Parkinson's disease voice intelligent auxiliary analysis optimization method based on images

Abstract

The invention provides an intelligent Parkinson's disease voice auxiliary analysis optimization method based on an image. The method comprises the following steps: voice recording: recording a voice sample of a Parkinson's disease patient through an acoustic recording pen; voice processing: using software to carry out denoising and standardization processing on the recorded voice samples, and carrying out visual analysis on time domain, frequency domain and loudness characteristics of the voice samples; acoustic parameter extraction: importing the processed voice sample into MATLAB, and calculating related acoustic parameters of the sample; and database establishment: constructing a voice database of the Parkinson's disease patient based on the acoustic parameters, and classifying and labeling the data.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent auxiliary analysis of Parkinson's disease voice, and in particular to an image-based optimization method for intelligent auxiliary analysis of Parkinson's disease voice. Background Art

[0002] Parkinson's disease is a common neurodegenerative disease, and its dysarthria is one of the important symptoms that affects patients' daily lives. According to the latest research, the incidence of dysarthria in Parkinson's patients is very high, and about 70%-90% of patients will experience dysarthria in the early stages of the disease or even in the prodromal stage. Dysarthria is mainly manifested as low-dynamic dysarthria, including coarse and hoarse voice, monotonous tone, disordered speech speed, imprecise consonant pronunciation, and increased unconscious pauses.

[0003] In recent years, the development of artificial intelligence and machine learning technology has gradually developed data-driven flow prediction models, which significantly reduce the calculation time while ensuring the accuracy of the calculation and are suitable for clinical applications. However, there are still few studies on non-invasive auxiliary diagnosis and treatment of dysarthria.

[0004] At present, the common diagnosis and treatment methods of Parkinson's disease mainly include:

[0005] Clinical diagnosis: based on special clinical features found in medical history and neurological examination, such as bradykinesia, resting tremor, muscle rigidity, etc. The advantage is that it directly reflects the disease status, but the disadvantage is that early symptoms may not be obvious and easy to miss the diagnosis.

[0006] Response to dopaminergic drug therapy: Parkinson's disease has a clear, significant and sustained response to dopaminergic drugs (such as levodopa), which can serve as a diagnostic support feature.

[0007] Unified Parkinson's Disease Rating Scale: A means of assessing the condition, but Parkinson's disease syndrome cannot be defined simply by the scores of each item in the scale.

[0008] Imaging tests such as single-photon emission computed tomography (SPECT) and positron emission tomography can detect brain abnormalities but are mainly used in research settings.

[0009] Biomarkers such as hyposmia and cardiac sympathetic denervation demonstrated by cardiac metaiodobenzylguanidine scintigraphy can help differentiate Parkinson's disease from atypical Parkinson's syndromes.

[0010] Genetic testing: About 10% to 25% of people with Parkinson's disease have a family history, and genetic testing can help with diagnosis.

[0011] Deep brain stimulation: For patients who do not respond to medication or whose side effects are intolerable, surgery may be considered.

[0012] However, in clinical diagnosis, it is easy to misdiagnose because early symptoms may not be obvious; while other forms of Parkinson's syndrome respond poorly to dopaminergic drug treatment;

[0013] When treating with the Unified Parkinson's Disease Assessment Scale, the diagnosis cannot rely solely on the scale score; when treating with imaging examinations, the cost is high and it cannot distinguish Parkinson's disease from other Parkinson's disease syndromes; among them, biomarker treatment has not yet been widely used in clinical practice; genetic testing currently only has diagnostic value for some hereditary Parkinson's disease; when treating with deep brain stimulation, the surgical risks and costs are relatively high. Summary of the invention

[0014] In view of the deficiencies in the prior art, the present invention provides an image-based intelligent auxiliary analysis and optimization method for Parkinson's disease voice. By adopting advanced signal processing technology and adaptive filtering algorithm, noise interference is effectively suppressed, thereby improving the accuracy of bowel sound feature parameter extraction and ensuring the reliability of clinical diagnosis.

[0015] In order to achieve the above object, the present invention adopts the following technical solutions:

[0016] An image-based intelligent auxiliary analysis and optimization method for Parkinson's disease voice, comprising the following steps:

[0017] Voice recording: Voice samples of Parkinson's disease patients were recorded using an acoustic recording pen;

[0018] Speech processing: Use software to denoise and normalize the recorded speech samples, and perform visual analysis of their time domain, frequency domain, and loudness characteristics;

[0019] Acoustic parameter extraction: import the processed speech samples into MATLAB and calculate the relevant acoustic parameters of the samples;

[0020] Database establishment: Based on the above acoustic parameters, a speech database of Parkinson's disease patients was constructed, and the data was classified and labeled.

