Image recognition and micro-current detection method and system for intercostal nerve plexus avoidance

Abstract

The embodiment of the invention provides an image recognition and micro-current detection method and system for intercostal nerve plexus avoidance, and belongs to the technical field of data processing. The method comprises the steps of collecting initial image information corresponding to a first target object, and performing edge detection on the initial image information to obtain a corresponding target recognition result; determining a corresponding current parameter during micro-current detection, and obtaining a corresponding reaction potential signal of a related area of the first target object under the current parameter; performing nerve classification according to the reaction potential information to obtain a target nerve type corresponding to the related area; aligning the related area with the initial image information to obtain a target alignment result, and determining related edge information corresponding to the related area according to the target alignment result and the target recognition result; and generating intercostal nerve plexus avoidance information according to the target nerve type and the related edge information, and displaying the intercostal nerve plexus avoidance information to a second target object.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to an image recognition and micro-current detection method and system for intercostal nerve plexus avoidance. Background Art

[0002] Surgeries in the field of thoracic surgery usually need to deal with complex anatomical structures, and the avoidance of the intercostal nerve plexus is an important step to ensure no nerve damage to the patient after surgery. Currently, the identification of the intercostal nerve plexus mainly relies on imaging techniques such as ultrasound, CT (Computed Tomography), and MRI (Magnetic Resonance Imaging). Although they can provide general anatomical structure information of the chest, the display of the intercostal nerves is not very clear. On the other hand, neuroelectrophysiological detection methods (such as micro-current detection) can be used for nerve localization and function assessment. These methods identify nerve tissues by applying micro-currents and recording the bioelectric responses of the nerves. However, the separate electrophysiological methods may be interfered by other tissues such as muscles and connective tissues when identifying nerves, resulting in a decrease in accuracy. In addition, the existing technology lacks the fusion of image information and electrophysiological data, leading to the inability to accurately display the intercostal nerve plexus avoidance information.

[0003] Therefore, there is an urgent need to design a technical solution to solve at least one of the above technical problems. Summary of the Invention

[0004] The main purpose of the embodiments of the present invention is to provide an image recognition and micro-current detection method and system for intercostal nerve plexus avoidance, aiming to solve the problem in the related technology that the lack of the fusion of image information and electrophysiological data leads to the inability to accurately display the intercostal nerve plexus avoidance information.

[0005] In a first aspect, the embodiments of the present invention provide an image recognition and micro-current detection method for intercostal nerve plexus avoidance, including:

[0006] Collect the initial image information corresponding to the first target object, and perform edge detection on the initial image information to obtain the corresponding target recognition result;

[0007] Determine the current parameters corresponding to the micro-current detection, and obtain the response potential signal corresponding to the relevant area of the first target object under the current parameters;

[0008] Perform nerve classification according to the response potential information to obtain the target nerve type corresponding to the relevant area;

[0009] Align the relevant area and the initial image information to obtain the target alignment result, and determine the relevant edge information corresponding to the relevant area according to the target alignment result and the target recognition result;

[0010] Generate intercostal nerve plexus avoidance information according to the target nerve type and the relevant edge information, and display the intercostal nerve plexus avoidance information to a second target object.

[0011] In a second aspect, an embodiment of the present invention provides an image recognition and microcurrent detection system for intercostal nerve plexus avoidance, including:

[0012] A first acquisition module, configured to acquire initial image information corresponding to a first target object, and perform edge detection on the initial image information to obtain a corresponding target recognition result;

[0013] A second acquisition module, configured to determine current parameters corresponding to microcurrent detection, and obtain a response potential signal corresponding to a relevant area of the first target object under the current parameters;

[0014] A data classification module, configured to perform nerve classification according to the response potential information to obtain a target nerve type corresponding to the relevant area;

[0015] An edge determination module, configured to align the relevant area and the initial image information to obtain a target alignment result, and determine relevant edge information corresponding to the relevant area according to the target alignment result and the target recognition result;

[0016] An information determination module, configured to generate intercostal nerve plexus avoidance information according to the target nerve type and the relevant edge information, and display the intercostal nerve plexus avoidance information to a second target object.

[0017] In a third aspect, an embodiment of the present invention further provides a terminal device, which includes a processor, a memory, a computer program stored on the memory and executable by the processor, and a data bus for realizing connection communication between the processor and the memory. When the computer program is executed by the processor, the steps of any one of the image recognition and microcurrent detection methods for intercostal nerve plexus avoidance provided in the specification of the present invention are implemented.

[0018] In a fourth aspect, an embodiment of the present invention further provides a storage medium for computer-readable storage, characterized in that the storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of any one of the image recognition and microcurrent detection systems for intercostal nerve plexus avoidance provided in the specification of the present invention.

[0019] An embodiment of the present invention provides an image recognition and microcurrent detection method and system for intercostal nerve plexus avoidance. The method includes: collecting initial image information corresponding to a first target object, and performing edge detection on the initial image information to obtain a corresponding target recognition result; determining current parameters corresponding to microcurrent detection, and obtaining a response potential signal corresponding to a relevant area of the first target object under the current parameters; classifying nerves according to the response potential information to obtain a target nerve type corresponding to the relevant area, so as to be able to clarify the target nerve type corresponding to the relevant area, which provides an important basis for the diagnosis and treatment of nerve-related diseases. Furthermore, aligning the relevant area and the initial image information to obtain a target alignment result, and determining relevant edge information corresponding to the relevant area according to the target alignment result and the target recognition result; generating intercostal nerve plexus avoidance information according to the target nerve type and the relevant edge information, and displaying the intercostal nerve plexus avoidance information to a second target object. This method can achieve precise positioning of the intercostal nerve plexus by combining image recognition and microcurrent detection, effectively reducing interference factors, improving the accuracy of nerve recognition, and thus accurately displaying the intercostal nerve plexus avoidance information. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0021] Figure 1 It is a flowchart of an image recognition and microcurrent detection method for intercostal nerve plexus avoidance provided by an embodiment of the present invention;

[0022] Figure 2 It is a module structure diagram of an image recognition and microcurrent detection system for intercostal nerve plexus avoidance provided by an embodiment of the present invention;

[0023] Figure 3 It is a structural schematic block diagram of a terminal device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the protection scope of the present invention.

[0025] The flowcharts shown in the accompanying drawings are merely illustrative examples, and do not necessarily include all content and operations / steps, nor do they necessarily need to be executed in the described order. For example, some operations / steps can be decomposed, combined, or partially merged, so the actual execution order may change according to the actual situation.

[0026] It should be understood that the terms used in the specification of the present invention are merely for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an", and "the" are intended to include the plural forms.

[0027] An embodiment of the present invention provides an image recognition and microcurrent detection method and system for intercostal nerve plexus avoidance. Among them, the image recognition and microcurrent detection method for intercostal nerve plexus avoidance can be applied to a terminal device, and the terminal device can be an electronic device such as a tablet computer, a notebook computer, a desktop computer, a personal digital assistant, and a wearable device. The terminal device can be a server or a server cluster.

[0028] The following will describe in detail some embodiments of the present invention with reference to the accompanying drawings. Without conflict, the following embodiments and the features in the embodiments can be combined with each other.

[0029] Please refer to Figure 1 , Figure 1 which is a schematic flowchart of an image recognition and microcurrent detection method for intercostal nerve plexus avoidance provided by an embodiment of the present invention.

[0030] As Figure 1 shown, the image recognition and microcurrent detection method for intercostal nerve plexus avoidance includes steps S101 to S105.

[0031] Step S101: Collect the initial image information corresponding to the first target object, and perform edge detection on the initial image information to obtain the corresponding target recognition result.

