range over 8 sec or 3.75 dB/sec). Since there is a 1 second window, the range for that data is 35 +/- 1.875 dB. Accordingly, the threshold could be estimated as between about 27 and 32 dB SPL. Alternatively, if physiological threshold is thought to occur at 10 dB above behavioral threshold, then the threshold could be calculated as 17 to 22 dB SPL. Smoothing of the amplitude or phase plots may occur prior to estimation of threshold. Other types of signal processing can be used upon the amplitude and phase plots, or on the actual spectrogram as well. As is known to those skilled in the art that the epoch length, the sweep length, the size of the moving window, and the number of points in the overlap can be changed without significantly deviating from the spirit of the invention.
Symmetrical ramp stimuli can provide 2 estimates of threshold, based upon the data from the upward ramp and the data from the downward ramp.
Alternatively, since the data are symmetrical around the maximum of the ramp, the amplitude and phase data from the second half of the test can be re-sorted in reverse, and added to the data obtained for the upward ramp to obtain mean values. If the tamping responses evoked by the downward ramp were different than those obtained to the upward slope due to, for example, louder stimuli being presented just prior to a softer stimuli (e.g. as may occur due to masking or hysteresis) then the first and second halves of the tamping results would be different, but initial work by the inventor has not found this to be the case.
Computing the mean amplitude from the first and second halves of the data will give a more reliable estimate of amplitude at a given intensity, but will not act to reduce the overall level of the noise floor. By obtaining the mean of the complex spectra from two data windows rather than merely the two amplitude values, better estimates can be obtained for both the signal and noise bins. In order to average the complex spectra, the data window should start at the same time for epochs on both sides of the ramp. This can be done by using integer sub-multiple of the epoch length (e.g., in 1024 point epoch, the data windows must be advanced by either 32, 64, 128, etc time-points). Otherwise only real rather than complex values are used for estimating amplitude. For phase data this stipulation 2~

