Online monitoring and early warning method and system

Abstract

The invention relates to the technical field of brain-computer interfaces, and discloses an online monitoring and early warning method and system. The method comprises the steps of determining the size of a spherical electrode according to a preset image and placing the spherical electrode into a preset cavity; collecting electrode impedance based on the spherical electrode to construct an electrode impedance characteristic pair so as to carry out impedance identification; when the impedance identification result is abnormal, early warning information is sent, a neural signal acquisition instruction is started to construct a feature vector of a neural signal, and neural signal identification is performed; when the neural signal identification result is abnormal, early warning information is sent, and a liquid extraction instruction is started to obtain a feature vector of the liquid; and obtaining a current state identification result based on the electrode impedance feature pair, the neural signal and the feature vector of the liquid. Various modal data of impedance, neural signals and tumor cavity fluid are integrated, the postoperative illness condition is judged, the current state recognition result is obtained, the whole process of marker collection, data analysis, early warning and the like is automated, the invasiveness of an operation is reduced, and treatment parameters are adjusted in time.

Description

Technical Field

[0001] The present invention relates to the technical field of brain-computer interfaces, and particularly to an online monitoring and warning method and system. Background Art

[0002] In the traditional method, diagnosis is carried out through CT or MRI, clinical indicators, or in combination with brain tissue impedance, cerebrospinal fluid analysis, etc. Impedance testing requires multiple puncture channels, a separate procedure, and is likely to cause additional pain to the user and increase the risk of complications. Summary of the Invention

[0003] The main object of the present invention is to provide an online monitoring and warning method and system, aiming to solve at least one of the above technical problems.

[0004] To achieve the above object, the present invention provides an online monitoring and warning method, which is characterized by including:

[0005] Determine the size of the spherical electrode according to a preset image, and place the spherical electrode into a preset cavity;

[0006] Collect electrode impedance based on the spherical electrode to construct an electrode impedance feature pair, and perform impedance recognition on the electrode impedance feature pair;

[0007] When the impedance recognition result is abnormal, send a warning message and start a nerve signal acquisition instruction to construct a feature vector of the nerve signal, and perform nerve signal recognition on the feature vector of the nerve signal;

[0008] When the nerve signal recognition result is abnormal, send a warning message and start a liquid extraction instruction for the preset cavity to obtain a feature vector of the liquid in the preset cavity;

[0009] Perform state recognition based on the electrode impedance feature pair, the feature vector of the nerve signal, and the feature vector of the liquid to obtain a current state recognition result.

[0010] In some embodiments, the determining the size of the spherical electrode according to a preset image and placing the spherical electrode into a preset cavity includes:

[0011] Obtain a preset image, and fuse the preset image to obtain a tomographic scan fusion image;

[0012] Train a preset segmentation model based on the tomographic scan fusion image to mark the target boundary of each tomographic scan fusion image;

[0013] Perform three-dimensional reconstruction based on the tomographic scan fusion image with the marked target boundary to obtain a three-dimensional head image with a target three-dimensional boundary;

[0014] Determine a spherical electrode based on the three-dimensional head image, and place the spherical electrode into a preset cavity.

[0015] In some embodiments, the determining the spherical electrode based on the three-dimensional head image includes:

[0016] Determine the volume ratio of the target volume in the three-dimensional head image;

[0017] According to the actual head size of a preset object and the volume ratio, obtain the physical space size of the cavity;

[0018] Determine the size and number of sheet-shaped electrodes according to the physical space size of the cavity;

[0019] Generate a spherical electrode according to the size and number of sheet-shaped electrodes.

[0020] In some embodiments, the collecting electrode impedance based on the spherical electrode to construct an electrode impedance feature pair and performing impedance identification on the electrode impedance feature pair includes:

[0021] When the spherical electrode is placed into the preset cavity, repeatedly test the electrode impedance of each electrode based on the spherical electrode;

[0022] Calculate the average value of the electrode impedance of each sheet-shaped electrode for the same test according to the electrode impedance, and fit the average value of the electrode impedance of each sheet-shaped electrode to obtain a slope;

[0023] Construct an electrode impedance feature pair according to the average value of the electrode impedance of the first test, the average value of the electrode impedance of the last test, and the slope;

[0024] Train a first classification model based on the electrode impedance feature pair, and perform impedance identification according to the trained first classification model.

[0025] In some embodiments, when the impedance identification result is abnormal, sending a warning message and starting a nerve signal acquisition instruction to construct a feature vector of the nerve signal, and performing nerve signal identification on the feature vector of the nerve signal includes:

[0026] When the impedance identification result is abnormal, send a warning message and start a nerve signal acquisition instruction to collect signals of each electrode for a preset duration and process the signals of the preset duration to obtain two-dimensional nerve signals;

[0027] Convert the two-dimensional nerve signals to the time domain and the frequency domain respectively to extract a time domain feature set and a frequency domain feature set of each sheet-shaped electrode;

[0028] Concatenate the electrode impedance feature pair, the time domain feature set, and the frequency domain feature set into a first feature vector;

[0029] Train a second classification model based on the first feature vector, and perform neural signal recognition according to the trained second classification model.

[0030] In some embodiments, when the neural signal recognition result is abnormal, sending a warning message and initiating a liquid extraction instruction for a preset cavity to obtain a feature vector of the liquid in the preset cavity includes:

[0031] When the neural signal recognition result is abnormal, send a warning message and initiate a liquid extraction instruction for the preset cavity to extract the liquid in the preset cavity;

[0032] Take an image of the liquid to obtain an image of the liquid;

[0033] Perform color quantization on the image to obtain a color quantization value;

[0034] Perform transparency quantization on the image to obtain an average pixel value;

[0035] Determine the protein quantification and white blood cell count of the liquid;

[0036] Use the color quantization value, average pixel value, protein quantification, and white blood cell count as the feature vector of the liquid.

[0037] In some embodiments, the performing color quantization on the image to obtain a color quantization value includes:

[0038] Perform palette control on the image to obtain a palette image;

[0039] Perform principal component analysis of colors on the palette image to obtain color features of the palette image;

[0040] Eliminate black color features from the color features to obtain updated color features;

[0041] Obtain the three-channel pixel mean of each color feature in the updated color features, and determine the target color feature corresponding to the median of the three-channel pixel means;

[0042] Obtain a color quantization value according to the target color feature.

[0043] In some embodiments, the performing transparency quantization on the image to obtain an average pixel value includes:

[0044] Convert the image to the Lab color mode, perform bilateral filtering on the L channel, and then convert it back to the BGR color mode to obtain a processed image;

[0045] Calculate the average pixel value of the processed image.

