An adaptive cochlear implant fitting method and system based on electrode-band mapping

Abstract

The application discloses an electrode-frequency band mapping-based self-adaptive cochlear implant debugging method and system, relates to the technical field of hearing aid and digital medical treatment, and effectively eliminates the frequency-position mismatch caused by individual cochlear anatomical differences by generating an electrode-specific reference vector through actual measurement, significantly improves the accuracy and repeatability of debugging, and overcomes the limitations of traditional experience or CT images; meanwhile, by adopting the strategy of mapping the initial consonants, final consonants and tones to electrodes according to acoustic frequency bands and independently optimizing, the three types of speech information are prevented from interfering with each other on the electrodes, and speech intelligibility and naturalness are fundamentally improved.

Description

Technical Field

[0001] This invention relates to an adaptive cochlear implant tuning method and system based on electrode-frequency band mapping, belonging to the fields of hearing assistance and digital medical technology. Background Technology

[0002] Cochlear implants are a primary rehabilitation method for patients with severe to profound hearing loss, and their postoperative adjustment directly affects patients' speech recognition ability and quality of life. However, the following long-standing technical challenges remain in the field of cochlear implant adjustment: (1) The default frequency allocation table for cochlear implants is based on average cochlear anatomy data. However, significant differences exist between individuals in cochlear length and electrode implantation depth, leading to a systematic deviation between the default assigned electrical stimulation frequency and the actual frequency-position mapping relationship of the patient's cochlea. Studies have shown that this mismatch can severely impair speech recognition ability, especially in noisy environments. Existing solutions rely on CT images or the experience of the audiologist. The former is costly, carries radiation risks, and is time-consuming, while the latter is highly subjective and has poor repeatability.

[0003] (2) The energy of initial consonants is concentrated in a narrow frequency band (e.g., / s / is concentrated in 4-8kHz), and their recognition is highly dependent on the electrical stimulation intensity of a specific electrode channel; the energy of final vowels is distributed in a wide frequency spectrum (e.g., / a / covers 0.5-1.5kHz), and their timbre integrity depends on multi-electrode coordination; the tone information is concentrated in the low-frequency fundamental frequency (50-500Hz). Existing tuning methods use a uniform electrical stimulation strategy for all speech signals, which leads to mutual interference of the three types of information on the electrodes.

[0004] (3) After the cochlear implant is activated, the patient's brain needs 6-12 months to relearn and decode the new frequency-position mapping relationship. Current technology cannot provide the initial "spectral balance" setting on the day of activation, resulting in up to 80% of patients complaining that the sound is "strange" or "unnatural" within the first three months after activation, and some patients give up using it as a result.

[0005] (4) The required electrical stimulation intensity for each electrode channel varies from person to person and is closely related to the precise position of the electrode in the cochlea and the survival density of the spiral ganglion at that location. Existing adjustment methods use fixed formulas based on audiograms for calculation, with an error of up to ±6 levels (relative current units), which cannot capture the user's micro-preferences for each electrode.

[0006] (5) The pitch recognition accuracy of cochlear implant recipients is on average 30-40 percentage points lower than that of people with normal hearing, which seriously affects the perception of suprasegmental information such as tone and emotion. Existing technology lacks a dedicated optimization mechanism for pitch.

[0007] (6) The dynamic range of electrical stimulation of cochlear implants is extremely narrow (usually only 10-20 dB), and patients have significantly different electrical stimulation needs at soft (such as whispering) and loud (such as shouting) levels. Existing methods are usually tuned at a single medium level, leading to the clinical problem of "soft sounds are inaudible and loud sounds are too stimulating".

[0008] (7) The speech recognition rate of cochlear implant users is significantly reduced in noisy environments (up to 20-30 percentage points) compared to hearing aid users, but the existing adjustment process lacks standardized evaluation and optimization methods for different noise types.

[0009] (8) The background noise of the circuit and the noise of the electrode-tissue interface of the cochlear implant are more significant than those of the hearing aid. Even a slight increase in the volume gain will expose the noise, which seriously affects the user experience. Summary of the Invention

[0010] To address the aforementioned technical problems, the objective of this invention is to propose an adaptive cochlear implant tuning method and system based on electrode-band mapping.

[0011] The technical solution of this invention is implemented as follows: an adaptive cochlear implant adjustment method based on electrode-frequency band mapping, comprising the following steps: S1. Construct and store a phoneme-electrode mapping library. The phoneme-electrode mapping library classifies phonemes, including at least initials, finals, and tones, and maps each phoneme to one or more electrode channels of the cochlear implant according to the acoustic characteristic frequency band of each phoneme. S2, retrieve the vowel-type test words from the phoneme-electrode mapping library, perform current intensity testing on each vowel-type test word, and obtain the optimal current value and result type of each vowel-type test word; based on the weight vector, optimal current value and result type of each vowel-type test word on the electrode channel, solve the electrode-specific reference vector; S3, on the device after applying the electrode-specific reference vector adjustment, call the initial consonant test words from the phoneme-electrode mapping library, perform current intensity test on each initial consonant test word, and obtain the optimal current value and result type of each initial consonant test word; based on the mapping relationship between each initial consonant test word and the electrode channel, obtain the electrode current demand vector; S4. Repeat step S2 under different input sound levels to obtain the electrode current demand vectors corresponding to low sound level, medium sound level and high sound level respectively. S5, based at least on the electrode-specific reference vector and the electrode current demand vector corresponding to the mid-sound level, calculate the final electrode current vector using the following formula and output it: in, This represents the electrode current demand vector corresponding to the mid-sound level. This is a reference vector specific to the electrode.

