Apparatus and method for automated measurement of auditory brainstem response

The adaptive averaging method in ABR testing dynamically adjusts recording iterations to improve signal-to-noise ratios, enabling efficient and accurate automated ABR analysis, addressing the limitations of subjective human judgment in existing ABR testing methods.

JP7584604B2Active Publication Date: 2024-11-15SHANGHAI NINTH PEOPLES HOSPITAL SHANGHAI JIAO TONG UNIV SCHOOL OF MEDICINE
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
JP2023178139
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2019-04-30
Filing Date
2023-10-16
Publication Date
2024-11-15
Estimated Expiration
2039-09-17

AI Technical Summary

Technical Problem

Current auditory brainstem response (ABR) testing methods rely heavily on subjective human judgment, leading to inconsistent results and difficulties in automating the analysis, especially for infants and young children, due to varying signal-to-noise ratios and waveform differences, which complicates hearing screening and lacks reliable commercial automation solutions.

Method used

An adaptive averaging method that dynamically adjusts the number of recordings through iterations to improve the signal-to-noise ratio by using cross-correlation functions to detect time-locked signals, determining the audibility threshold through function fitting and interpolation, allowing for automated and accurate ABR signal detection.

Benefits of technology

The method achieves highly reliable and efficient automated ABR testing with accuracy close to expert judgment, reducing the need for manual adjustments and minimizing unnecessary data collection, suitable for clinical and scientific research applications.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007584604000008
    Figure 0007584604000008
  • Figure 0007584604000009
    Figure 0007584604000009
  • Figure 0007584604000010
    Figure 0007584604000010
Patent Text Reader

Abstract

SOLUTION: An automatic measuring device, etc. for auditory brainstem response (ABR) are provided. An automatic measuring method comprises: a step A1 of acquiring a corresponding ABR record with sound intensity at start; a step A3 of determining whether an ABR signal is detected, when the ABR signal is detected, executing a step A4, and when the ABR signal is not detected, executing a step A5; a step 4 of determining whether current sound intensity reaches set minimum measurement sound intensity; a step 5 of, when the ABR signal is not detected at P consecutive sound intensities, executing a step A6, otherwise, proceeding to the step A4; and a step A6 of regarding the number of iterations used for each sound intensity, acquiring sound intensity corresponding to the audible threshold by function fitting and interpolation.EFFECT: An automatic measuring device and a method for an auditory brainstem response have high efficiency of detecting a threshold, have accuracy close to human determination, are further objective, have further high repeatability, and effectively reduce the number of repeated collections of ABR records.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to data processing technology, and in particular to an apparatus and method for automated testing of the auditory brainstem response. [Background technology]

[0002] The Auditory Brainstem Response (ABR) is a brainwave change induced by a sound stimulus, the characteristic waveform of which occurs within 10 milliseconds after the stimulus and is recorded by electrodes placed on the head of a person or animal. Due to its non-invasive nature, ABR is widely used clinically as a means of hearing testing and measurement, especially for infants, mentally disabled people, and surgical patients whose hearing cannot be tested due to speech communication or limb movements. The hearing threshold (the minimum sound intensity used to elicit a subject's ABR characteristic waveform) has already become one of the important indices for evaluating hearing function.

[0003] However, because the signal-to-noise ratio of ABR recordings varies greatly and the waveforms at each acoustic intensity vary greatly, currently, the threshold is basically judged based on the average of more than 100 recordings by professionally trained doctors. This subjective method of judging the threshold means that the accuracy of the results depends on personal experience and skills, and the reliance on professional skills increases, making it impossible to meet the ever-increasing clinical needs, and bringing great difficulties to the rollout of hearing screening for infants and young children nationwide. This also leads to the difficulty of automating ABR test data analysis. In the past 30 to 40 years, people have made a series of efforts and various proposals on automating ABR hearing tests, but no mature and reliable commercial products have yet appeared. Summary of the Invention [Problem to be solved by the invention]

[0004] The present invention provides an apparatus and method for automatic measurement of auditory brainstem response that is based on an adaptive averaging method and is capable of performing efficient and reliable automated processing and analysis of auditory brainstem response measurement data. [Means for solving the problem]

[0005] To achieve the above objective, one technical solution of the present invention provides an automatic measurement method for auditory brainstem response, including the following steps: A1, obtaining a corresponding ABR recording at a starting acoustic intensity; performing an adaptive averaging operation, stepwise increasing the number of averages of the records by iterations until the iterations are completed; For each iteration, group the ABR recordings corresponding to the current acoustic intensity at the current iteration, calculate the average curve of each group according to the current averaging number, respectively, obtain the time lag corresponding to the maximum value of the cross-correlation function between the groups, and determine whether an ABR signal having time-locking characteristics is detected according to whether the time lag deviation is within a predetermined range; If an ABR signal is not detected and the number of iterations has not reached the maximum number of iterations at the current sound intensity, perform the next iteration and continue obtaining an ABR recording at the current sound intensity; Step A2, if an ABR signal is detected or if no ABR signal is detected and the number of iterations has reached the maximum number of iterations at the current acoustic intensity, terminating the iterations and not continuing to obtain an ABR recording at the current acoustic intensity; Step A3: determining whether an ABR signal having a time-lock characteristic is detected in step A2, and if an ABR signal is detected, executing step A4, and if an ABR signal is not detected, executing step A5; step A4 of determining whether the current acoustic intensity reaches the set minimum test acoustic intensity, and if not, obtaining a corresponding ABR record at a new acoustic intensity and combining it with the original data, and re-executing step A2, the new acoustic intensity being obtained after subtracting the set interval from the current acoustic intensity, and if the minimum test acoustic intensity is reached, executing step A6; Step A5: determining whether or not an ABR signal is not detected at P consecutive acoustic intensities, and if an ABR signal is not detected at P consecutive acoustic intensities, executing step A6; otherwise, proceeding to execution of step A4; A method for automatically measuring an auditory brainstem response, comprising: a step A6 of performing function fitting and interpolation for the number of iterations used at each acoustic intensity in the calculation of the adaptive averaging method, obtaining a minimum acoustic intensity required to detect an ABR signal, and outputting the minimum acoustic intensity as an audible threshold.

[0006] In a preferred embodiment, the ABR record is a time curve in which the input type is a single record that is repeated, and in step A2, the adaptive averaging operation increases the number of records by iterations until the iterations are completed, and each iteration therein performs the following process on the received ABR record: Step S1' of randomly dividing the ABR recordings into two groups and calculating the mean curves respectively; a step S2' of calculating the cross-correlation function after each of the two groups has been averaged; A step S3' of obtaining a time lag corresponding to a maximum value of the cross-correlation function; Comparing the absolute value of the time lag with the k value to determine whether the time lag deviation is within k data points; If the absolute value of the time lag is smaller than the k value, it means that the time lag deviation is within k data points, indicating that a stable time-locked signal is detected. In this case, terminate the iteration and stop acquiring and collecting data, corresponding to the case where an ABR signal is detected. If the absolute value of the time lag is not smaller than the k value, it means that the time lag deviation is not within k data points, indicating that a stable time-locked signal has not been detected, in this case, step S4' further performs step S5'; determining whether the number of iterations has reached the maximum number of iterations for the current acoustic intensity; and step S5' of ending the iterations and stopping data acquisition and collection if the maximum number of iterations has been reached, and responding to the case where no ABR signal is detected; and if the maximum number of iterations has not been reached, performing the next iteration, acquiring new collected data and adding it to the received ABR record, and repeatedly performing the calculation process S1' to S5' at the current acoustic intensity until the iterations are ended. Here, k is a preset fixed value, or a value obtained after calculating all data points at a preset ratio.

