RU2248106C2 - Method for modeling auditory perception in patients after cochlear implantation - Google Patents

Abstract

FIELD: medicine, audiology, surdology.

SUBSTANCE: the present innovation deals with speech signal reprocessing dealing with analog-digital transformation of incoming signal, division of transformed signal into frequency strips, summing up of strips, digital-analog transformation of summed up signal and its presentation for patients under testing. Moreover, one should sum up 5 uneven strips of 50 Hz width, the lower border of the first uneven strip is equal to 200 Hz, the upper border of the last uneven strip is equal to 6250 Hz, moreover, the strips should be placed being at equal distance against each other corresponding to 5.2 mm by the length of basilar membrane in accordance with normal cochlear tonotopics. Processed signal should be applied for patients' self-instruction to have a perception of spectrally deprivated speech signal. As a result, the chance appears to model auditory perception in patients after cochlear implantation.

EFFECT: higher efficiency.

1 dwg

Description

The invention relates to medicine, namely to audiology and audiology, and may find application in the process of rehabilitation of implanted patients.

In the last decade, the introduction of the method of auditory rehabilitation of deaf patients, cochlear implantation (CI), has begun in Russia. CI systems created in a number of countries are used for implantation, and therefore the main issue of the CI program is the development of methods for auditory rehabilitation of patients after cochlear implantation.

It is known that in the process of auditory rehabilitation of children after cochlear implantation, a significant role is given to parents. This is due to the need for their direct participation in the rehabilitation of their children, but for their effective work it is necessary that they understand the problem deeply enough. Unfortunately, in most cases, parents' perceptions of cochlear implantation are very limited.

Moreover, bearing in mind the funds invested in the operation, they cannot understand why their child cannot immediately speak and understand the words addressed to him.

In the case when the speech processor of a cochlear implant is set up in 2-3 children at the same time, parents naturally see the differences between the children and also do not understand their cause. Knowing about the successes of some previously implanted patients, they can not understand why their situation is different. A clear physiological explanation of the peculiarities of the auditory perception of implanted patients and the resulting problems requires starting the presentation of the question from the basics of acoustics and physiology, which is hardly possible.

How can parents clearly demonstrate the challenges facing their implanted children? It is necessary to simulate the speech signal perceived by the implanted patients and present it to relatives and all those who are interested in the problem of CI.

To build a real program of rehabilitation for a specific contingent of deaf patients, namely patients who underwent cochlear implantation, a clear idea of what and how they hear is absolutely necessary. It is known that their auditory perception is significantly different from the perception of a person with normal hearing, and from the perception of hearing-impaired patients using hearing aids.

You can give at least such an example.

From the literature it is known that a person with normal hearing can determine 600 frequency gradations, in patients with sensorineural hearing loss the number of gradations is 10 times less, but an implanted patient can distinguish only a small number of frequency bands, which is limited by the number of electrodes introduced into the cochlea. Currently, it does not exceed 20.

Thus, the difference in perception only with respect to the spectral parameter is huge, and for this reason, modeling a speech signal, which, as far as possible, is close to that which the implanted patient perceives, is very important.

There is a method of processing a speech signal, where a sound image is perceived by an implanted patient by creating a speech signal limited in frequency representation and the intelligibility of speech processed in this way is evaluated in people with normal hearing [1].

The Combi 40 cochlear implantation system of the Austrian company Med El was used, where a chain of 8 electrodes is inserted into the cochlear patient. The width of the spectrum of the speech signal processed by the processor is in the range from 300 to 5500 Hz. The implant processor "Combi-40" divides this spectrum into 8 bands. In each of them, energy is measured and, in accordance with it, a pulse is generated that is supplied to the corresponding electrode. The program embedded in the speech processor allows you to change the boundary (and central) frequencies of these bands depending on the applicable processing law. There are 4 laws of spectrum separation in the implant program: logarithmic, linear-logarithmic, tonotopic and linearly descending.

Next, a computer program for the spectral processing of a speech signal was applied [2]. It makes it possible to cut, i.e. remove from the acoustic signal, various, specified by the researcher, parts of the spectrum. In this work, frequency bands of 50 Hz wide were used as test signals, which were located around the central frequencies of the bandpass filters of the speech processor. The rest of the spectrum was removed.

