Intelligent Recognition Method and System for Tissue Injury in Neurosurgery

Abstract

The present invention discloses an intelligent recognition method and system for tissue damage in neurosurgery, which relates to the field of medical technology and includes: performing ultrasonic monitoring on the target nerve tissue, collecting ultrasonic scattered echo signals, and preprocessing the collected signals to obtain the preprocessed ultrasonic scattered echo signals; performing Fourier transform on the preprocessed ultrasonic scattered echo signals to obtain the time-frequency of the Fourier transform; performing energy analysis on the time-frequency of the Fourier transform to obtain the time-frequency energy analysis result; positioning the nerve tissue damage area according to the time-frequency energy analysis result to obtain the echo signal of the damage area, and positioning the nerve tissue damage area according to the time-frequency energy analysis result to obtain the echo signal of the damage area. The present invention improves the accuracy and reliability of nerve tissue damage recognition through feature extraction and classification.

Description

Technical Field

[0001] The present invention relates to the field of medical technology, and particularly to an intelligent recognition method and system for tissue damage based on neurosurgery. Background Art

[0002] Clinical observations mainly rely on the self-reported symptoms of patients and the physical examinations of doctors. However, the symptom manifestations of different patients may vary due to individual differences, and some initial symptoms of nerve damage may not be obvious, which increases the difficulty of diagnosis.

[0003] Due to the limitations of traditional diagnostic methods, the field of neurosurgery faces a certain risk of misdiagnosis when diagnosing nerve tissue damage. Especially when faced with minor injuries, the recognition ability of these traditional methods is even more stretched. Minor injuries may only manifest as slight abnormalities in nerve tissue, and these abnormalities are often difficult to be accurately captured and recognized by traditional diagnostic methods. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide an intelligent recognition method and system for tissue damage based on neurosurgery, which improves the accuracy and reliability of nerve tissue damage recognition through feature extraction and classification.

[0005] To solve the above technical problem, the technical solution of the present invention is as follows:

[0006] In the first aspect, an intelligent recognition method for tissue damage based on neurosurgery, the method includes:

[0007] Performing ultrasonic monitoring on the target nerve tissue, collecting ultrasonic scattered echo signals, and preprocessing the collected signals to obtain preprocessed ultrasonic scattered echo signals;

[0008] Performing Fourier transform on the preprocessed ultrasonic scattered echo signals to obtain the time-frequency of the Fourier transform;

[0009] Performing energy analysis on the time-frequency of the Fourier transform to obtain the time-frequency energy analysis result;

[0010] According to the time-frequency energy analysis result, locating the nerve tissue damage area to obtain the echo signal of the damage area;

[0011] Performing feature extraction on the echo signal of the damage area to obtain feature parameters;

[0012] According to the feature parameters, classifying the ultrasonic scattered echo signals to obtain a classification result;

[0013] According to the classification result, identifying the damage state of the nerve tissue to obtain an identification result.

[0014] Further, perform ultrasonic monitoring on the target nerve tissue, collect ultrasonic scattered echo signals, and preprocess the collected signals to obtain preprocessed ultrasonic scattered echo signals, including:

[0015] Use a high-frequency ultrasonic probe to perform ultrasonic monitoring on the target nerve tissue to obtain ultrasonic scattered echo signals;

[0016] Amplify the ultrasonic scattered echo signals and filter the amplified signals through calculation to obtain filtered signals;

[0017] Perform time calibration on the filtered signals to obtain preprocessed ultrasonic scattered echo signals.

[0018] Further, perform Fourier transform on the preprocessed ultrasonic scattered echo signals to obtain the time-frequency of the Fourier transform, including:

[0019] Divide the preprocessed ultrasonic scattered echo signals into a series of overlapping frames and determine the length of the frames;

[0020] According to the length of the frames, calculate the Hamming window function for each frame of signals;

[0021] According to the Hamming window function, calculate to obtain the transformation result;

[0022] Summarize the transformation results to obtain the time-frequency of the Fourier transform.