[0012] Preferably, the current intensity test adopts: The two-stage optimal current comparison search method includes: a first stage of coarse positioning with 6 levels as the step to obtain a rough optimal point; and a second stage of fine positioning with 3 levels as the step within ±6 levels of the rough optimal point to obtain a precise optimal point. Alternatively, a progress bar test mode can be used, in which the user adjusts the current intensity in real time by dragging a slider, and the system records the current value finally confirmed by the user as the optimal point.

[0013] Preferably, the final test characters include at least one of / a / , / o / , / e / , / i / , / u / , / ü / , / an / , and / ang / ; the initial test characters include at least one of / s / , / sh / , / t / , and / x / .

[0014] Preferably, the result types of the test words for vowels and test words for consonants both include "most suitable for clear hearing", "most suitable for fuzzy hearing" and "insufficient loudness", and correspond to confidence weights of 1.0, 0.8 and 0.3 respectively.

[0015] Preferably, the electrode-specific reference vector is solved using the weighted least squares method, and the calculation formula is: Among them, This is the weight matrix of the test characters in the vowel category on the electrode channel. This is the confidence weight diagonal matrix. This is the optimal current value vector for each test character with a vowel.

[0016] Preferably, the electrode current demand vector is obtained by calculating the weighted median for each electrode channel, as shown in the following formula: in, For mapping to electrodes The set of initial consonants, The optimal current value for the initial consonant test word. For confidence level weights.

[0017] Preferably, in step S2, a representative electrode testing method is used. This method is based on the aggregation characteristics of initial consonants on the electrode channel. The initial consonant with the highest demand in each electrode region is selected as the representative for testing, and the test results are applied to other initial consonants in the same electrode region.

[0018] Preferably, the different input sound levels include a low sound level of -15dB relative to the medium sound level, a medium sound level of 0dB, and a high sound level of +15dB; the high sound level test also includes an discomfort threshold recording step, in which the current value is recorded as an upper limit constraint when the user reports discomfort.

[0019] Preferably, it also includes a tone electrode optimization step S6: The tone-type test words are retrieved from the phoneme-electrode mapping library. These tone-type test words include the four tones (high level, rising, falling-rising, and falling) carried by the same syllable. The tone-type test words are mapped to the low-frequency electrode channel. A current intensity test is performed on each tone-type test word to obtain the optimal current value and result type for each tone word. The results are mapped to the low-frequency electrode channel to obtain the tone electrode demand vector. This vector is then incorporated into the calculation of the final electrode current vector in step S5. in, This represents the electrode current demand vector corresponding to the mid-sound level. This is the demand vector for tone electrodes. This is a reference vector specific to the electrode.

[0020] Preferably, the method further includes a noise environment electrode optimization step S7: Under at least two different noise scenarios, step S2 is repeated to obtain a specific electrode current demand vector for each noise scenario; a general noise electrode current prescription is obtained by solving a weighted system of equations; and this prescription is incorporated into the fusion calculation in step S5. in, This represents the electrode demand vector for mid-range acoustic phonons. This is the demand vector for tone electrodes. For the electrode prescription vector in the noisy environment, This is a reference vector specific to the electrode.

[0021] Preferably, the noise scenario includes at least a speech noise scenario and a traffic noise scenario, and the test signal-to-noise ratio is +3dB to +8dB.

[0022] Preferably, the method further includes a background noise constraint step S8: testing the user's tolerance limit for background noise in a quiet environment and using it as a hard constraint for the low-volume electrode current. The channel values ​​in the low-volume electrode current demand vector must not exceed this tolerance limit.

[0023] Preferably, the process further includes a frequency allocation table remapping step S9: Detect whether the electrode current requirement for high-frequency initial consonants exceeds a preset threshold; If the value exceeds the preset threshold, the system outputs a remapping suggestion; After adjusting the frequency allocation table, the optician re-executes steps S1 and S2 to obtain a new electrode current prescription.

[0024] Preferably, it also includes a timbre adaptation test step S10: Based on user performance data on recognizing speakers of different genders during auditory training, the average recognition rate for each timbre was calculated. If the average recognition rate of a certain timbre is significantly lower than that of another timbre, then the timbre is identified as a weak timbre. In subsequent debugging, the speaker with the weaker timbre was used as the test benchmark.

[0025] This invention also provides an adaptive cochlear implant tuning system based on electrode-band mapping, which is based on the aforementioned adaptive cochlear implant tuning method based on electrode-band mapping, and includes... A phoneme-electrode mapping library is used to store pre-built phoneme classifications and their mapping relationships with electrode channels; The current intensity test module is used to perform current intensity tests on the test word and obtain the optimal current value and result type. Electrode reference calibration engine, used to solve for electrode-specific reference vectors based on vowel test data; The acoustic mother electrode optimization engine is used to solve the electrode current demand vector based on acoustic mother test data. The multi-input level test module is used to perform tests at different input sound levels. The prescription fusion optimization engine is used to calculate the final electrode current vector based on the electrode-specific reference vector and the electrode current demand vector; The debugging report generation module is used to generate a structured report containing recommended current values ​​for each electrode channel; The data export interface is used to export optimization parameters via QR code, file, or cloud. It also includes at least one of the following modules: The background noise test module is used to test the user's upper limit of tolerance to background noise in a quiet environment; The noise scene testing module is used to perform electrode optimization tests under different noise scenes; A tone electrode optimization engine is used to generate tone electrode demand vectors based on tone test data. The frequency allocation table remapping engine is used to detect abnormal high-frequency initial consonant demand and guide frequency allocation table remapping. Clinical adaptation layer engine for providing representative electrode testing methods and progress bar testing modes; The timbre adaptive module is used to identify weak timbres based on training data and adjust the test benchmark. The data interface module is used for data exchange with external auditory training systems.