[0032] Exemplarily, a high-resolution imaging device (such as an ultrasound probe, a CT scanner, or an MRI device) is used to obtain the initial image information of the first target object.

[0033] Exemplarily, the initial image information is subjected to Gaussian smoothing processing, then the gradient amplitude and direction are calculated, and then non-maximum suppression is performed, and only the pixels with the largest local gradient are retained as edge points to obtain the image edge information corresponding to the initial image information, and then the image edge information is determined as the target recognition result.

[0034] In some embodiments, obtaining the corresponding target recognition result by performing edge detection on the initial image information includes: performing edge detection on the initial image information to obtain the target edge information corresponding to the initial image information; using the Swin-Transformer network of the target recognition model to perform information encoding on the initial image information to obtain the corresponding first encoded information, second encoded information, third encoded information, and fourth encoded information under four branches; using the feature fusion layer of the target recognition model to perform fusion processing on two adjacent encoded information among the first encoded information, the second encoded information, the third encoded information, and the fourth encoded information to obtain the first fusion information, the second fusion information, and the third fusion information, where the first fusion information includes first edge information and first target information, the second fusion information includes second edge information and second target information, and the third fusion information includes third edge information and third target information; using the normalization network of the first spatial attention layer of the target recognition model to extract features from the first edge information to obtain first edge features, and using the first spatial attention network of the first spatial attention layer to enhance the first edge features with the first target information to obtain second edge features, and then fusing the first edge features and the second edge features according to the first channel attention network of the first spatial attention layer to obtain third edge features; using the edge enhancement layer of the target recognition model to enhance the third edge features in combination with the target edge information to obtain edge-enhanced features; using the normalization network of the second spatial attention layer of the target recognition model to extract features from the first target information to obtain first target features, and using the second spatial attention network of the second spatial attention layer to enhance the first target features with the first edge information to obtain second target features, and then fusing the first target features and the second target features according to the second channel attention network of the second spatial attention layer to obtain third target features; using the mutual enhancement layer of the target recognition model to fuse the edge-enhanced features and the third target features and then perform feature stitching to obtain the corresponding target output features; predicting edge information according to the target output features to obtain the target edge image corresponding to the initial image information; and performing neural recognition on the target edge image to obtain the target recognition result corresponding to the initial image information.

[0035] Exemplarily, an edge detection operator such as the Sobel operator is used to determine the position of the edge by calculating the gradients of the initial image information in the horizontal and vertical directions, so as to obtain the target edge information corresponding to the initial image information.

[0036] Exemplarily, the initial image information is input into a pre-trained Swin-Transformer network, and the network will perform multi-scale encoding processing on the initial image information. The Swin-Transformer network will divide the initial image information into multiple non-overlapping windows and perform self-attention calculations within each window, thereby obtaining the corresponding first encoded information, second encoded information, third encoded information, and fourth encoded information under four branches.

[0037] Exemplarily, the feature fusion layer of the target recognition model is used to splice and fuse the first encoded information with the second encoded information, the second encoded information with the third encoded information, and the third encoded information with the fourth encoded information respectively, so as to synthesize feature information of different scales and levels by fusing adjacent encoded information, and obtain the first fusion information, the second fusion information, and the third fusion information. Among them, the first fusion information includes first edge information and first target information, the second fusion information includes second edge information and second target information, and the third fusion information includes third edge information and third target information.

[0038] Exemplarily, the normalization network of the first spatial attention layer is used to extract features from the first edge information, remove data redundancy and noise, and obtain first edge features. According to the first spatial attention network of the first spatial attention layer, the first edge features are enhanced using the first target information. The first spatial attention network will focus on the part of the first target information related to the edge, highlight the edge features, and obtain second edge features. Through the first channel attention network of the first spatial attention layer, the first edge features and the second edge features are fused, and the information in the channel dimension is comprehensively considered to obtain third edge features.

[0039] Exemplarily, the target edge information and the third edge features are input into an edge enhancement layer, and the edge enhancement layer will further highlight and strengthen the edge information in the third edge features according to the target edge information to obtain edge enhanced features.

[0040] Exemplarily, the normalization network of the second spatial attention layer is used to extract features from the first target information to obtain first target features. According to the second spatial attention network of the second spatial attention layer, the first target features are enhanced using the first edge information. The second spatial attention network will focus on the part of the first edge information related to the target, highlight the target features, and obtain second target features. Through the second channel attention network of the second spatial attention layer, the first target features and the second target features are fused to obtain third target features.

[0041] Exemplarily, the edge enhancement feature and the third target feature are input into the mutual enhancement layer, and the mutual enhancement layer fuses the edge enhancement feature and the third target feature, enabling the edge information and the target information to complement and reinforce each other. Then, the fused features are concatenated to obtain the corresponding target output feature.

[0042] Exemplarily, according to the target output feature, the convolutional layer is used to predict the edge information, obtaining the target edge image corresponding to the initial image information. Then, neural recognition is performed on the target edge image. For example, using a pre-trained classifier or recognition model, the target nerve in the image is recognized based on the features of the edge image. Furthermore, only the target nerve in the target edge image is retained, thereby obtaining the target recognition result corresponding to the initial image information.

[0043] Specifically, edge detection is first performed to obtain the target edge information, and then the edge enhancement layer combines the target edge information to enhance the edge features, which can more accurately extract the edge information in the image and provide clearer boundary information for target recognition. The mutual enhancement layer fuses and concatenates the edge enhancement feature and the target feature, enabling the edge information and the target information to complement and reinforce each other, which helps to better recognize the shape and boundary of the target, and thus enables accurate recognition of the target in different scenarios and images.

[0044] In some embodiments, the obtaining the target edge information corresponding to the initial image information by performing edge detection on the initial image information includes: performing filtering processing on the initial image information to obtain the target image information corresponding to the initial image information; determining an initial segmentation threshold, and performing region division on the target image information according to the initial segmentation threshold to obtain the corresponding first background region and first foreground region; obtaining the target gray entropy corresponding to the target image information according to the first background region and the first foreground region; adjusting the initial segmentation threshold and simultaneously re-updating and calculating the target gray entropy to obtain the target segmentation threshold corresponding to the maximum target gray entropy; performing edge segmentation on the target image information according to the target segmentation threshold to obtain the target edge information corresponding to the target image information; wherein, the target gray entropy is obtained according to the following formula:

[0045]

[0046] Among them, S(th) represents the target gray entropy corresponding to when the initial segmentation threshold is th, th represents the initial segmentation threshold, p(i) represents the pixel distribution corresponding to when the pixel in the target image information is i, p(j) represents the pixel distribution corresponding to when the pixel in the target image information is j, max represents the maximum pixel value corresponding to in the target image information, p(k) represents the pixel distribution corresponding to when the pixel in the target image information is k, and p(g) represents the pixel distribution corresponding to when the pixel in the target image information is g.

[0047] Exemplarily, the initial image information is filtered by using a filtering algorithm such as mean filtering or median filtering, so as to remove the noise in the image and obtain the target image information corresponding to the initial image information.

[0048] Exemplarily, the median value of the image gray value corresponding to the target image information is used as the initial segmentation threshold, and then according to the determined initial segmentation threshold, the pixels in the target image information are divided into two categories. The pixels with pixel gray values less than the initial segmentation threshold are classified into the first background region, and the pixels with pixel gray values greater than or equal to the initial segmentation threshold are classified into the first foreground region.