[0021] Preferably, in the speech recording step, the speech samples include vowels / a / , / o / and / e / , each participant forms 12 monosyllabic samples, 8 disyllabic samples and 6 polysyllabic samples, and the acoustic recording pen is held by the participant at a distance of 25 to 35 cm from his lips.

[0022] Preferably, in the voice recording step, the participant maintains a habitual pitch and loudness in a relaxed state to ensure the naturalness and consistency of the voice sample.

[0023] Preferably, the speech processing step includes using Audacity software to perform denoising and standardization processing on the speech sample to improve the accuracy of subsequent acoustic parameter extraction.

[0024] Preferably, in the acoustic parameter extraction step, the denoised and standardized speech samples are analyzed using MATLAB software, and the extracted acoustic parameters include average fundamental frequency, maximum fundamental frequency, jitter frequency, signal-to-noise ratio and continuous pronunciation time.

[0025] Preferably, in the database establishment step, the constructed Parkinson's disease patient speech database contains classification information of speech samples, and realizes effective management and application of different types of speech samples through data annotation.

[0026] An image-based intelligent auxiliary analysis device for Parkinson's disease voice, comprising:

[0027] A voice recording module, used to record voice samples of Parkinson's disease patients through an acoustic recording pen;

[0028] The speech processing module is used to use Audacity software to denoise and standardize the recorded speech samples, and to perform visual analysis of their time domain, frequency domain and loudness characteristics;

[0029] The acoustic parameter extraction module is used to import the processed speech samples into MATLAB and calculate the relevant acoustic parameters of the samples;

[0030] The database building module is used to build a speech database of Parkinson's disease patients based on the above acoustic parameters.

[0031] Preferably, the voice recording module includes an acoustic recording system capable of recording high-quality voice samples when the participant is 25 to 35 cm away from the acoustic recording pen.

[0032] Preferably, the database establishment module is further configured to classify and label samples in the speech database.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] 1. The present invention realizes rapid and accurate diagnosis of Parkinson's disease through intelligent auxiliary analysis, thereby improving the efficiency and accuracy of diagnosis and treatment.

[0035] 2. The present invention can reduce patient discomfort and provide a more comfortable diagnosis and treatment experience through non-invasive examination.

[0036] 3. The present invention can provide a comprehensive evaluation through dynamic evaluation, making up for the shortcomings of two-dimensional evaluation.

[0037] 4. The present invention can provide personalized diagnosis and treatment plans based on the patient's specific condition, and has important theoretical research significance and social application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0039] Figure 1 It is a schematic diagram of the overall process of the present invention. DETAILED DESCRIPTION

[0040] The present invention is further described in detail below with reference to the accompanying drawings.

[0041] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art can think of other obvious variations. The basic principles of the present invention defined in the following description can be used for other embodiments, variations, improvements, equivalents, and other technical solutions that do not deviate from the spirit and scope of the present invention.

[0042] Those skilled in the art should understand that, in the disclosure of the present invention, the directions or positions indicated by the terms "longitudinal", "lateral", "up", "down", "left", "right", "front", "back", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc. are based on the directions or positional relationships shown in the accompanying drawings, which are only simplified descriptions for the convenience of describing the present invention, and do not indicate or imply that the device or component referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, the above terms should not be understood as limitations on the present invention.

[0043] It is to be understood that the term "one" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element may be one, while in another embodiment, the number of the element may be multiple, and the term "one" should not be understood as a limitation on the quantity.

[0044] Example:

[0045] The present invention provides an image-based intelligent auxiliary analysis and optimization method for Parkinson's disease voice, comprising the following steps:

[0046] S1. Voice recording: Voice samples of Parkinson's disease patients were recorded using an acoustic recording pen;

[0047] S2. Speech processing: Use software to denoise and standardize the recorded speech samples, and perform visual analysis of their time domain, frequency domain, and loudness characteristics;

[0048] S3, acoustic parameter extraction: importing the processed speech sample into MATLAB, and calculating the relevant acoustic parameters of the sample;

[0049] S4. Database establishment: Build a speech database of Parkinson's disease patients based on the above acoustic parameters, and classify and annotate the data.

