Instrument for detecting auditory evoked neural responses
Patent Information
- Application Number
- JP2024136968
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-08-18
- Filing Date
- 2024-08-16
- Publication Date
- 2025-03-03
Smart Images
Figure 2025028778000001_ABST
Abstract
Description
[Technical field]
[0001] The present application relates to an apparatus for detecting evoked responses.
[0002] The present application further relates to a method for detecting an evoked response.
[0003] The present application further relates to a detection unit.
[0004] The present application further relates to a data processing system comprising a processor and program code means for causing said processor to execute at least some of the steps of the method.
[0005] The present application further relates to a computer program comprising instructions which, when executed by an apparatus, cause said apparatus to carry out the steps of said method. [Background technology]
[0006] device Objective detection of auditory evoked neural activity plays an important role in the diagnosis and early intervention of hearing loss in neonates and infants, as well as in the neural assessment of hearing function during surgery (such as cochlear implantation or acoustic neuroma removal).
[0007] For newborns and infants, hearing screening is performed by auditory evoked neural responses (i.e., electrophysiological responses) recorded by electroencephalography (EEG) using electrodes placed on the infant's scalp. These responses include auditory brainstem responses (ABRs), cortical auditory potentials (CAPs), auditory steady-state responses (ASSRs), and otoacoustic emissions (OAEs). Evoked responses are obtained by presenting sounds (stimuli) such as clicks, narrowband chirps, or speech signals, and averaging many presentations of such stimuli, so that background noise is reduced and the evoked responses are observed.
[0008] For decades, detection of evoked neural responses has been performed by visual inspection by expert audiologists and clinicians, or by statistical methods (i.e., objective detection), which rely on statistical parameters and their associated probabilities that do not represent the actual evoked response but indicate the likelihood that the measured parameters are driven by background noise (null hypothesis).
[0009] These objective methods include parametric (F-test, Hotelling-T2), non-parametric (Q-sample), and machine learning-based approaches. Each of these methods has its own advantages and disadvantages. For example, the F-test measures the ratio of the energy of a noisy signal to the energy of the noise, making it very efficient, intuitive, and easy to understand for anyone. However, the disadvantage of the F-test is that it may not take into account individual types of background noise in the EEG recording, resulting in less accurate and slower detection.
[0010] Q-samples are non-parametric tests that rely on either bootstrap methods or pre-computed statistical tables (usually derived from simulations using artificial noise signals). Bootstrap methods can lead to very good estimates of the underlying background noise, but they can be computationally expensive and conservative; that is, they tend to detect fewer true positive responses if the number of replicates used to bootstrap the data is limited. As a result, Q-samples with bootstraps cannot be used in real-time measurements using standard clinical computers. This can be partially solved by comparing Q-sample statistics with pre-established tables obtained by simulations with different types of noise. However, this approach may not account for the individual characteristics of the background noise in real recordings.
[0011] Furthermore, the statistics derived by the Q-sample and Hotelling-T2 methods are abstract and difficult for clinicians to understand.
[0012] Finally, machine learning methods depend on the training dataset and tend to replicate biases inherited from the training dataset.
[0013] For many years, there has been an increasing interest in developing detectors (or detection methods) that are easy to interpret, reliable in terms of statistical detection, and applicable in real-time situations. The impact of such developments is crucial to perform optimal measurements (fast, reliable, with controlled false positive rates), which in turn directly impacts the accuracy, efficiency and costs of clinical centers.
[0014] Currently, no such detectors or methods for detecting neural responses have been reported, and objective detection of brain activity is achieved by the above methods despite their shortcomings.
[0015] Thus, there is a need for improved objective detection of auditory evoked neural responses, or other types of responses that rely on averaging to reduce the underlying background noise in order to detect a signal. Summary of the Invention
[0016] In one aspect of the present application, an apparatus (device / instrument, or detector of an apparatus) for detecting an evoked response is provided.
[0017] For example, the device (apparatus / instrument, detector of the device) may be suitable for detecting an auditory evoked neural response.
[0018] The device may further be configured to estimate the hearing ability of a test subject (patient).
[0019] The device may include a stimulus generator.
[0020] The stimulus generator may be configured to generate at least one stimulus. The at least one stimulus may be generated according to a test protocol.
[0021] The stimulus generator may be configured to generate multiple successive stimuli according to a test protocol.
[0022] The device may include at least one output unit.
[0023] The at least one output unit may include a transducer.
[0024] The output unit may be configured to receive the at least one stimulus or the plurality of successive stimuli from a stimulus generator.
[0025] An output unit may be configured to provide / present said at least one stimulus or a plurality of successive stimuli to an ear of a test subject via said transducer.
[0026] The output unit may be configured to provide / present said at least one stimulus or a plurality of successive stimuli as one or more tactile or visual stimuli / stimuli to the test subject.
[0027] The output unit may be configured to present an acoustic stimulus.
[0028] The output unit may be configured to present an electrical stimulus (stimulus).
[0029] The output unit may be configured to present a tactile stimulus.
[0030] The output unit may be configured to present a visual stimulus.
[0031] For example, an auditory nerve response may be evoked by a cochlear implant (electrical) or a bone conductor device.
[0032] For example, the transducer may include a receiver (speaker) for presenting the stimulus to the subject as an acoustic signal.
[0033] The output unit may include an ear probe, a (separate) speaker, or a headset, which may refer to an in-ear, supra-ear or over-the-ear headset, earphones, etc., configured to introduce sound into the user's ear.
[0034] The device may include at least one recording unit that includes one or more sensors that are placed on the scalp of the test subject.
[0035] For example, the sensor may include an electrode or optode, a magnetometer, or a microphone.
[0036] Thus, the device may include at least one recording unit comprising one or more electrodes or optodes, a magnetometer or a microphone, which are placed on the scalp of the test subject.
[0037] Thus, the one or more electrodes or optodes, magnetometer or microphone may be configured to measure one or more evoked responses of the test subject.