[0023] Further, perform energy analysis on the time-frequency of the Fourier transform to obtain the time-frequency energy analysis result, including:

[0024] According to the time-frequency of the Fourier transform, calculate the squared amplitude of each frequency component to obtain the energy spectrum of the signal;

[0025] Identify the frequency components in the ultrasonic scattered echo signals by analyzing the energy spectrum to obtain the time-frequency energy analysis result.

[0026] Further, according to the time-frequency energy analysis result, locate the damaged area of the nerve tissue to obtain the echo signals of the damaged area, including:

[0027] Perform edge detection on the time-frequency energy analysis result to identify the characteristics of the damaged area, and the characteristics include shape, size, and position;

[0028] Locate the corresponding damaged area according to the characteristics of the damaged area;

[0029] Extract the echo signals of the damaged area from the ultrasonic scattered echo signals according to the damaged area.

[0030] Further, extract the characteristics of the echo signals of the damaged area to obtain characteristic parameters, including:

[0031] Divide the echo signals of the damaged area into different granularity levels;

[0032] For each granularity level, calculate the entropy of the signal to obtain the characteristic parameter of signal complexity. L is the granularity level, is the number of signal values at the granularity level, is the probability of the signal value occurring, and i is the signal value;

[0033] For each granularity level, calculate the average value of the signal to obtain the characteristic parameter of the DC component of the signal;

[0034] For each granularity level, calculate the variance of the signal to obtain the characteristic parameter of the fluctuation degree of the signal;

[0035] For each granularity level, calculate the sum of the squares of the signal to obtain the characteristic parameter of the energy of the signal.

[0036] Furthermore, classify the ultrasonic scattering echo signals according to the characteristic parameters to obtain the classification results, including:

[0037] Set the initial cluster centers and the number of clusters according to the characteristic parameters of complexity, the characteristic parameters of the DC component, the characteristic parameters of the fluctuation degree, and the characteristic parameters of energy;

[0038] Calculate the membership degree of each cluster center according to the initial cluster centers and the number of clusters;

[0039] Update the cluster centers according to the membership degree of each cluster center;

[0040] Perform cluster analysis on the extracted characteristic parameters through iterative calculations of the membership degree and the cluster centers to obtain the classification results.

[0041] In the second aspect, an intelligent tissue damage recognition system based on neurosurgery includes:

[0042] An acquisition module for ultrasonically monitoring the target nerve tissue, collecting ultrasonic scattering echo signals, and preprocessing the collected signals to obtain the preprocessed ultrasonic scattering echo signals;

[0043] A transformation module for performing Fourier transform on the preprocessed ultrasonic scattering echo signals to obtain the time-frequency of the Fourier transform;

[0044] An energy analysis module for performing energy analysis on the time-frequency of the Fourier transform to obtain the time-frequency energy analysis result; and locating the damaged area of the nerve tissue according to the time-frequency energy analysis result to obtain the echo signals of the damaged area;

[0045] An extraction module, configured to extract features from the echo signals of the damaged area to obtain feature parameters;

[0046] A classification module, configured to classify the ultrasonic scattered echo signals according to the feature parameters to obtain a classification result;

[0047] An identification module, configured to identify the damage state of the nerve tissue according to the classification result to obtain an identification result.

[0048] In a third aspect, a computing device includes:

[0049] One or more processors;

[0050] A storage device, configured to store one or more programs, which when executed by the one or more processors cause the one or more processors to implement the method.

[0051] In a fourth aspect, a computer-readable storage medium stores a program, which when executed by a processor implements the method.

[0052] The above solution of the present invention has at least the following beneficial effects:

[0053] Through the high-precision ultrasonic monitoring technology, the present invention can accurately capture the minute changes of nerve tissue, thus significantly improving the accuracy of diagnosis; by using Fourier transform, time-frequency energy analysis and feature extraction technology, the detailed analysis of ultrasonic scattered echo signals can accurately identify nerve tissue damage at the initial stage; by optimizing the coarse-graining algorithm and GK fuzzy clustering algorithm, intelligent feature extraction and classification of ultrasonic scattered echo signals are realized, improving the efficiency and accuracy of data analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 is a schematic flowchart of an intelligent identification method for tissue damage under neurosurgery provided by an embodiment of the present invention.