[0007] In a preferred embodiment, the ABR record is a time curve in which the input type is a single record that is repeated, and in step A2, the adaptive averaging calculation stepwise increases the number of records by iterations until the iterations are completed, each iteration of which judges the received ABR record Q times in parallel; For each parallel judgment, first, the following steps S1 to S4 are executed, In step S1, the ABR recordings are randomly divided into two groups and the average curves are calculated respectively. In step S2, the cross-correlation function is calculated after each of the two groups is averaged; In step S3, a time lag corresponding to the maximum value of the cross-correlation function is obtained; In step S4, the absolute value of the time lag is compared with the k value to determine whether the time lag deviation is within k data points; If the absolute value of the time lag is smaller than the k value, it means that the time lag deviation is within k data points, indicating that a stable time-locked signal is detected; if the absolute value of the time lag is not smaller than the k value, it means that the time lag deviation is not within k data points, indicating that a stable time-locked signal is not detected; After completing step S4 for each parallel judgment, further execute the following step S5: In step S5, it is determined whether a stable time lock signal is detected in each of the Q parallel determinations; If a stable time-locked signal is detected in each of the Q iterations, the iterations end, and data acquisition and collection stops. Corresponding to the case where an ABR signal is detected, If a stable time-lock signal has not been detected in each of the Q times, step S6 is further carried out; In step S6, it is determined whether the number of iterations has reached the maximum number of iterations at the current sound intensity; If the maximum number of iterations has been reached, terminate the iterations and stop acquiring and collecting data, corresponding to when no ABR signal is detected. If the maximum number of iterations has not been reached, perform the next iteration, acquiring new collected data and adding it to the received ABR recording, and repeating step S1 for each of the Q parallel decisions at the current acoustic intensity; Here, k is a preset fixed value, or a value obtained after calculating all data points at a preset ratio.

[0008] In a preferred embodiment, the ABR record is a time curve whose input type is a multiple average record, and in step A2, the adaptive averaging calculation stepwise increases the number of average records by iterations until the iterations are completed, and each iteration of the adaptive averaging calculation performs the following process on the received ABR record: For the received ABR recording, the current average avgA is calculated according to the weights. 新 Calculate the line,

number

[0009] In a preferred embodiment, the averaging number used in calculating the average curve at each acoustic intensity is the number of iterations multiplied by a value of N, where N corresponds to the number of new groups of ABR recordings added at each iteration.

[0010] In a preferred embodiment, in step A6, the number of iterations used for each sound intensity is normalized, and then the normalized number of iterations is subjected to Sigmoid function fitting and interpolation to obtain the audible threshold.

[0011] In a preferred embodiment, the ABR records at each sound intensity are animal experimental data or clinical data, and are obtained by real-time testing or stored offline, and the data includes characteristic waveforms generated when sound stimuli induce electroencephalographic changes; The ABR recording is Signal amplification and Bandpass filtering and Adjusting the time interval for collecting the characteristic waveform by the electrodes, and selecting the time curve of the corresponding time interval after the onset of the sound stimulus as the analysis target; Elimination of the time curves corresponding to background noise, and The signal undergoes one or more pre-processing steps: removal of low frequency background noise by a smoothing spline fitting function;

[0012] In a preferred embodiment, the ABR records at each sound intensity are animal experimental data or clinical data, and are obtained by real-time testing or stored offline, and the data includes characteristic waveforms generated when sound stimuli induce electroencephalographic changes; The ABR recording is Signal amplification and Bandpass filtering and Adjusting the time interval for collecting the characteristic waveform by the electrodes, and selecting the time curve of the corresponding time interval after the onset of the sound stimulus as the analysis target; Elimination of the time curves corresponding to background noise, and The signal undergoes one or more pre-processing steps: removal of low frequency background noise by a smoothing spline fitting function;

[0013] In a preferred embodiment, the automatic testing device for auditory brainstem response further includes a function fitting module, which performs function fitting and interpolation processing for the number of iterations used in calculating the adaptive averaging method to obtain the minimum acoustic intensity required to detect an ABR signal, and outputs the minimum acoustic intensity as the audibility threshold.

[0014] In a preferred embodiment, the adaptive averaging module issues a command to the control module to collect data. When the data type input by the input module is a repeated single record, the adaptive averaging module performs one judgment or multiple parallel judgments corresponding to each repetition of each acoustic intensity. During each judgment of the one judgment or the multiple parallel judgments, the currently collected ABR records are randomly divided into two groups to calculate the average curves respectively, and then calculate the time lag corresponding to the maximum value of the cross-correlation function. The absolute value of the time lag is compared with a k value to determine whether the time lag deviation is within k data points. If the absolute value of the time lag is smaller than the k value, it indicates that a stable time-locked signal is detected. If the absolute value of the time lag is not smaller than the k value, it indicates that a stable time-locked signal is not detected. If the single judgment or each judgment among the multiple parallel judgments indicates that a stable time-locked signal is detected at the current acoustic intensity, the iteration is terminated, and the adaptive averaging calculation module issues a command to the control module to stop collecting data, and outputs a return information indicating that an ABR signal is detected at the current acoustic intensity, and stores the current iteration number by the storage module; If a determination indicates that a stable time-lock signal is not detected at the current acoustic intensity, or if none of the parallel determinations indicates that a stable time-lock signal is detected at the current acoustic intensity, the adaptive averaging calculation module further determines whether the number of iterations has reached a maximum number of iterations at the current acoustic intensity; If the maximum number of iterations has been reached, the iterations are terminated, and the adaptive averaging calculation module issues a command to the control module to stop collecting data, and sends back information indicating that the ABR signal cannot be detected at the current acoustic intensity, and stores the current number of iterations in the storage module; If the maximum number of iterations has not been reached, perform the next iteration, and the adaptive averaging calculation module further issues a command to the control module to add new data, and repeats the adaptive averaging calculation process with the current acoustic intensity after adding the newly collected data to the original data; Here, k is a preset fixed value, or a value obtained after calculating all data points at a preset ratio.

[0015] In a preferred embodiment, the adaptive averaging calculation module issues a command to the control module to collect data. When the data type input by the input module is a multiple average record, for each iteration, the adaptive averaging calculation module calculates:

number

[0016] In conventional ABR measurement, the sound intensity is measured from high to low, each sound intensity is repeated a certain number of times, and the signal obtained after averaging is judged by the doctor to be the lowest. The sound intensity at which the auditory brainstem response is judged to exist is the audible threshold (when the sound intensity is higher than the threshold, the cerebrum reacts and it is considered to be audible). In the past 30 to 40 years, in order to reduce misdiagnosis due to artificial and subjective judgment, the industry has developed automatic hearing measurement methods, but no mature commercial products have appeared, because the quantitative analysis of the signal is affected by the experimental conditions (e.g., the degree of anesthesia, the position of the electrode, etc.). Therefore, when the value is higher than the value, there is a signal, and when the value is lower than the value, there is no signal, and it is difficult to find an absolute value (such as signal-to-noise ratio or correlation coefficient) that can correspond to the audible threshold.