[0049] Exemplarily, traverse the target image information, count the frequency of each pixel value occurrence to obtain the pixel distribution, and then substitute the initial segmentation threshold and the statistically obtained pixel distribution into the given target gray entropy formula to calculate the target gray entropy corresponding to when the initial segmentation threshold is, where the target gray entropy is obtained according to the following formula:

[0050]

[0051] Among them, S(th) represents the target gray entropy corresponding to when the initial segmentation threshold is th, th represents the initial segmentation threshold, p(i) represents the pixel distribution corresponding to when the pixel in the target image information is i, p(j) represents the pixel distribution corresponding to when the pixel in the target image information is j, max represents the maximum pixel value corresponding to in the target image information, p(k) represents the pixel distribution corresponding to when the pixel in the target image information is k, and p(g) represents the pixel distribution corresponding to when the pixel in the target image information is g.

[0052] Exemplarily, determine the step size, and adjust the initial segmentation threshold according to this step size, for example, increase or decrease this step size each time, and then re-perform region division for each adjusted segmentation threshold and calculate the corresponding target gray entropy, and then record the target gray entropy obtained each time, find the segmentation threshold corresponding to when the target gray entropy is the largest, and use it as the target segmentation threshold.

[0053] Exemplarily, the target image information is segmented according to the obtained target segmentation threshold. Pixel points in the target image information whose gray values mutate near the target segmentation threshold are determined as edge points, thereby obtaining the target edge information corresponding to the target image information.

[0054] Specifically, by adjusting the segmentation threshold and finding the target segmentation threshold corresponding to the maximum gray entropy, the optimal segmentation threshold can be adaptively determined, making the division of the background and foreground regions of the image more reasonable, thereby improving the accuracy of edge detection. Moreover, the target gray entropy comprehensively considers the gray distribution information of the image and can better reflect the information entropy of the image under different image contents. By maximizing the gray entropy to determine the segmentation threshold, the edge detection method has strong robustness to illumination changes and image noise.

[0055] In some embodiments, the filtering the initial image information to obtain the target image information corresponding to the initial image information includes: obtaining a pixel to be processed from the initial image information and determining an initial window; obtaining a corresponding window region from the initial image information according to the pixel to be processed and the initial window; obtaining the maximum pixel value, minimum pixel value, and median pixel value corresponding to the initial window according to the window region; if the pixel to be processed is less than the maximum pixel value and the pixel to be processed is greater than the minimum pixel value, determining the pixel to be processed as a normal point, otherwise, judging whether the median pixel value is within the pixel range corresponding to the minimum pixel value and the maximum pixel value; if the median pixel value is less than the maximum pixel value and the median pixel value is greater than the minimum pixel value, determining the pixel to be processed as the normal point, otherwise, expanding the initial window to obtain a target window, and re-judging the abnormality of the pixel to be processed according to the target window. If the pixel to be processed is an abnormal point and the window size corresponding to the target window is greater than a preset value, performing pixel processing on the pixel to be processed according to the target window to obtain a target pixel result corresponding to the pixel to be processed; obtaining the target image information corresponding to the initial image information according to the target pixel result.

[0056] Exemplarily, a pixel is sequentially selected as the pixel to be processed from the initial image information by scanning row by row from left to right and from top to bottom, and a window with a fixed size is set as the initial window, which is a rectangular region centered on the pixel to be processed. For example, windows of sizes such as 3×3 and 5×5 can be selected, and then, centered on the selected pixel to be processed, a corresponding window region is intercepted from the initial image information according to the size of the initial window.

[0057] Exemplarily, all pixels within the window region are traversed to find the maximum pixel value, the minimum pixel value, and the median pixel value among them. The median pixel value can be obtained by sorting all pixel values within the window and taking the middle value.

[0058] Exemplarily, it is checked whether the pixel to be processed is less than the maximum pixel value and greater than the minimum pixel value. If this condition is met, the pixel to be processed is determined as a normal point. If the pixel to be processed does not meet the condition of being less than the maximum pixel value and greater than the minimum pixel value, it is further determined whether the median pixel value is within the pixel range corresponding to the minimum pixel value and the maximum pixel value. If it is met, the pixel to be processed is still determined as a normal point.

[0059] Exemplarily, if the median pixel value is also not within the pixel range corresponding to the minimum pixel value and the maximum pixel value, the initial window is enlarged. For example, the 3*3 window is enlarged to a 5*5 window to obtain a target window. Then the above judgment process is repeated, and the abnormal judgment of the pixel to be processed is performed again according to the target window. If the pixel to be processed is determined as an abnormal point and the window size corresponding to the target window is greater than a preset value (a preset upper limit of the window size), the average value of the pixels within the target window is used to replace the value of the pixel to be processed to obtain the target pixel result corresponding to the pixel to be processed.

[0060] Exemplarily, the above steps are repeated for each pixel in the initial image information, and the target pixel result corresponding to each pixel is obtained in turn. Combining all the target pixel results, the target image information corresponding to the initial image information is obtained.

[0061] Specifically, the above adaptive filtering method can accurately identify abnormal pixels in the initial image information and process them. By continuously adjusting the window size and judgment conditions, it can effectively remove noise under different noise intensities and distributions, while retaining the detail information of the image. When determining whether a pixel is a normal point, information such as the maximum value, minimum value, and median value of the pixels within the window region is fully considered, avoiding the problem of image detail blurring that may be caused by simple filtering methods. Only when it is determined that the pixel is an abnormal point and the window size reaches a certain degree is it processed, protecting the original details of the image as much as possible. This method can adaptively adjust the filtering strategy according to the local characteristics of the image. For an image with uneven noise distribution, by continuously enlarging the window for judgment, it can better adapt to the noise conditions in different regions and improve the filtering effect.

[0062] In some embodiments, obtaining the target pixel result corresponding to the pixel to be processed according to the target window includes: obtaining a corresponding target region from the initial image information according to the pixel to be processed and the target window; obtaining the target maximum value, target minimum value, and target median corresponding to the target window according to the target region; removing the target maximum value, the target minimum value, and the target median from the pixel values corresponding in the target window to obtain remaining pixel values; determining the pixel center corresponding to the remaining pixel values, and calculating the distance information between the remaining pixel values and the pixel center; determining the pixel weight corresponding to the remaining pixel values according to the distance information, and determining the target pixel result corresponding to the pixel to be processed according to the pixel weight and the remaining pixel values.

[0063] Exemplarily, centered on the pixel to be processed, a corresponding region is intercepted from the initial image information according to the size of the target window to obtain the target region. The target window is the window obtained after being enlarged in the previous steps, and through it, a larger range of pixel information can be obtained for subsequent processing.

[0064] Exemplarily, all pixel values in the target region are traversed and analyzed. The maximum value among them is found as the target maximum value, and the minimum value as the target minimum value. All pixel values in the target region are sorted, and the pixel value at the middle position is taken as the target median.

[0065] Exemplarily, the target maximum value, target minimum value, and target median are removed from the pixel values corresponding in the target window, and the remaining pixel values are the remaining pixel values for subsequent calculation. This step is to exclude possible extreme values and intermediate representative values, making the subsequent calculation focus more on general pixel information.

[0066] Exemplarily, the pixel center is obtained by calculating the average value of the remaining pixel values. For example, all remaining pixel values are added together, and then divided by the number of remaining pixel values to obtain the pixel center. Then, for each remaining pixel value, the distance information between it and the pixel center is calculated.

[0067] Exemplarily, the pixel weight corresponding to each remaining pixel value is determined according to the calculated distance information. Generally, the closer the remaining pixel value is to the pixel center, the greater its weight; the farther the distance, the smaller the weight. For example, a Gaussian function is used in combination with the distance information to assign pixel weights, and then each remaining pixel value is multiplied by its corresponding pixel weight, and then all the products are added together. The result obtained is the target pixel result corresponding to the pixel to be processed.