[0046] In some embodiments, the state recognition based on the electrode impedance feature pairs, the feature vectors of the neural signals, and the feature vectors of the liquid to obtain the current state recognition result includes:

[0047] Concatenate the electrode impedance feature pairs, the feature vectors of the neural signals, and the feature vectors of the liquid corresponding to each petal-shaped electrode into a second feature vector;

[0048] Train a third classification model based on the second feature vector and a preset label to obtain the target level corresponding to the signal acquisition area of each petal-shaped electrode;

[0049] Determine the second-level target area based on the target level, a preset amplification factor, the total area of the preset cavity, the normal tissue area, and the first-level target area;

[0050] Determine the first-level target variance and the second-level target variance according to the number of each petal-shaped electrode;

[0051] Obtain the current state recognition result according to the first-level target area, the second-level target area, the first-level target variance, the second-level target variance, and a preset weight.

[0052] In addition, to achieve the above object, the present invention also proposes an online monitoring and warning system, including:

[0053] An electrode determination module, configured to determine the size of the spherical electrode according to a preset image and place the spherical electrode into a preset cavity;

[0054] An impedance recognition module, configured to construct electrode impedance feature pairs based on the electrode impedance collected by the spherical electrode and perform impedance recognition on the electrode impedance feature pairs;

[0055] A neural signal recognition module, configured to send a warning message and initiate a neural signal acquisition instruction to construct a feature vector of the neural signal and perform neural signal recognition on the feature vector of the neural signal when the impedance recognition result is abnormal;

[0056] A liquid extraction module, configured to send a warning message and initiate a liquid extraction instruction for the preset cavity to obtain a feature vector of the liquid in the preset cavity when the neural signal recognition result is abnormal;

[0057] A state recognition module, configured to perform state recognition based on the electrode impedance feature pairs, the feature vectors of the neural signals, and the feature vectors of the liquid to obtain the current state recognition result. Description of the Drawings

[0058] Figure 1 It is a schematic flowchart of an embodiment of the online monitoring and warning method of the present invention;

[0059] Figure 2 Schematic diagram of the structure of the electrode device involved in the solution of the embodiment of the present invention;

[0060] Figure 3 Example diagram of the size and number of petal-shaped electrodes involved in the solution of the embodiment of the present invention;

[0061] Figure 4 Schematic diagram of the numbering of petal-shaped electrode sheets involved in the solution of the embodiment of the present invention;

[0062] Figure 5 Block diagram of the structure of an embodiment of the online monitoring and warning system of the present invention;

[0063] Figure 6 Schematic diagram of the technical process involved in the solution of the embodiment of the present invention;

[0064] Figure 7 Schematic diagram of the electrode style confirmation unit involved in the solution of the embodiment of the present invention;

[0065] Figure 8 Schematic diagram of the data acquisition unit involved in the solution of the embodiment of the present invention;

[0066] Figure 9 Schematic diagram of the condition monitoring unit involved in the solution of the embodiment of the present invention;

[0067] Figure 10 Schematic diagram of the warning unit involved in the solution of the embodiment of the present invention;

[0068] Figure 11 Schematic diagram of the structure of the electronic device of the hardware operating environment involved in the solution of the embodiment of the present invention.

[0069] The realization, functional features and advantages of the object of the present invention will be further described with reference to the embodiments and the accompanying drawings. Detailed implementation manners

[0070] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0071] It should be noted that all the directional indications (such as up, down, left, right, front, back...) in the embodiments of the present invention are only used to explain the relative positional relationship and movement conditions between components in a specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indications will also change accordingly.

[0072] In addition, the descriptions involving "first", "second", etc. in the present invention are for descriptive purposes only, and should not be construed as indicating or implying their relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one such feature. In addition, the technical solutions between various embodiments may be combined with each other, but it must be based on the fact that those of ordinary skill in the art can implement it. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0073] The present invention provides an online monitoring and warning method and system.

[0074] An embodiment of the present invention provides an online monitoring and warning method. Refer to Figure 1 , Figure 1 which is a schematic flowchart of an embodiment of the online monitoring and warning method of the present invention.

[0075] As Figure 1 shown, the online monitoring and warning method includes:

[0076] Step S100: Determine the size of the spherical electrode according to a preset image, and place the spherical electrode into a preset cavity;

[0077] Step S200: Collect electrode impedance based on the spherical electrode to construct an electrode impedance feature pair, and perform impedance identification on the electrode impedance feature pair;

[0078] Step S300: When the impedance identification result is abnormal, send a warning message and start a nerve signal acquisition instruction to construct a feature vector of the nerve signal, and perform nerve signal identification on the feature vector of the nerve signal;

[0079] Step S400: When the nerve signal identification result is abnormal, send a warning message and start a liquid extraction instruction for the preset cavity to obtain a feature vector of the liquid in the preset cavity;

[0080] Step S500: Perform state identification based on the electrode impedance feature pair, the feature vector of the nerve signal, and the feature vector of the liquid to obtain a current state identification result.

[0081] It should be noted that the execution subject in this embodiment may be an electronic device. The electronic device may be a computer device with data processing functions, or other devices that can achieve the same or similar functions. This embodiment does not make any restrictions in this regard. In this embodiment, a computer device is used as an example for illustration.

[0082] It can be understood that this embodiment is described by taking the postoperative stage of glioma as an example. After glioma surgery, regular head CT or MRI scans, blood routine, and tumor marker examinations are required to evaluate the surgical effect and monitor tumor recurrence. The method described in this embodiment is the application of brain-computer interface technology combined with artificial intelligence technology in the field of neurosurgery.

[0083] In one embodiment, determining the size of the spherical electrode according to a preset image and placing the spherical electrode into a preset cavity includes: obtaining the preset image and fusing the preset images to obtain a tomographic scan fusion image; training a preset segmentation model based on the tomographic scan fusion image to mark the target boundary of each tomographic scan fusion image; performing three-dimensional reconstruction based on the tomographic scan fusion image with the marked target boundary to obtain a three-dimensional head image with a target three-dimensional boundary; determining the spherical electrode based on the three-dimensional head image and placing the spherical electrode into the preset cavity.

[0084] It should be noted that referring to the Figure 2 electrode device shown in, including: a flexible electrode shaft; a plurality of petal-shaped electrodes arranged in sequence along the extension direction of the flexible electrode shaft, the petal-shaped electrodes surrounding to form a spherical electrode with a spherical hollow; a sampling tube, the sampling end of the sampling tube extending into the center of the spherical electrode; a drug delivery tube, the drug delivery end of the drug delivery tube extending into the inner cavity liquid of the center of the spherical electrode. Specifically, as shown in Figure 2 the left side, the spherical electrode unfolds in the shape of a petal, and each petal has a petal-shaped electrode. As shown in Figure 2 the right side, the petal-shaped electrodes surround to form a spherical electrode with a spherical hollow. The electrode device is also equipped with a sampling tube, the sampling end of the sampling tube extending into the center of the sphere, and the electrode device is also equipped with a drug delivery tube for introducing drugs from the outside into the tumor cavity liquid in the center of the sphere.

