DE4441127B4 - Automated hearing screening based on otoacoustic emissions with mathematically defined probability criteria - Google Patents

Abstract

Method for auditing (automatic hearing screening) by recording, evaluating and displaying otoacoustic emissions, wherein a presence ("pass") or a lack of otoacoustic emissions ("fail") is determined according to strictly mathematical-binomial-statistical criteria, with the following steps
- Generating a stimulus
Measuring a response signal by means of a probe microphone in the auditory canal,
Conversion of the response signal into a binominal vector,
Summing the various binominal vectors respectively determined after the presentation of a stimulus,
Calculation of a value crit = (sum of the binomial vectors) / SQRT (number of samples) at defined phase-fixed times of the summed binomial vector after a predetermined number of measurements,
Judgment on the presence ("pass") or the absence of otoacoustic emissions ("fail") by comparing the calculated value crit with a given significance level.

Description

On untreated hearing problem in children and infants leads to serious communication problems with the environment.

State of the art

literature

systems for otoacoustic Emissions see:

• / 1 / GOLD, T .: The physical basis of the action of cochlea, in The Zoological Laboratory, University of Cambridge, April 27th 1948, pp. 492-498.
• / 2 / KEMP, D.T .: Stimulated acoustic emmisions from within the human auditory system in J. Acoustic Soc. Am., Vol. 64, no. 5, November 1978
• / 3 / KEMP, DT: EP 00 15 258 B1
• / 4 / WOLTER, A: DE 43 14 757 A1

This Consequences can over affect the entire life. Therefore, the detection of hearing loss in the earliest potential Age more and more importance.

Lots Studies show insufficient sensitivity and specificity in conventional Screening methods in childhood (see STEVEN J.C., WEBB H.D., HUTCHINSO J., SMITH M.F, BUFFIN J .: Clicked evoked otoacoustic emissions in neonatal screening, at 3rd International Symposium on Cochlear Mechanisms and Otoacoustic Emissions at Rome, December 4-6, 1992.)

Currently available Measuring Systems are based on the principle of evoked potentials or otoacoustic Emissions by means of signal averaging and individual evaluation through the examiner and are for Screening little suitable.

• – sehr kostenintensive Anschaffungen notwendig sind
• – die Untersuchungsergebnisse von einem geschulten, erfahrenen Untersucher interpretiert werden müssen (subjektive Einflüsse), wie in dem in Druckschrift /3/ beschriebenen System
• – keine exakte Aussage über die Signifikanz des Meßergebnisses möglich ist (fehlendes oder unzureichendes mathematisches Kriterium), wie in Druckschrift /4/ beschrieben

Previous systems based on otoacoustic emissions are only partially operational because:

• - very costly purchases are necessary
• - The examination results must be interpreted by a trained, experienced examiner (subjective influences), as in the system described in document / 3 /
• - No exact statement about the significance of the measurement result is possible (missing or insufficient mathematical criterion), as described in publication / 4 /

• – die Notwendigkeit einer interaktiven Artefaktschwelleneinstellung während der Messung entsprechend dem aktuellen Rauschpegel
• – die Unmöglichkeit, ein Signifikanzniveau für das Vorliegen einer Antwort mathematisch exakt zu definieren

A healthy inner ear generates a sound signal in response to an acoustic stimulus through a mechanical activity of the outer hair cells, which can be measured using a miniature microphone in the external auditory canal (evoked otoacoustic emissions). This phenomenon has been described in publication / 1 / for pure tones and in publication / 2 / for short-term stimuli. The detection of such a signal therefore largely excludes a speech-related peripheral (ie, not caused by a malfunction of the nervous system) hearing impairment. Since, in general, the sound level of the signal thus generated is significantly lower than the spontaneous frequencies measurable in the auditory canal (hereinafter referred to as noise), the signal-to-noise ratio (SNR) must be improved so far by signal processing methods that the signal emitted by the inner ear becomes identifiable. The decisive statement consists in the simple determination of whether such a signal is present or not. The systems described in the publications / 3 / and / 4 / apply for the method of signal averaging, in which by a phase-synchronous addition of a certain number of sweeps continuously improved the SNR. The decision on the existence of an emission had to be made by the experienced investigator. Publication / 4 / shows a method for signal-statistical analysis of the measured stimulus responses, which should ensure an examiner independence, the result of which, however, can be influenced considerably by the respective examination conditions. Decisive disadvantages of all previously used methods, however, were

• The need for an interactive artifact threshold adjustment during the measurement according to the current noise level
• The impossibility of defining mathematically a level of significance for the presence of a response

In order to was the intertesterreliability the method a crucial obstacle to the use of the method in a screening program for children especially in the first years of life. But just the method offers crucial Opportunities.

basis The invention is the physiological phenomenon of emission production of the inner ear so in a technical device to implement that the participation an experienced examiner becomes redundant and at the same time a precise indication of the significance of the result can be made.

task the invention described below is the currently existing measurement methods so to complete and change, that screening automatically according to strictly mathematical evaluation criteria and with mathematically defined probability lead to a result and thus also be carried out by not specially trained personnel can.