[0038] The one or more evoked responses can be measured in response to at least one stimulus or to multiple successive stimuli provided.
[0039] The evoked response may be an electrophysiological response (ie, an auditory evoked neural response), a visual evoked potential, an acoustic evoked response (such as an OAE), or the like.
[0040] The one or more electrodes can be configured to measure one or more electrophysiological responses of the test subject.
[0041] The one or more electrodes may be configured to measure one or more electrophysiological responses of the test subject in response to (or as a result of) the at least one stimulus or multiple successive stimuli provided.
[0042] In other words, the one or more sensors may be configured (or prompted, for example, by the control unit) to measure said response after a stimulus is provided / presented to the ear of the test subject.
[0043] For example, one or more sensors can refer to anywhere from one to hundreds of sensors.
[0044] For example, one or more optodes can be used for functional near-infrared spectroscopy (fNIRS).
[0045] For example, one or more magnetometers may be used for Magnetoencephalography (MEG).
[0046] At least one recording unit may be configured to receive (and record) one or more evoked responses measured by one or more electrodes.
[0047] For example, the response may be measured / recorded by electroencephalography (EEG) using the one or more electrodes placed on the test subject's scalp.
[0048] For example, a device containing one or more electrodes can be used with any EEG system / device and can be extended to any other region where a response is obtained by averaging measured responses (i.e., input data) to repeated presentation of a stimulus (vestibular evoked response, evoked otoacoustic emissions, magnetoencephalography, etc.).
[0049] The instrument may include an analysis unit.
[0050] The analysis unit may be configured to receive said measured one or more evoked responses.
[0051] An analysis unit may be configured to receive said one or more measured evoked responses from said recording unit.
[0052] The analysis unit may be configured to analyze said measured one or more evoked responses.
[0053] The analysis unit may be configured to determine, during operation of said device (ie online), the probability p that each of said responses is driven by underlying background noise.
[0054] The analysis unit may be configured to determine the probability p based on an F-value of the measured one or more evoked responses determined as the ratio between the variance of the averaged one or more evoked responses and the variance of the residual background noise, each having degrees of statistical freedom (DOF) v and r.
[0055] In other words, the analysis unit can be configured to estimate the probability p (i.e. the likelihood) that said measured response is the result of underlying background noise in the measured response, rather than being evoked by the stimulus presented to the test subject's ear.
[0056] In other words, the analysis unit can be configured to determine said probability p from an F-value of one or more measured evoked responses determined as the ratio between the variance of the average evoked response and the variance of the residual background noise, each with degrees of statistical freedom (DOF) v and r.
[0057] More specifically, the ratio may be the ratio of the variance of the averaged one or more evoked responses (degrees of freedom (DOF) v) to the variance of the residual background noise (statistical DOF r).
[0058] This allows accounting for background noise (obtained from the test subject) during the measurement and provides improved objective detection of the evoked response, reducing the required detection / measurement time.
[0059] The solution provided can be implemented in real time and provides a better estimate of the underlying background noise signal than existing techniques.
[0060] The provided solution can be applied to any time-domain based evoked response (brainstem and cortical responses) and can be used for diagnosis with higher accuracy and improved detection time compared to existing solutions.
[0061] Furthermore, the provided solution can be applied to any frequency domain based evoked responses (brainstem cortical responses, e.g., Auditory Steady-State Responses (ASSRs), Frequency Following Responses (FFR), Otoacoustics Emissions (OAEs), and other responses) and can be used for diagnosis with higher accuracy and improved detection time compared to existing solutions.
[0062] In other words, the above-mentioned determination of probability p can be performed while the device is in operation.
[0063] One or more sensors (electrodes or optodes or magnetometers or microphones) may be configured to measure one or more evoked responses, including the ratio of the variance of the mean response evoked by the stimulus to the variance of the residual background noise.
[0064] In other words, the evoked response measured following presentation of a stimulus may reflect the underlying background noise (signal) combined with the response evoked by the stimulus (if the stimulus is perceived by the test subject). Thus, the measured evoked response may include partly the stimulus-evoked response and partly the underlying background noise, the term "partly" covering percentages of 0 or more.
[0065] For example, evoked responses (e.g., EEG data) can be measured / recorded in response to a presented sound. In each time segment, a sound is presented and a trial of duration T can be recorded using one or several electrodes.
[0066] A trial can refer to the presentation of one stimulus to the test subject's ear followed by measurement of the evoked response.
[0067] Thus, the one or more evoked responses measured can have a duration T.
[0068] For example, a stimulus can be presented 1000 times with 40 ms between each (e.g., 0 ms, 40 ms, 80 ms, ...). The duration of the evoked response can be set to a length of 20 ms after the onset of each stimulus. In this case, there are 1000 individual trials, each lasting 20 ms.
[0069] The individual trials from a continuous data stream are denoted as s1(t), s2(t), ..., s N The resulting stimulus-evoked response, denoted as (t), is obtained for analysis, where N is the total number of trials recorded. We are interested in obtaining the stimulus-evoked response (i.e., the evoked neural response), X(t), which, for simplicity, can be assumed to be the same for each stimulus presented. By doing so, the problem can be expressed as:
number
[0070] To reduce the underlying background noise, the mean evoked response of all individual trials is estimated as follows:
number
[0071] time domain For standard averaging, in the time domain w is given by
number
number
[0072] The variance of the background noise (σ 2 BNK ) can be estimated as either 1) the average variance of the individual variances of fixation time points tracked across trials, or 2) the variance of points within each trial.
[0073] In the first approach, σ 2 BNK is estimated from a subset of trials by
number
number
number
[0074] σ 2 BNK As a result, the weight (w sOr w w ) is obtained, the residual background noise in the average evoked response can be estimated by
number
number
number
[0075] The importance of SNR is in the numerator, σ 2 s (v) statistical degrees of freedom (DOF) and the denominator σ 2 RN It depends entirely on the F-value, which is determined by the DOF of (r).