is not really necessary because combining phase from slightly different areas will not really affect calculations of phase stability or "coherence".
The data in Figure 5 can be analyzed in several manners in order to determine the hearing threshold of a subject. For example only amplitude values whose squared values are larger than the sum of the squares of the noise estimate are considered significant and used to estimate threshold. A
regression line is fit to amplitude values that correspond to the significant points of the ascending slope of the response function. The point where the regression line intersects the x-axis is taken to be the behavioral threshold. Alternatively, a second regression line is fit to amplitude values that correspond to the significant points of the descending slope of the response function. The average of the x-intercepts for the regression lines fit rising and falling functions can then be used as an estimate of threshold. Alternatively, rather than using an amplitude criteria that compares signal to noise, the points of the response function used to fit the regression lines can be chosen based upon the phase, slope of the phase, or phase variance of the points meeting a specified criteria. Alternatively, both the amplitude and phase or phase variance can be used to identify the points in the spectrogram that should be used in the fitting of the regression line.
The use of a spectrogram to look at time varying spectral data is well known in the art. However, certain rules should be followed when using the ramp stimulus technique, since the time series data is time locked to a continuous stimulus, which will cause the technique to fail or severely under-perform when these are not followed.
The phase data produced by the spectrogram is defined arbitrarily in relation to the beginning of the current data window, rather than in relation to the phase of the stimulus. While the phase of the stimulus is invariant, for example, zero at the beginning of the recorded epoch, the phase of the response evaluated in the current data window will be a function of the point,in time of the beginning of the window and the modulation rate of the stimuli. The phase data can be adjusted according to the current region in the response data from which it is taken. Actual phase can be computed by adding the phase data obtained by the spectrogram to phase data related to the offset of the window from the beginning of the recorded data using the equation:
ea = e~ ( (T~ / L) * 360 Where, ea is the actual phase value of the response frequency being measured, e~ is the current phase value of the current data window in the spectrogram, T~ is cumulative time for the total number of points that have occurred in the response data prior to the first point of the current window in the spectrogram, and L
is the cycle length of the modulation frequency, or the duration of the inter-stimulus interval, of said at least one tamping stimulus that was presented to the subject.
Data Quality Control Techniques.
As can be seen in figure 5, the columns of sweeps are locked to a particular intensity range of the stimulus. All the epochs in the first column will have evoked responses which where elicited by the lowest intensity stimuli, the second column contains data evoked by a higher intensity, and so forth. In steady-state recording techniques and epoch may be rejected if the noise is above some threshold value, and a subsequent epoch can be used in its place.
However, if this is done in the tamping technique then the columns of the data matrix will contain evoked potentials elicited by stimuli of many different ranges and will not be sensible. Weighted averaging can be used instead somewhat successfully. However, if there is a large amount of noise, then when the epoch is normalized, the estimate of the signal will be diminished. In the zero replacement technique, a noisy epoch is replaced with zeros so that the averaged sweep is not affected by that epoch, and the average of the column data is divided by n-1 rather than n, where n is the number of total sweeps collected. A threshold value can be set as the absolute amplitudes of the time series data, or the amount of acceptable high frequency energy in an epoch (e.g., spectral energy from 70 to 200 Hz). However, because the slope of R-AEP
responses over time is used to determine threshold, homogeneity criteria may ensure that the data can be optimally evaluated. An example of homogeneity criteria is an inter-epoch noise criteria which can be created wherein the amount of noise in any given epoch can not be more than, for example, 200% of the average amount of noise measured in other epochs of that sweep.
.Alternatively, the inter-epoch criteria may be applied across sweeps, where as more data is collected, the average level of noise for that subject is more stable. Intra epoch criteria may be used both within sweeps, across sweeps, or both. Intra sweep criteria may also be used, where an entire sweep is rejected if it has substantially more noise than the other sweeps. Homogeneity criteria are more useful than merely rejecting an epoch of data because it has more noise than a cutoff value defined based upon normative data. If a subject is producing data that is well below the noise cut-off used to define acceptable limits of noise for a population, since the intra-subject noise may still vary considerably (e.g., due to state of arousal) and may still be below this limit.
The stability of the R-AEPs are important in the estimation of threshold.
As the intensity of the ramp stimulus decreases, the variance of the evoked responses will likely become larger, and the SNR will become worse. This variance will effect threshold estimation both when it occurs using simple SNR
criteria and when the slope of the responses is used to predict threshold, for example, using regression or other means. Rather than using all the response data contained in the amplitude plots when predicting threshold, only points in the amplitude plots which exceed a certain SNR, with respect to the noise estimate, may be used in the computation of threshold. This will obviously tend to cause the threshold estimate to be based upon the responses evoked by the higher intensity sections of the tamping stimulus. Alternatively, two or more amplitude plots may be obtained, dividing the sweeps in an A-B-A manner, and computing the data on both sub-averages. While these sub-average sweeps are also combined into a single average sweep upon which the threshold is determined, these sub-averaged sweeps may provide estimates of the stability of specific regions of the data. By computing the cross correlation between, for example, 1 second of data from the amplitude plots that are computed for the two (or more) sub-average sweeps, only the sections of the amplitude plots which have a correlation above a certain value will used in the final averaged sweep from which the thresholds are estimated.
Method 3: Multiple Intensity Technique for Estimating Threshold.
In an alternative embodiment, a method is used to achieve rapid estimate of a subject's hearing threshold, whereby a single type of stimulus is presented at several intensities. This type of stimulus is referred to as a multiple intensity stimulus. For example, in the case of amplitude modulated stimuli, several modulation functions are used each having their own intensity envelopes and modulation frequencies. These can be used to shape an amplitude modulated white noise stimulus. An example of how to make a multiple intensity stimulus is as follows: 2 or more modulation envelopes of different amplitudes (i.e., intensities) are summed together and multiplied with a noise stimulus. The response data evoked by this stimulus will contain responses evoked by stimuli at the different intensities. One difficulty with this technique is that the 2 or more modulation envelopes may serve to activate the cochlea at the same time and thereby reduce the response to each of the envelope components. In an additional embodiment of the technique this problem is diminished by using exponential envelopes with closely spaced modulation frequencies and correctly chosen phase values. For example, adding 2 envelope functions generated by exponential envelopes with the exponential set at 10, which are 180 degrees out of phase, and which have closely spaced modulation frequencies produces a multiple intensity envelope in which the different functions stimulate the cochlea at different times for at least a portion of the stimulus. Another solution to the problem of overlap is to rely on rapid transient stimuli rather than steady-state noise. Multiple transient stimuli can be simultaneously presented at different repetition rates, where each rate has a different intensity. Each of the repetition rates must have inter-stimulus intervals that are integer sub-multiples of the epoch length. Figure 6 shows multiple intensity envelopes created using both conventional AM envelopes (top row) and noise and using well chosen envelope parameters and AM noise (middle row). The third row shows a multiple intensity .
stimulus using clicks at 2 intensities. Each column in the figure shows the stimulus over a 50 msec period. In the upper graph of Figure 6 the 3 stimuli show considerable overlap, while in the middle graph the individual 3 envelope functions of different intensity are more discrete. The bottom graph shows the least amount of overlap due to the rapid presentation time of the stimuli. The responses to this type of multiple intensity stimuli can be used as a screening test, by examining whether the response to a stimulus at the screening intensity is significant. For example, a screening test can consist of 3 intensities being tested at once, where the middle intensity is the screening intensity. If the subject does not show a response to the middle intensity but does show a response to the higher intensity stimulus, which may be 7 dB above the middle intensity stimulus, then this may indicate that while the subject failed the screening test, hearing is almost normal. This may avoid the necessity of a subsequent test being necessary. The multiple intensity test can also be used to estimate threshold either by considering the significance of the evoked responses directly or by fitting the response data with appropriate regression equations. Because some masking of the simultaneously presented stimuli may occur due to the temporal overlap and temporal proximity of the stimuli, a weighting factor, or set of weighting factors may be multiplied with each of the evoked responses from this test. The issue of masking may also be addressed by presenting the multiple intensity stimuli at higher intensity ranges than are normally used in screening tests, if it can be shown that the information obtained at these higher intensities is relevant to an estimation of actual behavioral thresholds.
The systems and methods described here can be used in initial screening evaluations to provide a rapid, reliable, and automatic test of hearing. Novel stimuli such as amplitude modulated noise and transient stimuli, presented at certain repetition rates, will evoke large steady-state responses and thereby increase the speed of the automatic testing procedure. Novel statistical methods are described which use a significance series to decrease the occurrence of false positives and false negatives. Additionally, the use of ramping tests which rely on rapidly presented tamping stimuli to evoke tamping auditory evoked potentials is also described. Ramping tests can provide a rapid and objective estimate of threshold for either frequency specific or non-frequency specific stimuli. By performing homogeneity testing on the data, rather than simple artifact rejection criteria, the phase and amplitude plots, which provide a measure of the signal at different moments in time, can be used to obtain a better estimate. of threshold.
Additionally, the use of novel equation is used in order to make the phase data of the spectrogram useful in the detection of the response. Further, multiple intensity tests are described which can be used to obtain a quick screening test, as well as providing some information about a subject's threshold. All of the tests described can be performed with multiple stimuli and can be used to test both ears simultaneously. The methods described above can be implemented and incorporated into software programs by programmers of ordinary skill in the art, based upon this disclosure.
The presently described embodiments of the hearing evaluation systems and methods offer advantages over prior art. Although modifications and changes may be suggested by those skilled in the art, it is the intention of the inventors to embody within the patent warranted herein all changes and modifications as reasonably and properly come within the scope of their contribution to the art.
A number of references have been described in the specification. A full citation is presented below.
REFERENCES:
Patents:
PCT/CA 01/00715 John and Picton, System and Methods For Objective Evaluation Of Hearing Using Auditory Steady-State Responses.
Scientific Publications:
John M.S, Dimitrijevic, A., and Picton, T. W. Efficient Stimuli for Evoking Auditory Steady-State Responses, Ear and Hearing, submitted.