[0085] Exemplarily, when a glioma is removed from the intracranial cavity, a tumor cavity is left after the glioma is removed, and the spherical electrode is placed into the tumor cavity. The tumor cavity liquid can be obtained through the cerebrospinal fluid acquisition device, that is, the spherical electrode, and the current condition of the patient can be judged based on the electrode impedance and other clinical data. In this embodiment, the spherical electrode is adopted, which is convenient for reaching deep into the tumor and directly performing electric field inhibition on tumor cells. During the electric field inhibition, drug administration is combined, which is convenient and fast, and online treatment can be performed in a timely manner according to the evaluation of the patient's condition.

[0086] Specifically, step S1, electrode style confirmation: step S1.1, tumor boundary (target boundary) confirmation: step S1.1.1: obtaining preoperative preset images of a preset object (such as a patient), such as MRI and CT tomographic scan images; step S1.1.2: fusing the MRI and CT corresponding single-layer scan layer images one by one to obtain a tomographic scan fusion image img MCi, where the image fusion method includes but is not limited to methods based on feature matching such as SIFT, and methods based on deep learning models such as IFCNN; Step S1.1.3: Based on the tomographic fusion image img MCi , train a preset segmentation model (such as a glioma segmentation model). The preset segmentation model includes but is not limited to Unet, Unet++, Mask-Rcnn, etc., and mark the target boundary (such as the glioma boundary) on each tomographic fusion image img MCi ; Step S1.1.4: Based on the tomographic fusion image img with the glioma boundary MCi perform three-dimensional reconstruction to obtain a 3D head image 3D with the three-dimensional boundary of the tumor MC (3D head image with the target three-dimensional boundary). Here, the three-dimensional reconstruction can be performed using commercial software such as VTK.

[0087] In one embodiment, determining the spherical electrode based on the 3D head image includes: determining the volume ratio of the target volume in the 3D head image; obtaining the physical space size of the cavity according to the actual size of the patient's head and the volume ratio; determining the size and number of sheet electrodes according to the physical space size of the cavity; generating a spherical electrode according to the size and number of the sheet electrodes.

[0088] Specifically, Step S1.2: Electrode size selection: Step S1.2.1: Based on the volume ratio of the glioma volume to the 3D head image 3D with the three-dimensional boundary of the tumor in Step S1.1.4 MC and the actual size of the patient's head, obtain the true physical space size V of the tumor cavity LQ (physical space size of the cavity); Step S1.2.2: According to the true physical size V of the tumor cavity LQ , select a sheet electrode with appropriate size and number of sheets. As Figure 3 shown in the sheet electrodes of different sizes and different numbers of sheets, in this embodiment, a spherical electrode is generated according to the selected size and number of sheets of the sheet electrode.

[0089] In one embodiment, collecting electrode impedance based on the spherical electrode to construct an electrode impedance feature pair and performing impedance identification on the electrode impedance feature pair includes: when the spherical electrode is placed in the preset cavity, repeatedly testing the electrode impedance of each electrode based on the spherical electrode; calculating the average value of the electrode impedance of each sheet electrode for the same test according to the electrode impedance, and fitting the average value of the electrode impedance of each sheet electrode to obtain a slope; constituting an electrode impedance feature pair according to the average value of the electrode impedance of the first test, the average value of the electrode impedance of the last test, and the slope; training a first classification model based on the electrode impedance feature pair, and performing impedance identification according to the trained first classification model.

[0090] Exemplarily, step S2, data acquisition, includes step S2.1 electrode acquisition and step S2.2 tumor cavity fluid acquisition. After the glioma is removed from the intracranial cavity, for the tumor cavity left after the glioma is removed, the spherical electrode selected according to the above step S1.2.2 is placed into the tumor cavity.

[0091] Specifically, step S2.1, electrode acquisition: Step S2.1.1: Impedance value acquisition: ① After the spherical electrode is placed into the tumor cavity, the impedance of each electrode is synchronously tested each time the tumor cavity fluid is extracted; ② As Figure 4 shown, number the lobed electrode pieces. Since the spherical electrode is composed of N lobed electrode pieces, obtain the electrode impedance of each electrode and arrange them in chronological order, and calculate the average value (electrode impedance average value) of all the electrodes included in each lobed electrode piece during the same impedance acquisition. Among them, the average value of the impedance of each lobed electrode piece at the first test is Ω 1i ; ③ Fit a straight line to the multiple average values (electrode impedance average values) obtained for each lobed electrode piece by the least squares method to obtain its slope k ji ; ④ Based on the electrode impedance average value Ω 1ji , slope k ji and the average value of the electrode impedance Ω lastji collected at the last time, form a feature pair list Ωji = (Ω 1ji , Ω lastji , k ji )(electrode impedance feature pair); ⑤ Obtain the electrode impedance feature pair lists Ω of multiple patients, and set the label as abnormal or normal, and train a classification model model1 (the first classification model). Here, the first classification model includes but is not limited to decision trees, random forests, and BP neural networks, etc., so as to perform impedance identification according to the trained first classification model.

[0092] In one embodiment, when the impedance identification result is abnormal, send a warning message and start a nerve signal acquisition instruction to construct a feature vector of the nerve signal, and perform nerve signal identification on the feature vector of the nerve signal, including: when the impedance identification result is abnormal, send a warning message and start a nerve signal acquisition instruction to collect the signals of each electrode for a preset duration and process the signals of the preset duration to obtain two-dimensional nerve signals; respectively convert the two-dimensional nerve signals to the time domain and the frequency domain to extract the time domain feature set and the frequency domain feature set of each lobed electrode; splice the electrode impedance feature pair, the time domain feature set and the frequency domain feature set into a first feature vector; train a second classification model based on the first feature vector, and perform nerve signal identification according to the trained second classification model.