[0050] When recording voice samples, the present invention places the acoustic recording pen 25 to 35 cm away from the lips of the participant for recording. In order to ensure the authenticity and reliability of the data, the patient's emotional state, fatigue level, and whether the patient has taken drugs that may affect the voice should also be taken into account during recording. These factors may affect the voice samples. After completing the voice recording, the collected voice samples need to be properly stored and prepared for subsequent processing and analysis.

[0051] When collecting speech samples using a voice recorder, the speech samples included vowels / a / , / o / and / e / in the speech recording step. Each participant formed 12 monosyllabic samples, 8 disyllabic samples and 6 polysyllabic samples so that subsequent analysis could cover different speech characteristics and pathological manifestations.

[0052] Furthermore, the recorded speech samples are subjected to denoising and standardization processing by software, and visual analysis processing of the time domain, frequency domain and loudness characteristics is performed. The software used includes Audacity software. The recorded speech is processed by Audacity software, and the processing steps are as follows:

[0053] Speech denoising: Audacity software is used to remove noise from speech samples while retaining the important features of the speech signal, thereby eliminating unnecessary interference factors such as environmental noise and electronic equipment noise that may be introduced during the recording process;

[0054] Standardization: Adjust the volume of the audio signal to a uniform level to make the speech signals between different samples comparable;

[0055] Time domain analysis: Through the time domain diagram, we can observe the waveform of the speech, the change of amplitude, and the start and end points of the pronunciation. Through time domain analysis, we can directly evaluate the duration, pauses and breathing patterns of the pronunciation of Parkinson's patients. And through time domain analysis, we can directly observe the changes of speech signals over time.

[0056] Frequency domain analysis: Through the frequency domain diagram, the distribution and energy of different frequency components can be clearly seen. Frequency domain analysis can be used to evaluate fundamental frequency (i.e., the vibration frequency of the vocal cords), formant (frequency related to the shape of the vocal tract), and spectrum tilt. These characteristics are often closely related to pathological changes in the voice. Frequency domain analysis converts speech signals from the time domain to the frequency domain through methods such as Fourier transform to observe the frequency components of the signal.

[0057] Loudness characteristic analysis: This is achieved by calculating the short-time energy or short-time average amplitude of the speech signal. The analysis of loudness characteristics can be used to evaluate the strength and stability of the voice, as well as whether there are sudden changes in volume.

[0058] Visual analysis: Displaying the analysis results of the above time domain, frequency domain and loudness characteristics in the form of charts or images can help to more intuitively understand the characteristics and changes of speech signals, help researchers and clinicians quickly identify abnormal phenomena in speech samples, and provide intuitive reference for subsequent acoustic parameter extraction and model training.

[0059] Then import the processed speech samples into MATLAB to calculate the relevant acoustic parameters of the segment: average fundamental frequency, maximum fundamental frequency, jitter frequency, signal-to-noise ratio and continuous pronunciation time;

[0060] Average fundamental frequency: 1. Read audio file: Read audio signal and sampling rate:

[0061] 2. Fundamental frequency estimation: Use the fundamental frequency estimation algorithm Ali to estimate the fundamental frequency of each frame in the audio signal;

[0062] 3. Calculate the average value: filter the estimated fundamental frequency sequence (remove invalid or extreme values), and then calculate the average value of the remaining valid fundamental frequencies.

[0063] Maximum base frequency: 1. Read audio file: read audio signal and sampling rate;

[0064] 2. Fundamental frequency estimation: Use the fundamental frequency estimation algorithm Ali to estimate the fundamental frequency of each frame in the audio signal;

[0065] 3. Find the maximum value: Find and return the maximum value in the estimated fundamental frequency sequence.

[0066] Jitter frequency: 1. Read the audio file and estimate the fundamental frequency: estimate the fundamental frequency of each frame in the audio signal through the fundamental frequency estimation algorithm;

[0067] 2. Clean up the base frequency sequence: remove invalid values ​​(such as NaN or extreme values) in the base frequency sequence;

[0068] 3. Calculate the fundamental frequency difference: Differencing the cleaned fundamental frequency sequence to obtain the change in fundamental frequency between adjacent frames;

[0069] 4. Calculate jitter: Calculate the mean or standard deviation of the absolute value of the difference sequence and compare it with the average fundamental frequency to obtain the relative jitter value.

[0070] Signal-to-noise ratio: It is the ratio of signal power to noise power, and its formula is Where Psignal is the average power of the signal and Pnoise is the average power of the noise.

[0071] Sustain: The total duration of an audio clip, which can be calculated by dividing the signal length by the sampling rate.