[0026] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements an adaptive cochlear implant tuning method based on electrode-band mapping.

[0027] Due to the application of the above technical solution, the present invention has the following advantages compared with the prior art: 1. This invention generates electrode-specific reference vectors through actual measurements, eliminating the frequency-position systematic deviation caused by individual cochlear anatomy differences without the need for CT images, significantly improving the accuracy and repeatability of tuning; at the same time, by mapping consonants, vowels and tones to electrodes according to acoustic frequency bands and optimizing them independently, it avoids mutual interference of the three types of speech information on the electrodes, fundamentally improving speech intelligibility and naturalness.

[0028] 2. This invention establishes an initial spectral balance setting through vowel reference calibration, enabling patients to adapt to the sound of the cochlear implant in a shorter time after activation, significantly shortening the postoperative adaptation period.

[0029] 3. After applying this invention, the user's initial consonant recognition rate is significantly improved, especially the recognition of high-frequency initial consonants (such as / s / and / sh / ); at the same time, the accuracy of tone perception is also substantially improved, thereby enhancing the user's sense of language and comprehension of Chinese.

[0030] 4. This invention establishes an independent optimization framework for three levels of input: soft, medium, and loud. It simultaneously introduces a loud discomfort threshold and background noise constraints, which solves the problem of "not being able to hear soft sounds and being too stimulating to hear loud sounds" caused by the narrow dynamic range of cochlear implants. It can adapt to all kinds of everyday sounds. At the same time, it independently tests and optimizes typical scenarios such as speech noise and traffic noise, and generates a universal noise electrode prescription, which improves the speech recognition rate by 10-20 percentage points in noisy environments and significantly improves the communication experience in noisy environments.

[0031] 5. This invention can monitor abnormal high-frequency consonant current demand in real time, automatically trigger frequency allocation table remapping, avoid high-frequency speech information falling into the insensitive electrode area of ​​the cochlea, achieve frequency shift equivalent optimization, and further enhance the high-frequency speech perception capability.

[0032] 6. This invention uses a representative electrode testing method and a progress bar interaction mode, which reduces the entire debugging time from the traditional 30-40 minutes to within 15 minutes, making it suitable for children and people with limited attention spans, and greatly improving the efficiency and accessibility of clinical implementation. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the content of the embodiments of the present invention and these drawings without creative effort.

[0034] Appendix Figure 1 This is an overall flowchart of an adaptive cochlear implant tuning method based on electrode-band mapping as described in this invention; Appendix Figure 2 This is a schematic diagram illustrating the construction of the phoneme-electrode mapping library described in this invention; Appendix Figure 3 This is a flowchart of the electrode reference calibration and phonon electrode optimization described in this invention; Appendix Figure 4 This is a schematic diagram of the optimized tone-modulation electrode described in this invention; Appendix Figure 5 This is a schematic diagram of the multi-input level dynamic range optimization described in this invention; Appendix Figure 6 This is a flowchart of the frequency allocation table remapping process described in this invention; Appendix Figure 7 This is a schematic diagram illustrating the principle of the representative electrode testing method described in this invention; Appendix Figure 8 This is an overall architecture diagram of an adaptive cochlear implant tuning based on electrode-band mapping as described in this invention. Detailed Implementation

[0035] To make the technical problems solved by the present invention, the technical solutions adopted, and the technical effects achieved clearer, the technical solutions of the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "straight," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance. The terms "first position" and "second position" refer to two different positions.

[0037] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to fixed connections or detachable connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; and internal connections between two components. Those skilled in the art can understand the specific meaning of these terms in this invention based on the specific circumstances.

[0038] As attached Figure 1-3 As shown, this invention provides an adaptive cochlear implant tuning method and system based on electrode-band mapping. Example

[0039] An adaptive cochlear implant adjustment method based on electrode-band mapping includes the following steps: S1. Construct and store a phoneme-electrode mapping library. The phoneme-electrode mapping library classifies phonemes, including at least initials, finals, and tones, and maps each phoneme to one or more electrode channels of the cochlea according to its acoustic characteristic frequency band.

[0040] Commercially available cochlear implants typically have 12-22 electrode channels, each corresponding to a frequency range. This invention, based on CT images of cochlear implant recipients or standard cochlear models, pre-establishes a phoneme-electrode mapping library before any user undergoes adjustment. It precisely binds Chinese initials, finals, and tones to electrode channels according to acoustic characteristic frequency bands, providing a unified, quantitative, and reusable mapping basis for subsequent calibration and optimization.

[0041] In this embodiment, a cochlear implant with 22 electrode channels is used. The correspondence between electrode channels and frequency bands is shown in Table 1 below.