Thus, after processing the speech signal, only 8 frequency bands with a width of 50 Hz remain informative. Therefore, from the entire spectrum of speech (300-5500 Hz) processed by the speech processor of the implant, only 8 · 50 = 400 Hz is left, that is, only 400/5200 = 1/13 of the spectrum is used to evaluate speech perception.

After processing the speech signal in accordance with the 4 laws of separation of the spectrum, speech intelligibility was measured. For this, Greenberg-Zinder tables were used, with 30 words each. The speech signal from the IBM PC goes to the Amphiton amplifier and then to the TDH-39 telephone. Amplifier tone controls are set to mid-position, providing uniform amplification of all frequencies in the speech spectrum. The subject selects only the level of comfortable volume of the speech signal. Prior to registration, they conduct a short training session (5-10 minutes) to master the new sound of Russian speech.

Speech intelligibility was measured in 6 hearing-impaired individuals with monaural presentation of speech stimuli in the right ear. All subjects showed 100% speech intelligibility in all four speech processing options. Based on the fact that after processing only 1/13 remained in the speech signal, i.e. 7.7% of the spectrum, we can conclude that the redundancy of speech in terms of frequency representation exceeds 92.3%.

Based on the results of such a significant speech redundancy (in terms of frequency representation), it becomes clear why implanted patients with such low frequency selectivity successfully learn the new Russian language.

It should be noted that at the beginning of the study, all subjects were lost from such an unusual speech signal - from the first training dozens of words (regardless of the law of separation of the spectrum), they distinguished only 2-3. However, they quickly mastered with a changed picture of speech and in the future testing no longer caused significant difficulties. To achieve 100% speech intelligibility, only a short training session was enough [1]. That is, such a representation of a speech signal can be considered as a model of perception in the first approximation, but it is unsuitable for demonstration.

There is also a method of processing a speech signal, where 4 bands of 50 Hz are used in the frequency range 200-6250 Hz [3]. Comb filtering of speech is carried out using the developed computer program for the spectral processing of the speech signal [2].

After comb filtering the speech signal and conditionally numbering the frequency bands, the odd-numbered bands are added up, and the sum of these bands is used as a test signal to measure speech intelligibility. The lower cutoff frequency of the first band is 200 Hz. The sum of the widths of the even and odd bands is the period of the comb filter. In the study, the bandwidth with odd numbers was the same and equal to 50 Hz. In the first part of the work, the filtration period was a variable, and in the second, speech material was used, processed only with one period value equal to 2000 Hz.

Based on the results obtained on a group of inexperienced subjects in the first part of the work, the parameters of the comb filter were chosen such that even the best of them did not have 100 percent speech intelligibility, but it was achievable by experienced auditors with these filter parameters. Namely, the width of the test bands was 50 Hz; the period of the comb filter was 2000 Hz.

In the second part of the work, we compared two ways of teaching normally hearing people how to perceive a spectrally deprived speech signal. The subjects were divided into 2 subgroups of 5 people. The processed words were sent monaurally to the right ear. According to the instructions, the subject set a comfortable volume level for the speech signal and repeated the words he heard.

Subjects of the first subgroup were subsequently presented with word tables, registering the correct answers. The first table was used for training and self-training, and the results obtained upon presentation were not taken into account.

Subjects of the second subgroup also consistently presented tables of words, but they listened to the first table for the purpose of learning while reading the same words from the sheet. And they should first read the word, and then listen to it.

Next, each test subject was presented, without taking into account the first training session, 4 test tables of 30 words each. It was found that some hearing impaired from the first group of subjects could not identify any of the 120 words presented to them. However, after learning the perception of processed speech with visual accompaniment, it turned out that, firstly, all the subjects of the second group began to understand some words and, secondly, in comparison with the results of the first group, speech intelligibility after training significantly increased.

Based on the results obtained by the above methods, it seems possible to draw parallels between speech perception treated with a comb filter, normally hearing and speech perception by implanted patients, and based on this, we can simulate the auditory perception of patients after cochlear implantation.

It should be noted that in the method of processing a speech signal [1], the result is not demonstrative, since it is not difficult to master the perception of speech processed in this way, and in the method [3] the task was too complicated. However, the results of the second method prove that the perception of speech so poor in spectral content can be learned. Therefore, to create a working model, it is necessary to select the appropriate parameters of the comb filter.