[0076] Therefore, the present solution provides a novel estimation of these parameters. The solution presented estimates these DOFs (v and r) for each individual measurement online while the data is being recorded (i.e., during the operation of the device).
[0077] The analysis unit may be configured to determine said probability based on at least a statistical degree of freedom (DOF) of said variance. The probability may be determined during operation of said device.
[0078] In other words, the present application provides that the statistical DOF v is estimated online.
[0079] The analysis unit may be configured to determine the probability based on a statistical DOF of the average variance in at least one or more evoked responses and a statistical DOF of the variance of residual background noise in the measured one or more evoked responses.
[0080] The analysis unit may be configured to determine an F-value of the measured one or more evoked responses as a ratio of the variance of the averaged one or more evoked responses, each having statistical DOFs v and r, to the variance of the residual background noise.
[0081] For example, without going into formal mathematical proofs, the following DOFs are derived for the numerator of the F-ratio:
number
[0082] G ^ xx (ω) is the power spectral density, given by Equation 10. It corresponds to the weighted average of the magnitude of the squared Fourier transform of each trial.
number
[0083] Weight w i is given by Equation 3 or Equation 4.
[0084] The DOF in the denominator is given by:
number
[0085] G ^ xx Note that (ω) can be estimated using a longer time window, T′, and then scaled down for a smaller analysis window of length T.
number
[0086] Once the numerator and denominator DOFs are obtained, the probability that the calculated F-value is driven by the underlying background noise can be derived by:
number
[0087] The analysis unit may therefore be configured to determine the probability p according to the following formula:
number
[0088] The provided solution can measure the F-ratio (F-value, i.e. the energy of the average measurement relative to the energy of the estimated residual background noise). The ongoing activity can be weighted based on the change in the underlying background activity. These weights are obtained by consistent tracking of multiple points between and / or within trials in the measurement and can be used to estimate the energy of the residual background noise in the measurement (denominator of the F-ratio). The energy of the numerator can be estimated from the variance of the ongoing average evoked response. The provided solution uses a novel method to obtain the correct statistical degrees of freedom (DOF), which determines the sensitivity to find a statistically significant evoked response. Moreover, the provided solution can be used in real time, and its sensitivity is automatically adjusted for each individual case, providing an optimal individualized accuracy. The F-value is a parameter quantity, and since the parameters are the DOF in the numerator of the ratio (here called v) and the DOF in the denominator (called r), its performance depends entirely on the accuracy of the estimation of these two parameters.
[0089] Frequency Domain For standard averaging, in the frequency domain w is given by
number
number
[0090] An important property used in the following equation is that the variance of a signal in the time domain can be estimated in the frequency domain as the sum of the power spectral density of all positive frequencies by Parseval's theorem. This is shown below.
number
[0091] This property can be used to estimate the frequency-specific underlying background noise (σ 2 BNN ) can be estimated in the frequency domain as the sum of the power spectral density components at frequencies of interest (e.g., angular frequencies).
number
[0092] If we use all frequencies to estimate the variance of Equation 18, we get the same variance as in the time domain. However, if we only use a subset of the frequency components, we get the variance due to only those frequencies, i.e., the frequency-specific variance.
[0093] σ 2 BNN As a result, the weight (w s Or w w ) is obtained, the residual background noise in the average response can be estimated by
number
number
number
number
[0094] The importance of SNR is in the numerator, σ 2 s (v) Statistical DOF and denominator σ 2 RN It depends entirely on the F-value, which is determined by the DOF of (r).
[0095] The novelty of the presented solution lies in the estimation of these parameters: the analysis unit is able to estimate these DOFs (v and r) from the specific frequencies of interest, for each individual recording / measurement, online while the data are being recorded.
[0096] The following DOF for the numerator of the F value (i.e., the F ratio) can be derived:
number
[0097] G ^ xx (ω) is the power spectral density, given by Equation 23. It corresponds to the weighted average of the magnitude of the Fourier transform of each trial.
number
[0098] Weight w i is given by Equation 15 or Equation 16.
[0099] The DOF in the denominator is given by:
number
[0100] Once the numerator and denominator DOFs are obtained, the probability that the calculated F-value is driven by the underlying background noise can be derived by:
number
[0101] The analysis unit may therefore be configured to determine the probability p according to the following formula:
number
[0102] There are many applications where one or more frequencies need to be evaluated to assess whether their power is significantly greater than the power of the residual background noise, such as ASSR, steady-state visual evoked potentials, etc.
[0103] To address this issue, the above formula can be used. However, when the number of frequencies to be tested is small (i.e., the frequencies of interest are few and sparse), the estimate of the residual background noise may be inaccurate due to the limited number of data points (frequency samples) that contribute to the background noise estimate. To address this issue, the following adaptation can be used:
[0104] For example, consider L sparsely distributed frequencies of interest (ω1, ω2, ..., ω L )
[0105] We then consider assessing whether the pooled power of these frequencies is significantly greater than the underlying background noise for the same frequencies.
[0106] Due to the small number of frequencies to be analyzed over the full spectrum of data, the estimate of the variance of the background noise (Equation 18) may be inaccurate. To avoid this problem, we can make the assumption that the power of the frequencies of interest is similar to the power of the neighboring frequencies. Under this assumption, we can obtain a good approximation of the power of the underlying background noise at each frequency of interest by averaging the power of the frequencies in the neighborhood of each frequency of interest. That is,
number
number
[0107] Note that in Equation 27, the number of samples, and therefore the number of DOFs, will be higher the more samples that contribute to this estimate.
[0108] The number of DOFs given by Equation 25 needs to include all frequencies used to estimate the variance of the underlying background noise in Equation 27. That is, r can be determined by:
number
[0109] In other words, if the instrument is used in ASSR measurements, use Equation 27 to find σ 2 BNN For example, if the response frequency is 40 Hz, σ 2 BNN can be estimated as the average energy within the frequency interval 40 Hz ± 5 Hz.