[0093] Specifically, step S2.1.2: Neural signal acquisition: ① When the first classification model model1 identifies an anomaly, the acquisition impedance synchronously initiates a neural signal acquisition instruction, measures the signals of each electrode for a duration of Δt (predetermined duration signals), and stores them according to the petal-shaped pieces to form a two-dimensional signal (two-dimensional neural signal) of W ji =(w = Δt, h = m); ② For each two-dimensional signal (two-dimensional neural signal), features in the time domain can be considered for extraction, such as kurtosis, skewness, maximum value, minimum value, mean value, median value, mean absolute error, and root mean square, etc., to obtain a time-domain feature set list of each petal-shaped piece electrode Tji ; ③ Consider converting the signal (two-dimensional neural signal) to the frequency domain, and extracting frequency-domain features such as the maximum power of the spectrum, power spectral bandwidth, fundamental frequency, maximum peak, frequency-domain kurtosis, and frequency-domain skewness in the frequency domain to obtain a frequency-domain feature set list of each petal-shaped piece electrode Pji ; ④ Concatenate the electrode impedance feature pair list Ωji , the time-domain feature set list Tji , and the frequency-domain feature set list Pji into a one-dimensional feature vector (the first feature vector) through concat(). Collect multiple patients and train a classification model model2 (the second classification model) with the disease conditions (normal / abnormal) corresponding to different time points of each petal-shaped electrode as labels to obtain the disease conditions of the signal acquisition areas of each petal-shaped electrode. Here, the second classification model includes, but is not limited to, decision trees, random forests, and BP neural networks, etc.

[0094] In one embodiment, when the neural signal recognition result is abnormal, send a warning message and initiate a liquid extraction instruction for the preset cavity to obtain the feature vector of the liquid in the preset cavity, including: when the neural signal recognition result is abnormal, send a warning message and initiate a liquid extraction instruction for the preset cavity to extract the liquid in the preset cavity; perform imaging on the liquid to obtain an image of the liquid; perform color quantization on the image to obtain a color quantization value; perform transparency quantization on the image to obtain an average pixel value; determine the protein quantification and white blood cell count of the liquid; use the color quantization value, average pixel value, protein quantification, and white blood cell count as the feature vector of the liquid

[0095] It can be understood that according to the impedance recognition result (the recognition result of the first classification model model1), it is decided whether to initiate neural signal acquisition, and according to the impedance recognition result (the recognition result of the first classification model model1) and the neural signal recognition result (the recognition result of the second classification model model2), it is decided whether to extract the tumor cavity fluid

[0096] Exemplarily, step S2.2: Obtaining tumor cavity fluid: Step S2.2.1, when the second classification model model2 identifies an abnormality, start the instruction to extract tumor cavity fluid, and obtain cerebrospinal fluid through a collection tube; Step S2.2.2, taking the first extraction of tumor cavity fluid as a reference, extract a certain amount of tumor cavity fluid L1 for the first time (it is necessary to ensure that no drug is administered during the first extraction of tumor cavity fluid); Step S2.2.3, routine examination of tumor cavity fluid, including the following contents: 1) Color quantization; 2) Transparency quantization; 3) Regularly extract a certain amount of tumor cavity fluid through a sampling tube, and refer to step S1.1.2 to obtain color i and Color i ; 4) 5) Protein quantification label 3i ; 6) White blood cell count label 4i .

[0097] In one embodiment, color quantization is performed on the image to obtain a color quantization value, including: performing palette control on the image to obtain a palette image; performing color principal component analysis on the palette image to obtain the color characteristics of the palette image; removing the black color characteristics from the color characteristics to obtain updated color characteristics; obtaining the three-channel pixel mean of each color characteristic in the updated color characteristics, determining the target color characteristic corresponding to the median of the three-channel pixel means; and obtaining the color quantization value according to the target color characteristic.

[0098] Specifically, 1) Color quantization: ① Palette control, image the first extracted certain amount of tumor cavity fluid L1 to obtain an initial image img 11 , and convert the initial image img 11 to the P color mode, and after adding color dithering, control the number of palette colors so that the picture is expressed by a certain number of color characteristics to obtain the image img P11 . In this embodiment, the number of palette color controls can be set to 10. ② Color principal component analysis, all color characteristics color_list = [(r1, g1, b1), (r2, g2, b2) … (r P11 in the image img n , g n , b n )] can be obtained through the getcolors() method provided by PIL. After removing the black color characteristics, the new color characteristics color_list_new = [(r1, g1, b1), (r2, g2, b2) … (r i , g i , b i)](Update color features); ③ Calculate the three-channel pixel mean of each color feature in the updated color feature color_list_new, and obtain the color feature (r m1 , g m1 , b m1 )(Target color feature) corresponding to the median value among all pixel means (three-channel pixel means), and obtain the color quantization value through the target color feature (r m1 , g m1 , b m1 )

[0099] In one embodiment, the transparency quantization of the image is performed to obtain the average pixel value, including: converting the image to the Lab color mode, performing bilateral filtering on the L channel, and then converting it to the BGR color mode to obtain the processed image; calculating the average pixel value of the processed image.

[0100] Specifically, 2) Transparency quantization: ① First, draw a certain amount of tumor cavity fluid L1 into a transparent glass beaker, irradiate it with visible light of brightness α at one end, and image it with a camera at the other end to obtain an image img L12 ; ② Convert the image img L12 from BGR to the Lab color mode, perform bilateral filtering on the L channel, and then convert it to the BGR color mode to obtain the image L 12 -img; ③ Calculate the average pixel value Color1 of the image L 12 -img.

[0101] It should be noted that here, the Lab color mode is a mode that simulates the human eye's recognition of colors. There are three types of photoreceptor cells in the human eye. One recognizes brightness (L channel), one distinguishes red and green, and one distinguishes yellow and blue. In the Lab color mode, there are three channels, namely the brightness channel (L channel), the a channel, and the b channel. Bilateral filtering is a non-linear filtering method, which is a compromise treatment that combines the spatial proximity of the image and the similarity of pixel values, taking into account both spatial domain information and gray-scale similarity to achieve the purpose of edge-preserving denoising. The reason why the bilateral filter can preserve the edges well while smoothing and denoising is that the kernel of its filter is generated by two functions: one function determines the coefficients of the filter template by the Euclidean distance of the pixels, and the other function determines the coefficients of the filter by the gray-scale difference of the pixels. The principle of bilateral filtering is as follows:

[0102]

[0103] Among them, g(i, j) represents the output point; S(i, j) refers to the size range of (2N + 1)×(2N + 1) centered on (i, j); f(k, l) represents the (multiple) input points; w(i, j, k, l) = w s *w r , w s is a spatial proximity Gaussian function, w r is a Gaussian function of pixel value similarity,

[0104] It can be understood that the bilateral filter is controlled by three parameters: the filter half-width N, the parameter δ s and δ r . The larger the filter half-width N, the stronger the smoothing effect; the parameters δ s and δ r control the attenuation degrees of the spatial proximity factor w s and the luminance similarity factor w r respectively. Here, in this embodiment, N = 5, δ r = 200, δ s = 200 can be set to calculate the average pixel value Color1 of the image L 12 -img.