Picture captions:

1 : Measurement arrangement for the automatic registration of OA emissions

2 : Easy to clean and sterilize sound probe

3 : Statistical algorithm for the automatic evaluation of OA emissions according to mathematically defined probability criteria

Device description (see 1 ):

Next the already mentioned and further specifically described matemathic Evaluation method is the device especially as easy to handle, portable, battery operated System for designed for mobile use.

The Device can held in one hand and the measurement process is with a one-button operation executable, so that at the same time with the other hand the probe can be placed and adjusted.

Operating instructions, measurement sequence and measurement result become the user over a display is displayed continuously.

A multicolored LED additionally signals measurement errors and end of measurement.

The overall system, according to 1 , consists of a digital control and computation unit described in the following ( 1 ) for stimulus generation and verification, for measurement data acquisition and for statistical evaluation, a stimulus generator ( 2 ), a measuring amplifier with filter ( 3 ) for stimulus and signal conditioning, a display unit ( 4 ) for status display and guidance of the user, an interface ( 5 ) for possible data transmission to a personal computer, a sound probe ( 6 ) with sounder and sound sensor and a pluggable accumulator with voltage regulation and monitoring ( 7 ).

Of the Digital part is divided into a central control and processing unit (16 bit microcontroller), a program memory, a data memory, a two-color LED indicator (for pass / fail), a single-line LCD display for menu navigation and to display intermediate information (such as measurement unstable, artifact, ..), a serial and a parallel RS232 interface to the optional presentation of data or by personal computer.

Of the Analog part consists of a D / A converter with downstream filter and switchable output amplifier for stimulus output, a dual-channel stimulus input amplifier and measuring signal, a two-channel filter and the downstream analog / digital converter with input multiplexer for selection of stimulus or input signal.

The sound probe, according to 2 , consists of an acoustic sounder ( 10 ), a sound sensor ( 11 ), which together in a housing ( 12 ) are housed. Depending on the version, the cable is either connected directly to the sound sensor and the sounder with appropriate strain relief ( 13 ) in the housing feedthrough or by means of a miniature plug to the probe housing.

In addition, the probe has an easily removable probe tip ( 14 ) which ensures cleaning and sterilization without damaging the sensitive probe elements.

The Adaptation to the different auditory canals takes place by means of flexible rubbery spacers.

The small dimensions of the probe allow use in both infants, in children and also in adults (adaptation through spacers).

• 1) Reizgenerierung und Kalibrierung
• 2) Meßphase (Meßwerterfassung/Auswertung)
• 3) Auswertung/Anzeige
• 4) Datenübertragung nach PC oder Drucker

The device works in 4 phases during measurement procedure and measuring procedure:

• 1) Stimulus generation and calibration
• 2) Measurement phase (measured value acquisition / evaluation)
• 3) Evaluation / Display
• 4) Data transfer to PC or printer

1. stimulus generation and calibration

1.1 Programmable stimulus generation (Selectable Stimulus shape and stimulus rate)

The Stimulus generation occurs via a memory in which the waveforms and amplitudes of various Stimuli are stored.

through Microcontroller is first a certain stimulus (in shape and amplitude) is selected and to the downstream digital / analog converter with filter. This converts the digital stimulus values into an analog signal, its polarity and amplitude of software program can be controlled.

Around to speed up the measurement in troubled patients (infants) is the stimulus repetition frequency selectable and can be in a range of 60Hz to 100Hz.

1.2 Linear or nonlinear irritation

With This program control is the device for both linear irritation as well for nonlinear irritation designed.

at of linear stimulation all emitted stimuli have the same shape, polarity and phase, in the case of nonlinear irritation, a large positive alternate with three small negative stimuli in a row.

The Nonlinear stimulation has the advantage of extinguishing passive acoustic Reflections, the linear irritation has the advantage of a better Signal to noise ratio.

Linear and non-linear irritation see:
GRANDORI F., and RAVAZZANI P .: Non-linearities of clicked-evoked otoacoustic emissions and the derived non-linear technique in British Journal of Audiologogy, 1993, 27, 97-102.