[0110] The provided solution can furthermore measure the F-ratio (energy of the averaged measurement to the energy of the estimated residual background noise) in a frequency-specific manner. The ongoing activity can be weighted based on the change in the underlying background activity. These weights can be derived from the power in the frequency domain and can be used to estimate the frequency-specific energy of the residual background noise during the measurement / recording (the denominator of the F-ratio). The energy of the numerator can be estimated from the power in the frequency domain of the ongoing averaged response. The provided solution can estimate the correct statistical degrees of freedom (DOF) based on the frequency components of interest, which in turn determines the sensitivity to find statistically significant evoked responses. Furthermore, the provided solution can be used in real time, and its sensitivity is automatically adjusted for each individual case, providing optimal individualized accuracy. Since the F-ratio is a parametric quantity, the parameters being the DOF in the numerator of the ratio (here called v) and the DOF in the denominator (called r), its performance depends entirely on the accuracy of the estimation of these two parameters.
[0111] Thus, the analysis unit extracts the frequency-specific underlying background noise (σ 2 BNN ) as a sum of power spectral density components at frequencies of interest (e.g., angular frequencies).
[0112] The analysis unit extracts the frequency-specific underlying background noise (σ 2 BNN ) and the variance of the estimated weights (w s Or w w ) vector. In other words, the residual background noise variance (σ 2 RN ) can be estimated by Equation 7 or Equation 19.
[0113] The analysis unit may be configured to estimate the power of the residual background noise at each frequency of interest by averaging the power of the residual background noise of neighbouring (adjacent) frequencies of the frequency of interest.
[0114] The analysis unit is the angular frequency of interest, ω i σ for 2 s The numerator of the F-value can be determined based on the DOF of the object (see Equation 9 (time domain) and Equation 22 (frequency domain)).
[0115] The analysis unit is the angular frequency of interest, ω i σ for 2 s The denominator of the F-number can be determined based on the DOF of the object (see Equation 12 (time domain) and Equations 25 and 28 (frequency domain)).
[0116] The analysis unit is σ 2 s and σ 2 RN It can be constructed to estimate the SNR, and hence the F-measure, based on the ratio of
[0117] For example, σ 2 s can be calculated from equation 21(2).
[0118] For example, σ 2 RN can be calculated from Equation 19.
[0119] For example, the frequencies of interest may be between 2 Hz and 30 Hz for CAP measurements.
[0120] For example, frequencies of interest may be 40 Hz, 80 Hz, and further harmonics for ASSR measurements.
[0121] The analysis unit may be configured to adaptively determine the probabilities during operation of the device.
[0122] The probability can be determined in either the time domain and / or the frequency domain.
[0123] This provides a user of the device with rapid, continuous determination of probabilities during operation of the device.
[0124] At least one of the stimuli or the plurality of successive stimuli may comprise an acoustic stimulus.
[0125] At least one of the stimuli or the plurality of successive stimuli may comprise a click.
[0126] At least one of the stimuli or the plurality of successive stimuli may comprise a narrowband chirp.
[0127] At least one of the stimuli or the plurality of successive stimuli may comprise an audio signal.
[0128] The at least one stimulus or the plurality of successive stimuli may comprise an electrical stimulus or any repetitive auditory stimulus.
[0129] For example, at least one of the stimuli or the plurality of successive stimuli may be based on an auditory change complex (ACC) stimulus.
[0130] The one or more evoked responses measured can include auditory brainstem responses (ABRs).
[0131] The one or more evoked responses measured can include cortical auditory potentials (CAPs).
[0132] The one or more evoked responses measured can include auditory steady-state responses (ASSRs).
[0133] For example, the one or more electrophysiological responses measured can include any response that relies on averaging to reduce background noise to detect a target signal, such as otoacoustic emissions, visual evoked responses, vestibular responses, among others.
[0134] The analysis unit may be configured to determine the hearing ability of the test subject in response to the probability being below a predefined threshold.
[0135] The predetermined threshold may be, for example, the minimum stimulus amplitude at which a sound can be heard, the minimum difference that distinguishes acoustic signatures between sounds, the minimum difference that distinguishes speech in the presence of noise, the minimum difference that distinguishes the spatial location of a sound, or the minimum difference that detects odd sounds from repeated sounds.
[0136] The stimulus generator may be configured to modify the at least one stimulus or plurality of successive stimuli in response to the determined probability.
[0137] The stimulus generator may be configured to modify the test protocol in response to the determined probability.
[0138] The analysis unit may be configured to determine said stimulus or a number of successive stimuli based on an analysis of previously measured evoked responses.
[0139] The analysis unit may include a neural network. The analysis unit may be configured to determine the probability p by using said neural network.
[0140] Detection Unit In one aspect, the present application further provides a detection unit.
[0141] The detection unit may be configured to determine the probability p whether the measured evoked response is driven by an underlying background noise (signal).
[0142] For example, the above-mentioned analysis unit may include the detection unit.
[0143] The detection unit may be configured to determine said probability p based on an F-value of the measured evoked response determined as the ratio between the variance of the average evoked response and the variance of the residual background noise, each having statistical degrees of freedom (DOF) v and r.
[0144] In other words, the detector may be configured to determine the probability p according to:
number
[0145] use In one aspect there is further provided the use of an apparatus as described above and in the claims.
[0146] method In one aspect, the present application further provides a method for detecting an evoked response.
[0147] For example, the method may be a method of detecting an auditory evoked neural response.
[0148] The method may further include estimating the hearing ability of the test subject.
[0149] The method may include generating at least one stimulus according to a test protocol.
[0150] The method may include generating a plurality of successive stimuli according to a test protocol.
[0151] The method may include providing at least one stimulus to the test subject.
[0152] The method may include providing said at least one stimulus to an ear of the test subject via a transducer.
[0153] The method may include providing said plurality of successive stimuli to a test subject.