[0105] In one embodiment, state recognition is performed on the electrode impedance feature pairs, the feature vectors of the nerve signals, and the feature vectors of the liquid to obtain the current state recognition result, including: splicing the electrode impedance feature pairs, the feature vectors of the nerve signals, and the feature vectors of the liquid corresponding to each petal-shaped electrode into a second feature vector; training a third classification model based on the second feature vector and a preset label to obtain the target level corresponding to the signal acquisition area of each petal-shaped electrode; determining the second-level target area based on the target level, a preset magnification factor, the total area of the preset cavity, the normal tissue area, and the first-level target area; determining the first-level target variance and the second-level target variance according to the number of each petal-shaped electrode; and obtaining the current state recognition result according to the first-level target area, the second-level target area, the first-level target variance, the second-level target variance, and a preset weight.

[0106] Exemplarily, step S3, condition monitoring, includes: pattern recognition, area determination, and level determination.

[0107] Specifically, step S3.1: Pattern recognition: Step S3.1.1, the obtained label 1i , label 2i , protein quantification label 3i , white blood cell count label 4i , electrode impedance feature pair listΩji The time-domain feature set list Tji The frequency-domain feature set list Pji They are concatenated into a one-dimensional feature vector (the second feature vector) by concat(), where Color quantization value Color1 is the average pixel value. In step S3.1.2, multiple patients are collected and a classification model (the third classification model) is trained with the glioma status (normal brain tissue / low-grade glioma / high-grade glioma) corresponding to different time points of each petal-shaped electrode as the label, so as to obtain the glioma grade situation in the area where the signal is collected by each petal-shaped electrode. Among them, the third classification model includes but is not limited to decision tree, random forest, BP neural network, etc.

[0108] Specifically, in step S3.2: Area determination: If a certain petal-shaped electrode piece is identified as high-grade glioma, then only extremely abnormal signals of very few channels it contains are required; but if a certain petal-shaped electrode piece is identified as normal brain tissue, then all the signals it contains must be normal. From this perspective, when a certain petal-shaped electrode piece is identified as normal brain tissue, the avoidance area in its corresponding tumor cavity is appropriately enlarged; on the contrary, the avoidance area in the tumor cavity corresponding to high-grade glioma is appropriately reduced, and the remaining tumor cavity wall corresponds to the petal-shaped electrode piece of low-grade glioma. Here, the avoidance area in the tumor cavity means that in brain tumor surgery, a surgical resection area is usually planned, which includes the tumor itself and a small part of normal tissue around it (referred to as the safety margin or avoidance area) to prevent tumor recurrence. If a petal-shaped electrode piece is identified as normal brain tissue, then when planning the surgery, the avoidance area in the corresponding tumor cavity should be appropriately enlarged to ensure that enough normal tissue is removed as the safety margin. On the contrary, if a petal-shaped electrode piece is identified as high-grade glioma, the avoidance area in the corresponding tumor cavity can be appropriately reduced because this area is considered to have a higher tumor risk and needs to be removed as much as possible.

[0109] In this embodiment, the magnification factor for normal brain tissue is α, the magnification factor for high-grade glioma is β, the total area of the tumor cavity is S (the total area of the preset cavity), the area occupied by normal brain tissue is s0 (the area of normal tissue), the area occupied by high-grade glioma is s2 (the area of the first-grade target), then the area occupied by low-grade glioma (the area of the second-grade target) is s1 = S - α·s0 - β·s2; where α > 1, for example, it can be set to 1.2 / 1.3, etc.; β < 1, for example, it can be set to 0.7 / 0.65, etc.

[0110] Specifically, step S3.3: level determination: Step S3.3.1: Number each high-grade glioma flap electrode sheet obtained in step S2.1, and calculate the variance std2; and number each low-grade glioma flap electrode sheet and calculate the variance std1. Step S3.3.2: Obtain the current state recognition result according to the first-level target area s2, the second-level target area s1, the first-level target variance std2, the second-level target variance std1, the total area S of the preset cavity, and the preset weight. At this moment, the condition of the disease (current state recognition result)

[0111]

[0112] In one example, step S4: early warning includes four functions: impedance value early warning, nerve signal early warning, tumor cavity fluid early warning, and disease condition (status) early warning.

[0113] Specifically, step S4.1, impedance value early warning: When the recognition result of the first classification model model1 indicates an abnormal impedance value, an early warning is issued. Exemplarily, an early warning message of "There may be a recurrence of glioma. Please turn on the spherical electrode nerve signal acquisition function" is issued, and nerve signals are synchronously acquired while collecting the impedance.

[0114] Specifically, step S4.2, nerve signal early warning: Step S4.2.1, when the second classification model model2 indicates an abnormality (for example, when the nerve signal recognition result is abnormal), an early warning is issued. Exemplarily, an early warning message of "There may be a recurrence of glioma. Please turn on the collection tube collection function" is issued; step S4.2.2, according to the true physical space size (cavity physical space size) V of the tumor cavity LQ and the current condition of the disease (current state recognition result) le, determine the extraction of tumor cavity fluid. The extraction volume L of the tumor cavity fluid can be determined through a multi-label regression algorithm, such as random forest, etc. i and the collection speed v i ; step S4.2.3, according to the collection speed v it ≤0.8×v i , indicating that the volume of the tumor cavity fluid is too small, and an early warning (tumor cavity fluid early warning) should be issued; the extraction volume L of the tumor cavity fluid it ≥0.9×L i , indicating that the current tumor cavity fluid collection task is about to be completed, and a reminder should be issued.

[0115] Specifically, step S4.3: disease condition early warning: Step S4.3.1, when the ratio of the occupied area s2 of high-grade glioma to the total area S of the tumor cavity is greater than or equal to the preset parameter, that is , a high-priority early warning signal is issued, where ε is a preset parameter set by experience and can be set by oneself. This embodiment does not limit this.

[0116] In this embodiment, by comprehensively analyzing various modal data such as electrode impedance, nerve signals, and tumor cavity fluid, the current state recognition result of the postoperative condition of glioma is comprehensively determined, and the entire process of biomarker collection, data analysis, early warning, etc. is automated. According to the impedance recognition result, it is decided whether to start nerve signal collection. According to the impedance recognition result combined with the nerve signal recognition result, it is decided whether to extract tumor cavity fluid. For the online monitoring of the postoperative condition, real-time monitoring and early warning are realized through a variety of modal data in combination with computer processing. In addition, the size of the novel spherical electrode proposed in this embodiment can be customized according to the size of the tumor cavity; the identification of normal tissue, low-grade glioma, and high-grade glioma is carried out by dividing each petal-shaped electrode piece into units, and the final condition is considered comprehensively based on all electrode pieces.