[0017] Compared with the prior art, the adaptive averaging algorithm provided by the present invention is used to analyze data generated in ABR-based hearing test, and can automatically obtain the hearing threshold of the test subject (the main index of hearing test) instead of human, where the number of times each sound intensity is repeated can be dynamically adjusted, that is, since there is no exact value corresponding to the hearing threshold, a standard is set first, and the problem of how many times it takes on average to reach this standard is solved.

[0018] For this reason, the adaptive averaging algorithm of the present invention increases the number of averaging recordings stepwise by iterations for ABR recordings collected in batches at each measured sound intensity from high to low, until it meets the condition for detecting an ABR signal, which is determined by randomly grouping the ABR recordings collected in batches to obtain average curves of two groups, calculating the time lag at which the maximum value of the cross-correlation function is located, and judging whether the ABR signal has a time-locked characteristic based on whether the time lag deviation is within a predetermined range, and the hearing threshold is obtained as the minimum sound intensity required for signal detection, or the sound intensity corresponding to the hearing threshold obtained by function fitting and interpolation of the number of iterations performed at each sound intensity.

[0019] The cross-correlation function is used to make the judgment. If there is a time-locked signal (ABR), the theoretical time lag is 0 (the actual calculation is k data points shifted, k=1 in the example, which corresponds to the black dots in FIG. 6), and if there is no signal, the time lag is an arbitrary value (corresponding to the white dots in FIG. 6). When using the adaptive averaging algorithm of the present invention, if the signal-to-noise ratio is high, the signal can be detected without multiple averaging, and if the signal-to-noise ratio is low, more averaging times are required to detect the signal. The algorithm can dynamically adjust the number of averaging times. If there is no signal, no signal can be detected (an upper limit is set) no matter how many averaging times are performed, as shown in FIG. 7.

[0020] Meanwhile, the number of iterations required to detect ABR signals at each measured sound intensity obtained by the present invention can obtain a more accurate audible threshold through function fitting and interpolation, improve the accuracy of threshold judgment, and effectively reduce the number of repeated collections of ABR recordings. When the measured sound intensity decreases by 5 dB, the conventional method only has an accuracy of 5 dB, but the present invention can reach 1 dB by function fitting (Figure 8).

[0021] The present invention can furthermore accommodate two different input data types (FIGS. 2 and 3), namely, data from single repeated recordings and data after averaging. These two data formats cover all clinical models, and the algorithms of the present invention can be applied to either one.

[0022] According to the method of the present invention, the accuracy is essentially 100% within an error of ±5 dB, and since each acoustic intensity detects a signal, the next acoustic intensity is executed, thereby reducing repeated recordings by up to 69% (Figures 10 and 11). Effect of the Invention

[0023] To sum up, the present invention provides an adaptive averaging calculation model, which realizes an efficient and reliable automatic measurement method of auditory brainstem response, and the accuracy of the audibility threshold obtained by this method is close to the human judgment of experts, and is more objective and reproducible.More importantly, the method of the present invention does not need to adjust the model based on data quality in the use process, so it has a wide range of future applications in various clinical and scientific research ABR measurements.It is also worth mentioning that the method of the present invention can end the iteration based on the judgment result and feedback to the hardware in real time, not only shorten the measurement time and avoid the waste of memory capacity, but also can be used as a core module in an unmanned fully automatic ABR hearing detection device, which is expected to completely free clinicians from the work of hearing measurement. [Brief description of the drawings]

[0024] [Figure 1] 1 is a flowchart of the method for automatically measuring an auditory brainstem response according to the present invention. [Diagram 2] 13 is a flow chart of module 1 when the input data type is a single record. [Diagram 3] 1 is a flow chart of module 1 when the input data type is averaged data. [Figure 4]These are mouse ABR time curves after averaging 350 times at each acoustic intensity. [Diagram 5] The results are obtained after the data of the same group has been subjected to the adaptive averaging method, and the two curves corresponding to each sound intensity are the average curves for calculating the cross-correlation function. [Figure 6] is the time lag corresponding to the maximum correlation coefficient obtained in an iteration. [Figure 7] The number of iterations required to detect a signal at each acoustic intensity. [Figure 8] This is the corresponding threshold value obtained by fitting the normalized iteration count with a Sigmoid function. [Figure 9] 1 is a schematic block diagram of an automatic auditory brainstem response measuring device according to the present invention. [Figure 10] This is a comparison of the deviations of the threshold values ​​obtained by three methods: when a doctor reads the measurement data using the conventional averaging method, when a doctor reads the data required for the adaptive algorithm, and when a machine applies the adaptive algorithm of the present invention. [Figure 11] A comparison of the normalized total number of records obtained when a doctor reads the measurement data using the conventional averaging method and when a machine applies the adaptive algorithm of the present invention is shown. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0025] In response to a given sound stimulus (a short sound, often a short pure tone), the auditory system generates a series of potential responses. Auditory brainstem response (ABR) measurement records the waveform changes of these potentials and uses various digital signal processing algorithms to extract them from various strong noise backgrounds to obtain the auditory brainstem evoked potentials, which are an important index for evaluating the integrity of the auditory conduction system and monitoring the function of the nervous system.

[0026] The present invention provides an algorithm model based on adaptive averaging to realize an automatic measurement device and method for auditory brainstem response by performing automated processing on data signals acquired in ABR measurement.

[0027] First, a conventional measurement method and the principle according to the present invention will be briefly described.

[0028] In the conventional method, the doctor analyzes the original recording obtained by ABR measurement, repeats each measurement sound intensity a certain number of times, and obtains a number of time curves after smoothing. Originally, the doctor determines which sound intensity is the lowest but has an auditory brainstem response based on the above data, and sets this as the audible threshold, and when the sound is higher than this threshold, the cerebrum reacts and it is considered to indicate that it can be heard.

[0029] After analyzing the data, it can be seen that when the stimulus intensity is below the hearing threshold, the correlation coefficients between ABR recordings are distributed symmetrically around 0 (no correlation), and when the stimulus intensity is above the hearing threshold, the distribution of correlation coefficients moves toward +1 (positive correlation). The minimum correlation coefficient required to detect a stable ABR waveform can be determined by experience, which can serve as a threshold criterion (described in detail below).

[0030] In the actual operation process, the difference in the signal-to-noise ratio of data due to different experimental conditions (such as the placement of electrodes, the degree of anesthesia of animals, etc.) will affect the modeling of the correlation coefficient, so each recording needs to be calibrated to ensure the accuracy of its threshold judgment, which will affect its practical value. In response to the above problem, the present invention proposes to use the classical judgment standard of the time-locked signal (i.e., the time lag at the maximum of the cross-correlation function is 0) to adjust the signal-to-noise ratio by changing the number of averaging, and finally realize the dynamic detection of the ABR signal, which is called the adaptive averaging method. That is, the present invention dynamically adjusts the number of repetitions of each acoustic intensity to set a standard corresponding to the audible threshold, and determines how many averaging times are required to reach this standard and obtain the audible threshold through the adaptive averaging method.

[0031] Therefore, in the specific algorithm of the adaptive averaging method, the present invention uses a cross-correlation function to judge whether there is a time-locked signal (ABR), and if there is a signal, the theoretical time lag is 0 (the actual calculation is shifted by k data points, exemplarily k=1), or if there is no signal, the time lag is an arbitrary value.