[0154] The method may include presenting said plurality of successive stimuli to an ear of a test subject via a transducer.
[0155] The method may include placing one or more electrodes on the scalp of the test subject.
[0156] The method may include placing one or more optodes on the scalp of the test subject.
[0157] The method may include placing one or more microphones at the ears of a test subject.
[0158] The method may include placing one or more magnetometers on the scalp of the test subject.
[0159] The method may include measuring one or more evoked responses in the test subject.
[0160] One or more evoked responses of the test subject may be measured / recorded in response to (or as a result of) said provided at least one stimulus or a number of successive stimuli.
[0161] The method may include receiving, by an analysis unit, said measured one or more evoked responses.
[0162] The method may comprise analysing said measured one or more evoked responses by an analysis unit.
[0163] The analysis unit may determine, during operation of the device, the probability p that each of said responses is driven by residual (underlying) background noise.
[0164] The analysis unit may determine said probability p based on an F-value of the measured one or more evoked responses, determined as the ratio between the variance of the averaged one or more evoked responses and the variance of the residual background noise.
[0165] The analysis unit can determine said probability p based on an F-value of the measured one or more evoked responses determined as the ratio between the variance of the averaged one or more evoked responses and the variance of the residual background noise, each having statistical degrees of freedom v and r.
[0166] For example, the analysis unit may determine the probability (in the time domain) according to at least equations 7, 8, 9, 12 and 14.
[0167] For example, the analysis unit may determine the probabilities (in the frequency domain) according to at least equations 19, 20, 21, 22, 25, 28 and 26.
[0168] It is intended that some or all of the structural features of the devices described above or in the claims can be combined with the method embodiments when appropriately replaced by the corresponding process and vice versa, which have the same advantages as the corresponding devices.
[0169] Computer-readable medium or data carrier In one aspect, the present application further provides a tangible computer readable medium (data carrier) storing a computer program, the computer program comprising program code means (instructions) for causing a data processing system (computer) to execute (perform) at least some (e.g. most or all) of the (steps of) the methods described above and in the claims when the computer program is executed on a data processing system.
[0170] Computer Programs The computer program (product) comprises instructions that, when the program is executed by an apparatus (instrument, an apparatus detector, a computer, etc.), cause the apparatus to carry out (the steps of) the methods described above and in the claims provided herein.
[0171] Data Processing System In one aspect, the present application further provides a data processing system comprising a processor and program code means for causing said processor to perform at least some (eg, most or all) of the steps of the method.
[0172] system In another aspect, there is further provided a system including the apparatus described above and claimed herein, the system including an auxiliary device.
[0173] The system may establish a communications link between the appliance and the auxiliary device to allow information (eg, control and / or status signals) to be exchanged or transferred from one to the other.
[0174] The auxiliary device may include a remote control, a mobile device, a smart phone, or other portable electronic device such as a tablet, or may include a server device, or the like.
[0175] The communication link can be a wired or wireless link, for example based on Bluetooth or other standardized schemes.
[0176] APP In a further aspect, the present disclosure further provides non-transitory applications, referred to as APPs, which may include executable instructions configured to be executed on, for example, an appliance or auxiliary device, for example, to implement a control interface. [Brief description of the drawings]
[0177] Aspects of the present disclosure will be best understood from the following detailed description in conjunction with the accompanying drawings. The drawings are schematic and simplified for clarity, showing only details and omitting other details to improve understanding of the claims. The same reference numerals are used throughout for identical or corresponding parts. Each individual feature of each aspect may be combined with any or all features of the other aspects. These and other aspects, features and / or technical advantages will be apparent from and elucidated with reference to the figures described below. [Figure 1] 1 shows an example of an apparatus for detecting auditory evoked neural responses. [Diagram 2] An example of measuring and recording electrophysiological responses with a recording unit containing four electrodes is shown. [Diagram 3] An example of the distribution of p-values as a function of trials in the presence of Gaussian noise is shown. [Figure 4] An example of the distribution of F-measures in the presence of Gaussian noise as a function of trials is shown. [Diagram 5] An example of the distribution of F-measures as a function of trial in the presence of Gaussian noise and a simulated "nerve" signal is shown. [Figure 6A] 4 shows an example of another distribution of p-values as a function of trials in the presence of Gaussian noise. [Figure 6B] 4 shows an example of another distribution of p-values as a function of trials in the presence of Gaussian noise. [Figure 7A] 4 shows an example of another distribution of p-values as a function of trials in the presence of Gaussian noise. [Figure 7B] 4 shows an example of another distribution of p-values as a function of trials in the presence of Gaussian noise. [Figure 8A] An example of detection rate in the presence of simulated ASSR as a function of epoch number is shown. [Figure 8B] An example of detection rate in the presence of simulated ASSR as a function of epoch number is shown.
[0178] Further scope of applicability of the present disclosure will become apparent from the detailed description given hereinafter. It should be understood, however, that the detailed description and specific examples, while indicating preferred embodiments of the present disclosure, are given by way of example only. Other embodiments will be apparent to those skilled in the art from the following detailed description. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0179] The detailed description set forth below in conjunction with the accompanying drawings is intended as a description of various configurations. The detailed description includes specific details for the purpose of providing a thorough understanding of various concepts. However, it will be apparent to those skilled in the art that these concepts may be practiced without these specific details. Some aspects of the apparatus, systems and methods are described by various blocks, functional units, modules, components, steps, processes, algorithms, etc. (collectively referred to as "elements"). Depending on a particular application, design constraints, or other reasons, these elements may be implemented using electronic hardware, computer programs, or any combination thereof.
[0180] FIG. 1 shows an example of an instrument for detecting auditory evoked neural responses.
[0181] 1, it is shown that the device 1 may include a stimulus generator (SG), an output unit (OU), a recording unit (RU), and an analysis unit (AU). The stimulus generator SG, the output unit OU, the recording unit RU, and the analysis unit AU may be located in a base unit 2.