[0068] Specifically, by removing the maximum, minimum, and median values in the target window, the interference of these potential outliers to the target pixel result is avoided, making the calculated target pixel result more reliable and stable. Moreover, weights are assigned according to the distance between the remaining pixel values and the pixel center, which can highlight the pixel information that is closer to the overall pixel distribution. In this way, while retaining the main features of the image, unnecessary interference can be reduced, the quality and clarity of the image can be improved, and the influence degree of each pixel can be adaptively adjusted according to the actual distribution of pixels within the target window. For different image regions and different noise situations, reasonable processing can be carried out by calculating the distance and assigning weights, which has strong generality and adaptability.

[0069] Step S102: Determine the current parameters corresponding to the microcurrent detection, and obtain the response potential signal corresponding to the relevant region of the first target object under the current parameters.

[0070] Exemplarily, a series of preliminary experiments with different current parameters are designed. The current parameters usually include current intensity, current frequency, current duration, etc. For example, different current intensity levels (such as gradually increasing from a small intensity in the microampere level), different frequencies (such as low frequency, medium frequency, high frequency), and different durations (such as short - time pulses, long - time continuous current) are set, and then the responses of biological tissues corresponding to different current parameters are observed. Thus, based on the responses, the parameter range that can cause obvious and observable responses while not causing excessive damage or adverse effects to biological tissues is evaluated and screened.

[0071] Exemplarily, according to the parameter range, determine the current parameters corresponding to the microcurrent detection of the first target object, and use potential signal detection devices, such as electrodes, amplifiers, data acquisition cards, etc., to collect in real - time the response potential signal generated by the relevant region of the first target object under the action of the current.

[0072] Step S103: Perform nerve classification based on the response potential information to obtain the target nerve type corresponding to the relevant region.

[0073] Exemplarily, prepare samples containing these different nerve types, which can be obtained from experimental animals or suitable tissue models. Ensure the stable physiological state of the samples to guarantee the reliability of the experimental results. Place the samples in a suitable experimental environment, apply the set current parameters to different nerve types in the samples using devices such as electrodes, and then collect the current response information of different nerve types under this current stimulation. Then, cluster the current response information and the response potential information according to the clustering algorithm to form different clustering clusters, so as to obtain the clustering cluster corresponding to the response unit information. Furthermore, determine the type with the largest number of nerve types in the clustering cluster as the target nerve type corresponding to the relevant region.

[0074] In some embodiments, obtaining the target nerve type corresponding to the relevant region according to the reaction potential information includes: determining an initial nerve type and the relevant reaction potential corresponding to the initial nerve type; calculating the information similarity between the relevant reaction potential and the reaction potential information; and obtaining the target nerve type corresponding to the relevant region from the initial nerve types according to the information similarity.

[0075] Exemplarily, the initial nerve type and its corresponding relevant reaction potential are obtained from a database, and then the collected reaction potential information and the relevant reaction potential corresponding to the initial nerve type in the database are preprocessed, including operations such as noise removal, filtering, and normalization, to ensure the accuracy and comparability of the data.

[0076] Exemplarily, the Euclidean distance or cosine similarity is used to compare the collected reaction potential information with the relevant reaction potential corresponding to each initial nerve type in the database one by one to calculate the information similarity between them.

[0077] Exemplarily, the calculated information similarities are sorted in descending order, and then the initial nerve type with the highest information similarity is selected as the target nerve type corresponding to the relevant region. If there are multiple nerve types with the highest similarity and the similarity values are very close, other information (such as the anatomical location and functional characteristics of the nerves) can be further combined for comprehensive judgment.

[0078] In some embodiments, calculating the information similarity between the relevant reaction potential and the reaction potential information includes: obtaining first discrete data by discretizing the relevant reaction potential, and obtaining second discrete data by discretizing the reaction potential information; obtaining a first value by performing data statistics on the first discrete data, and obtaining a second value by performing data statistics on the second discrete data; establishing an initial matrix according to the first value and the second value, and calculating the data distance between a first sub-data in the first discrete data and a second sub-data in the second discrete data, and then filling the data distance into the initial matrix according to the first position corresponding to the first sub-data and the second position corresponding to the second sub-data; calculating a first standard deviation for the first discrete data and a second standard deviation for the second discrete data; obtaining the maximum standard deviation corresponding to between the first standard deviation and the second standard deviation, comparing the maximum standard deviation with the initial matrix, and when the matrix value corresponding to the initial matrix is greater than the preset maximum standard deviation, modifying the matrix value to a first target value; when the matrix value corresponding to the initial matrix is less than or equal to the preset maximum standard deviation, modifying the matrix value to a second target value; obtaining a target matrix according to the first target value and the second target value, and obtaining the maximum similarity path corresponding to between the first discrete data and the second discrete data according to the target matrix; determining the information similarity between the relevant reaction potential and the reaction potential information according to the maximum similarity path.

[0079] Exemplarily, the data range of the relevant reaction potential is divided into a plurality of intervals with equal widths, and then each data point is mapped into the corresponding interval to obtain the first discrete data, and the reaction potential information is processed by using the same discretization method as the relevant reaction potential. Similarly, the data range of the reaction potential information is divided according to the selected method, and the data points are mapped into the corresponding intervals to obtain the second discrete data.

[0080] Exemplarily, the data volume of the first discrete data is statistically obtained to obtain the data volume corresponding to the first discrete data, and the data volume is determined as the first value, and the second value is obtained by statistically processing the second discrete data with the same statistical index.

[0081] Exemplarily, an initial matrix is constructed according to a first numerical value and a second numerical value. The number of rows and columns of the initial matrix respectively correspond to the first numerical value of the first discrete data and the second numerical value of the second discrete data. Furthermore, for each first sub-data in the first discrete data and each second sub-data in the second discrete data, the Euclidean distance is used to calculate the data distance between them, and according to the first position of the first sub-data in the first discrete data and the second position of the second sub-data in the second discrete data, the calculated data distance is filled into the corresponding position of the initial matrix.

[0082] Exemplarily, the standard deviations of the first discrete data and the second discrete data are calculated respectively to obtain a first standard deviation and a second standard deviation. The magnitudes of the first standard deviation and the second standard deviation are compared, and the maximum value among them is taken as the maximum standard deviation.

[0083] Exemplarily, a first target value and a second target value are preset. Usually, the first target value represents a situation with low similarity, and the second target value represents a situation with high similarity. The initial matrix is traversed. When the matrix value in the matrix is greater than the maximum standard deviation, the matrix value is modified to the first target value, and the first target value is 0; when the matrix value is less than or equal to the maximum standard deviation, it is modified to the second target value, and the second target value is 1. Thus, after the above matrix value modification operation, a target matrix is obtained.

[0084] Exemplarily, the dynamic programming algorithm is used to find the maximum similarity path from the upper left corner to the lower right corner in the target matrix. This path represents the corresponding relationship with the highest similarity between the first discrete data and the second discrete data. Thus, according to the found maximum similarity path, combined with information such as the element values on the path, the information similarity between the relevant reaction potential and the reaction potential information is determined. For example, the information similarity can be represented by the proportion of the number of element values on the path that are the second target value to the total number of path elements.