[0117] It should be noted that for the online monitoring of the postoperative condition of glioma, real-time monitoring and early warning are carried out through a variety of modal data in combination with computer processing. Real-time monitoring helps to reduce unnecessary treatments and examinations, thereby reducing the overall treatment cost. Compared with using multiple puncture needle track impedance tests, using a spherical electrode combined with a computer processing system for online treatment and real-time monitoring, since no additional puncture steps are required, the treatment process can be more rapid, reducing the patient's treatment time and correspondingly reducing related complications such as infection and bleeding. There is no need to penetrate the brain tissue multiple times like a puncture needle track, reducing the invasiveness of the surgery, which can reduce damage to normal brain tissue, reduce the side effects of treatment, and improve the patient's tolerance. The computer processing system can monitor the changes in the electric field distribution and tumor tissue in real time, providing higher monitoring accuracy and real-time feedback, which helps to adjust treatment parameters in a timely manner. In addition, real-time monitoring can detect possible abnormal situations in a timely manner, such as damage to normal brain tissue caused by improper electrode position or excessive electric field strength, so as to adjust the treatment plan in a timely manner and ensure the safety of treatment. The real-time monitoring carried out by combining a spherical electrode with computer processing provides a more accurate, safe, and efficient method for the postoperative treatment of glioma.

[0118] In this embodiment, the size of the spherical electrode is determined according to a preset image, and the spherical electrode is placed into a preset cavity; based on the spherical electrode, electrode impedance is collected to construct an electrode impedance feature pair, and impedance identification is performed on the electrode impedance feature pair; when the impedance identification result is abnormal, a warning message is sent and a nerve signal acquisition instruction is initiated to construct a feature vector of the nerve signal, and nerve signal identification is performed on the feature vector of the nerve signal; when the nerve signal identification result is abnormal, a warning message is sent and a liquid extraction instruction for the preset cavity is initiated to obtain a feature vector of the liquid in the preset cavity; based on the electrode impedance feature pair, the feature vector of the nerve signal, and the feature vector of the liquid, state identification is performed to obtain a current state identification result. In this embodiment, by comprehensively considering various modal data such as electrode impedance, nerve signals, and tumor cavity fluid, the current state identification result is obtained by comprehensively judging the condition after glioma surgery, and the entire process of marker collection, data analysis, warning, etc. is automated, which helps to reduce the invasiveness of the surgery, facilitates timely adjustment of treatment parameters, reduces unnecessary treatments and examinations, and thus reduces the overall treatment cost.

[0119] In addition, an embodiment of the present invention further provides a storage medium, on which an online monitoring and warning program is stored. When the online monitoring and warning program is executed by a processor, the steps of the online monitoring and warning method described above are implemented.

[0120] Refer to Figure 5 , Figure 5 which is a structural block diagram of an embodiment of the online monitoring and warning system of the present invention.

[0121] As Figure 5 shown, the online monitoring and warning system includes:

[0122] An electrode determination module 10, configured to determine the size of the spherical electrode according to a preset image and place the spherical electrode into a preset cavity;

[0123] An impedance identification module 20, configured to collect electrode impedance based on the spherical electrode to construct an electrode impedance feature pair and perform impedance identification on the electrode impedance feature pair;

[0124] A nerve signal identification module 30, configured to send a warning message and initiate a nerve signal acquisition instruction to construct a feature vector of the nerve signal and perform nerve signal identification on the feature vector of the nerve signal when the impedance identification result is abnormal;

[0125] A liquid extraction module 40, configured to send a warning message and initiate a liquid extraction instruction for the preset cavity to obtain a feature vector of the liquid in the preset cavity when the nerve signal identification result is abnormal;

[0126] A state recognition module 50 is configured to perform state recognition based on the electrode impedance feature pairs, the feature vectors of neural signals, and the feature vectors of the liquid to obtain a current state recognition result.

[0127] Exemplarily, referring to the Figure 6 technical process shown, the online monitoring and warning system can be used for online monitoring of the condition after glioma surgery. The online monitoring and warning system can include an electrode style confirmation unit, a data acquisition unit, a condition confirmation unit, and a warning unit.

[0128] Exemplarily, as Figure 7 shown, the electrode style confirmation unit can include a tumor boundary confirmation module and an electrode size selection module.

[0129] In one embodiment, the electrode determination module 10 can include a tumor boundary confirmation module and an electrode size selection module. The electrode determination module 10 is specifically configured to obtain a preset image, and fuse the preset images to obtain a tomographic scan fusion image; train a preset segmentation model based on the tomographic scan fusion image to mark the target boundary of each tomographic scan fusion image; perform three-dimensional reconstruction based on the tomographic scan fusion image with the marked target boundary to obtain a three-dimensional head image with a target three-dimensional boundary; determine a spherical electrode based on the three-dimensional head image, and place the spherical electrode into a preset cavity.

[0130] In one embodiment, the electrode size selection module is configured to determine the volume ratio of the target volume in the three-dimensional head image; obtain the physical space size of the cavity according to the actual size of the head of a preset object and the volume ratio; determine the size and number of sheet-shaped electrodes according to the physical space size of the cavity; generate a spherical electrode according to the size and number of sheet-shaped electrodes.

[0131] Exemplarily, as Figure 8 shown, the data acquisition unit can include an electrode acquisition module and a tumor cavity fluid acquisition module.

[0132] In one embodiment, the electrode acquisition module can include an impedance recognition module 20. The impedance recognition module 20 is specifically configured to: when the spherical electrode is placed into the preset cavity, test the electrode impedance of each electrode based on the spherical electrode multiple times; calculate the average value of the electrode impedance of each sheet-shaped electrode for the same test according to the electrode impedance, and fit the average value of the electrode impedance of each sheet-shaped electrode to obtain a slope; form an electrode impedance feature pair according to the average value of the electrode impedance of the first test, the average value of the electrode impedance of the last test, and the slope; train a first classification model based on the electrode impedance feature pair, and perform impedance recognition according to the trained first classification model.

[0133] In one embodiment, the electrode acquisition module may include a neural signal recognition module 30. The neural signal recognition module 30 is specifically configured to: when the impedance recognition result is abnormal, send a warning message and initiate a neural signal acquisition instruction to acquire signals of a preset duration for each electrode and process the signals of the preset duration to obtain two-dimensional neural signals; convert the two-dimensional neural signals to the time domain and the frequency domain respectively to extract a time-domain feature set and a frequency-domain feature set for each petal-shaped electrode; splice the electrode impedance feature pair, the time-domain feature set, and the frequency-domain feature set into a first feature vector; train a second classification model based on the first feature vector, and perform neural signal recognition according to the trained second classification model.