[0055] Figure 2 is a schematic diagram of an intelligent identification system for tissue damage under neurosurgery provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0056] Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be completely conveyed to those skilled in the art.

[0057] As Figure 1As shown in the figure, an embodiment of the present invention proposes an intelligent recognition method for tissue damage in neurosurgery, and the method includes the following steps:

[0058] Step 11, perform ultrasonic monitoring on the target nerve tissue, collect ultrasonic scattered echo signals, and preprocess the collected signals to obtain preprocessed ultrasonic scattered echo signals;

[0059] Step 12, perform Fourier transform on the preprocessed ultrasonic scattered echo signals to obtain the time-frequency of the Fourier transform;

[0060] Step 13, perform energy analysis on the time-frequency of the Fourier transform to obtain the time-frequency energy analysis result;

[0061] Step 14, locate the damaged area of the nerve tissue according to the time-frequency energy analysis result to obtain the echo signal of the damaged area;

[0062] Step 15, extract features from the echo signal of the damaged area to obtain feature parameters;

[0063] Step 16, classify the ultrasonic scattered echo signals according to the feature parameters to obtain a classification result;

[0064] Step 17, identify the damage state of the nerve tissue according to the classification result to obtain an identification result.

[0065] In the embodiment of the present invention, through ultrasonic monitoring technology, non-invasive, real-time, and continuous monitoring of the target nerve tissue is realized, avoiding destructive operations such as tissue sectioning in traditional methods; the collected ultrasonic scattered echo signals are preprocessed to effectively remove noise interference factors, improve the signal-to-noise ratio and accuracy of the signals. The preprocessed ultrasonic scattered echo signals are transformed from the time domain to the frequency domain through Fourier transform to obtain the time-frequency characteristics of the signals. Energy analysis is performed on the time-frequency after Fourier transform to obtain the time-frequency energy analysis result. According to the time-frequency energy analysis result, the damaged area of the nerve tissue is successfully located, and the echo signal of the damaged area is obtained; features are extracted from the echo signal of the damaged area to obtain feature parameters that can reflect the damage state of the nerve tissue. According to the feature parameters, a suitable classifier is used to classify the ultrasonic scattered echo signals to obtain a classification result, which can accurately reflect the damage state of the nerve tissue.

[0066] In a preferred embodiment of the present invention, the above step 11 may include:

[0067] An ultrasonic high-frequency probe performs ultrasonic monitoring on the target nerve tissue to obtain ultrasonic scattered echo signals;

[0068] Amplify the ultrasonic scattered echo signals and pass the amplified signals through Perform filtering to obtain a filtered signal, where, is the filtered signal, is the original signal, is the number of in-domain points, is each time point, is the index;

[0069] Perform time calibration on the filtered signal to obtain the preprocessed ultrasonic scattered echo signal.

[0070] In the embodiment of the present invention, ultrasonic monitoring of the target nerve tissue is performed through a high-frequency ultrasonic probe, achieving high-resolution imaging of the nerve tissue, amplifying the ultrasonic scattered echo signal, and enhancing the signal intensity; through filtering processing, noise and interference factors in the ultrasonic scattered echo signal are effectively removed, and time calibration is performed on the filtered signal to eliminate the time deviation during signal acquisition, ensuring the accuracy and consistency of the signal.