[0032] In the example of Figure 2, the present invention performs a random group averaging operation on the data recorded by a single audio stimulus (another example of Figure 3 applies to a situation where only average data is derived, as described in detail below), and uses "whether the time lag of the maximum value of the cross-correlation function shifts within one data recording point" (typically <50 microseconds per data point due to systematic errors) to determine whether a time-locked ABR signal appears, and if no signal is detected, increases the amount of data stepwise by repetition until a signal appears or a preset maximum value is reached.

[0033] According to the actual measurement results of the above detection method, when the stimulus sound intensity is higher than the audible threshold, the number of averaging times required to meet the time lag condition is maintained at a relatively low level, when it is close to the threshold, the number of averaging times required increases exponentially, and when it is lower than the threshold, it reaches a preset maximum number, proving that the audible threshold is the maximum sound intensity at which no signal is detected.As can be seen from the above, when using the adaptive averaging method of the present invention, when the signal-to-noise ratio is high, a signal is detected without averaging many times, and when the signal-to-noise ratio is low, more averaging times are required to detect a signal, and an upper limit is set, so when there is no signal, no signal will be detected no matter how many times averaging is performed.

[0034] As shown in FIG. 1, the method for automatically measuring an auditory brainstem response of the present invention includes the following steps. Present speech stimuli from a starting acoustic intensity and collect corresponding ABR data by the device. In this example, mouse ABR measurement data is used, and the acoustic intensity of the stimuli during the experiment ranges from high to low (90 dB to 0 dB) with an interval of 5 dB. Step A1: The adaptive averaging operation (corresponding to module 1) determines whether data collection needs to continue in order to increase the number of averages, or whether collection should be discontinued. The operation of the adaptive averaging method (described in detail below) involves gradually increasing the number of recordings, dividing the current ABR data into two arbitrary groups using a random grouping and averaging method, and calculating the average curves respectively, obtaining the time lag corresponding to the maximum value of the cross-correlation function of the average curves of the two groups, and judging whether the absolute value of the time lag is smaller than k data points, up to the maximum number of iterations (the iterations are terminated when a signal is detected or there is no signal even if the maximum number of iterations is reached), to judge whether a stable time-locked signal exists. k is a fixed value, corresponding to a fixed number of data recording points, or k is a value obtained by calculating based on a certain ratio (e.g., all data points are 1%), and the following explanation and experimental verification are both performed using k=1 as an example). The experimental data can demonstrate that when the absolute value of the time lag is greater than the audible threshold, it is maintained at a level lower than k=1 (FIG. 6) Step A2; Step A3 of determining whether an ABR signal is detected in step A2, executing step A4 if the signal is detected, and executing step A5 if the signal is not detected; A step A4 of determining whether the current sound intensity has reached a set minimum value of the measured sound intensity (0 dB in this example), and if the minimum value of the measured sound intensity has not been reached, providing a corresponding sound stimulus for a new sound intensity obtained by subtracting the interval (5 dB) from the current sound intensity and re-executing the operation of A2, and if the minimum value of the measured sound intensity has been reached, executing A6; Step A5: determining whether or not an ABR signal is not detected at P consecutive acoustic intensities (P value in this example is 2, and can be adjusted according to the actual situation); if no signal is detected at P consecutive acoustic intensities, execute step A6; otherwise, proceed to execute step A4; The number of iterations used in computing the adaptive average is called, and the threshold is obtained by function fitting. In this example, the number of iterations at each acoustic intensity is the output (Figure 7), which is normalized and then fitted with a Sigmoid function (Figure 8).

number

[0035] In step A1, acquisition and pre-processing of ABR data includes:

[0036] (1) Animal experiment: ABR data was collected by using the TDTRZ6 / BioSigRZ system to excite mice with short pure tones (frequency 16 kHz, duration 3 ms) from 90 to 0 dB (5 dB intervals) in both ears (the original signal was amplified 20,000 times and filtered by a 50-5,000 Hz bandpass filter). Each sound intensity was recorded 500 times, the stimulation signal speed was 21 times / sec, the data collection frequency was 21 kHz, and the collection time interval was 0-15 ms after the sound stimulation.

[0037] In the TDTRZ6 / BioSigRZ system, the TDT auditory evoked potential workstation is composed of an RZ6 processor, a preamplifier and BioSigRZ experimental design and analysis software. The workstation can record various bioelectrical signals including brainstem evoked potentials, otoacoustic emissions, etc., in conjunction with low impedance electrodes, needle electrodes and surface electrodes.

[0038] Clinical: The data are from a shared recording (www.physionet.org) funded by the NIH. The experimental conditions are that the ABR is recorded by one ear at a stimulus sound (a short pure tone of 1 kHz or 4 kHz, with a stimulus rate of 24 times / s) from 100 dB to 30 dB (5 dB intervals), and the electrode placement of the subject is that the recording electrode is placed on the forehead, the reference electrode is placed on the postauricular mastoid on the same side of the stimulated ear, and the earth electrode is placed on the mastoid on the opposite side of the stimulated ear. Each acoustic intensity is recorded 1000 times, and the ABR signal is band-pass filtered from 30 Hz to 3000 Hz and recorded after being amplified 50,000 times by an amplifier, the sampling rate of the recording equipment is 48 kHz, and the time interval of collection is 0 to 15 ms after the sound stimulus. The above animal experiments and clinical ABR data acquisition, and each parameter during preprocessing described are only examples, and in reality, may be adjusted according to the specific application situation.

[0039] (2) Based on the time interval corresponding to the ABR waveform, select the ABR time curve between 0-6 ms (mouse) or 5-15 ms (clinical) after the onset of the sound stimulus as the analysis target. The time interval here is an empirical value and may be adjusted according to the actual situation.

[0040] (3) Background noise caused by muscle activity or respiration can be removed by deleting time curves greater than 11 microvolts or less than 11 microvolts (this step is optional, the parameter value depends on the hardware installation used).

[0041] (4) Remove low frequency background noise by a smoothingspline fitting function (this step is optional, an exemplary smoothing parameter is 0.5).

[0042] For example, after performing the above processing on the acquired mouse ABR data, 350 curves are taken at each acoustic intensity, and the correlation coefficient distribution between any two groups of curves can be calculated here. The calculation formula is as follows:

number

[0043] The median correlation coefficients for each sound intensity were plotted, and it was found that the correlation coefficients corresponding to sound intensities higher than the hearing threshold were larger than those corresponding to sound intensities lower than the hearing threshold.

[0044] In the conventional method, the median of the correlation coefficient corresponding to the vicinity of the threshold is calculated and plotted based on the threshold determined by the clinician, and the curve corresponding to the normal mouse is aligned with the curve corresponding to the mouse at high risk of hearing loss, so that the correlation coefficient corresponding to the hearing threshold (approximately 0.01 in this example) can be known. However, since the signal-to-noise ratios obtained in different experiments are different (correlation coefficients measured by reducing background noise), the reproducibility between the obtained curves is low, and the accuracy of determining the threshold is limited. In the present invention, the minimum number of averages required to detect ABR signals at different signal-to-noise ratios is not equal, and the average number is dynamically adjusted based on the signal-to-noise ratios of different measured acoustic intensities by calculating the adaptive averaging method, and the automatic measurement of the auditory brainstem response realized thereby can effectively solve the above problems that exist in the case of artificial judgment.