[0042] Table 1; Partial Mapping Table of 22-Channel Cochlear Implant Electrodes and Frequency Bands Based on Table 1 above, a consonant-electrode mapping table and a vowel electrode weight vector table are constructed. In this embodiment, the vowel test characters must include at least one of / a / , / o / , / e / , / i / , / u / , / ü / , / an / , / ang / ; the consonant test characters must include at least one of / s / , / sh / , / t / , / x / .

[0043] Taking the core vowel / a / as an example, its energy is concentrated in the E3-E9 electrode channels, and the weight vector is: [E3:0.05,E4:0.10,E5:0.20,E6:0.25,E7:0.20,E8:0.10,E9:0.10].

[0044] Table 2: High-Frequency Initial Consonant-Electrode Mapping Table Table: Weight Distribution of Core Vowel Electrodes S2, Electrode reference calibration, calls the vowel class test words from the phoneme-electrode mapping library, performs current intensity test on each vowel class test word, and obtains the optimal current value and result type of each vowel class test word; based on the weight vector, optimal current value and result type of each vowel class test word on the electrode channel, solves the electrode-specific reference vector.

[0045] After the user powers on, the system uses the default frequency allocation table and calls up 8 test words of the vowel category: / a / (ah), / o / (oh), / e / (e), / i / (one), / u / (house), / ü / (silt), / an / (peace), / ang / (ang).

[0046] In this embodiment, the current intensity test is performed on the above 8 vowel-type test characters in the following two ways: Method 1: Two-stage optimal current comparison search method, which includes: In the first stage, coarse positioning is performed in 6-level increments. Versions with current levels of -6, 0, +6, and +12 are played in sequence, and the user selects "which is more comfortable". Assuming that the user feedback is that +6 level is better than 0 level, and +12 level is better than +6 level, but +12 level is slightly noisy, then the approximate optimal point is +6 level, and the approximate optimal point is obtained. In the second stage, fine positioning is performed within a rough optimal range of ±6 levels, with 3 levels in increments. Versions at +3, +6, and +9 levels are played. After careful comparison, the user selects +6 level as the optimal point and records the best current value and result type for that character.

[0047] Method 2: Progress bar test mode, where users can adjust the current intensity in real time by dragging the slider, and the system records the current value finally confirmed by the user as the optimal point.

[0048] The specific operating procedure is as follows: The device screen displays the current test word, such as "silk". Below it is a horizontal slider. The left end indicates "too light / too low current" and the right end indicates "too loud / too high current". The current current level (range -6 to +12) is displayed in real time next to the slider.

[0049] First, the system plays the "silk" sound at the default setting (0). Next, the user drags the slider to the left or right to adjust, and the system plays the version corresponding to the current level in real time; After finding the "most comfortable" position, the user clicks the "confirm" button; Finally, the system records the current current value and automatically proceeds to the next test word.

[0050] By adopting a progress bar testing mode, users do not need to make abstract comparisons of "which is more comfortable". They only need to find their own preference by continuously adjusting. The testing time is shortened by 40% compared with the two-stage search method, making it more suitable for children and first-time users.

[0051] Based on user feedback, the results were divided into three categories, as shown in Table 4 below.

[0052] Table 4: Classification of Electrode Test Result Types After collecting data for 8 vowel characters, the following equation was constructed: in, The optimal current value vector (8×1) for each vowel character. The matrix is ​​a vowel × electrode weight matrix (8×22). The electrode-specific reference vector (22×1) to be solved. This is the error term.

[0053] Then, the weighted least squares method is used to solve the problem. : in, This is an 8×8 diagonal matrix of confidence weights.

[0054] The electrode-specific reference vector obtained by solving This eliminates the frequency-position mismatch in the electrode-specific reference vector, which represents the baseline electrical stimulation intensity required for each electrode channel to provide a "spectrally balanced" auditory experience for the user across all electrode channels under the current frequency allocation table.

[0055] S3, Initial Consonant Electrode Optimization: On devices with adjusted electrode-specific reference vectors, initial consonant test words are retrieved from the phoneme-electrode mapping library. Current intensity tests are performed on each initial consonant test word to obtain the optimal current value and result type for each initial consonant test word. Based on the mapping relationship between each initial consonant test word and the electrode channel, the electrode current requirement vector is obtained.

[0056] After applying the electrode-specific reference vector to the device, the mother electrode is optimized.

[0057] Please see the appendix Figure 7 In this step, a representative electrode testing method is used to shorten the testing time. This method selects representative initials based on electrode aggregation. Specifically: The principle of the representative electrode test method is as follows: based on the aggregation characteristics of initial consonants on the electrode channel, the demand trends of initial consonants in the same electrode area are highly correlated. Therefore, by testing only the representative initial consonant with the "maximum" demand in each area, the demand of the entire area can be inferred.

[0058] The operation process is as follows: ① The system determines the representative initial consonant list based on the electrode region division: / t / (E14–E19 region), / sh / (E16–E18 region), / s / (E18–E22 region). ② The user performs a complete current intensity test on / t / , / sh / , and / s / respectively; ③ Extend the test results of / t / to initials in the same region such as / p / , / k / , and / g / ; ④ Randomly select a non-representative initial consonant (such as / p / ) for verification testing. If the deviation is within ±2 levels, the promotion is confirmed to be effective.