[0071] In a preferred embodiment of the present invention, step 12 above may include:

[0072] Divide the preprocessed ultrasonic scattered echo signal into a series of overlapping frames and determine the length of the frames;

[0073] According to the length of the frames, for each frame of signal, through calculate to obtain the Hamming window function, where, is the window length; is the index of the current sampling point, with a value range of 0 to ;

[0074] According to the Hamming window function, through calculate to obtain the transformation result, where, is a discrete time series, w[n] is the window function, m is the frame index, R is the sample interval between two adjacent frames, n is the index of the current sampling point, is the angular frequency, is the complex exponential function, n is the time index, is the imaginary unit, is at the given window index and the angular frequency the signal spectrum value at;

[0075] Summarize the transformation results to obtain the time-frequency of the Fourier transform.

[0076] In an embodiment of the present invention, the preprocessed ultrasonic scattered echo signal is divided into a series of overlapping frames, and the length of the frames is determined, which can capture local features in the signal and improve the accuracy of feature extraction. For each frame of the signal, a Hamming window function is obtained through calculation. The Hamming window function can effectively reduce the sudden change of the signal at the frame edge, reduce spectral leakage, and improve the accuracy of the Fourier transform. According to the Hamming window function, a transformation result is obtained through calculation, and the transformation result can reflect the energy distribution of the signal at different frequencies and times. The transformation results are summarized to obtain the time-frequency of the Fourier transform, so as to understand the frequency and time characteristics of the signal, which improves the accuracy and reliability of identifying the injury state of nerve tissue for realizing high-resolution imaging and precise monitoring of nerve tissue.

[0077] In a specific embodiment of the present invention, according to the frequency content of the signal and the fineness of the analysis, the frame length is determined. In order to make the signal processing smoother and capture the dynamic changes in the signal, overlapping frames are usually used, which means that when the signal is segmented, there will be a certain number of shared samples between two adjacent frames. For example, if the frame shift (i.e., the sample interval R between two adjacent frames) is less than the frame length, there will be an overlap between frames. For each frame of the signal, a window function needs to be applied to reduce spectral leakage. The Hamming window is a commonly used window function. According to the formula, the signal samples of each frame are multiplied by the corresponding Hamming window coefficients to obtain the windowed signal frame.

[0078] According to the length of the frame, the range and resolution of the angular frequency, and the frame index m are set, where m represents the frame currently being processed. For each windowed frame of the signal, the short-time Fourier transform (STFT) formula is used to calculate its spectrum. For each frame of the signal and each value of the angular frequency, the above STFT calculation is performed, so as to obtain the spectrum value of the frame at that frequency; the STFT results of each frame are stored, usually stored as a complex matrix, where the rows represent different frames and the columns represent different frequency components.

[0079] In a preferred embodiment of the present invention, the above step 13 may include:

[0080] According to the time-frequency of the Fourier transform, by calculating the square of the amplitude of each frequency component, the energy spectrum of the signal is obtained;

[0081] By analyzing the energy spectrum, the frequency components in the ultrasonic scattered echo signal are identified to obtain the time-frequency energy analysis result.

[0082] In an embodiment of the present invention, according to the time-frequency of the Fourier transform, by calculating the squared amplitude of each frequency component, the energy spectrum of the signal is obtained. By analyzing the energy spectrum, the frequency components in the ultrasonic scattered echo signal are identified, and the time-frequency energy analysis result is obtained; the nerve tissue is monitored and analyzed non-invasively, avoiding destructive operations such as tissue sectioning in traditional methods and reducing the risk of damage to the nerve tissue.

[0083] In a specific embodiment of the present invention, according to the time-frequency representation of the Fourier transform, the time-frequency representation is usually a complex matrix, where the rows represent time and the columns represent frequency. For each complex value at the time-frequency points in the Fourier transform result, the square of its amplitude is calculated. The formula for the squared amplitude is , where is the amplitude of the complex value at the time-frequency point, is the time index, is the frequency index, represents the energy spectrum value at the frame index and frequency. Summing up the squared amplitude values of all time-frequency points forms a new matrix, and this matrix is the energy spectrum of the signal. The rows of the energy spectrum still represent time, and the columns still represent frequency, but the value of each element is now the energy of the corresponding time-frequency point;

[0084] Draw a heat map or a three-dimensional surface plot of the energy spectrum to visually observe how the energy in the signal is distributed at different times and frequencies. By observing the energy spectrum, the main frequency components in the signal are identified. The main frequency components usually correspond to the high-energy regions in the energy spectrum, shown as regions with higher brightness or darker colors. Analyze the changes in the signal energy at different times and frequencies, perform quantization processing on the energy spectrum, and calculate certain statistical characteristics of the energy spectrum.