[0182] The stimulus generator SG may be configured to generate at least one stimulus 3 or a number of successive stimuli 3 according to a test protocol.
[0183] The output unit OU can be connected to a stimulus generator SG. The output unit OU can include a transducer. In Fig. 1, the transducer can include a receiver (speaker) arranged in an ear probe 4 and providing one or more stimuli 3 as acoustic signals to an ear 5 of a test subject 6.
[0184] The output unit OU may be configured to receive one or more stimuli 3 from the stimulus generator SG and provide one or more stimuli 3 via a transducer to an ear 4 of a test subject 6. It is shown that multiple successive stimuli 3 may be presented to the ear canal of the test subject 6 via an ear probe 4.
[0185] In FIG. 1, ear probe 4 is shown inserted into the right ear of test subject 6, however it is envisaged that either the right or left ear can be tested, or both ears can be tested simultaneously.
[0186] The recording unit RU may include one or more electrodes 7a, 7b, 7c, 7d shown to be placed on the scalp of the test subject 6. The one or more electrodes 7a, 7b, 7c, 7d may measure (during a time segment T) one or more electrophysiological responses 8 of the test subject 6 in response to one or more stimuli 3 being presented to the ear 5 of the test subject 6. The recording unit RU may record the electrophysiological responses 8.
[0187] The analysis unit AU can be configured to receive one or more measured (and recorded) electrophysiological responses 8 from the recording unit RU (or alternatively directly from the electrodes 7a, 7b, 7c, 7d) (see solid lines between RU and AU).
[0188] The analysis unit AU may further be configured to receive data / information (eg time, type, etc.) relating to one or more stimuli 3 from a stimulus generator SG (see solid line between SG and AU).
[0189] The analysis unit AU may be configured to analyze one or more electrophysiological responses 8 measured by the electrodes 7a, 7b, 7c, 7d. The analysis may for example be based on data / information regarding one or more stimuli 3 received from a stimulus generator SG.
[0190] The analysis unit AU may be configured to determine a probability p that each of said responses is or is not driven by an underlying background noise. The determination may further be based on at least the variance of the measured one or more electrophysiological responses 8.
[0191] 1 shows that the device 1 may further comprise a processor PR (e.g. located in the base unit 2). The processor PR may be configured to control the operation of the device 1. In other words, it may control such that test protocols are executed and measured data are analysed.
[0192] As indicated by dotted lines, the processor PR may be connected to at least a stimulus generator SG, a recording unit RU, and an analysis unit AU.
[0193] As shown by the dashed lines, the output unit OU may transmit stimuli 3 to the ear probe 4 via a wired or wireless connection 9, and the recording unit RU may receive the measured electrophysiological responses 8 from the electrodes 7a, 7b, 7c, 7d via a wired or wireless connection 10.
[0194] FIG. 2 shows an example of measuring and recording electrophysiological responses with a recording unit containing four electrodes.
[0195] In Fig. 2, the electrophysiological response may be EEG data, which may be measured by four electrodes 7a, 7b, 7c, 7d placed on the scalp of a test subject 6, substantially similar to the example shown in Fig. 1.
[0196] Thus, four channels may extend from the electrodes 7a, 7b, 7c, 7d to the recording unit RU for transmitting the EEG data. The time series of the EEG data is shown in a graph within the recording unit RU.
[0197] The measured electrophysiological response 8 (ie, the target neural response) is shown by the (black) waveform X(t) at the top of the graph.
[0198] The measured and recorded data, contaminated by the underlying background noise, are shown by separate colored traces for each channel, i.e. the first (blue, CH1) and third (turquoise, CH3) curves from the top measured at the left ear, and the second (red, CH2) and fourth (orange, CH4) curves from the top measured at the right ear. Electrodes 7a, 7b placed on the left ear may represent channels CH1 and CH3, and electrodes 7a, 7b placed on the right ear may represent channels CH2 and CH4. Electrode 7c placed on the side and electrode 7d placed on the top of the scalp may represent ground and reference channels / electrodes.
[0199] The onset of a new stimulus presentation is indicated by a vertical line in the graph, within which trials of duration T (s1(t), s2(t), ..., s N (t)) is obtained.
[0200] P, that is, P1...P m indicates that m points are tracked and used to determine a probability p.
[0201] FIG. 3 shows the distribution of p-values in the presence of Gaussian noise as a function of trials.
[0202] Since the F-value is estimated using multiple (measurement) points m and DOFs (characterizing the underlying background noise of each recording individually), the determination of the F-value is called Fmpi.
[0203] To demonstrate the accuracy of the Fmpi detector (or detection method), a total of 10,000 simulations were acquired with 360 trials each. The simulations were performed using a typical setup for cortical response recordings. In the simulated recordings, there were no neural responses and the underlying background noise was Gaussian. To compare the performance of the Fmpi detector, the same dataset was processed using a standard single-point F detector [1][2] and a standard average (Fsp-sa). This detector estimates the residual noise of the F ratio using a single point and assumes a DOF of r=N (total number of trials) and v=5 (based on an empirical analysis of human EEG noise).
[0204] In Figure 3, three different instruments (including three different detectors / methods for estimating the F-value) are used. The three different detectors (methods / algorithms) include Fmpi using weighted average (Fmpi-wa), Fmpi using standard average (Fmpi-sa), and Fsp using standard average with v=5 [1][2]. The trial length was 1 second, the sampling rate was 500Hz, the analysis window was 300ms, and the data was bandpass filtered between 2Hz and 30Hz. All points within the range of the trial were used by Fmpi. Using the Fmpi detector, the p-values are uniformly distributed across all trials (which is the ultimate goal for a correct estimation of the underlying noise distribution).
[0205] FIG. 4 shows an example of the distribution of F-measures in the presence of Gaussian noise as a function of trials.