The closest method for processing a speech signal is also known, including analog-to-digital conversion of the input signal, then separation of the converted signal using band-pass filters into frequency bands, which are summed up and, after digital-to-analog conversion, are fed to the output device [4].

The disadvantage of this method is that with such a conversion of the speech signal the spectral representation of the speech signal is excluded, i.e. the real spectrum of the speech signal is completely absent. Speech is represented by a set of sinusoids that are not contained in the original speech signal.

The technical result of the present invention is the modeling of the speech perception of implanted patients by obtaining a speech signal limited in frequency representation, which makes it possible to demonstrate to the relatives of the implanted patients the complexity of the problem of auditory rehabilitation of their children and to make sure that it is possible to solve it first hand.

This result is achieved by the fact that in the method of modeling the auditory perception of patients after cochlear implantation, which includes processing the speech signal, which consists in analog-to-digital conversion of the input signal, dividing the converted signal into frequency bands, summing the bands, digital-to-analog conversion of the total signal and presenting it to subjects with normal hearing, according to the invention summarize 5 odd bands with a width of 50 Hz, where the lower boundary of the first odd band is 200 Hz, the upper Itza odd last band is 6250 Hz, the strip being spaced at an equal distance from each other, is 5.2 mm along the length of the basilar membrane in accordance with normal tonotopikoy cochlea. The processed signal is used for self-study of subjects with the perception of a spectrally deprived speech signal.

The presence of distinctive features, namely the summation of 5 bands with a width of 50 Hz, where the first strip has boundaries of 200 Hz, and the fifth strip is located at 6,250 Hz, the separation of the bands at an equal distance from each other, amounting to 5.2 mm along the length of the basilar membrane, when the coordinates of the point on the basilar membrane and the center frequency of the band perceived at this point correspond, the use of the received signal for measuring speech intelligibility indicates the conformity of the claimed technical solution with the patentability criterion novelty. "

The specified method is based on studies conducted at the St. Petersburg Research Institute of ENT with the participation of 7 materials of implanted patients. The study is based on the conclusion that there is a sensation of a frequency band during stimulation of the auditory nerve by electrical impulses [5] and on data on the possibility of learning to perceive a spectrally deprived speech signal [3].

The speech signal is fed through an analog-to-digital converter to an IBM PC, and the converted signal is divided into even and odd bands. In this case, the lower boundary of the first odd band is 200 Hz, and the upper boundary of the last band is 6250 Hz.

The odd bands have a width of 50 Hz and, in accordance with the normal tonotopy of the cochlea, they are located at an equal distance from each other (5.2 mm) along the length of the basilar membrane, namely, when the coordinate along the cochlear axis and the perceived frequency correspond. These calculations were carried out in accordance with the Hartmann W. formula on the dependence of the characteristic neuron frequency on the coordinate of a point on the basilar membrane [6]. The signals of the five odd bands are added up and then output to the output device through a digital-to-analog converter.

As a characteristic of bandpass filters, the filter bandwidth was used at the level of I = -20 dB, i.e. the width of the test frequency bands at the intensity level is 20 dB lower than the maximum intensity value in this band. The passband of all filters was equal to 170-200 Hz, i.e. the computer program used ensured the steepness of the decay of the fronts of the band-pass filter, equal to 0.3-0.35 dB / Hz, both in the direction of high and low frequencies.

As an illustration, the drawing shows the spectrum of the speech signal used in the work and processed by a comb filter. In the drawing: the abscissa axis is the frequency in Hz, the ordinate axis is the intensity level in dB attenuation.

As a result of the studies, it was found that with successive presentations of two speech tables (60 words), self-learning is conducted for the perception of spectrally deprived speech, represented by five spectral bands 50 Hz wide in the frequency range 200-6250 Hz, which is reflected in an increase in the intelligibility of the converted speech signal to saturation level. All subjects on the 7th-9th day of examinations, according to the results of speech intelligibility measurements, reached their intelligibility maximum (plateau level).