[0059] The validation data is as follows: Pre-test data from 50 cochlear implant users showed that the correlation coefficient between the representative electrode method and the measured values ​​of non-representative initial consonants reached over 0.89, with an average deviation of ±1.2 levels, which is clinically acceptable.

[0060] This yields the following representative initials and other initials in the same region: Region E14–E19: Represents the initial consonant / t / , and other initial consonants in the same region: / p / , / k / , / g / , / h / ; E16–E18 region: represents the initial consonant / sh / , and other initial consonants in the same region are / ch / , / zh / , and / j / ; Region E18–E22: Represents the initial consonant / s / , and other initial consonants in the same region: / x / , / z / , / c / , / q / .

[0061] For each representative initial consonant, perform the above two-stage optimal current comparison search or progress bar test to obtain the optimal current value and result type. For example, the optimal current value for / s / is +10, and the result type is "optimal for clear hearing". According to Table 2 above, / s / is mapped to five electrode channels E18–E22, and the contribution weight of each channel is 0.2. Therefore, the demand for +10 is allocated to these five electrode channels according to the weight, and one test data is recorded for each channel.

[0062] For each electrode channel Collect all initial current values ​​mapped to this location and their confidence weights, and calculate the weighted median as the initial current demand value for this electrode: in For mapping to electrodes The set of initial consonants, The optimal current value for the initial consonant test word. For confidence level weights.

[0063] Please see the appendix Figure 5S4, Multi-input level dynamic range optimization: Repeat step S2 under different input sound levels to obtain the electrode current demand vectors corresponding to the low sound level, medium sound level, and high sound level, respectively.

[0064] In this embodiment, different input sound levels include a low sound level of -15dB relative to the medium sound level, a medium sound level of 0dB, and a high sound level of +15dB; the high sound level test also includes an discomfort threshold recording step, in which the current value is recorded as an upper limit constraint when the user reports discomfort.

[0065] It should be noted that the relative mid-range level mentioned here is 65dB. The actual standard is based on the most comfortable and clear volume. The soft, mid, and loud levels are all increased or decreased in 15dB increments. Therefore, the soft, mid, and loud levels are all relative. For example, when the relative mid-range level is 65dB, the soft level is reduced by 15dB, which is actually 50dB. The mid-range and loud levels are 65dB and 80dB, respectively.

[0066] Step S2 was repeated at relative mid-range sound levels of -15dB (soft sound, typical scenario: whispering), 0dB (medium sound, everyday conversation), and +15dB (loud sound, shouting) to obtain three independent electrode current demand vectors, and a three-input-level test framework table was constructed: : Electrode requirements for low-volume sounding consonants; : Mid-range acoustic electrode requirements; : Requirements for loud-volume acoustic electrode.

[0067] Table 5: Three-Input Level Test Framework If a user reports "clear sound but uncomfortable sound" during the loudness level test, the system will automatically record this current value as an upper limit constraint, and the loudness level must not exceed this value in subsequent tests.

[0068] Please see the appendix Figure 4 Furthermore, the tuning method of this application also includes a tone electrode optimization step S6: The tone-type test words are retrieved from the phoneme-electrode mapping library. The tone-type test words include the four tones (high level, rising, falling-rising, and falling) carried by the same syllable. The tone-type test words are mapped to the low-frequency electrode channel. A current intensity test is performed on each tone-type test word to obtain the optimal current value and result type for each tone word. The results are mapped to the low-frequency electrode channel to obtain the tone electrode demand vector. In step S5, this vector is incorporated into the calculation of the final electrode current vector.

[0069] In this embodiment, tone test characters are called from the phoneme-electrode mapping library: / mā / (妈, high level tone), / má / (麻, rising tone), / mǎ / (马, falling-rising tone), / mà / (骂, falling tone). The main mapping electrodes for each character are E1-E5 (low-frequency electrode channels).

[0070] A two-stage optimal current comparison search is performed for each tone character to obtain the optimal current value. For example: · / mā / (high level tone): optimal current value +3 levels; · / má / (麻, rising tone, mid-rising tone): optimal current value +6 levels; · / mǎ / (马, falling-rising tone, fall-rise tone): optimal current value +9 levels; / mà / (骂, falling tone, full-falling tone): optimal current value +6 levels.

[0071] For the E1-E5 electrodes, the data of the four tone characters are aggregated, and the weighted median is calculated to obtain the tone electrode demand vector .

[0072] Table 6: Tone-electrode mapping table Further, the debugging method of the present application further includes a background noise constraint step S8: testing the user's tolerance upper limit for background noise in a quiet environment, and taking it as a hard constraint for the low-volume electrode current, wherein each channel value in the low-volume electrode current demand vector shall not exceed the tolerance upper limit: That is:

[0073] Wherein, is the electrode channel.

[0074] In this embodiment, the quiet environment is an environment with ambient noise less than 20dB SPL. In this environment, the user wears the cochlear implant speech processor without playing the test sound, and the user slowly increases the low-volume electrode current. The instruction is: "Please adjust to the position where you can just hear the device's own noise but it is not annoying".

[0075] The current value of each electrode channel at this time is recorded as , and the final low-volume electrode current demand shall not be higher than , if it originally exceeds, it will be forcibly truncated: That is:

[0076] S5, at least based on the electrode-specific reference vector and the electrode current demand vector corresponding to the medium volume level, calculate and output the final electrode current vector.