John M. S., Dimitrijevic, A., and Picton, T. W. Weighted averaging of steady-state responses. Clinical Neurophysiology, 112:555-562, 2001.
John, M. S., and Picton, T. W. MASTER: A Windows program for recording multiple auditory steady-state responses. Computer Methods and Programs in Biomedicine, 61, 125-150, 2000.
John, M.S., Lins, O.G., Boucher, B.L., and Picton, T.W. Multiple auditory steady state responses (MASTER): Stimulus and recording parameters. Audiology, 37:59-82, 1998.
Linden RD, Campbell KB, Hamel G, Picton TW. Human auditory steady state evoked potentials during sleep. Ear Hear. 1985 May-Jun;6(3):167-74.
Norcia AM, Tyler CW. Spatial frequency sweep VEP: visual acuity during the first year of life. Vision Res. 1985;25(10):1399-408.
Picton TW, Dimitrijevic A, John MS, Van Roon P. The use of phase in the detection of auditory steady-state responses. Clin Neurophysiol. 2001 Sep;112(9):1698-711.
Rees A, Green GG, Kay RH. Steady-state evoked responses to sinusoidally amplitude-modulated sounds recorded in man. Hear Res. 1986;23(2):123-33.
Stapells DR, Oates P. Estimation of the pure-tone audiogram by the auditory brainstem response: a review. Audiol Neurootol. 1997 Sep-Oct;2(5):257-80.
Review.
Zar JH. Biostatistical Analysis. Fourth edition. Upper Saddle River: prentice Hall, 1999.
All above references are hereby incorporated by reference in this disclosure.