[0045] Two examples of the adaptive averaging calculation (corresponding to module 1) in step A2 of the present invention are introduced below.

[0046] In one embodiment of a single record with repeating input data types as shown in FIG. 2, the operation of the adaptive averaging method of the present invention is as follows: Step S1: randomly divide the currently collected ABR data into two groups, calculate the average curve of each group, and obtain the ABR time curve corresponding to the average curve of each group of data (e.g., two curves at each acoustic intensity in FIG. 5); a step S2 of calculating a cross-correlation function after each of the two groups has been averaged; Step S3 of obtaining a time lag corresponding to a maximum value of the cross-correlation function; The method includes step S4, in which the absolute value of the time lag is compared with k=1, and if the absolute value of the time lag is smaller than 1, it indicates that a stable time-locked signal has been detected (corresponding to the black dots in Figure 6), and if the absolute value of the time lag is not smaller than 1, it indicates that a stable time-locked signal has not been detected (corresponding to the white dots in Figure 6).

[0047] The present invention can perform parallel judgment multiple times, and by executing the above steps S1 to S4 each time, the problem of "background noise appears a time lag that meets the requirement when it is very accidental" is avoided. In this example, parallel judgment is performed three times, and the three times are randomly grouped and averaged to avgA, avgB, avgA', avgB', avgA" and avgB", and the cross-correlation coefficients of avgA and avgB, avgA' and avgB', and avgA" and avgB" are calculated, and the corresponding time lag is obtained after the judgment result of the absolute value of the time lag being smaller than 1 (data point) is set to "1", and the judgment result of the absolute value of the time lag being 1 or more (data point) is set to "0". In this case, the following flow is further included.

[0048] S5, if all three judgments indicate that a stable time-lock signal is detected at the current acoustic intensity, i.e., the judgment results corresponding to the three judgments A, B, and C are all "1", data collection may be stopped, and a stable time-lock signal can be detected at the current acoustic intensity, indicating that the measurement is passed at the current acoustic intensity; Otherwise, if none of the judgment results corresponding to the three judgments A, B, and C are "1", further execute the following steps.

[0049] S6, determining whether the maximum number of iterations at the acoustic intensity has been reached; That is, M=M max Here, we determine whether or not the result matches M max is a preset maximum number of iterations, M is the current number of iterations, and M is incremented by 1 each time step S1 is executed. <M max If so, continue iterating; If the maximum number of iterations has been reached, data collection may be stopped, indicating that a stable time-locked signal cannot be detected at the current acoustic intensity and the measurement has not been passed at the current acoustic intensity; if the maximum number of iterations has not been reached, N groups of newly collected data are further added to the original data (N=50 in this example), and the process of steps S1 to S6 is repeated at the current acoustic intensity.

[0050] Figure 4 shows the mouse ABR time curves after averaging 350 times at each acoustic intensity. Figure 5 shows the results obtained after adaptive averaging using the same group data of Figure 4. The black / gray lines at each acoustic intensity are the average curves of the two groups to calculate the cross-correlation function. The number of averages used at each acoustic intensity is the product of the current number of iterations M and N shown in Figure 7 (×50 in this example). Figure 6 shows the time lag (absolute value) corresponding to the maximum correlation coefficient obtained in the iterations. The black dots represent the detection of a stable time-locked signal, and the white dots represent the detection of a stable time-locked signal. Based on the acoustic intensity range corresponding to the jump from the black dots to the white dots in Figure 6, it can be seen that the hearing threshold in this example is between 25 and 30 dB. Figure 7 shows the number of iterations required to detect a signal at each acoustic intensity. If there are two consecutive instances of no signal (the maximum number of repetitions has been reached), this indicates that the sound intensity is below the audible threshold and the test is automatically terminated; the solid dots in the figure are the actual measured sound intensities, and the open dots are sound intensities that do not require measurement (because in principle the measurement can be terminated if no signal is detected twice).

[0051] As shown in Fig. 10 (the left and right groups correspond to animal experimental data and clinical experimental data, respectively), the doctor reads the average curve of all data and performs threshold judgment by the conventional method, the doctor reads the average number of times of the adaptive algorithm of the present invention and performs threshold judgment on the data obtained, and the device performs threshold judgment on the data obtained by the average number of times of the adaptive algorithm of the present invention. As shown in Fig. 11 (the left and right groups correspond to animal experimental data and clinical experimental data, respectively), the normalized number of data (i.e., the total number of records) of all data required when the doctor performs threshold judgment by the conventional method is compared with the normalized number of data required when performing threshold judgment by the adaptive algorithm of the present invention.

[0052] As shown in Figures 10 and 11, compared with the average curve of the hearing threshold judged by multiple doctors, the method of the present invention has a basically 100% accuracy within an error of ±5 dB. Also, compared with using the maximum number of averages preset for each sound intensity, the present invention performs measurement of the next sound intensity immediately after each sound intensity detects a signal, thereby reducing the number of records that do not contribute to threshold judgment by up to 69%.

[0053] The present invention can further improve the accuracy of the audibility threshold judgment by function fitting. Figure 8 shows that the normalized number of iterations is fitted with a sigmoid function and interpolated (corresponding to step A6) to obtain the ABR detection result of the sound intensity corresponding to the audibility threshold, (inset) 10 dB above and below the threshold, with an interval of 1 dB, and the threshold obtained in this example is 26 dB. That is, when the measured sound intensity decreases by 5 dB, the conventional method only has an accuracy of 5 dB, but the present invention can reach 1 dB by function fitting.

[0054] For situations where only average data is derived, in an alternative embodiment shown in FIG. 3, the operation of the adaptive averaging method of the present invention (where the input data type is multiple average records) differs from the previous embodiment in the following ways:

[0055] Add the average data of the three groups collected consecutively for each iteration (for example, the average number of each group is 50, and the number of groups corresponds to the number of parallel judgments), combine with the previous data, and recalculate the average curve according to the weights. For example, for the fourth iteration,

number

[0056] As shown in FIG. 9 , the present invention further provides an automatic measuring device for auditory brainstem response, including: An input module for acquiring ABR data collected at each acoustic intensity and inputting it into the device, the type of input data being either a repeated single recording or an average recording of multiple recordings. An input module connected to an auditory brainstem evoked potential acquisition device for acquiring ABR data detected from a subject in real time, or for acquiring experimental data from some storage device for offline processing; A control module, which drives the input module to obtain ABR data with appropriate acoustic intensity stimulation, for example, corresponding to the ABR time curves recorded respectively at the minimum acoustic intensity measured from the starting acoustic intensity (in this example, 90 dB to 0 dB, 5 dB intervals) according to different stages of measurement. When obtaining real-time data, the control module drives the sound stimulus generator to adjust the acoustic intensity step by step to provide a corresponding sound stimulus signal to the test subject; when obtaining offline data, the control module drives the input module to obtain more groups of experimental data according to the measurement process; An adaptive averaging calculation module (corresponding to module 1), which determines whether data collection needs to be continued to increase the number of averages, or whether collection should be interrupted. According to different types of input data, the adaptive averaging calculation module may execute the calculation flow shown in FIG. 2 based on a single repeated record, or execute the calculation flow shown in FIG. 3 based on multiple average records; a storage module for storing the number of iterations used for each sound intensity during the adaptive averaging; A main judgment module judges whether an ABR signal is detected during the adaptive averaging calculation based on the result output by the adaptive averaging calculation module, and if it is judged that a signal is detected, it further judges whether the current acoustic intensity reaches the set minimum value of the measured acoustic intensity, or if it is judged that a signal is not detected, it further judges whether an ABR signal is not detected at P consecutive acoustic intensities. Here, if it is determined that the minimum value of the measured acoustic intensity has not been reached, the main judgment module issues a command to the control module, which then drives the input module to obtain corresponding ABR data at a new acoustic intensity (the new acoustic intensity is obtained after subtracting an interval (5 dB) from the current acoustic intensity), or if it is determined that the minimum value of the measured acoustic intensity has been reached, the main judgment module issues a command to the control module, which then retrieves the number of iterations used in calculating the adaptive averaging method from the memory module and sends it to the function fitting module to obtain a threshold value by function fitting.