[0077] The calculation formula of the final electrode current vector is: in, This represents the electrode demand vector for mid-range acoustic phonons. This is the demand vector for tone electrodes. For the electrode prescription vector in the noisy environment, This is a reference vector specific to the electrode.

[0078] If tone optimization and noise optimization are not included, then it is simplified to: In this embodiment, the final prescription calculation for the user's sound level scenario is shown in Table 7 below (taking some electrodes as an example).

[0079] Table 7: Examples of Prescription Fusion Please see the appendix Figure 6 As a preferred embodiment, the debugging method of this application further includes a frequency allocation table remapping step S9: Detect whether the electrode current requirement for high-frequency initial consonants exceeds a preset threshold; If the value exceeds the preset threshold, the system outputs a remapping suggestion; After adjusting the frequency allocation table, the optician re-executes steps S1 and S2 to obtain a new electrode current prescription.

[0080] The triggering condition is as follows: In the initial consonant electrode optimization, the user achieves the optimal current values ​​of +14 and +12 for the high-frequency initial consonants / s / and / sh / respectively, which are significantly higher than other initial consonants (average +6), and the result type is "optimal fuzzy" (can be heard but cannot be accurately identified).

[0081] After the above conditions are triggered, the system responds by outputting the message "An abnormal demand for high-frequency initials has been detected. It is recommended to remap the frequency allocation table. Possible reason: High-frequency information falls in the insensitive electrode area. Please have the audiologist evaluate and adjust the frequency allocation table."

[0082] After receiving the system's response, the audiologist, in the adjustment software, shifted the original 4–8kHz frequency range mapped to E18–E22 towards a direction closer to the base of the cochlea. For example, the frequency starting point was adjusted from 4kHz to 3kHz, so that the 4–8kHz characteristic frequency band of / s / falls within the E17–E22 range.

[0083] After the optometrist remapping: Step S1 is repeated, and the user retests the 8 core vowels to obtain new results. .

[0084] Step S2 is repeated, and the user retests the representative initial consonant to obtain a new result. .

[0085] The system outputs a new prescription, and the user feedback indicates that the optimal current value for / s / has been reduced to +8, and the result type has been upgraded to "Most Suitable for Clear Hearing".

[0086] Furthermore, in a preferred embodiment, the debugging method of this application further includes a timbre adaptive testing step S10: Based on user performance data in auditory training on recognizing speakers of different genders, the average recognition rate for each timbre is calculated. If the average recognition rate of a certain timbre is significantly lower than that of another timbre, then that timbre is identified as a weak timbre. In subsequent adjustments, the speaker with the weak timbre is used as the test benchmark. Example

[0087] Please see the appendix Figure 8 This invention provides an adaptive cochlear implant adjustment system based on electrode-band mapping, comprising: A phoneme-electrode mapping library is used to store pre-built phoneme classifications and their mapping relationships with electrode channels.

[0088] The current intensity test module is used to perform current intensity tests on the test word and obtain the optimal current value and result type.

[0089] Electrode reference calibration engine, used to solve for electrode-specific reference vectors based on vowel test data.

[0090] The acoustic mother electrode optimization engine is used to solve the electrode current demand vector based on acoustic mother test data.

[0091] The multi-input level test module is used to perform tests at different input sound levels.

[0092] The prescription fusion optimization engine is used to calculate the final electrode current vector based on the electrode-specific reference vector and the electrode current demand vector.

[0093] The debugging report generation module is used to generate a structured report containing recommended current values ​​for each electrode channel.

[0094] In this embodiment, as shown in the table below, the debugging report is a structured report of the recommended current values ​​for each electrode channel.

[0095] Table 8: Example of a Cochlear Implant Optimization Recommendation Report The data export interface is used to export optimization parameters via QR code, file, or cloud.

[0096] It also includes at least one of the following modules: The background noise test module is used to test the user's upper limit of tolerance to background noise in a quiet environment.

[0097] The noise scene testing module is used to perform electrode optimization tests under different noise scenarios.

[0098] A tone electrode optimization engine is used to generate tone electrode demand vectors based on tone test data.

[0099] The frequency allocation table remapping engine is used to detect abnormal high-frequency consonant demand and guide the frequency allocation table remapping.

[0100] Clinical adaptation layer engine for providing representative electrode testing methods and progress bar testing modes.

[0101] The timbre adaptive module is used to identify weak timbres based on training data and adjust the test benchmark.

[0102] The data interface module is used for data exchange with external auditory training systems.

[0103] In summary, the operation process of the adaptive cochlear implant tuning system based on electrode-band mapping in this application is as follows: First, the phoneme-electrode mapping library serves as the underlying support, pre-storing phoneme classifications, electrode mapping relationships, test character libraries, and weight parameters to provide standard data for all modules. The clinical adaptation layer engine is embedded in the process, providing representative electrode testing methods and progress bar testing modes to adapt to clinical scenarios and shorten testing time, supporting the efficient execution of subsequent testing steps.

[0104] Next, the current intensity optimization test module serves as a unified data acquisition entry point, receiving test words for vowels, initials, tones, and noise scenarios from the mapping library. Combined with the test modes of the clinical adaptation layer, it completes the current intensity test for each test word, simultaneously acquiring the optimal current value, result type, and confidence weight, providing raw test data for subsequent calculation modules.