Claims (14)

What is claimed is:
1. A method of performing an automatic and rapid screening test comprising:
a. acoustically presenting at least one modulated white noise stimulus to at least one ear of a subject;
b. recording response data;
c. performing signal analysis on said response data in order to generate result data;
d. using result data to statistically evaluate the presence of at least one auditory steady-state response, and e. providing a pass/fail result to the user, which indicates whether said subject has passed or failed said screening test.
2. The method of claim 1 in which step a comprises acoustically presenting at least one periodic acoustic stimulus to at least one ear of a subject, said stimulus being at least one of the following types of stimuli, an amplitude modulated broadband noise stimulus (BBN), an amplitude modulated high-pass noise stimulus (HPN), or an enhanced high-pass noise stimulus (EHPN).
3. The method of claim 1 in which step a comprises acoustically presenting at least one transient stimulus at a rapid periodic rate, such that the periodic rate causes an inter-stimulus interval that is a sub-multiple of the epoch length;
and step d comprises using result data to statistically evaluate the presence of at least one auditory steady-state response using a significance series and statistical conditional criteria of at least, the absolute count, consecutive count, or relative count.
4. The method of claim 1 in which step d comprises using result data to statistically evaluate the presence of at least one auditory steady-state response using a significance series and statistical conditional criteria of at least one of, the absolute count, consecutive count, relative count, and Bonferroni adjusted critical values.
5. The method of claim 1 in which step c comprises using weighted averaging in said signal analysis.
6. The method claim 1 in which step b, said recording response includes a. recording an epoch of data b. rejecting the epoch of data if said epoch fails to meet one or more of the following criteria; i) having an SNR level that is above a specified value for at least one specified bin of an amplitude spectrum, ii) failing a homogeneity test
7. The method of claim 1, wherein a threshold of a patient is obtained by performing steps a-d and then presenting said stimulus at a lower intensity and repeating steps a-d until said response of a specified number of a significance series fails to meet a specified criteria and indicates the absence of a response.
8. A method of testing auditory function comprising:
a. acoustically presenting at least one ramp stimulus to at least one ear of a subject;
b. recording data epochs of ramping evoked potential response data;
c. classifying said response data epochs into accepted epochs and rejected epochs, said response data being classified as rejected if said epoch fails in meeting a homogeneity criteria;
d. performing signal analysis on said acceptable response data in order to generate result data, and e. using result data to evaluate temporal changes in the acceptable response data at different instants in time and compute a subject's threshold for said at least one ramp stimulus.
9. The method of claim 8, wherein a base signal for the ramp stimulus is at least one of a periodic transient stimulus, a periodically modulated tone, and a periodically modulated noise stimulus.
10. The method of claim 9 wherein the base signal is the periodic transient stimulus, and the repetition rate of said periodic transient stimulus is equal to a duration which is an integer sub-multiple of an epoch of data.
11. The method of claim 8 wherein said performing signal analysis includes the steps of:
I. Computing a spectrogram using said response data;
II. Obtaining phase values from the spectrogram for at least one R-AEP;
III. Obtaining a phase plot from said phase values using the equation:
.theta.a = .theta.c ((T c/ L)*360 Where, .theta.a is the actual phase value of the response frequency being measured, .theta. c is a current phase value of a current data window in a spectrogram, T c is cumulative time for a total number of points that have occurred in a response data prior to the first point of the current data window, and L is a cycle length of the modulation frequency, or a duration of an inter-stimulus interval, of said at least one ramping stimulus that was presented to the subject.
12. A method of claim 8 wherein the ramp stimulus is created using a ramp function that is at least symmetrical, having an upward and downward ramp, said upward and downward ramp each being comprised of at least one slope.
13. A rapid and automatic screening method for evaluating auditory function comprising:
a. acoustically presenting at least two periodic acoustic stimuli of different intensities to at least one ear of a subject;
b. recording steady-state response data;

c. performing signal analysis on said response data in order to generate result data, and d. using result data to at least compare the amplitudes of at least two steady-state responses or to statistically evaluate the presence of at least two auditory steady-state responses.
14. An apparatus for performing an automatic and rapid screening test comprising:
a. means for acoustically presenting at least one modulated white noise stimulus to at least one ear of a subject;
b. means for recording steady-state response data;
c. means for performing signal analysis on said response data in order to generate result data;
d. means for using result data to statistically evaluate the presence of at least one auditory steady-state response, and e. means for providing a pass/fail result to the user, which indicates whether said subject has passed or failed said screening test.
CA002475368A 2002-02-08 2003-02-07 Rapid hearing screening and threshold evaluation Abandoned CA2475368A1 (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
US35499102P 2002-02-08 2002-02-08
US60/354,991 2002-02-08
PCT/US2003/003895 WO2003088836A1 (en) 2002-02-08 2003-02-07 Rapid hearing screening and threshold evaluation