1.3 Automatic Stimulus Calibration

• – Der Schalldruck im Gehörgang wird durch Amplitudenmessung über das Sondenmikrophon bestimmt und adaptiv bis zum optimalen Meßschalldruck verändert.
• – Die Stabilität des Stimulus wird über eine kurze Zeitspanne bestimmt. Die Meßphase beginnt nur unter der Voraussetzung einer ausreichenden Stabilität.
• – Das Nachschwingverhalten wird gemessen und nur bei ausreichender Dämpfung wird die Messung gestartet. Diese Kontrolle ist wesentlich, um auszuschließen, daß reizbedingte, passiv durch Gehörgangsreflexionen entstandene Echosignale im Auswerteintervall vorhanden sind und vom Auswertealgorithmus erkannt werden.

The calibration is based on 3 aspects:

• - The sound pressure in the ear canal is determined by amplitude measurement via the probe microphone and adaptively changed to the optimal measurement sound pressure.
• - The stability of the stimulus is determined over a short period of time. The measuring phase begins only under the condition of sufficient stability.
• - The ringing behavior is measured and only with sufficient damping the measurement is started. This control is essential in order to rule out that stimulus-related, echo signals which are produced passively by auditory canal reflections are present in the evaluation interval and are recognized by the evaluation algorithm.

After the beginning of the measurement (pressing the Start button) is the stimulus issued by the acoustic probe (Sound pressure) by means of a microphone located in the probe taken over an input amplifier supplied to the analog / digital converter and from the microprocessor via a defined period according to stability criteria. The System fits automatically adjusts to the different volume of the outer ear and presents itself to the optimal signal amplitude. First, a minimal stimulus delivered to the ear and measured.

at non-fulfillment the predetermined criteria, the signal amplitude is iteratively fixed increments until the stability criteria are met or the probe is relocated if the criteria are not met and the measurement must be restarted. As stability criteria counting Signal shape constancy and signal amplitude which for a minimum number of measurements must be constant in a defined sound pressure range.

Becomes no stable stimulus after a given number of measurements measured, the information 'unstable stimulus' is displayed to the user and the probe must be new be placed.

Of Further, this automatic sound pressure control will be several times while running a measurement, about the stimulus stability over a entire measuring time of up to 2 minutes to guarantee.

• – das Auftreten akustischer Resonanzen ausgeschlossen
• – die Signalamplitude bestmöglichst dem Volumen und den Gegebenheiten des äußeren Ohres angepaßt
• – die Stabilität des Stimulus während einer Meßperiode garantiert
• – und somit die Reproduzierbarkeit der Meßergebnisses garantiert

With this automatic sound pressure control is

• - The occurrence of acoustic resonances excluded
• - The signal amplitude best matched to the volume and conditions of the outer ear
• - Guarantees the stability of the stimulus during a measurement period
• - And thus guarantees the reproducibility of the measurement result

2. The measuring phase includes 3 tasks:

• 2.1 control of the stimulus stability, indication of lights, Interruption of evaluation in case of loss of stability, if necessary return into the calibration phase
• 2.2 Creation of the signal statistics, according to 3 A binomial vector is calculated for every phase-fixed point within the evaluated time: After each sweep, the field content of each component increases or decreases by one, depending on whether the signal amplitude at the corresponding position is positive or negative is negative. This vector is now evaluated according to statistical laws: A statistical frequency distribution of individual events from a population of two different events follows a binomial distribution. As the sample number grows, it converges to a Gaussian distribution. The mean of this distribution is defined by the relative probability of the occurrence of each of the two events. The half-width sigma of the distribution depends only on the number N of samples: σ 2 = 1 / N A DC-free stochastic and ergodic noise will therefore fluctuate around the zero point at any given time in accordance with this frequency statistic. This also applies to the acoustic spontaneous activity in the external auditory canal. For a phase-constant additional event that always adds the same contribution to the signal at one point, this distribution shifts by an amount corresponding to the signal-to-noise ratio (SNR) at that exact point. To prove such an additional signal, one starts from the null hypothesis that there is an equal distribution around the zero line at every distance from the stimulus, ie that no additional signal is present. If the current value for a vector component SUM / √ N is greater than a predetermined level of significance (for example, 3), this null hypothesis can be considered refuted, a phase-fixed signal is thus detected to a defined residual uncertainty. A given number of points in the evaluation interval must reach this criterion (positive and negative) in order to trigger a pass. The evaluation is done separately for two frequency bands. Both bands must pass.
• 2.3 Artifact Control: The noise level of the single signal is determined according to an amplitude criterion and depending on the result different quality levels assigned. Statistical evaluation is done for the different ones quality levels separated. Dependent SNR can improve the signal statistics with a single measurement or worsened. However, because the SNR is unknown during the measurement is, in principle, unpredictable, whether it is signal-statistical better is to include a specific measurement in the evaluation or Not. Since the amplitude of the useful signal to be measured (emission) while each individual measurement remains constant, it is therefore sufficient, the measured Divide sections into different "noise classes". For the result "pass" it is ultimately irrelevant in which class it was reached. Every class except the "best" (ie the one with the lowest mean noise amplitude) contains all values of their own and additionally all values of the next better Class. The estimation of the noise level can be done in several ways:
• a. Determination of the maximum amplitude of the single sweep
• b. Determination of the variance of the single sweep
• c. Determination of the variance of a phase fixed point in a Group of sweeps (e.g., 10). In this case all sweeps of this group must be the same quality class be assigned.
• - Of the State of statistical evaluation (deviation from the neutral line in sigma units} quasianalog is displayed.
• - The Measurement can be interrupted at any time by the examiner, which in any case leads to the message repeat. Otherwise, the measurement is after Achievement of the significance criterion (pass) or after a certain Number of valid sweeps finished (fail). An alternative to the measurement abort with the result fail is the drop below a minimum noise amplitude at each one phase fixed point, whereby the significance of the statement "no answer" also on the Based on their significance test can be made.