[0057] The function fitting module in this example uses the number of iterations at each sound intensity in the adaptive averaging calculation process (Figure 7), normalizes, and then performs fitting using a Sigmoid function (Figure 8), and obtains the sound intensity corresponding to the audible threshold through interpolation.

[0058] The output module can obtain information such as corresponding process parameters, operation results or judgment conclusions from other modules, and output data or graphs (Figures 4 to 8) for each step of the measurement as necessary, which is convenient for engineers to observe or save records. For example, based on the ABR data collected with the stimulation of each acoustic intensity obtained by the input module, it can provide an ABR time curve after averaging the data of each group. Based on the result of the operation of the adaptive averaging method, it can provide the average curve of two groups for calculating the cross-correlation function at each acoustic intensity, the time lag corresponding to the maximum correlation coefficient obtained by iteratively obtaining when a stable time-locked signal is detected or not detected, the interval of acoustic intensity corresponding to the audible threshold, the number of iterations required to detect a signal at each acoustic intensity, the threshold at which the normalized iteration number function is obtained by fitting and interpolation, etc.

[0059] Here, when the adaptive averaging calculation module executes the calculation flow shown in Figure 2 based on repeated single recordings, it issues a command to the control module to collect data (the control module thereby drives the input module to obtain measurement data of appropriate acoustic intensity and appropriate type), and the adaptive averaging calculation module randomly divides the currently collected ABR data into two groups and calculates the average curves respectively, and then calculates the time lag corresponding to the maximum value of the cross-correlation function, and compares the absolute value of the time lag with k=1 (data point). If the absolute value of the time lag is smaller than 1, it indicates that a stable time-locked signal has been detected, and if the absolute value of the time lag is greater than 1, it indicates that a stable time-locked signal has not been detected.

[0060] Preferably, the adaptive averaging calculation module can perform multiple parallel judgments, and if each parallel judgment indicates that a stable time-lock signal is detected at the current acoustic intensity, it commands the control module to stop collecting data (the control module thereby drives the input module to stop acquiring data), and returns a signal indicating that a stable time-lock signal cannot be detected at the current acoustic intensity, indicating that the measurement was not successful at the current acoustic intensity; Otherwise, if the parallel judgments of each time do not necessarily indicate that a stable time-lock signal has been detected at the current acoustic intensity, it further judges whether the maximum number of iterations at the acoustic intensity has been reached, and if the maximum number of iterations has been reached, it issues a command to the control module to stop collecting data, and can return a signal indicating that a stable time-lock signal cannot be detected at the current acoustic intensity and that the measurement was not successful at the current acoustic intensity; if the maximum number of iterations has not been reached, it further issues a command to the control module to add new data (the control module thereby drives the input module to continue acquiring data, and in this example adds 50 groups of newly collected data), and after adding the newly collected data to the original data, the adaptive averaging calculation module repeatedly performs the above calculation process at the current acoustic intensity.

[0061] Alternatively, when the adaptive averaging calculation module executes the calculation flow shown in FIG. 3 based on multiple average records, it issues a command to the control module to collect data (the control module then drives the input module to obtain measurement data of appropriate acoustic intensity and appropriate type), and the adaptive averaging calculation module adds three groups of average data collected consecutively in each iteration (for example, the average number of each group is 50, and the number of groups corresponds to the number of parallel judgments), combines it with the previous data, recalculates the average curve according to the weight, and calculates the cross-correlation function between every two groups for the updated average curve of each current group, obtains the time lag of multiple groups corresponding to the maximum value of the cross-correlation function, and performs a judgment between the absolute value of the time lag and k=1 (data point).

[0062] If the judgments of each group indicate that a stable time-lock signal is detected at the current acoustic intensity, issue a command to the control module to stop collecting data (the control module drives the input module to stop acquiring data), and return a signal indicating that a stable time-lock signal can be detected at the current acoustic intensity and a successful measurement at the current acoustic intensity; otherwise, further determine whether the maximum number of iterations at the acoustic intensity has been reached. If the maximum number of iterations has been reached, issue a command to the control module to stop collecting data, and return a signal indicating that a stable time-lock signal cannot be detected at the current acoustic intensity and a successful measurement at the current acoustic intensity; if the maximum number of iterations has not been reached, issue a command to the control module to add new data, and after adding the new average data to the original average data, the adaptive averaging calculation module repeats the above calculation process at the current acoustic intensity.

[0063] The input module, control module, adaptive averaging calculation module, memory module, main judgment module, function fitting module, output module, etc. of the automatic auditory brainstem response measurement device of the present invention can be understood to be used as follows: In this case, the function of each module can be realized independently through the cooperation of necessary software and hardware, or some powerful processing units can realize the functions of multiple modules, but the present invention is not limited thereto.

[0064] Although the contents of the present invention have been described in detail through the above preferred embodiment forms, it should be appreciated that the above description should not be considered as limiting the present invention. After reading the above contents, those skilled in the art will understand various modifications and substitutions made to the present invention. Therefore, the protection scope of the present invention should be defined by the appended claims.

Claims

1. A1, obtaining a corresponding ABR recording at a starting acoustic intensity; Performing an adaptive averaging operation, incrementally increasing the number of records averaged by iteration until the iterations are completed; For each iteration, randomly divide the ABR recordings corresponding to the current acoustic intensity at the current iteration into two groups, then calculate the average curves of each group according to the current averaging number, respectively, obtain the time lag corresponding to the maximum value of the cross-correlation function between the groups, and determine whether an ABR signal with time-locking characteristics is detected based on whether the absolute value of the time lag is smaller than k data points; If an ABR signal is not detected and the number of iterations has not reached the maximum number of iterations at the current acoustic intensity, perform a next iteration and continue obtaining an ABR recording at the current acoustic intensity; If an ABR signal is detected or if the number of iterations reaches a maximum number of iterations at the current acoustic intensity but an ABR signal is not detected, then terminate the iterations and do not continue to obtain an ABR recording at the current acoustic intensity, step A2; A step A3 of determining whether an ABR signal having a time lock characteristic is detected in step A2, and if an ABR signal is detected, executing step A4, and if an ABR signal is not detected, executing step A5; step A4 of determining whether the current acoustic intensity reaches the set minimum test acoustic intensity, and if not, obtaining the corresponding ABR record at a new acoustic intensity and combining it with the original data, and re-executing step A2, the new acoustic intensity being obtained after subtracting the set interval from the current acoustic intensity, and if the minimum test acoustic intensity is reached, executing step A6; Step A5: determining whether or not an ABR signal is not detected at P consecutive acoustic intensities, and if an ABR signal is not detected at P consecutive acoustic intensities, executing step A6; otherwise, proceeding to execution of step A4; and step A6 of performing function fitting and interpolation for the number of iterations used for each acoustic intensity in the adaptive averaging calculation to obtain a minimum acoustic intensity required to detect an ABR signal, and outputting the minimum acoustic intensity as an audible threshold. The adaptive averaging calculation involves gradually increasing the number of recordings, randomly grouping and averaging the current ABR data into two arbitrary groups, and calculating the average curves respectively, to obtain the time lag corresponding to the maximum value of the cross-correlation function of the average curves of the two groups, and judging whether the absolute value of the time lag is smaller than k data points, until the maximum number of iterations (i.e., the iterations are terminated when a signal is detected or no signal is detected even if the maximum number of iterations is reached) to judge whether a stable time-locked signal exists; k is a fixed value or a value obtained after calculating all data points with a preset ratio; 13. A method for automated testing of auditory brainstem response, comprising:

2. The ABR record is a time curve whose input type is a single record that is repeated, and in step A2, the adaptive averaging calculation stepwise increases the number of records by iterations until the iterations are completed, and each iteration therein performs the calculation process of steps S1' to S5' on the received ABR record; A step S1' of randomly dividing the ABR recordings into two groups and calculating the mean curves respectively; a step S2' of calculating the cross-correlation function after each of the two groups has been averaged; A step S3' of obtaining a time lag corresponding to the maximum value of the cross-correlation function; Comparing the absolute value of the time lag with the k value to determine whether the time lag deviation is within k data points; If the absolute value of the time lag is smaller than the k value, it means that the time lag deviation is within k data points, indicating that a stable time-locked signal is detected. In this case, the iteration is terminated, and data acquisition and collection are stopped, corresponding to the case where an ABR signal is detected; If the absolute value of the time lag is not smaller than the k value, it means that the time lag deviation is not within k data points, indicating that a stable time-locked signal is not detected, in this case, step S4' further executes step S5'; determining whether the number of iterations has reached the maximum number of iterations for the current acoustic intensity; and S5', if the maximum number of iterations has been reached, terminate the iterations, stop acquiring and collecting data, and respond to the case where no ABR signal is detected; if the maximum number of iterations has not been reached, perform the next iteration, acquire new collected data and add it to the received ABR record, and repeatedly perform the calculation process S1' to S5' with the current acoustic intensity until the iterations are completed.

2. A method for automatic testing of auditory brainstem responses according to claim 1.

3. The ABR record is a time curve in which an input type is a single record that is repeated, and in step A2, the adaptive averaging calculation stepwise increases the number of records averaged by iterations until an iteration is completed, where each iteration is performed in parallel Q times on the received ABR record; For each parallel judgment, first, the following steps S1 to S4 are executed, In step S1, the ABR recordings are randomly divided into two groups and the average curves are calculated respectively. In step S2, a cross-correlation function is calculated after each of the two groups is averaged; In step S3, a time lag corresponding to the maximum value of the cross-correlation function is obtained; In step S4, the absolute value of the time lag is compared with the k value to determine whether the time lag deviation is within k data points; If the absolute value of the time lag is smaller than the k value, it means that the time lag deviation is within k data points, indicating that a stable time-locked signal is detected; if the absolute value of the time lag is not smaller than the k value, it means that the time lag deviation is not within k data points, indicating that a stable time-locked signal is not detected; After completing step S4 for each parallel judgment, the following step S5 is further executed: In step S5, it is determined whether a stable time lock signal is detected in each of the Q parallel determinations; If a stable time-lock signal is detected in each of the Q iterations, terminate the iterations and stop acquiring and collecting data, corresponding to the case where an ABR signal is detected; If a stable time-lock signal has not been detected in any of the Q times, step S6 is further executed; In step S6, it is determined whether the number of iterations has reached the maximum number of iterations at the current sound intensity; If the maximum number of iterations has been reached, terminate the iterations, stop acquiring and collecting data, and respond when no ABR signal is detected; If the maximum number of iterations has not been reached, perform the next iteration, acquiring new collected data and adding it to the received ABR recording, and repeating step S1 for each of the Q parallel decisions at the current acoustic intensity.

2. A method for automatic testing of auditory brainstem responses according to claim 1.

4. The ABR record is a time curve whose input type is a multiple average record, and in step A2, the adaptive averaging calculation stepwise increases the number of records by iterations until the iterations are completed, and each iteration of the adaptive averaging calculation step S1"-S6" is performed on the received ABR record; For the received ABR records, the current average curve avgA is calculated according to the weights. 新 Calculate [0010] Here, avgA 新 , avgA 旧 , avgA 加 are the average data of the ABR test of group Q, M is the number of the current iteration, avgAadd represents the average curve newly added in the current iteration, and avgAadd represents the average curve newly added in the current iteration. 旧 represents the average curve of the previous iteration, step S1"; Current average curve of Q group avgA 新 a step S2" of calculating a cross-correlation function between each two groups for Step S3" of acquiring the time lag of the Q group corresponding to the maximum value of the cross-correlation function; Perform Q parallel decisions, Step S4": comparing the absolute value of the time lag of each group with a k value to determine whether the time lag deviation is within k data points; if the absolute value of the time lag is smaller than the k value, it means that the time lag deviation is within k data points, indicating that a stable time-locked signal is detected; if the absolute value of the time lag is not smaller than the k value, it means that the time lag deviation is not within k data points, indicating that a stable time-locked signal is not detected; determining whether a stable time-lock signal is detected in each of the Q parallel determinations; If a stable time-lock signal is detected in each of the Q iterations, terminate the iterations and stop acquiring and collecting data, corresponding to the case where an ABR signal is detected; If a stable time-lock signal has not been detected in any of the Q times, step S5'' further includes step S6''; determining whether the current number of iterations has reached the maximum number of iterations for the current acoustic intensity; and step S6'', if the maximum number of iterations has been reached, then terminate the iterations and stop acquiring and collecting data, and respond to the case where no ABR signal is detected; if the maximum number of iterations has not been reached, then perform the next iteration, acquire new collected data and add it to the received ABR record, and repeatedly perform the calculation process S1''-S6'' with the current acoustic intensity until the iterations are completed.

2. A method for automatic testing of auditory brainstem responses according to claim 1.

5. 5. The method for automatic testing of auditory brainstem response according to claim 1, wherein the averaging number used in calculating the average curve at each acoustic intensity is the multiplication value of the number of iterations and the value N, the value N corresponding to the number of groups of ABR recordings newly added at each iteration.

6. The method for automatic testing of auditory brainstem response according to any one of claims 1 to 4, characterized in that in step A6, a normalization process is performed on the number of iterations used for each acoustic intensity, and then a Sigmoid function fitting and an interpolation process are performed on the number of iterations after the normalization process to obtain the audibility threshold.

7. The ABR recordings at each sound intensity are animal experimental data or clinical data, and are obtained by real-time testing or stored offline, and the data includes characteristic waveforms generated when sound stimuli induce electroencephalographic changes; The ABR recording is Signal amplification and Bandpass filtering and Adjusting the time interval for collecting the characteristic waveform by the electrodes, and selecting the time curve of the corresponding time interval after the onset of the sound stimulus as the analysis target; Elimination of the time curves corresponding to background noise, and The method for automatic testing of auditory brainstem response according to any one of claims 1 to 4, further comprising one or more pre-processing steps: removing low-frequency background noise by a smoothing spline fitting function.