[0085] In a preferred embodiment of the present invention, the above step 14 may include:

[0086] Perform edge detection on the time-frequency energy analysis result to identify the characteristics of the damaged area, and the characteristics include shape, size, and position;

[0087] Locate the corresponding damaged area according to the characteristics of the damaged area;

[0088] Extract the echo signal of the damaged area from the ultrasonic scattered echo signal according to the damaged area;

[0089] In the embodiments of the present invention, edge detection technology can accurately outline the contour of the damaged area, thereby precisely identifying the shape, size, and location of the damaged area; according to the characteristics of the damaged area identified by edge detection, the corresponding damaged area can be accurately located; by precisely locating the damaged area, the echo signal of the damaged area can be accurately extracted from a large number of ultrasonic scattered echo signals; accurate identification and location of the damaged area provide assistance for diagnosis and treatment; edge detection and damaged area location reduce the time for manual analysis and location, improving the efficiency of medical services. At the same time, it also reduces the risk of misjudgment caused by human factors.

[0090] In a specific embodiment of the present invention, according to the time-frequency energy analysis results, the feature point matching technology is used to locate the damaged area on the time-frequency energy distribution map, and the position of the damaged area in the time-frequency domain is converted into the time or space coordinates in the original ultrasonic signal; according to the located coordinates of the damaged area, the signal in the corresponding time period or space segment is intercepted from the original ultrasonic scattered echo signal, and the intercepted signal is enhanced and processed.

[0091] In a preferred embodiment of the present invention, the above step 15 may include:

[0092] Dividing the echo signal of the damaged area into different granularity levels;

[0093] For each granularity level, by Calculating the entropy of the signal to obtain the characteristic parameter of the signal complexity, where L is the granularity level, is the number of signal values at the granularity level, is the probability of the signal value appearing, is the signal value;

[0094] For each granularity level, by Calculating the average value of the signal to obtain the characteristic parameter of the DC component of the signal, where is the signal length at the granularity level L, is the time, is the signal value at the granularity level and time, is the mean value;

[0095] For each granularity level, by Calculating the variance of the signal to obtain the characteristic parameter of the fluctuation degree of the signal, where is the mean value of the signal at the granularity level;

[0096] For each granularity level, by Calculating the sum of the squares of the signal to obtain the characteristic parameter of the energy of the signal, where is the energy of the signal.

[0097] In the embodiments of the present invention, the echo signals of the damaged area are divided into different granularity levels, realizing multi-scale analysis of the signals. By calculating the entropy, mean, variance, and energy of each granularity level, a variety of characteristic parameters are obtained. These characteristic parameters describe the characteristics of the signals from multiple aspects such as signal complexity, DC component, fluctuation degree, and energy. The extraction of a variety of characteristic parameters provides more information for the classifier, which helps to improve the classification performance and accuracy of the classifier.

[0098] In a specific embodiment of the present invention, the echo signals of the damaged area are segmented according to the defined granularity levels. For example, if we define the granularity level as 1 millisecond, the signal for every 1 millisecond will be regarded as a separate segment. For the signal segments at each granularity level, by counting the number of times each value appears in the signal segment and then dividing by the length of the signal segment to determine the probability of each value appearing in the signal, the entropy of the signal segment at each granularity level is calculated using the entropy formula. Entropy is a measure of signal complexity and can reflect the amount of information in the signal. For the signal segments at each granularity level, all the values in the segment are accumulated, and the accumulated signal value is divided by the length of the signal segment to obtain the mean of the signal segment at that granularity level. The mean reflects the DC component of the signal. For the signal segments at each granularity level, the difference between each signal value and the mean is calculated, each difference is squared, and all the squared values are accumulated. The accumulated squared difference is divided by the length of the signal segment to obtain the variance of the signal segment at that granularity level. The variance reflects the fluctuation degree of the signal. For the signal segments at each granularity level, all the squared signal values are accumulated, and the accumulated squared value is the energy of the signal segment at that granularity level. The energy reflects the intensity of the signal.