At the very beginning of testing, parents learned the examination procedure with a full range of speech. They had no problems. Hearing, at the first presentation, a speech signal processed by a comb filter, and knowing from the instruction that the words sound, they themselves say that it is difficult to believe. The most common assessment of sound quality that they give on their first examination: "Awful." Based on a preliminary explanation of the essence of the study, they more clearly assess the tasks facing their children and conclude that their requirements for auditory perception of children, presented immediately after the first inclusion of the implant were overstated.

It should be noted that the increase in speech intelligibility occurred to the level of the plateau, and it was different for different mothers (from 63 to 93%). 100% intelligibility was not achieved by any of the mothers. The different final (plateau level) results of measuring speech intelligibility in parents serve as a vivid demonstration of the different abilities of people to understand spectrally transformed speech and, on the basis of analogy with cochlear implantation, enable the audiologist to explain to parents the differences in the outcomes of their children's rehabilitation.

The absence of 100% speech intelligibility in all subjects makes it possible to explain to mothers that for a more complete development of new sound pictures, special approaches must be used. It should be emphasized that in this work we considered only self-study without the use of additional sound-pedagogical techniques, and therefore, for the successful rehabilitation of their children, special classes and the direct participation of parents in the rehabilitation process are necessary.

It should be especially noted that at the end of the study, all subjects change their initial assessment of the sound quality of the speech material and do not make negative judgments about the perception of processed speech. Moreover, they themselves wonder why it was so difficult to understand the words at the beginning of the survey. And here again it is possible to draw parallels with implanted postlingual patients, i.e. such patients who have lost their hearing after mastering a speech. After the implant is first turned on, it is difficult for them to believe that later on they will perceive this incomprehensible sound picture as speech. As they gain auditory experience, they experience less and less difficulty in communicating. At the end of the processor setup, most post-lingual patients are asked, "How do you hear now?" answer: "As before." Those. they restore an unusual (significantly distorted!) sound picture, similar to the one that they had before. Patients themselves believe that they perceive speech as before, which they are convinced of because of their ability to communicate on the phone.

The successful development of the perception of a new sound picture of speech is evidenced by the result obtained on the last day of the experiment. With a single presentation of the third control table processed by a comb filter with the same parameters as the training table, no significant differences were found between the control result and the final one obtained upon presentation of repeatedly repeated two tables (p <0.05).

It should be noted that after several days of examination, some parents are already aware that the study is not quite like the perception of their children, since children who are deaf from birth do not know what they hear. We cannot but agree with the parents of the implanted children that this examination is far from a complete analogy of the auditory perception of their children. Indeed, their children are probably in a more difficult position - they, unlike their parents, have no strongholds for the completion or reconstruction of their first-heard paintings. But, firstly, such questions and reasoning of parents already indicate the beginning of a sound understanding of the problem and, secondly, this allows us to explain and emphasize the difficulties that their child faces. And parents already have a great understanding of the explanation of the problems of implanted patients.

In the first tests, some parents note that they do not have enough time to analyze and identify some words (apparently those whose spectral picture is not adequately reflected in the remaining spectral bands). The time interval between words was 5 seconds. As the experiment progresses, the process of recognizing words is simplified for them. And again, relying on the analogy with cochlear implantation, one can explain to parents that their children are also difficult, and perhaps harder, to perceive all the words, and therefore it is not surprising that they do not immediately understand the speech addressed to them and often ask again.

It should be noted that as the experiment progresses, parents more and more penetrate the problem of cochlear implantation and the questions asked by them indicate an increasingly full understanding of this problem. The mothers of implanted children who took part in the experiment positively assess the situation simulated by him and themselves recommend that all mothers conduct such an examination. Other mothers, having heard from the examined parents about a similar study, themselves express a desire to participate in it.

The proposed method was carried out in the Department of Ear Pathophysiology of the St. Petersburg Research Institute of Ear, Throat, Nose and Speech.

Using the proposed method for processing a speech signal allows to obtain the following technical result:

1. The results of the study indicate the presence of parallels between the perception of a spectrally deprived speech signal by subjects with normal hearing and speech perception by implanted patients and, on this basis, allow us to simulate the auditory perception of implanted patients using the developed program.

2. The participation of parents of implanted children in experiments on the perception of processed speech serves as a vivid demonstration for them of how their child learns how to perceive speech processed by the processor of a cochlear implant and what he can achieve.