[0085] Specifically, through data statistics and standard deviation calculation, the overall characteristics and distribution of the data can be better grasped. When processing data, modifying the matrix values with the standard deviation as a reference can effectively reduce the influence of noise and outliers in the data on the similarity calculation, and improve the stability and reliability of the data. And using the method of matrix and path search to calculate the similarity can comprehensively consider the corresponding relationship between each element in the two data sequences, rather than simply comparing the overall statistical indicators. This method can more accurately reflect the similarity degree between the relevant reaction potential and the reaction potential information, providing a more accurate basis for subsequent work such as neural classification.

[0086] Step S104: Align the relevant region and the initial image information to obtain a target alignment result, and determine the relevant edge information corresponding to the relevant region according to the target alignment result and the target recognition result.

[0087] Exemplarily, the Scale-Invariant Feature Transform (SIFT) or Speeded-Up Robust Features (SURF) algorithm is used to extract features from the initial image information to obtain first feature information. Also, the Scale-Invariant Feature Transform (SIFT) or Speeded-Up Robust Features (SURF) algorithm is used to extract second feature information corresponding to the relevant region. Then, the second feature information corresponding to the relevant region is translated and rotated in the second feature information corresponding to the initial image information, and the least squares method is used to solve the transformation matrix so that the relevant region can be aligned with the initial image information through this transformation matrix, thereby obtaining the target alignment result, which is the position information of the relevant region in the initial image information.

[0088] Exemplarily, the target alignment result and the target recognition result are integrated. The position information of the relevant region in the initial image information is marked in the target recognition result, and then the relevant edge information corresponding to the relevant region is obtained from the target recognition result according to this position information.

[0089] Step S105: Generate intercostal nerve plexus avoidance information according to the target nerve type and the relevant edge information, and display the intercostal nerve plexus avoidance information to the second target object.

[0090] Exemplarily, the intercostal nerves are screened according to the target nerve type, and the edge information corresponding to the intercostal nerves is accurately extracted from the relevant edge information. Then, the region defined by this part of the edge information is determined, so as to accurately obtain the region information corresponding to the intercostal nerves. This region information can be represented in forms such as coordinates and ranges. The region information is marked and distinguished with different colors to form the intercostal nerve plexus avoidance information and displayed for the second target object to view.

[0091] Please refer to Figure 2 , Figure 2An image recognition and micro-current detection system 200 for intercostal nerve plexus avoidance provided by an embodiment of the present application. The image recognition and micro-current detection system 200 for intercostal nerve plexus avoidance includes a first acquisition module 201, a second acquisition module 202, a data classification module 203, an edge determination module 204, and an information determination module 205. Among them, the first acquisition module 201 is used to acquire initial image information corresponding to a first target object, and perform edge detection on the initial image information to obtain a corresponding target recognition result; the second acquisition module 202 is used to determine current parameters corresponding to micro-current detection, and obtain a response potential signal corresponding to a relevant area of the first target object under the current parameters; the data classification module 203 is used to perform nerve classification according to the response potential information to obtain a target nerve type corresponding to the relevant area; the edge determination module 204 is used to align the relevant area and the initial image information to obtain a target alignment result, and determine relevant edge information corresponding to the relevant area according to the target alignment result and the target recognition result; the information determination module 205 is used to generate intercostal nerve plexus avoidance information according to the target nerve type and the relevant edge information, and display the intercostal nerve plexus avoidance information to a second target object.

[0092] In some embodiments, during the process of performing edge detection on the initial image information by the first acquisition module 201 to obtain a corresponding target recognition result, it performs:

[0093] Perform edge detection on the initial image information to obtain target edge information corresponding to the initial image information;

[0094] Use the Swin-Transformer network of the target recognition model to perform information encoding on the initial image information to obtain first encoded information, second encoded information, third encoded information, and fourth encoded information corresponding to four branches;

[0095] Use the feature fusion layer of the target recognition model to perform fusion processing on adjacent two of the first encoded information, the second encoded information, the third encoded information, and the fourth encoded information to obtain first fusion information, second fusion information, and third fusion information. Among them, the first fusion information includes first edge information and first target information, the second fusion information includes second edge information and second target information, and the third fusion information includes third edge information and third target information;

[0096] Feature extraction is performed on the first edge information by using the normalization network of the first spatial attention layer of the target recognition model to obtain first edge features, and feature enhancement is performed on the first edge features by using the first target information according to the first spatial attention network of the first spatial attention layer to obtain second edge features. Furthermore, the first edge features and the second edge features are fused according to the first channel attention network of the first spatial attention layer to obtain third edge features;

[0097] The edge enhancement layer of the target recognition model is used to enhance the third edge features in combination with the target edge information to obtain edge-enhanced features;

[0098] Feature extraction is performed on the first target information by using the normalization network of the second spatial attention layer of the target recognition model to obtain first target features, and feature enhancement is performed on the first target features by using the first edge information according to the second spatial attention network of the second spatial attention layer to obtain second target features. Furthermore, the first target features and the second target features are fused according to the second channel attention network of the second spatial attention layer to obtain third target features;

[0099] The mutual enhancement layer of the target recognition model is used to perform feature fusion on the edge-enhanced features and the third target features and then perform feature stitching to obtain corresponding target output features;

[0100] Edge information prediction is performed according to the target output features to obtain a target edge image corresponding to the initial image information;

[0101] Neural recognition is performed on the target edge image to obtain the target recognition result corresponding to the initial image information.

[0102] In some embodiments, during the process of performing edge detection on the initial image information to obtain the target edge information corresponding to the initial image information, the first acquisition module 201 performs:

[0103] Filter processing is performed on the initial image information to obtain target image information corresponding to the initial image information;

[0104] An initial segmentation threshold is determined, and the target image information is divided into corresponding first background region and first foreground region according to the initial segmentation threshold;

[0105] The target gray entropy corresponding to the target image information is obtained according to the first background region and the first foreground region;

[0106] The initial segmentation threshold is adjusted and at the same time the target gray entropy is recalculated to obtain the target segmentation threshold corresponding to the maximum target gray entropy;

[0107] Edge segmentation is performed on the target image information according to the target segmentation threshold to obtain the target edge information corresponding to the target image information;

[0108] Among them, the target gray entropy is obtained according to the following formula:

[0109]

[0110] Among them, S(th) represents the target gray entropy corresponding to the initial segmentation threshold being th, th represents the initial segmentation threshold, p(i) represents the pixel distribution corresponding to the pixel being i in the target image information, p(j) represents the pixel distribution corresponding to the pixel being j in the target image information, max represents the maximum pixel value corresponding to the target image information, p(k) represents the pixel distribution corresponding to the pixel being k in the target image information, and p(g) represents the pixel distribution corresponding to the pixel being g in the target image information.

[0111] In some embodiments, during the process of filtering the initial image information to obtain the target image information corresponding to the initial image information, the first acquisition module 201 performs:

[0112] Obtain the pixel to be processed from the initial image information and determine the initial window;

[0113] Obtain the corresponding window area from the initial image information according to the pixel to be processed and the initial window;

[0114] Obtain the maximum pixel value, minimum pixel value, and median pixel value corresponding to the initial window according to the window area;

[0115] If the pixel to be processed is less than the maximum pixel value and the pixel to be processed is greater than the minimum pixel value, determine that the pixel to be processed is a normal point; otherwise, determine whether the median pixel value is within the pixel range corresponding to the minimum pixel value and the maximum pixel value;

[0116] If the median pixel value is less than the maximum pixel value and the median pixel value is greater than the minimum pixel value, determine that the pixel to be processed is the normal point; otherwise, expand the initial window to obtain a target window, and re-determine whether the pixel to be processed is an abnormal point according to the target window. If the pixel to be processed is an abnormal point and the window size corresponding to the target window is greater than a preset value, perform pixel processing on the pixel to be processed according to the target window to obtain the target pixel result corresponding to the pixel to be processed;

[0117] Obtain the target image information corresponding to the initial image information according to the target pixel result.