• – Bildung einer Teilsumme aus einer vorgegebenen Anzahl n von sweeps an jeder phasenfesten Stelle des sweeps
• – Berechnung der Einzelpunktvarianz an jeder dieser Stellen
• – Einteilung der Signalqualität nach der Einzelpunktvarianz für jeden Punkt
• – Erstellung der Signalstatistik für die angegebene Gruppensumme aus n sweeps nach dem oben angegebenen Algorithmus. Dabei wird die Gruppe genau so behandelt wie ein einzelner sweep nach dem obigen Verfahren

Modification of the method for small signal amplitudes:
For small signal amplitudes, the specified algorithm becomes less effective because the bit noise of the least significant bit of the ADC must be taken into account. For improvement, the following modification of the method is possible:

• - Forming a partial sum from a predetermined number n of sweeps at each phase-fixed point of sweeps
• - Calculation of the single point variance at each of these points
• - Classification of the signal quality according to the single-point variance for each point
• - Generate the signal statistics for the specified group sum of n sweeps according to the algorithm given above. The group is treated exactly like a single sweep according to the above procedure

3. Evaluation / Display

from pass or fail. In the case of a fail, in addition, the stimulus stability control and evaluated the relative artifact number. When falling below the minimum quality requirements (average noise level above a certain limit) will appear on Display an additional Message repeat, which shows the examiner that a Repetition of the measurement due to insufficient measuring conditions (if necessary at another time) is worthwhile.

4) Data transfer to PC or printer

Possible subsequent evaluation, Editing and documentation by transferring the data to one Personal computer or printer.

Claims (10)

Method for hearing test (automatic hearing screening) by acquisition, evaluation and representation of otoacoustic Emissions, with a presence ("pass") or a lack of otoacoustic Emissions ("fail") according to strictly mathematical binomial statistics Criteria is determined, with the following steps - Generation a stimulus - Measure up a response signal using a probe microphone in the ear canal, - Transformation the response signal into a binominal vector, - Sum up of the different ones each after the presentation of a stimulus Binominalvektoren, - Calculation of a value crit = (sum of binomial vectors) / SQRT (number of Random samples) at defined phase-fixed times of the summed binominal vector after a predetermined number of measurements, - Assessment of the Presence ("pass") or absence of otoacoustic emissions ("fail") by means of a comparison of the calculated Value crit with a given significance level. Method according to claim 1, characterized in that that the predetermined significance level of the calculated value crit must be achieved in two different frequency bands, to produce a "pass" result. Method according to claim 1, characterized in that that each signal section corresponding to its noise level one quality level (Noise class) is assigned and the statistical evaluation independently done in these classes. Method according to claim 3, characterized that a result "pass" from the noise class can be derived, the first to the statistical criteria fulfilled for a "pass". Method according to claim 1, characterized in that that the result "fail" after falling short of both the average noise amplitude of the response signal at each one phase-fixed point as well as after exceeding a certain point Number of valid evaluated intervals (sweeps) after a stimulus can be. Method according to claim 1, characterized in that that in stimulus generation both linear and non-linear Stimulus pattern can be used. Method according to claim 1, characterized in that that for the determination of for the measurement of optimal sound pressure an analysis of the stimulus in the ear canal stability criteria takes place, both amplitude and form criteria, in particular an investigation of the reverberation of the stimulus. Method according to claim 7, characterized in that that the analysis of the stimulus to an automatic stimulus calibration is used, with the stimulus calibration automatically at the beginning every measurement and as well several times during carried out the measurement becomes. Method according to claim 8, characterized in that that in case of failure to achieve predetermined stability criteria with respect to amplitude and shape of the stimulus the probe microphone will be repositioned in the ear canal got to. Method according to claim 1, characterized in that that the data is about an interface for evaluation, processing and documentation transferred to a PC or printer become.