Publications (1)

Publication Number Publication Date
CA2475368A1 true CA2475368A1 (en) 2003-10-30

Family

ID=29250453

Family Applications (1)

Application Number Title Priority Date Filing Date
CA002475368A Abandoned CA2475368A1 (en) 2002-02-08 2003-02-07 Rapid hearing screening and threshold evaluation

Country Status (4)

Country Link
AU (1) AU2003215123A1 (en)
CA (1) CA2475368A1 (en)
GB (1) GB2402067B (en)
WO (1) WO2003088836A1 (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DK3588984T3 (en) * 2018-06-29 2022-07-04 Interacoustics As SYSTEM FOR VALIDATION OF HEARING AID FOR INFANTS USING A SPEECH SIGNAL
CN110101395B (en) * 2019-04-24 2024-03-29 张语轩 Self-service hearing rapid measurement system

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
AUPP313798A0 (en) * 1998-04-22 1998-05-14 University Of Melbourne, The Improved evoked response audiometer
US6200273B1 (en) * 1999-04-26 2001-03-13 House Ear Institute Power-optimized cumulative, sequential statistical method for detection of auditory evoked potentials
DE19954666B4 (en) * 1999-11-13 2004-05-06 Pilot Blankenfelde Medizinisch-Elektronische Geräte GmbH Method for objective frequency-specific hearing threshold determination using the amplitude modulation following response (AMFR)
US6602202B2 (en) * 2000-05-19 2003-08-05 Baycrest Centre For Geriatric Care System and methods for objective evaluation of hearing using auditory steady-state responses

Also Published As

Publication number Publication date
GB0419924D0 (en) 2004-10-13
AU2003215123A1 (en) 2003-11-03
GB2402067B (en) 2006-07-19
WO2003088836A1 (en) 2003-10-30
GB2402067A (en) 2004-12-01

Similar Documents

Publication Publication Date Title
US8177726B2 (en) Rapid screening, threshold, and diagnostic tests for evaluation of hearing
US7014613B2 (en) System and method for objective evaluation of hearing using auditory steady-state responses
EP1284647B1 (en) Apparatus for objective evaluation of hearing using auditory steady-state responses
Purcell et al. Human temporal auditory acuity as assessed by envelope following responses
Martin et al. The effects of decreased audibility produced by high-pass noise masking on cortical event-related potentials to speech sounds/ba/and/da
Linden et al. Human auditory steady state evoked potentials during sleep
US6071246A (en) Process for automatic determination of hearing acuity, particularly of newborns and infants
Smurzynski et al. Distortion-product and click-evoked otoacoustic emissions of preterm and full-term infants
RU2481789C2 (en) Method and device for objective detection of hearing impairment
Whitehead et al. Measurement of otoacoustic emissions for hearing assessment
Harkrider et al. Effects of age and spectral shaping on perception and neural representation of stop consonant stimuli
Dimitrijevic et al. Human envelope following responses to amplitude modulation: effects of aging and modulation depth
Ménard et al. Relationship between loudness growth function and auditory steady-state response in normal-hearing subjects
US6524258B1 (en) Method for an objective frequency-specific determination of an audible threshold value using an amplitude modulation following response (AMFR)
Keefe et al. Detecting high-frequency hearing loss with click-evoked otoacoustic emissions
Van Dun et al. Optimal electrode selection for multi-channel electroencephalogram based detection of auditory steady-state responses
John et al. Introduction to technical principles of auditory steady-state response testing
Puel et al. Effect of visual selective attention on otoacoustic emissions
Purcell et al. Evaluating the modulation transfer function of auditory steady state responses in the 65 Hz to 120 Hz range
CA2475368A1 (en) Rapid hearing screening and threshold evaluation
Purcell et al. Concurrent measurement of distortion product otoacoustic emissions and auditory steady state evoked potentials
JP3613794B2 (en) Auditory measurement device
Wilson et al. Assessment of low-frequency hearing with narrow-band chirp-evoked 40-Hz sinusoidal auditory steady-state response
AU2001236922A1 (en) Audiometric apparatus and associated screening method
Small et al. Physiological Mechanisms Underlying ASSRs

Legal Events

Date Code Title Description
FZDE Discontinued