[0099] In a preferred embodiment of the present invention, the above step 16 may include:

[0100] Setting the initial cluster centers and the number of clusters according to the characteristic parameters of complexity, the characteristic parameters of DC component, the characteristic parameters of fluctuation degree, and the characteristic parameters of energy;

[0101] According to the initial cluster centers and the number of clusters, by Calculating the membership degree of each cluster center, where, is the membership degree of the cluster center, x j is the data point, is the distance from the data point to the cluster center, is the index, c is the number of clusters, is the distance between data point j and another cluster center k, is the fuzzy coefficient;

[0102] According to the membership degree of each cluster center, by Updating the cluster centers, is the center of the i-th updated cluster, is the membership degree of the cluster center, x j is the data point, is the total number of data points, is the fuzzy parameter;

[0103] Through the iterative calculation of the membership degree and the cluster center, the extracted feature parameters are subjected to cluster analysis to obtain the classification result.

[0104] In the embodiment of the present invention, through cluster analysis, the extracted feature parameters can be accurately divided into different categories, so as to realize the accurate identification of the damage state of nerve tissue. The method can adaptively determine the optimal number of clusters and the cluster center through iterative calculation of the membership degree and the cluster center, without the need to specify the number of clusters in advance, making the clustering result more in line with the actual distribution of the data. In the clustering process, the feature parameters of complexity, the feature parameters of the DC component, the feature parameters of the fluctuation degree, and the energy feature parameters are comprehensively considered, and the extracted feature information is fully utilized, improving the accuracy and reliability of classification.

[0105] In a specific embodiment of the present invention, according to the distribution of the feature parameters of complexity, the feature parameters of the DC component, the feature parameters of the fluctuation degree, and the energy feature parameters, a suitable number of clusters c is set, and c points are randomly selected in the feature space as the initial cluster centers. These points are as scattered as possible to cover the entire data space. For each data point, calculate its distance to each cluster center, and calculate the membership degree of each data point to each cluster center through the formula. The membership degree represents the degree to which the data point belongs to a certain cluster; according to the membership degree and the position of the data point, use the weighted average method to calculate the new cluster center. The contribution of each data point to the cluster center is proportional to its membership degree, ensuring that the sum of the membership degrees of each data point to all clusters is 1.

[0106] Continuously repeat the process of calculating the membership degree and updating the cluster center until the maximum number of iterations is reached. After the iteration ends, the obtained cluster center is the final cluster center; according to the final membership degree matrix, the data points are assigned to the cluster with the highest membership degree to obtain the data points and cluster centers of each cluster.

[0107] In a preferred embodiment of the present invention, the above step 17 may include:

[0108] Obtain the nerve damage states corresponding to different cluster centers;

[0109] Compare the nerve damage states corresponding to different cluster centers with the cluster centers corresponding to the echo signal characteristics of each damage area to identify the nerve damage states, and the states include but are not limited to: normal, mild damage, moderate damage, severe damage;

[0110] Based on the identified nerve injury status, obtain the recognition result.

[0111] In the embodiments of the present invention, by comparing the cluster centers corresponding to the echo signal characteristics of each injury region with the nerve injury status corresponding to different cluster centers, the injury status of nerve tissue can be accurately identified, including normal, mild injury, moderate injury, and severe injury. Considering various injury states of nerve tissue, a more comprehensive injury assessment is provided; the degree and type of nerve tissue injury can be judged based on the recognition result, and a more precise treatment plan can be formulated to improve the treatment effect.