[0118] In some embodiments, during the process of obtaining the target pixel result corresponding to the pixel to be processed by performing pixel processing on the pixel to be processed according to the target window, the first acquisition module 201 performs the following:

[0119] Obtain a corresponding target region from the initial image information according to the pixel to be processed and the target window;

[0120] Obtain the target maximum value, target minimum value, and target median corresponding to the target window according to the target region;

[0121] Remove the target maximum value, the target minimum value, and the target median from the pixel values corresponding in the target window to obtain the remaining pixel values;

[0122] Determine the pixel center corresponding to the remaining pixel values, and calculate the distance information between the remaining pixel values and the pixel center;

[0123] Determine the pixel weight corresponding to the remaining pixel values according to the distance information, and determine the target pixel result corresponding to the pixel to be processed according to the pixel weight and the remaining pixel values.

[0124] In some embodiments, during the process of obtaining the target nerve type corresponding to the relevant region by performing nerve classification according to the reaction potential information, the first acquisition module 201 performs the following:

[0125] Determine the initial nerve type and the relevant reaction potential corresponding to the initial nerve type;

[0126] Calculate the information similarity between the relevant reaction potential and the reaction potential information;

[0127] Obtain the target nerve type corresponding to the relevant region from the initial nerve types according to the information similarity.

[0128] In some embodiments, during the process of calculating the information similarity between the relevant reaction potential and the reaction potential information, the first acquisition module 201 performs the following:

[0129] Perform data discretization on the relevant reaction potential to obtain first discrete data, and perform data discretization on the reaction potential information to obtain second discrete data;

[0130] Perform data statistics on the first discrete data to obtain a first value, and perform data statistics on the second discrete data to obtain a second value;

[0131] An initial matrix is established according to the first numerical value and the second numerical value, and the data distance between the first sub-data in the first discrete data and the second sub-data in the second discrete data is calculated. Then, the data distance is filled into the initial matrix according to the first position corresponding to the first sub-data and the second position corresponding to the second sub-data.

[0132] The standard deviation of the first discrete data is calculated to obtain the first standard deviation, and the standard deviation of the second discrete data is calculated to obtain the second standard deviation.

[0133] The maximum standard deviation corresponding to the first standard deviation and the second standard deviation is obtained, and the maximum standard deviation is compared with the initial matrix. Then, when the matrix value corresponding to the initial matrix is greater than the preset maximum standard deviation, the matrix value is modified to the first target value; when the matrix value corresponding to the initial matrix is less than or equal to the preset maximum standard deviation, the matrix value is modified to the second target value.

[0134] A target matrix is obtained according to the first target value and the second target value, and the maximum similarity path corresponding to the first discrete data and the second discrete data is obtained according to the target matrix.

[0135] The information similarity between the relevant reaction potential and the reaction potential information is determined according to the maximum similarity path.

[0136] In some embodiments, the image recognition and microcurrent detection system 200 for intercostal nerve plexus avoidance can be applied to a terminal device.

[0137] It should be noted that those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working process of the above-described image recognition and microcurrent detection system 200 for intercostal nerve plexus avoidance can refer to the corresponding process in the foregoing embodiment of the image recognition and microcurrent detection method for intercostal nerve plexus avoidance, and will not be described herein again.

[0138] Please refer to Figure 3 , Figure 3 which is a schematic block diagram of the structure of a terminal device provided by an embodiment of the present invention.

[0139] As Figure 3 shown, the terminal device 300 includes a processor 301 and a memory 302. The processor 301 and the memory 302 are connected through a bus 303, and the bus is, for example, an I2C (Inter-integrated Circuit) bus.

[0140] Specifically, the processor 301 is configured to provide computing and control capabilities to support the operation of the entire terminal device. The processor 301 may be a central processing unit (CPU), and the processor 301 may also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0141] Specifically, the memory 302 may be a Flash chip, read-only memory (ROM), magnetic disk, optical disc, USB flash drive, or mobile hard disk, etc.

[0142] Those skilled in the art can understand that Figure 3 the structure shown in is only a block diagram of some structures related to the solution of the embodiment of the present invention, and does not constitute a limitation on the terminal device to which the solution of the embodiment of the present invention is applied. The specific server may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0143] Among them, the processor is configured to run a computer program stored in the memory and implement any one of the image recognition and microcurrent detection methods for intercostal nerve plexus avoidance provided by the embodiments of the present invention when executing the computer program.

[0144] In one embodiment, the processor is configured to run a computer program stored in the memory and implement the following steps when executing the computer program:

[0145] Collect the initial image information corresponding to the first target object, and perform edge detection on the initial image information to obtain the corresponding target recognition result;

[0146] Determine the current parameters corresponding to the microcurrent detection, and obtain the response potential signal corresponding to the relevant area of the first target object under the current parameters;

[0147] Perform nerve classification according to the response potential information to obtain the target nerve type corresponding to the relevant area;

[0148] Align the relevant region and the initial image information to obtain a target alignment result, and determine the relevant edge information corresponding to the relevant region according to the target alignment result and the target recognition result;

[0149] Generate intercostal nerve plexus avoidance information according to the target nerve type and the relevant edge information, and display the intercostal nerve plexus avoidance information to a second target object.

[0150] In some embodiments, during the process of performing edge detection on the initial image information to obtain a corresponding target recognition result, the processor 301 executes:

[0151] Perform edge detection on the initial image information to obtain target edge information corresponding to the initial image information;

[0152] Use the Swin-Transformer network of the target recognition model to perform information encoding on the initial image information to obtain first encoded information, second encoded information, third encoded information, and fourth encoded information corresponding to four branches;

[0153] Use the feature fusion layer of the target recognition model to perform fusion processing on adjacent two of the first encoded information, the second encoded information, the third encoded information, and the fourth encoded information to obtain first fusion information, second fusion information, and third fusion information, wherein the first fusion information includes first edge information and first target information, the second fusion information includes second edge information and second target information, and the third fusion information includes third edge information and third target information;

[0154] Use the normalization network of the first spatial attention layer of the target recognition model to extract features from the first edge information to obtain a first edge feature, and use the first target information to perform feature enhancement on the first edge feature according to the first spatial attention network of the first spatial attention layer to obtain a second edge feature, and then fuse the first edge feature and the second edge feature according to the first channel attention network of the first spatial attention layer to obtain a third edge feature;

[0155] Use the edge enhancement layer of the target recognition model to enhance the third edge feature in combination with the target edge information to obtain an edge enhanced feature;

[0156] Feature extraction is performed on the first target information by using the normalization network of the second spatial attention layer of the target recognition model to obtain first target features, and feature enhancement is performed on the first target features by using the first edge information according to the second spatial attention network of the second spatial attention layer to obtain second target features. Furthermore, the first target features and the second target features are fused according to the second channel attention network of the second spatial attention layer to obtain third target features;

[0157] The edge enhancement features and the third target features are fused by using the mutual enhancement layer of the target recognition model, and then feature splicing is performed to obtain corresponding target output features;

[0158] Edge information prediction is performed according to the target output features to obtain a target edge image corresponding to the initial image information;

[0159] Neural recognition is performed on the target edge image to obtain the target recognition result corresponding to the initial image information.