From the foregoing, it follows that the introduced distinguishing features together with the known features are the reason for achieving the specified technical result, which indicates the compliance of the claimed technical solution with the patentability criterion of “inventive step”.

The essence of the method is illustrated by example.

The speech signal was recorded on the hard drive and processed according to the developed program. In accordance with the formula of Hartmann W. (Z = 7.24 ln (1 + F / 165)) on the dependence of the characteristic frequency of a neuron on the coordinate of a point on the basilar membrane, 5 frequencies were chosen that are located at an equal distance from each other (5.2 mm) along the length of the basilar membrane in the frequency range from 200 to 6250 Hz. Around these frequencies, 5 odd bands with a width of 50 Hz were allocated. So the first band had boundaries of 200-250 Hz, and the fifth 6200-6250 Hz. The sum of the five spectral bands was used to examine subject K.

After explaining the essence of the examination and training with the full spectrum of the speech signal, a study of speech intelligibility, processed according to our program, began. On the first day, the test subject correctly identified 20 out of 60 words (33%), but according to her, the sound of the speech was terrible. In the following days, with the presentation of the same 60 words, speech intelligibility gradually increased. Moreover, it should be noted that in sequential tests from 90 to 100% of the words perceived in the previous series, the test subject recognized in the next examination. If I may say so, then from test to test, the subject experienced "accumulation" of vocabulary.

At examination 7, 8 and 9, the results of speech intelligibility measurements differed by one word, i.e. the subject reached her maximum intelligibility. Upon presentation of another (control) table processed by a comb filter with the same parameters, the testee identified 83% of the words. The control (83%) and the result of self-training (87%) did not differ. The absence of differences allowed the subject to make sure that she not only remembered the repeated words, but acquired a new skill. And this again gives patients the opportunity to understand the problem of mastering speech perception by their implanted children. It should be especially noted that at the end of the study, the subject radically changed her initial assessment of the sound quality of the speech material and did not have negative judgments about the perception of processed speech. Moreover, she herself was surprised why at the beginning of the study it was so difficult to understand the words. And here again it is possible to draw parallels with implanted patients.

According to this scheme, 7 parents of implanted children were examined. As a result of the study, a significant increase in speech intelligibility (p <0.05) was found upon presentation of a speech signal processed by the proposed method on average in the group from 28 to 83%.

This work was supported by the Russian Foundation for Basic Research (grant No. 01-04-49800).

The proposed method provides a technical result and can be carried out using means known in the art. From the foregoing, it follows that the claimed technical solution meets the patentability criterion of "industrial applicability".

LIST OF REFERENCES

1. Lantsov A.A., Petrov S.M., Pudov V.I. To the question of modeling the auditory perception of implanted patients // News, otorhinolar. - 1999. - N6. - S.21-23.

2. Petrov S.M. A method of processing a speech signal // Patent N2121242. - 1999.

3. Petrov S.M. Perception of a spectrally deprived speech signal // Human Physiology. - 2003. - 1. - P.72-74.

4. Loizou S., Dorman M. and Tu Z. On the number of channels needed to understand speech // J. Acoust. Soc. Amer. - 1999. - Vol.106, N 4. - P.2097-2103.

5. Eddington O.K., Dobelle W.H., Blackman D.E. e.a. Auditory prosthesis with multiple channel intracochlear stimulation // Ann-Otol-Rhinol-Laryngol. - 1978. - Vol. 87, N6. - Part 2. Suppl. 53. - P.l-39.

6. Hartmann W. Pitch, periodicity and auditory organization // J. Acoust. Soc. Amer. - 1996. - Vol. 100, N6. - P.3491-3502.

Claims (1)

A method for simulating the auditory perception of patients after cochlear implantation, including processing a speech signal, which consists in analog-to-digital conversion of the input signal, dividing the converted signal into frequency bands, summing the bands, digital-to-analog conversion of the total signal and presenting it to the subjects, while adding five odd bands with a width of 50 Hz, the lower boundary of the first odd band is 200 Hz, the upper boundary of the last odd band is 6250 Hz, and the bands are spaced apart by p vnoe distance from each other, is 5.2 mm along the length of the basilar membrane in accordance with normal tonotopikoy cochlea, the processed signal is used for self test perception spectrally deprivirovannyh speech signal.