[0206] In Figure 4, a histogram of the obtained F-measures (stat values, 11) is shown in grey. Probability density functions estimated by three detectors are shown (blue, 12): Fmpi using weighted average (Fmpi-wa), Fmpi using standard average (Fmpi-sa) and Fsp using standard average with v=5 [1][2]. The empirical probability distribution is shown in red 13. The trial length was 1 s, the sampling rate was 500 Hz, the analysis window was 300 ms and the data was bandpass filtered between 2 Hz and 30 Hz. All points within the range of the trial were used by the Fmpi detector.
[0207] The estimated F-distribution closely matched the empirical distribution of the data using either the standard (Fmpi-sa, Eq. 3) or weighted (Fmpi-wa, Eq. 4) averaging Fmpi detector, whereas the classical F detector gave a skewed distribution (Fig. 3, Fsp-sa) and a poor fit (Fig. 4, Fsp-sa).
[0208] FIG. 5 shows an example of the distribution of F-measures in the presence of Gaussian noise and a simulated "elicitation" signal as a function of trial.
[0209] Simulations were performed under the same conditions as in the previous examples (Figs. 3 and 4), but now including the target "evoked" signal. The results show that the estimate of the underlying distribution of the noise is identical to that obtained without including the "neural" response (compare solid and dashed black lines). The Fmpi detector correctly estimated the 5% chance point of false positive detection (vertical line), allowing earlier and more accurate detections than the standard Fsp-sa detector. The standard Fsp-sa detector overestimated the "5%" chance point (compared to the empirical distribution) and resulted in fewer detections in fewer trials.
[0210] For comparison, histograms of F-values (stat values) obtained without target signals are shown in light blue 14 and histograms of F-values obtained in the presence of target "elicited" signals are shown in red 15 for all three detectors: Fmpi using weighted average (Fmpi-wa), Fmpi using standard average (Fmpi-sa) and Fsp using standard average of [1][2]. The distribution underlying the noise obtained without target neural signals is shown in black dashed line 16 and the distribution obtained in the presence of target neural signals is shown in black solid line 17. The vertical lines indicate the critical F-values for determining the presence of a response. The values on the right are considered as detected responses. The trial length was 1 s, the sampling rate was 500 Hz, the analysis window was 300 ms and the data was bandpass filtered between 2 Hz and 30 Hz. The pre-determined target SNR for the average of 360 trials was 6 dB. All points within the range of the trials were used by the Fmpi detector.
[0211] 6A and 6B show an example of another distribution of p-values (probability values) in the presence of Gaussian noise as a function of trials.
[0212] To illustrate the precision determination of F-measures (and probabilities) in the frequency domain (e.g. by detection units), a typical recording session of 200 epochs was simulated. The probability of detecting a response was estimated in blocks of 20 epochs. Each block was run 1000 times, resulting in a total of 10,000 random trials throughout the simulation. The simulation was performed using a typical setup for cortical response recordings. No neural responses were included, allowing us to show how well the characteristics of the underlying background noise, which is Gaussian distributed, can be predicted.
[0213] Figure 6A shows the distribution of p-values for these simulations, where the p-values are uniformly distributed across all trials (which is the ultimate goal for a good estimation of the underlying distribution).
[0214] The distribution of F-values (stat values) in the presence of Gaussian noise as a function of trials is shown in Figure 6B. A histogram of the obtained F-values (stat values) is shown in grey 19. The probability density function estimated using the modified detector is shown in blue (line 20). The empirical probability density is shown in red (line 21). The trial length was 1 s, the sampling rate was 500 Hz, the analysis window was 500 ms, and the data were bandpass filtered between 2 Hz and 30 Hz. The estimated F-distribution closely matches the empirical distribution of the data.
[0215] 7A and 7B show an example of another distribution of p-values as a function of trials in the presence of Gaussian noise.
[0216] In Figure 7A, the probability distribution (in its frequency-specific version, in the frequency domain) is tested by simulating the auditory steady-state response (ASSR). The simulation is performed by convolving the template response at a rate of 41 Hz in the presence of Gaussian noise.
[0217] The results of this simulation are compared with the proposed solution (in the frequency domain) using the Hotelling-T-squared (HT2) test, a method commonly used for this type of analysis.
[0218] Both tested whether the composite power of 12 frequencies (multiples of 41 Hz) was significantly greater than that of the underlying background noise. Two simulations were performed, first without an "evoked response", i.e., noise only (Fig. 7A and 7B), and then in the presence of a simulated evoked response (Fig. 8A).
[0219] Noise simulation results show that both our solution and the HT2 test can predict the underlying distribution well. However, the HT2 method cannot provide reliable estimates when the number of frequencies of interest is larger than the number of trials included, limiting its use especially when the number of trials is small.
[0220] Figure 7A shows the distribution of p-values in the presence of Gaussian noise as a function of trials using the frequency-specific implementation of the proposed method (Fmpi-FFT) (red, 22) and the Hotelling-T-squared (HT2, blue, 23) test. Figure 7B shows the distribution of F-values for the Fmpi-FFT and HT2 tests in the presence of Gaussian noise as a function of trials. The histogram of the obtained F-values (stat values) is shown in grey 24. The probability density function estimated using Fmpi-FFT and HT2 is shown in blue (line 25). The empirical probability density is shown in red (line 26). The HT2 method provides a reliable estimate of the underlying distribution for more than 30 trials, but cannot provide a reliable estimate for fewer than 30 epochs, so these data points are not included. The trial length was 1 s, the sampling rate was 2000 Hz, the analysis window was 1000 ms, and the data were not filtered.
[0221] 8A and 8B show an example of detection rate in the presence of simulated ASSR as a function of epoch number.
[0222] Figure 8A shows the detection rates of Fmpi-FFT27 and HT228 in the presence of a simulated 41 Hz ASSR as a function of epoch number. The right graph (Figure 8B) shows the false positive rates (FPR) of both methods (same noise, but without ASSR), Fmpi-FFT29 and HT230, as a function of epoch number. A total of 12 frequencies were included in the analysis (41 Hz and its harmonics). The trial length was 1 s, the sampling rate was 2000 Hz, the analysis window was 1000 ms, and the data were filtered. The horizontal line in the right panel indicates the target 5% false positive rate used in this simulation.