[0160] In some embodiments, during the process of performing edge detection on the initial image information to obtain the target edge information corresponding to the initial image information, the processor 301 executes:

[0161] Filtering processing is performed on the initial image information to obtain target image information corresponding to the initial image information;

[0162] An initial segmentation threshold is determined, and region division is performed on the target image information according to the initial segmentation threshold to obtain corresponding first background region and first foreground region;

[0163] The target gray entropy corresponding to the target image information is obtained according to the first background region and the first foreground region;

[0164] The initial segmentation threshold is adjusted and the target gray entropy is recalculated simultaneously to obtain a target segmentation threshold corresponding to the maximum target gray entropy;

[0165] Edge segmentation is performed on the target image information according to the target segmentation threshold to obtain the target edge information corresponding to the target image information;

[0166] Among them, the target gray entropy is obtained according to the following formula:

[0167]

[0168] Wherein, S(th) represents the target gray entropy corresponding to when the initial segmentation threshold is th, th represents the initial segmentation threshold, p(i) represents the pixel distribution corresponding to when the pixel in the target image information is i, p(j) represents the pixel distribution corresponding to when the pixel in the target image information is j, max represents the maximum pixel value corresponding to in the target image information, p(k) represents the pixel distribution corresponding to when the pixel in the target image information is k, and p(g) represents the pixel distribution corresponding to when the pixel in the target image information is g.

[0169] In some embodiments, during the process of filtering the initial image information to obtain the target image information corresponding to the initial image information, the processor 301 performs:

[0170] Obtain the pixel to be processed from the initial image information and determine the initial window;

[0171] Obtain the corresponding window area from the initial image information according to the pixel to be processed and the initial window;

[0172] Obtain the maximum pixel value, minimum pixel value, and median pixel value corresponding to the initial window according to the window area;

[0173] If the pixel to be processed is less than the maximum pixel value and the pixel to be processed is greater than the minimum pixel value, determine that the pixel to be processed is a normal point; otherwise, determine whether the median pixel value is within the pixel range corresponding to the minimum pixel value and the maximum pixel value;

[0174] If the median pixel value is less than the maximum pixel value and the median pixel value is greater than the minimum pixel value, determine that the pixel to be processed is the normal point; otherwise, expand the initial window to obtain a target window, and re-judge whether the pixel to be processed is abnormal according to the target window. If the pixel to be processed is an abnormal point and the window size corresponding to the target window is greater than a preset value, perform pixel processing on the pixel to be processed according to the target window to obtain the target pixel result corresponding to the pixel to be processed;

[0175] Obtain the target image information corresponding to the initial image information according to the target pixel result.

[0176] In some embodiments, during the process of performing pixel processing on the pixel to be processed according to the target window to obtain the target pixel result corresponding to the pixel to be processed, the processor 301 performs:

[0177] Obtain the corresponding target area from the initial image information according to the pixel to be processed and the target window;

[0178] Obtain the target maximum value, target minimum value, and target median value corresponding to the target window according to the target region;

[0179] Remove the target maximum value, the target minimum value, and the target median value from the corresponding pixel values in the target window to obtain the remaining pixel values;

[0180] Determine the pixel center corresponding to the remaining pixel values, and calculate the distance information between the remaining pixel values and the pixel center;

[0181] Determine the pixel weight corresponding to the remaining pixel values according to the distance information, and determine the target pixel result corresponding to the pixel to be processed according to the pixel weight and the remaining pixel values.

[0182] In some embodiments, during the process of obtaining the target nerve type corresponding to the relevant region by performing nerve classification according to the reaction potential information, the processor 301 executes:

[0183] Determine the initial nerve type and the relevant reaction potential corresponding to the initial nerve type;

[0184] Calculate the information similarity between the relevant reaction potential and the reaction potential information;

[0185] Obtain the target nerve type corresponding to the relevant region from the initial nerve types according to the information similarity.

[0186] In some embodiments, during the process of calculating the information similarity between the relevant reaction potential and the reaction potential information, the processor 301 executes:

[0187] Perform data discretization on the relevant reaction potential to obtain first discrete data, and perform data discretization on the reaction potential information to obtain second discrete data;

[0188] Perform data statistics on the first discrete data to obtain a first value, and perform data statistics on the second discrete data to obtain a second value;

[0189] Establish an initial matrix according to the first value and the second value, calculate the data distance between the first sub-data in the first discrete data and the second sub-data in the second discrete data, and then fill the data distance into the initial matrix according to the first position corresponding to the first sub-data and the second position corresponding to the second sub-data;

[0190] Calculate the first standard deviation for the first discrete data and calculate the second standard deviation for the second discrete data;

[0191] Obtain the corresponding maximum standard deviation between the first standard deviation and the second standard deviation, compare the maximum standard deviation with the initial matrix, and then when the matrix value corresponding to the initial matrix is greater than the preset maximum standard deviation, modify the matrix value to the first target value; when the matrix value corresponding to the initial matrix is less than or equal to the preset maximum standard deviation, modify the matrix value to the second target value;

[0192] Obtain a target matrix according to the first target value and the second target value, and obtain the corresponding maximum similarity path between the first discrete data and the second discrete data according to the target matrix;

[0193] Determine the information similarity between the relevant reaction potential and the reaction potential information according to the maximum similarity path.

[0194] It should be noted that those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working process of the above-described terminal device can refer to the corresponding process in the foregoing method embodiments for image recognition and microcurrent detection for intercostal nerve plexus avoidance, which will not be elaborated herein.

[0195] The embodiment of the present invention further provides a storage medium for computer-readable storage. The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of any image recognition and microcurrent detection system for intercostal nerve plexus avoidance provided in the specification of the embodiment of the present invention.

[0196] Among them, the storage medium may be an internal storage unit of the terminal device described in the foregoing embodiment, such as the hard disk or memory of the terminal device. The storage medium may also be an external storage device of the terminal device, such as a plug-in hard disk equipped on the terminal device, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc.

[0197] Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, and their appropriate combinations. In the hardware embodiments, the division between the functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, one physical component can have multiple functions, or one function or step can be executed by several physical components in cooperation. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or be implemented as hardware, or be implemented as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include a computer storage medium (or non-transitory medium) and a communication medium (or transitory medium). As is well known to those of ordinary skill in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information, such as computer-readable instructions, data structures, program modules, or other data. The computer storage medium includes, but is not limited to, RAM, ROM, EEPROM, flash memory, or other memory technologies, CD-ROM, digital versatile disk (DVD), or other optical disk storage, magnetic cassette, tape, magnetic disk storage, or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, as is well known to those of ordinary skill in the art, the communication medium generally contains computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transmission mechanism, and can include any information delivery medium.

[0198] It should be understood that the term "and / or" used in the specification and appended claims of the present invention refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations. It should be noted that, in this document, the terms "include", "comprise", or any other variant thereof are intended to cover 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 also includes elements inherent to such process, method, article, or system. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of another identical element in the process, method, article, or system including that element.

[0199] The serial numbers of the embodiments of the present invention above are only for description and do not represent the superiority or inferiority of the embodiments. As described above, the above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. An image recognition and microcurrent detection method for intercostal nerve plexus avoidance, characterized in that, The method includes: Collecting initial image information corresponding to a first target object, and performing edge detection on the initial image information to obtain a corresponding target recognition result; Determining current parameters corresponding to microcurrent detection, and obtaining a response potential signal corresponding to a relevant area of the first target object under the current parameters; Performing neural classification according to the response potential information to obtain a target nerve type corresponding to the relevant area; Aligning the relevant area and the initial image information to obtain a target alignment result, and determining relevant edge information corresponding to the relevant area according to the target alignment result and the target recognition result; Generating intercostal nerve plexus avoidance information according to the target nerve type and the relevant edge information, and displaying the intercostal nerve plexus avoidance information to a second target object.