[0072] The signal processing method based on the above-mentioned Parkinson's disease voice includes the following steps:

[0073] Place the acoustic recording pen 25 to 35 cm away from the patient's lips to collect voice signals;

[0074] Preprocessing the collected voice signal;

[0075] Extracting the processed acoustic parameters;

[0076] Establish a database for the extracted parameters.

[0077] The preprocessing steps include:

[0078] The voice signal is denoised and normalized, and its time domain, frequency domain and loudness characteristics are visualized and analyzed.

[0079] The analysis device based on the above-mentioned image-based intelligent auxiliary analysis and optimization method for Parkinson's disease voice includes a voice recording module, a voice processing module, an acoustic parameter extraction module, and a database establishment module:

[0080] The voice recording module records the voice samples of Parkinson's disease patients through an acoustic recording pen. The acoustic recording pen used usually has a high-sensitivity microphone that can capture clear and accurate voice signals. It may also have a noise reduction function to reduce the interference of environmental noise. During the recording process, it is necessary to ensure that the patient is in a relatively quiet environment in order to obtain a purer voice sample. At the same time, in order to obtain comprehensive data, the patient can be asked to read a specific text, have a free conversation, or perform specific voice exercises. The acoustic recording pen is placed 25 to 35 cm away from the participant's lips for recording;

[0081] The speech processing module is used to use Audacity software to denoise and standardize the recorded speech samples, and to perform visual analysis of their time domain, frequency domain and loudness characteristics;

[0082] The acoustic parameter extraction module is used to import the processed speech samples into MATLAB and calculate the relevant acoustic parameters of the samples;

[0083] A database building module is used to construct a speech database of Parkinson's disease patients using the above acoustic parameters. It should be understood by those skilled in the art that the embodiments of the present invention described above and shown in the accompanying drawings are only examples and do not limit the present invention. The purpose of the present invention has been fully and effectively achieved. The functions and structural principles of the present invention have been demonstrated and explained in the embodiments, and the embodiments of the present invention may be deformed or modified in any way without departing from the principles.

Claims

1. An image-based intelligent auxiliary analysis and optimization method for Parkinson's disease voice, characterized in that: The following steps are involved: Voice recording: Voice samples of Parkinson's disease patients were recorded using an acoustic recording pen; Speech processing: Use software to denoise and normalize the recorded speech samples, and perform visual analysis of their time domain, frequency domain, and loudness characteristics; Acoustic parameter extraction: import the processed speech samples into MATLAB and calculate the relevant acoustic parameters of the samples; Database establishment: Based on the above acoustic parameters, a speech database of Parkinson's disease patients was constructed, and the data was classified and labeled.

2. The method according to claim 1, characterized in that In the speech recording step, the speech samples include vowels / a / , / o / and / e / , and each participant forms 12 monosyllabic samples, 8 disyllabic samples and 6 polysyllabic samples. The acoustic recording pen is held by the participant at a distance of 25 to 35 cm from his lips.

3. The method according to claim 1, characterized in that During the voice recording step, the participants maintained their accustomed pitch and loudness in a relaxed state to ensure the naturalness and consistency of the voice samples.

4. The method according to claim 1, characterized in that: The speech processing step includes using Audacity software to perform denoising and standardization processing on the speech sample to improve the accuracy of subsequent acoustic parameter extraction.

5. The method according to claim 1, characterized in that In the acoustic parameter extraction step, the MATLAB software is used to analyze the speech samples after denoising and standardization, and the extracted acoustic parameters include average fundamental frequency, maximum fundamental frequency, jitter frequency, signal-to-noise ratio and continuous pronunciation time.

6. The method according to claim 1, characterized in that In the database establishment step, the constructed Parkinson's disease patient speech database contains classification information of speech samples, and realizes effective management and application of different types of speech samples through data annotation.

7. An image-based intelligent auxiliary analysis device for Parkinson's disease voice, characterized in that: include: A voice recording module, used to record voice samples of Parkinson's disease patients through an acoustic recording pen; The speech processing module is used to use Audacity software to denoise and standardize the recorded speech samples, and to perform visual analysis of their time domain, frequency domain and loudness characteristics; The acoustic parameter extraction module is used to import the processed speech samples into MATLAB and calculate the relevant acoustic parameters of the samples; The database building module is used to build a speech database of Parkinson's disease patients based on the above acoustic parameters.

8. The device according to claim 7, characterized in that The voice recording module includes an acoustic recording system that can record high-quality voice samples when the participant is 25 to 35 cm away from the acoustic recording pen.

9. The device according to claim 7, characterized in that The database building module is further configured to classify and label samples in the speech database.

Images