[0134] In one embodiment, the tumor cavity fluid acquisition module is specifically configured to: when the neural signal recognition result is abnormal, send a warning message and initiate a liquid extraction instruction for a preset cavity to extract the liquid in the preset cavity; perform imaging on the liquid to obtain an image of the liquid; perform color quantization on the image to obtain a color quantization value; perform transparency quantization on the image to obtain an average pixel value; determine the protein quantification and the number of white blood cells of the liquid; use the color quantization value, the average pixel value, the protein quantification, and the number of white blood cells as the feature vector of the liquid.

[0135] In one embodiment, when the tumor cavity fluid acquisition module performs color quantization, it specifically includes: performing palette control on the image to obtain a palette image; performing color principal component analysis on the palette image to obtain the color features of the palette image; removing the black color features from the color features to obtain updated color features; obtaining the three-channel pixel mean of each color feature in the updated color features, and determining the target color feature corresponding to the median of the three-channel pixel means; obtaining the color quantization value according to the target color feature.

[0136] In one embodiment, when the tumor cavity fluid acquisition module performs transparency quantization, it specifically includes: converting the image to the Lab color mode, performing bilateral filtering on the L channel and then converting it to the BGR color mode to obtain a processed image; calculating the average pixel value of the processed image.

[0137] Exemplarily, as Figure 9 shown, the condition confirmation / monitoring unit (status recognition module 50) may include a pattern recognition module, an area determination module, and a level determination module.

[0138] In one embodiment, the state recognition module 50 is specifically configured to: (pattern recognition module) splice the electrode impedance feature pairs corresponding to each petal-shaped electrode, the feature vectors of the nerve signals, and the feature vectors of the liquid into a second feature vector; train a third classification model based on the second feature vector and a preset label to obtain the target level corresponding to the signal acquisition area of each petal-shaped electrode; (area determination module) determine the second-level target area based on the target level, a preset magnification factor, the total area of the preset cavity, the normal tissue area, and the first-level target area; (level determination module) determine the first-level target variance and the second-level target variance according to the number of each petal-shaped electrode; obtain the current state recognition result according to the first-level target area, the second-level target area, the first-level target variance, the second-level target variance, and a preset weight.

[0139] Exemplarily, as Figure 10 shown, the warning unit may include an impedance value warning module, a nerve signal warning module, a tumor cavity fluid warning module, and a disease condition warning module.

[0140] Specifically, the impedance value warning module is configured to: when the recognition result of the first classification model model1 indicates an abnormal impedance value, issue a warning. Exemplarily, issue a warning message of "There may be a recurrence of glioma. Please turn on the function of collecting nerve signals with spherical electrodes" and collect nerve signals synchronously while collecting the impedance.

[0141] Specifically, the nerve signal warning module is configured to: when the second classification model model2 indicates an abnormality (for example, when the nerve signal recognition result is abnormal), issue a warning. Exemplarily, issue a warning message of "There may be a recurrence of glioma. Please turn on the function of collecting fluid with the collection tube"; Step S4.2.2, determine the extraction of the tumor cavity fluid according to the true physical space size (cavity physical space size) V LQ of the tumor cavity and the current disease condition (current state recognition result) le. The extraction volume L of the tumor cavity fluid can be determined by a multi-label regression algorithm, such as random forest, etc. i and the collection speed v i ; the tumor cavity fluid warning module is configured to issue a warning (tumor cavity fluid warning) according to the collection speed v it ≤0.8×v i , indicating that the volume of the tumor cavity fluid is too small; the extraction volume L it ≥0.9×L i , indicating that the current tumor cavity fluid collection task is about to be completed, and a reminder should be issued.

[0142] Specifically, the disease condition warning module is configured to: when the ratio of the area s2 occupied by high-grade glioma to the total area S of the tumor cavity is greater than or equal to a preset parameter, that is When a highly attention warning signal is issued, where ε is a preset parameter set according to experience and can be set by oneself, and this embodiment does not limit this.

[0143] This embodiment proposes an online monitoring and warning system. The system in this embodiment comprehensively determines the current state recognition result of the condition after glioma surgery by integrating various modal data such as electrode impedance, nerve signals, and tumor cavity fluid, and automates the entire process of marker collection, data analysis, warning, etc., to realize the online monitoring of the condition after glioma surgery, facilitate timely adjustment of treatment parameters, help reduce unnecessary treatments and examinations, and thus reduce the overall treatment cost.

[0144] It should be noted that for the technical details not described in detail in this embodiment of the online monitoring and warning system, reference can be made to the online monitoring and warning method applied in any embodiment of the present invention as described above, and details will not be repeated here.

[0145] Refer to Figure 11 , Figure 11 is a schematic structural diagram of an electronic device for the hardware operating environment of the solution of the embodiment of the present invention.

[0146] As Figure 11 shown, the electronic device may include: a processor 1001, such as a Central Processing Unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display) and an input unit such as a keyboard (Keyboard). Optionally, the user interface 1003 may further include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a Wireless-Fidelity (Wi-Fi) interface). The memory 1005 may be a high-speed random access memory (Random Access Memory, RAM memory), or a stable non-volatile memory (Non-Volatile Memory, NVM), such as a disk memory. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0147] Those skilled in the art can understand that Figure 11 the structure shown in

[0148] does not constitute a limitation on the electronic device, and may include more or fewer components than shown in the figure, or combine some components, or have different component arrangements. Figure 11As shown, the memory 1005, which is a storage medium, may include an operating system, a network communication module, a user interface module, and an online monitoring and warning program.

[0149] In Figure 11 In the electronic device shown, the network interface 1004 is mainly used for data communication with a network server; the user interface 1003 is mainly used for data interaction with a user; the processor 1001 and the memory 1005 in the electronic device of the present invention can be arranged in the electronic device. The electronic device calls the online monitoring and warning program stored in the memory 1005 through the processor 1001 and executes the online monitoring and warning method provided by the embodiments of the present invention.

[0150] It should be understood that the above is only an example for illustration and does not constitute any limitation to the technical solution of the present invention. In specific applications, those skilled in the art can set it according to needs, and the present invention does not make any restrictions in this regard.

[0151] It should be noted that the above-described work process is only illustrative and does not constitute a limitation to the protection scope of the present invention. In actual applications, those skilled in the art can select some or all of them according to actual needs to achieve the purpose of the solution of this embodiment, and no restrictions are made here.

[0152] In addition, it should be noted that in this article, the term "comprising", "including" or any other variant thereof is intended to cover a non-exclusive inclusion, so that a process, method, article or system including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or system. Without further limitations, an element defined by the statement "including a..." does not exclude the existence of another identical element in the process, method, article or system including that element.