[0105] The process then proceeds to the hierarchical calculation stage: The electrode reference calibration engine prioritizes calling the vowel test data and the mapping library weight matrix, and solves the electrode-specific reference vector using the weighted least squares method to eliminate frequency-position mismatch caused by individual cochlear anatomical differences; After the reference vector is output, it is simultaneously supplied to the initial consonant electrode optimization engine, which conducts tests based on the calibrated reference. Combining the initial consonant test data, mapping library relationships, and representative electrode rules of the clinical adaptation layer, the initial consonant electrode current demand vector is generated by weighted median fusion; The multi-input level test module coordinates the soft, medium, and loud sound levels, driving the initial consonant electrode optimization engine to repeat tests at different sound levels, outputting the current demand vectors corresponding to the three sound levels, and recording the loud sound level discomfort threshold as a constraint basis.

[0106] While performing layered calculations, the tone electrode optimization engine and the noise scene testing module conduct specialized optimizations in parallel: the tone optimization engine calls tone test data to generate tone requirement vectors for low-frequency electrodes; the noise testing module completes initial consonant tests under different noise scenarios, outputs noise-specific electrode vectors, and merges them into a general noise prescription; the background noise testing module simultaneously collects the noise tolerance upper limit in a quiet environment, providing hard constraints for subsequent prescriptions.

[0107] All calculation and optimization results are aggregated into the prescription fusion optimization engine. This engine substitutes the fusion formula to complete the final electrode current vector calculation, and after superimposing various constraint corrections, outputs a standardized electrode current prescription.

[0108] Finally, in the output stage, the debugging report generation module receives the final prescription and intermediate data from the entire process to generate a structured debugging report; the data export interface supports multiple ways to export optimization parameters; at the same time, the frequency allocation table remapping engine monitors high-frequency consonant demand anomalies in real time, and after triggering, it links the calibration engine and consonant optimization engine to re-execute the calibration and optimization process; the timbre adaptation module identifies weak timbres based on external training data and feeds them back to the testing module to adjust the subsequent test benchmarks, forming a continuously iterative closed-loop adaptation mechanism.

[0109] A computer-readable storage medium having a computer program stored thereon, characterized in that, when the program is executed by a processor, it implements an adaptive cochlear implant tuning method based on electrode-band mapping.

[0110] The above embodiments are only for illustrating the technical concept and features of the present invention. Their purpose is to enable those skilled in the art to understand the content of the present invention and implement it. They should not be used to limit the scope of protection of the present invention. All equivalent changes or modifications made in accordance with the spirit and essence of the present invention should be covered within the scope of protection of the present invention.

Claims

1. An adaptive cochlear implant adjustment method based on electrode-frequency band mapping, characterized in that: Includes the following steps: S1. Construct and store a phoneme-electrode mapping library. The phoneme-electrode mapping library classifies phonemes, including at least initials, finals, and tones, and maps each phoneme to one or more electrode channels of the cochlear implant according to the acoustic characteristic frequency band of each phoneme. S2, retrieve the vowel-type test characters from the phoneme-electrode mapping library, perform current intensity testing on each vowel-type test character, and obtain the optimal current value and result type for each vowel-type test character; Based on the weight vector, optimal current value and result type of each vowel test character on the electrode channel, the electrode-specific reference vector is solved. S3, on the device after applying the electrode-specific reference vector adjustment, call the initial consonant test words from the phoneme-electrode mapping library, perform current intensity test on each initial consonant test word, and obtain the optimal current value and result type of each initial consonant test word; Based on the mapping relationship between test words of each initial consonant class and electrode channels, the electrode current demand vector is obtained; S4. Repeat step S2 under different input sound levels to obtain the electrode current demand vectors corresponding to low sound level, medium sound level and high sound level respectively. S5, based at least on the electrode-specific reference vector and the electrode current demand vector corresponding to the mid-sound level, calculate the final electrode current vector using the following formula and output it: in, This represents the electrode current demand vector corresponding to the mid-sound level. This is a reference vector specific to the electrode.

2. The adaptive cochlear implant adjustment method based on electrode-band mapping according to claim 1, characterized in that: The current intensity test is performed using: The two-stage optimal current comparison search method includes: a first stage of coarse positioning with 6 levels as the step to obtain a rough optimal point; and a second stage of fine positioning with 3 levels as the step within ±6 levels of the rough optimal point to obtain a precise optimal point. Alternatively, a progress bar test mode can be used, in which the user adjusts the current intensity in real time by dragging a slider, and the system records the current value finally confirmed by the user as the optimal point.

3. The adaptive cochlear implant tuning method based on electrode-frequency band mapping according to claim 1, characterized in that: The test words for the vowel category must include at least one of / a / , / o / , / e / , / i / , / u / , / ü / , / an / , / ang / ; the test words for the consonant category must include at least one of / s / , / sh / , / t / , / x / .

4. The adaptive cochlear implant adjustment method based on electrode-band mapping according to claim 1, characterized in that: The result types for both the vowel and initial test characters include "most suitable for clear hearing", "most suitable for fuzzy hearing" and "insufficient loudness", with corresponding confidence weights of 1.0, 0.8 and 0.3, respectively.

5. The adaptive cochlear implant adjustment method based on electrode-band mapping according to claim 4, characterized in that: The electrode-specific reference vector is obtained by weighted least squares method, and the calculation formula is as follows: Among them, This is the weight matrix of the test characters in the vowel category on the electrode channel. This is the confidence weight diagonal matrix. This is the optimal current value vector for each test character with a vowel.