[0112] In a specific embodiment of the present invention, according to medical knowledge, the nerve injury status corresponding to different cluster centers, for example, a certain cluster center may mainly correspond to data points of "mild injury", and the corresponding relationship between such a cluster center and the nerve injury status is formed into a mapping table or database. For a new injury region, first extract the characteristics of its ultrasonic scattered echo signal, including complexity characteristic parameters, characteristic parameters of the DC component, characteristic parameters of the fluctuation degree, and energy characteristic parameters, perform cluster analysis on these characteristics, and determine the cluster center corresponding to the characteristics of the injury region. According to the comparison result, determine the nerve injury status of the injury region, such as "normal", "mild injury", "moderate injury", or "severe injury".

[0113] As Figure 2 shown, the embodiments of the present invention also provide an intelligent recognition system 20 for tissue injury under neurosurgery, including:

[0114] An acquisition module 21, configured to perform ultrasonic monitoring on the target nerve tissue, collect the ultrasonic scattered echo signal, and preprocess the collected signal to obtain the preprocessed ultrasonic scattered echo signal;

[0115] A transformation module 22, configured to perform Fourier transform on the preprocessed ultrasonic scattered echo signal to obtain the time-frequency of the Fourier transform;

[0116] An energy analysis module 23, configured to perform energy analysis on the time-frequency of the Fourier transform to obtain the time-frequency energy analysis result; according to the time-frequency energy analysis result, locate the nerve tissue injury region to obtain the echo signal of the injury region;

[0117] An extraction module 24, configured to extract the characteristics of the echo signal of the injury region to obtain the characteristic parameters;

[0118] A classification module 25, configured to classify the ultrasonic scattered echo signal according to the characteristic parameters to obtain the classification result;

[0119] An identification module 26, configured to identify the injury status of the nerve tissue according to the classification result to obtain the recognition result.

[0120] It should be noted that this system corresponds to the above method, and all implementation manners in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0121] An embodiment of the present invention further provides a computing device, including: a processor and a memory storing a computer program. When the computer program is run by the processor, it executes the method as described above. All implementation manners in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0122] An embodiment of the present invention further provides a computer-readable storage medium storing instructions. When the instructions are run on a computer, the computer is caused to execute the method as described above. All implementation manners in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0123] The above is the preferred implementation manner of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. An intelligent recognition method for tissue damage in neurosurgery, characterized in that The method includes: S11. Conduct ultrasonic monitoring on the target nerve tissue, collect ultrasonic scattered echo signals, and preprocess the collected signals to obtain preprocessed ultrasonic scattered echo signals; S12. Perform Fourier transform on the preprocessed ultrasonic scattered echo signals to obtain the time-frequency of the Fourier transform; S13. Conduct energy analysis on the time-frequency of the Fourier transform to obtain the time-frequency energy analysis result; Based on the time-frequency of the Fourier transform, the energy spectrum of the signal is obtained by calculating the square of the amplitude of each frequency component. By analyzing the energy spectrum, the frequency components in the ultrasonic scattered echo signal are identified, and the time-frequency energy analysis result is obtained; S14. Locate the damaged area of the nerve tissue according to the time-frequency energy analysis result to obtain the echo signal of the damaged area; It includes: Using the feature point matching technology according to the time-frequency energy analysis result to locate the damaged area on the time-frequency energy distribution map, and converting the position of the damaged area in the time-frequency domain into the time or space coordinates in the original ultrasonic signal; According to the coordinates of the located damaged area, intercept the signal corresponding to the corresponding time period or space segment in the original ultrasonic scattered echo signal, and enhance and process the intercepted signal; S15. Extract features from the echo signals of the damaged area to obtain characteristic parameters, including: dividing the echo signals of the damaged area into different granularity levels. For each granularity level, by calculating the entropy of the signal to obtain the characteristic parameter of signal complexity, where L is the granularity level, N L is the number of signal values at the granularity level, and p L (i) is the probability of the signal value occurring, and i is the signal value. For each granularity level, by calculating the average value of the signal to obtain the characteristic parameter of the DC component of the signal, where T L is the signal length at the granularity level L, t is the time, and x L (t) is the signal value at the granularity level and at the time, and Me L is the mean value; For each granularity level, by calculating the variance of the signal to obtain the characteristic parameter of the signal fluctuation degree, where Me L is the mean value of the signal at the granularity level; For each granularity level, by calculating the sum of the squares of the signals to obtain the energy characteristic parameter of the signal, where En L is the energy of the signal; S16. Classify the ultrasonic scattered echo signals according to the characteristic parameters to obtain the classification result; It includes: Setting the initial clustering centers and the number of clusters according to the characteristic parameters of complexity, the characteristic parameters of the DC component, the characteristic parameter of the fluctuation degree, and the energy characteristic parameters; Calculating the membership degree of each clustering center according to the initial clustering centers and the number of clusters; Updating the clustering centers according to the membership degree of each clustering center; Through the iterative calculation of the membership degree and the clustering centers, perform clustering analysis on the extracted characteristic parameters to obtain the classification result; S17. Identify the damaged state of the nerve tissue according to the classification result to obtain the identification result, which includes: Obtaining the nerve damage states corresponding to different clustering centers; Comparing the nerve damage states corresponding to different clustering centers with the clustering centers corresponding to the echo signal characteristics of each damaged area to identify the nerve damage states, and the states include: normal, mild damage, moderate damage, severe damage; Obtaining the identification result according to the identified nerve damage states.