8. An automatic auditory brainstem response test device used in the automatic auditory brainstem response test method according to any one of claims 1 to 7, The automated auditory brainstem response test device comprises: an input module that obtains the ABR recordings collected at each acoustic intensity stimulus; a control module for driving the input module to acquire ABR recordings collected in batches for each sound intensity stimulus from a starting sound intensity through a lowest test sound intensity; an adaptive averaging calculation module, which performs adaptive averaging calculation, gradually increasing the number of averaging records by iteration until the iteration is completed; for each iteration, grouping the ABR records corresponding to the current acoustic intensity at the current iteration, and then calculating the average curve of each group according to the current averaging number, respectively, obtaining a time lag corresponding to the maximum value of the cross-correlation function between the groups; judging whether an ABR signal with time-locked characteristics is detected according to whether the time lag deviation is within a predetermined data point range; if an ABR signal is not detected and the iteration number has not reached the maximum iteration number at the current acoustic intensity, performing the next iteration, and continuing to obtain ABR records at the current acoustic intensity; if an ABR signal is detected or if an ABR signal is not detected and the iteration number has reached the maximum iteration number at the current acoustic intensity, ending the iteration, and not continuing to obtain ABR records at the current acoustic intensity; a storage module for storing the number of iterations used for each sound intensity during the adaptive averaging calculation; a main judgment module for judging whether an ABR signal having a time-lock characteristic is detected during the calculation of the adaptive averaging based on the result output by the adaptive averaging calculation module, and if it is judged that the ABR signal is detected, further judging whether the current acoustic intensity reaches a set minimum test acoustic intensity; If it is determined that the minimum test acoustic intensity has not been reached, the main decision module issues an instruction to the control module, which in turn drives the input module to obtain the corresponding ABR recording at a new acoustic intensity, the new acoustic intensity being obtained after subtracting the interval from the current acoustic intensity; if it is determined that the minimum test acoustic intensity has been reached, the main decision module issues an instruction to the control module, which in turn drives the control module to retrieve the iteration number used in the calculation of the adaptive averaging from the storage module; If it is determined that the ABR is not detected, further determine whether the ABR signal is not detected at P consecutive sound intensities; if not, the main judgment module instructs the control module to drive the input module to further obtain corresponding ABR data at a new sound intensity (the new sound intensity is obtained after subtracting the interval from the current sound intensity); if the ABR signal is not detected at P consecutive sound intensities, the main judgment module instructs the control module to retrieve the iteration number used in the calculation of the adaptive averaging method from the storage module; The automatic testing device for auditory brainstem response further includes a function fitting module, which performs function fitting and interpolation for the number of iterations used in calculating the adaptive averaging method to obtain a minimum acoustic intensity required to detect an ABR signal, and outputs the minimum acoustic intensity as an audible threshold.

9. The adaptive averaging calculation module issues a command to the control module to collect data. When the data type input by the input module is a repeated single record, the adaptive averaging calculation module performs one judgment or multiple parallel judgments corresponding to each repetition of each acoustic intensity. During each judgment of the single judgment or multiple parallel judgments, the currently collected ABR records are randomly divided into two groups to calculate average curves respectively, and then calculate a time lag corresponding to the maximum value of the cross-correlation function. The absolute value of the time lag is compared with a k value to determine whether the time lag deviation is within k data points. If the absolute value of the time lag is smaller than the k value, it indicates that a stable time-locked signal is detected. If the absolute value of the time lag is not smaller than the k value, it indicates that a stable time-locked signal is not detected. If the single judgment or each judgment among the multiple parallel judgments indicates that a stable time-lock signal is detected at the current acoustic intensity, the iteration is terminated, and the adaptive averaging calculation module issues a command to the control module to stop collecting data, and outputs a return information indicating that an ABR signal is detected at the current acoustic intensity, and stores the current iteration number by the storage module; If a determination indicates that a stable time-lock signal is not detected at the current acoustic intensity, or none of the parallel determinations indicates that a stable time-lock signal is detected at the current acoustic intensity, the adaptive averaging calculation module further determines whether the number of iterations has reached a maximum number of iterations at the current acoustic intensity; If the maximum number of iterations has been reached, the iterations are terminated, and the adaptive averaging calculation module issues a command to the control module to stop collecting data, and sends back information indicating that the ABR signal cannot be detected at the current acoustic intensity, and stores the current number of iterations in the storage module; If the maximum number of iterations has not been reached, perform the next iteration, and the adaptive averaging calculation module further issues a command to the control module to add new data, and repeats the adaptive averaging calculation process with the current acoustic intensity after adding the newly collected data to the original data; Here, k is a preset fixed value or a value obtained after calculating all data points at a preset ratio, according to claim 8 (excluding claim 8 which refers to claim 4).

10. The adaptive average calculation module issues a command to the control module to collect data. When the data type input by the input module is a multiple average record, for each iteration, the adaptive average calculation module calculates as follows: [0025] Here, avgA 新 , avgA 旧 , avgA 加 are the average data of the ABR test of group Q, M is the number of the current iteration, and avgA is the average curve newly added in the current iteration. 加 avgA 旧 Combined with The adaptive averaging calculation module calculates the current average curve avgA 新 Calculate the cross-correlation function between each two groups, and obtain the time lag of the Q group corresponding to the maximum value of the cross-correlation function; The adaptive averaging calculation module performs Q parallel judgments as follows: compare the absolute value of the time lag of each group with a k value respectively to judge whether the time lag deviation is within k data points; if the absolute value of the time lag is smaller than k, it means that the time lag deviation is within k data points, which indicates that a stable time-locked signal is detected; if the absolute value of the time lag is not smaller than the k value, it means that the time lag deviation is not within k data points, which indicates that a stable time-locked signal is not detected; The adaptive averaging calculation module judges whether a stable time-lock signal is detected in each of the Q parallel judgments, and if a stable time-lock signal is detected in each of the Q parallel judgments, ends the iteration; the adaptive averaging calculation module issues a command to the control module to stop collecting data, and outputs a return information indicating that an ABR signal is detected at the current acoustic intensity, and stores the current iteration number by the storage module; If a stable time-lock signal is not detected in each of the Q iterations, further determine whether the iteration number has reached the maximum iteration number at the current acoustic intensity, and if so, end the iteration, and the adaptive averaging calculation module issues a command to the control module to stop collecting data, sends a return information indicating that an ABR signal is detected at the current acoustic intensity, and stores the current iteration number in the storage module; If the maximum number of iterations has not been reached, the next iteration is performed, and the adaptive averaging calculation module issues a command to the control module to add new data, and after adding the newly collected data to the original data, the adaptive averaging calculation process is repeatedly performed with the current sound intensity, where k is a preset fixed value or a value obtained after calculating all data points with a preset ratio as described in claim 8 (excluding claim 8 that relies on claims 2 and 3).

Citation Information

Patent Citations

  • Method and apparatus for objective and automated analysis of auditory brainstem response to determine hearing capacity

    US5697379A

  • Process for automatic determination of hearing acuity, particularly of newborns and infants

    US6071246A