[0223] In Figure 8A, we see that in the presence of simulated neural signals, Fmpi-FFT outperforms HT2, achieving more true detections (left graph) and fewer false positives (Figure 8B) than the HT2 test.
[0224] The structural features of the above-described devices and systems described in the detailed description and / or claims may be combined with method steps when appropriately replaced by corresponding processes.
[0225] When used, the singular forms "a," "an," and "the" are intended to include the plural (i.e., meaning "at least one") unless otherwise indicated. Furthermore, the terms "includes," "comprises," "including," and / or "comprising," as used herein, specify the presence of stated features, integers, steps, operations, elements, and / or components, but should be understood to not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It is also understood that when an element is referred to as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, but intervening elements may be present unless otherwise indicated. Furthermore, as used herein, "connected" or "coupled" may also be wirelessly connected or coupled. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items. The steps of any disclosed method are not limited to the exact order described herein, unless otherwise indicated.
[0226] The claims are not intended to be limited to the embodiments set forth herein, but are to be accorded the full scope consistent with the language of the claims, and references to elements in the singular are not intended to mean "only one," unless specifically stated otherwise, but rather "one or more." Unless specifically stated otherwise, the term "some" means one or more.
[0227] References [1] Don, M. and Elberling, C. (1994) 'Evaluating residual background noise in human auditory brain-stem responses.', The Journal of the Acoustical Society of America, 96(5 Pt 1), pp. 2746-57. [2] Elberling, C. and Don, M. (1984) 'Quality estimation of averaged auditory brainstem responses.', Scandinavian audiology, 13(3), pp. 187-97. [Explanation of symbols]
[0228] 1 equipment 2 Base Unit 3 stimulation 4 ear probe 5. Ears 6. Test Subjects 7 electrodes 8. Electrophysiological Responses SG Stimulus Generator OU Output unit RU Recording Unit AU Analysis Unit
Claims
1. 1. An apparatus for detecting an evoked response, the apparatus comprising: a stimulus generator configured to generate at least one stimulus or a plurality of successive stimuli according to a test protocol; at least one output unit including a transducer configured to receive the at least one stimulus or a plurality of successive stimuli from the stimulus generator and to provide the at least one stimulus or a plurality of successive stimuli to a test subject; at least one recording unit including one or more sensors for measuring one or more evoked responses of the test subject in response to the at least one stimulus or a plurality of successive stimuli provided; an analysis unit configured to receive and analyze the measured one or more evoked responses, the analysis unit configured to determine a probability p that each of the responses is driven by underlying background noise during operation of the device, the analysis unit configured to determine the probability p based on an F-value of the measured one or more evoked responses, the F-value being determined as the ratio between the variance of the mean of the one or more evoked responses and the variance of the residual background noise; Including equipment.
2. The device of claim 1 , wherein the one or more sensors are configured to measure one or more evoked responses comprising a ratio of a variance of a mean response evoked by a stimulus to a variance of the residual background noise.
3. The analysis unit is configured to determine the probability p according to the following formula: p=1-F CDF (F,v,r) Here, F CDF 2. The apparatus of claim 1, wherein (F,v,r) is the cumulative distribution function of an F-distributed random variable, v is the statistical DOF of the variance of the mean of one or more evoked responses, and r is the statistical DOF of the variance of residual background noise.
4. The analysis unit detects the frequency of interest ω i σ for ２ s Determine the numerator of F based on the DOF of ω and the frequency of interest ω i σ for ２ RN 2. The apparatus of claim 1, configured to determine a denominator of the F-number based on a DOF of 0.
01.
5. The apparatus of claim 1 , wherein the analysis unit is configured to estimate the power of the residual background noise at each frequency of interest by averaging the power of the residual background noise at neighboring frequencies of the frequency of interest.
6. The device of claim 1 , wherein the analysis unit is configured to adaptively determine the probability during operation of the device.
7. 2. The device of claim 1, wherein the at least one stimulus or sequence of stimuli comprises an acoustic stimulus such as a click, a narrowband chirp, an auditory change complex (ACC) stimulus, and / or a speech signal, and / or comprises an electrical stimulus.
8. 2. The device of claim 1, wherein the one or more evoked responses measured include an auditory brainstem response (ABR), a cortical auditory potential (CAP), or an auditory steady-state response (ASSR), or an otoacoustic emission (OAE), or any response that relies on averaging to reduce underlying background noise to detect a target signal.
9. The apparatus of claim 1 , wherein the analysis unit is configured to determine the hearing ability of the test subject in response to the probability being below a predetermined threshold.
10. The device of claim 1 , wherein the stimulus generator is configured to modify the at least one stimulus or plurality of successive stimuli in response to the determined probability.
11. 1. A method for detecting an evoked response, the method comprising: generating at least one stimulus or a plurality of successive stimuli according to a test protocol; providing said at least one stimulus or a plurality of successive stimuli to a test subject; placing one or more sensors on a test subject; measuring one or more evoked responses in the test subject in response to the at least one stimulus or a plurality of successive stimuli provided; receiving and analyzing, by an analysis unit, the one or more measured evoked responses; determining a probability p that each of the responses is driven by underlying background noise during operation of the device by determining the probability p based on an F-value of the measured one or more evoked responses, determined as the ratio between the variance of the mean of the one or more evoked responses and the variance of the residual background noise, each having statistical degrees of freedom v and r; The method includes:
12. A data processing system comprising a processor and program code means for causing said processor to perform at least some of the steps of the method according to claim 11.
13. A computer program comprising instructions which, when said computer program is executed by an apparatus according to any one of claims 1 to 10, cause said apparatus to carry out the steps of the method according to claim 11.