2. The intelligent tissue damage recognition method based on neurosurgery according to claim 1, wherein Conduct ultrasonic monitoring on the target nerve tissue, collect ultrasonic scattered echo signals, and preprocess the collected signals to obtain preprocessed ultrasonic scattered echo signals, including: Using a high-frequency ultrasonic probe to conduct ultrasonic monitoring on the target nerve tissue to obtain ultrasonic scattered echo signals; Amplify the ultrasonic scattered echo signals and filter the amplified signals through calculation to obtain filtered signals; Calibrate the time of the filtered signals to obtain preprocessed ultrasonic scattered echo signals.

3. The intelligent tissue damage recognition method based on neurosurgery according to claim 2, wherein Perform Fourier transform on the preprocessed ultrasonic scattered echo signals to obtain the time-frequency of the Fourier transform, including: Divide the preprocessed ultrasonic scattered echo signals into a series of overlapping frames and determine the length of the frames; According to the length of the frames, calculate the Hamming window function for each frame of signal; Obtain the transformation result through calculation according to the Hamming window function; Summarize the transformation results to obtain the time-frequency of the Fourier transform.

4. An intelligent tissue damage recognition system based on neurosurgery, which implements the method described in claim 1, characterized in that, Including: An acquisition module, configured to perform ultrasonic monitoring on a target nerve tissue, collect ultrasonic scattered echo signals, and preprocess the collected signals to obtain preprocessed ultrasonic scattered echo signals; A transformation module, configured to perform Fourier transform on the preprocessed ultrasonic scattered echo signals to obtain the time-frequency of the Fourier transform; An energy analysis module, configured to perform energy analysis on the time-frequency of the Fourier transform to obtain a time-frequency energy analysis result; According to the time-frequency energy analysis result, locate the damaged area of the nerve tissue to obtain the echo signal of the damaged area; An extraction module, configured to extract features from the echo signal of the damaged area to obtain feature parameters; A classification module, configured to classify the ultrasonic scattered echo signals according to the feature parameters to obtain a classification result; An identification module, configured to identify the damage state of the nerve tissue according to the classification result to obtain an identification result.

5. A computing device, characterized in that, Including: One or more processors; A storage device, configured to store one or more programs, which when executed by the one or more processors, cause the one or more processors to implement the method according to any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that, A program is stored in the computer-readable storage medium, and when the program is executed by a processor, the method according to any one of claims 1 to 3 is implemented.

Images