6. The adaptive cochlear implant adjustment method based on electrode-band mapping according to claim 4, characterized in that: The electrode current demand vector is obtained by calculating the weighted median for each electrode channel, as shown in the following formula: in, For mapping to electrodes The set of initial consonants, The optimal current value for the initial consonant test word. For confidence level weights.

7. The adaptive cochlear implant adjustment method based on electrode-band mapping according to claim 1, characterized in that: In step S2, a representative electrode test method is used. This method is based on the aggregation characteristics of initial consonants on the electrode channel. The initial consonant with the highest demand in each electrode region is selected as the representative for testing, and the test results are applied to other initial consonants in the same electrode region.

8. The adaptive cochlear implant tuning method based on electrode-band mapping according to claim 1, characterized in that: The different input sound levels include a low sound level of -15dB relative to the medium sound level, a medium sound level of 0dB, and a high sound level of +15dB; the high sound level test also includes an discomfort threshold recording step, in which the current value is recorded as an upper limit constraint when the user reports discomfort.

9. The adaptive cochlear implant adjustment method based on electrode-band mapping according to claim 1, characterized in that: It also includes tone electrode optimization step S6: Call the tone-type test words from the phoneme-electrode mapping library. The tone-type test words include the four tones carried by the same syllable: first tone, second tone, third tone, and fourth tone. Map the tone-type test words to the low-frequency electrode channel. A current intensity test is performed on each tone-type test word to obtain the optimal current value and result type for each tone word. The results are then mapped to the low-frequency electrode channel to obtain the tone electrode demand vector. This vector is then incorporated into the calculation of the final electrode current vector in step S5. in, This represents the electrode current demand vector corresponding to the mid-sound level. This is the demand vector for tone electrodes. This is a reference vector specific to the electrode.

10. The adaptive cochlear implant tuning method based on electrode-band mapping according to claim 9, characterized in that: The process also includes a noise environment electrode optimization step S7: Under at least two different noise scenarios, step S2 is repeated to obtain the specific electrode current demand vector for each noise scenario. A weighted solution of simultaneous equations is then used to obtain a general noise electrode current prescription; this prescription is then incorporated into the fusion calculation in step S5. in, This represents the electrode demand vector for mid-range acoustic phonons. This is the demand vector for tone electrodes. For the electrode prescription vector in the noisy environment, This is a reference vector specific to the electrode.

11. The adaptive cochlear implant adjustment method based on electrode-band mapping according to claim 1, characterized in that: It also includes the background noise constraint step S8: test the user's tolerance limit for background noise in a quiet environment, and use it as a hard constraint for the low-volume electrode current. The channel values ​​in the low-volume electrode current demand vector must not exceed this tolerance limit.

12. The adaptive cochlear implant tuning method based on electrode-band mapping according to claim 1, characterized in that: It also includes the frequency allocation table remapping step S9: Detect whether the electrode current requirement for high-frequency initial consonants exceeds a preset threshold; If the value exceeds the preset threshold, the system outputs a remapping suggestion; After adjusting the frequency allocation table, the optician re-executes steps S1 and S2 to obtain a new electrode current prescription.

13. The adaptive cochlear implant tuning method based on electrode-band mapping according to claim 1, characterized in that: It also includes tone adaptation test step S10: Based on user performance data on recognizing speakers of different genders during auditory training, the average recognition rate for each timbre was calculated. If the average recognition rate of a certain timbre is significantly lower than that of another timbre, then the timbre is identified as a weak timbre. In subsequent debugging, the speaker with the weaker timbre was used as the test benchmark.

14. An adaptive cochlear implant tuning system based on electrode-band mapping, used to implement the adaptive cochlear implant tuning method based on electrode-band mapping as described in any one of claims 1-13, characterized in that: Include A phoneme-electrode mapping library is used to store pre-built phoneme classifications and their mapping relationships with electrode channels; The current intensity test module is used to perform current intensity tests on the test word and obtain the optimal current value and result type. Electrode reference calibration engine, used to solve for electrode-specific reference vectors based on vowel test data; The acoustic mother electrode optimization engine is used to solve the electrode current demand vector based on acoustic mother test data. The multi-input level test module is used to perform tests at different input sound levels. The prescription fusion optimization engine is used to calculate the final electrode current vector based on the electrode-specific reference vector and the electrode current demand vector; The debugging report generation module is used to generate a structured report containing recommended current values ​​for each electrode channel; The data export interface is used to export optimization parameters via QR code, file, or cloud. It also includes at least one of the following modules: The background noise test module is used to test the user's upper limit of tolerance to background noise in a quiet environment; The noise scene testing module is used to perform electrode optimization tests under different noise scenes; A tone electrode optimization engine is used to generate tone electrode demand vectors based on tone test data. The frequency allocation table remapping engine is used to detect abnormal high-frequency initial consonant demand and guide frequency allocation table remapping. Clinical adaptation layer engine for providing representative electrode testing methods and progress bar testing modes; The timbre adaptive module is used to identify weak timbres based on training data and adjust the test benchmark. The data interface module is used for data exchange with external auditory training systems.

15. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the adaptive cochlear implant tuning method based on electrode-band mapping as described in claim 1 or 13.

Images