[0153] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0154] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-described embodiment methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as a read-only memory (ROM) / RAM, magnetic disk, optical disk), and includes several instructions for causing a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in various embodiments of the present invention.

[0155] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the description and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall equally be included in the patent protection scope of the present invention.

Claims

1. An online monitoring and early warning method, characterized in that: include: Determining the size of the spherical electrode according to the preset image, and placing the spherical electrode into the preset cavity; Collecting electrode impedance based on the spherical electrode to construct an electrode impedance feature pair, and performing impedance identification on the electrode impedance feature pair; When the impedance identification result is abnormal, a warning message is sent and a neural signal acquisition instruction is started to construct a feature vector of the neural signal, and neural signal identification is performed on the feature vector of the neural signal; When the neural signal recognition result is abnormal, a warning message is sent and a liquid extraction instruction of the preset cavity is started to obtain a characteristic vector of the liquid in the preset cavity; State recognition is performed based on the electrode impedance feature pair, the feature vector of the neural signal, and the feature vector of the liquid to obtain a current state recognition result.

2. The method according to claim 1, characterized in that The step of determining the size of the spherical electrode according to the preset image and placing the spherical electrode into the preset cavity comprises: Acquire a preset image, and fuse the preset image to obtain a tomography fusion image; Training a preset segmentation model based on the tomographic fusion image to mark the target boundary of each tomographic fusion image; Performing three-dimensional reconstruction based on the tomographic fusion image with the target boundary marked, and obtaining a three-dimensional head image with the target three-dimensional boundary; A spherical electrode is determined based on the three-dimensional head image, and the spherical electrode is placed into a preset cavity.

3. The method according to claim 2, characterized in that The step of determining a spherical electrode based on the three-dimensional head image comprises: Determine the volume ratio of the target volume in the three-dimensional image of the head; According to the actual size of the head of the preset object and the volume ratio, the physical space size of the cavity is obtained; Determine the size and number of petal-shaped electrodes according to the physical space size of the cavity; A spherical electrode is generated according to the size and number of the petal-shaped electrodes.

4. The method according to claim 1, characterized in that The step of collecting electrode impedance based on the spherical electrode to construct an electrode impedance feature pair, and performing impedance identification on the electrode impedance feature pair includes: When the spherical electrode is placed in the preset cavity, the electrode impedance of each electrode is tested multiple times based on the spherical electrode; Calculating the average electrode impedance of each petal electrode in the same test according to the electrode impedance, and fitting the average electrode impedance of each petal electrode to obtain a slope; An electrode impedance characteristic pair is formed according to the average electrode impedance of the first test, the average electrode impedance of the last test and the slope; A first classification model is trained based on the electrode impedance characteristics, and impedance recognition is performed according to the trained first classification model.

5. The method according to claim 1, characterized in that When the impedance identification result is abnormal, sending a warning message and starting a neural signal acquisition instruction to construct a feature vector of the neural signal, and performing neural signal identification on the feature vector of the neural signal, includes: When the impedance identification result is abnormal, a warning message is sent and a neural signal acquisition instruction is started to collect a preset time length signal of each electrode and process the preset time length signal to obtain a two-dimensional neural signal; Converting the two-dimensional neural signal into the time domain and the frequency domain respectively to extract the time domain feature set and the frequency domain feature set of each petal electrode; splicing the electrode impedance feature pair, the time domain feature set and the frequency domain feature set into a first feature vector; A second classification model is trained based on the first feature vector, and neural signal recognition is performed according to the trained second classification model.

6. The method according to claim 1, characterized in that When the neural signal recognition result is abnormal, sending a warning message and starting a liquid extraction instruction of the preset cavity to obtain a characteristic vector of the liquid in the preset cavity includes: When the neural signal recognition result is abnormal, a warning message is sent and a liquid extraction instruction of the preset cavity is started to extract the liquid in the preset cavity; Imaging the liquid to obtain an image of the liquid; Performing color quantization on the image to obtain a color quantization value; quantifying the transparency of the image to obtain an average pixel value; determining the protein quantity and white blood cell count of the fluid; The color quantification value, pixel value average, protein quantification and white blood cell count are used as feature vectors of the liquid.

7. The method according to claim 6, characterized in that The step of performing color quantization on the image to obtain a color quantization value includes: Performing palette control on the image to obtain a palette image; Performing a color principal component analysis on the palette image to obtain color features of the palette image; Eliminating the black color feature in the color feature to obtain an updated color feature; Obtaining a three-channel pixel mean of each color feature in the updated color feature, and determining a target color feature corresponding to a median of the three-channel pixel mean; A color quantization value is obtained according to the target color feature.

8. The method according to claim 6, characterized in that The step of quantifying the transparency of the image to obtain an average pixel value includes: The image is converted to Lab color mode, and the L channel is subjected to bilateral filtering and then converted to BGR color mode to obtain a processed image; The average pixel value of the processed image is calculated.

9. The method according to any one of claims 1 to 8, characterized in that The state recognition based on the electrode impedance feature pair, the feature vector of the neural signal and the feature vector of the liquid to obtain the current state recognition result includes: splicing the electrode impedance characteristic pair corresponding to each petal electrode, the characteristic vector of the neural signal, and the characteristic vector of the liquid into a second characteristic vector; Based on the second feature vector and the preset label, a third classification model is trained to obtain a target level corresponding to each petal electrode signal collection area; Determine a second-level target area based on the target level, a preset magnification factor, a total area of ​​a preset cavity, a normal tissue area, and a first-level target area; Determine the first level target variance and the second level target variance according to the number of each petal electrode; A current state recognition result is obtained according to the first-level target area, the second-level target area, the first-level target variance, the second-level target variance and preset weights.

10. An online monitoring and early warning system, characterized in that: include: An electrode determination module, used to determine the size of the spherical electrode according to a preset image, and place the spherical electrode into a preset cavity; An impedance identification module, used to collect electrode impedance based on the spherical electrode to construct an electrode impedance characteristic pair, and perform impedance identification on the electrode impedance characteristic pair; A neural signal recognition module, used for sending a warning message and starting a neural signal acquisition instruction to construct a neural signal feature vector when the impedance recognition result is abnormal, and performing neural signal recognition on the neural signal feature vector; A liquid extraction module, used to send a warning message and start a liquid extraction instruction of a preset cavity to obtain a characteristic vector of the liquid in the preset cavity when the neural signal recognition result is abnormal; The state recognition module is used to perform state recognition based on the electrode impedance feature pair, the feature vector of the neural signal and the feature vector of the liquid to obtain a current state recognition result.

Images