2. The method according to claim 1, wherein The performing edge detection on the initial image information to obtain a corresponding target recognition result includes: Performing edge detection on the initial image information to obtain target edge information corresponding to the initial image information; Using the Swin-Transformer network of the target recognition model to perform information encoding on the initial image information to obtain first encoding information, second encoding information, third encoding information, and fourth encoding information corresponding to four branches; Using the feature fusion layer of the target recognition model to perform fusion processing on adjacent two of the first encoding information, the second encoding information, the third encoding information, and the fourth encoding information to obtain first fusion information, second fusion information, and third fusion information, wherein the first fusion information includes first edge information and first target information, the second fusion information includes second edge information and second target information, and the third fusion information includes third edge information and third target information; Using the normalization network of the first spatial attention layer of the target recognition model to extract features from the first edge information to obtain a first edge feature, and using the first spatial attention network of the first spatial attention layer to perform feature enhancement on the first edge feature according to the first target information to obtain a second edge feature, and then fusing the first edge feature and the second edge feature according to the first channel attention network of the first spatial attention layer to obtain a third edge feature; Using the edge enhancement layer of the target recognition model to enhance the third edge feature by combining the target edge information to obtain an edge enhancement feature; Using the normalization network of the second spatial attention layer of the target recognition model to extract features from the first target information to obtain a first target feature, and using the second spatial attention network of the second spatial attention layer to perform feature enhancement on the first target feature according to the first edge information to obtain a second target feature, and then fusing the first target feature and the second target feature according to the second channel attention network of the second spatial attention layer to obtain a third target feature; Using the mutual enhancement layer of the target recognition model to perform feature fusion on the edge enhancement feature and the third target feature and then perform feature splicing to obtain a corresponding target output feature; Edge information prediction is performed according to the target output feature to obtain a target edge image corresponding to the initial image information; Neural recognition is performed on the target edge image to obtain the target recognition result corresponding to the initial image information.

3. The method according to claim 2, wherein The performing edge detection on the initial image information to obtain target edge information corresponding to the initial image information includes: Performing filtering processing on the initial image information to obtain target image information corresponding to the initial image information; Determining an initial segmentation threshold, and performing region division on the target image information according to the initial segmentation threshold to obtain a corresponding first background region and a first foreground region; Obtaining a target gray entropy corresponding to the target image information according to the first background region and the first foreground region; Adjusting the initial segmentation threshold and simultaneously re-updating and calculating the target gray entropy to obtain a target segmentation threshold corresponding to the maximum target gray entropy; Performing edge segmentation on the target image information according to the target segmentation threshold to obtain the target edge information corresponding to the target image information.

4. The method according to claim 3, characterized in that, The performing filtering processing on the initial image information to obtain target image information corresponding to the initial image information includes: Obtaining a pixel to be processed from the initial image information and determining an initial window; Obtaining a corresponding window region from the initial image information according to the pixel to be processed and the initial window; Obtaining a pixel maximum value, a pixel minimum value, and a pixel median value corresponding to the initial window according to the window region; If the pixel to be processed is less than the pixel maximum value and the pixel to be processed is greater than the pixel minimum value, determining that the pixel to be processed is a normal point; otherwise, judging whether the pixel median value is within the pixel range corresponding to the pixel minimum value and the pixel maximum value; If the pixel median value is less than the pixel maximum value and the pixel median value is greater than the pixel minimum value, determining that the pixel to be processed is the normal point; otherwise, expanding the initial window to obtain a target window, and re-performing abnormal judgment on the pixel to be processed according to the target window. If the pixel to be processed is an abnormal point and the window size corresponding to the target window is greater than a preset value, performing pixel processing on the pixel to be processed according to the target window to obtain a target pixel result corresponding to the pixel to be processed; Obtaining the target image information corresponding to the initial image information according to the target pixel result.

5. The method according to claim 4, wherein The performing pixel processing on the pixel to be processed according to the target window to obtain a target pixel result corresponding to the pixel to be processed includes: Obtaining a corresponding target region from the initial image information according to the pixel to be processed and the target window; Obtaining a target maximum value, a target minimum value, and a target median value corresponding to the target window according to the target region; Removing the target maximum value, the target minimum value, and the target median value from the pixel values corresponding to the target window to obtain remaining pixel values; Determining a pixel center corresponding to the remaining pixel values and calculating distance information between the remaining pixel values and the pixel center; Determine the pixel weight corresponding to the remaining pixel value according to the distance information, and determine the target pixel result corresponding to the pixel to be processed according to the pixel weight and the remaining pixel value.

6. The method according to claim 1, wherein, The obtaining of the target nerve type corresponding to the relevant region by performing nerve classification according to the reaction potential information includes: Determine the initial nerve type and the relevant reaction potential corresponding to the initial nerve type; Calculate the information similarity between the relevant reaction potential and the reaction potential information; Obtain the target nerve type corresponding to the relevant region from the initial nerve types according to the information similarity.

7. The method according to claim 6, characterized in that The calculating of the information similarity between the relevant reaction potential and the reaction potential information includes: Perform data discretization on the relevant reaction potential to obtain first discrete data, and perform data discretization on the reaction potential information to obtain second discrete data; Perform data statistics on the first discrete data to obtain a first value, and perform data statistics on the second discrete data to obtain a second value; Establish an initial matrix according to the first value and the second value, and calculate the data distance between the first sub-data in the first discrete data and the second sub-data in the second discrete data, and then fill the data distance into the initial matrix according to the first position corresponding to the first sub-data and the second position corresponding to the second sub-data; Calculate the standard deviation of the first discrete data to obtain a first standard deviation, and calculate the standard deviation of the second discrete data to obtain a second standard deviation; Obtain the maximum standard deviation corresponding to the first standard deviation and the second standard deviation, compare the maximum standard deviation with the initial matrix, and then when the matrix value corresponding to the initial matrix is greater than the preset maximum standard deviation, modify the matrix value to a first target value; when the matrix value corresponding to the initial matrix is less than or equal to the preset maximum standard deviation, modify the matrix value to a second target value; Obtain a target matrix according to the first target value and the second target value, and obtain the maximum similarity path corresponding to the first discrete data and the second discrete data according to the target matrix; Determine the information similarity between the relevant reaction potential and the reaction potential information according to the maximum similarity path.

8. An image recognition and microcurrent detection system for intercostal nerve plexus avoidance, characterized in that Including: A first acquisition module, configured to acquire initial image information corresponding to a first target object, and perform edge detection on the initial image information to obtain a corresponding target recognition result; A second acquisition module, configured to determine the current parameters corresponding to micro-current detection, and obtain the reaction potential signal corresponding to the relevant region of the first target object under the current parameters; A data classification module, configured to perform nerve classification according to the reaction potential information to obtain the target nerve type corresponding to the relevant region; An edge determination module, configured to align the relevant region with the initial image information to obtain a target alignment result, and determine the relevant edge information corresponding to the relevant region according to the target alignment result and the target recognition result; An information determination module, configured to generate intercostal nerve plexus avoidance information according to the target nerve type and the relevant edge information, and display the intercostal nerve plexus avoidance information to a second target object.

9. A terminal device, characterized in that, The terminal device includes a processor and a memory; The memory is used to store a computer program; The processor is configured to execute the computer program and, when executing the computer program, implement the image recognition and microcurrent detection method for intercostal nerve plexus avoidance according to any one of claims 1 to 7.

10. A computer storage medium for computer storage, characterized in that, The computer storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of the image recognition and microcurrent detection method for intercostal nerve plexus avoidance according to any one of claims 1 to 7.

Images