Contact selection method and device, electronic equipment and storage medium

Abstract

The invention provides a contact selection method and device, electronic equipment and a storage medium, and relates to the technical field of brain image processing. The method comprises the following steps: carrying out multi-modal image acquisition on a detection target to obtain a whole-brain fiber bundle imaging image; performing activated tissue volume estimation on the whole brain fiber bundle imaging image based on an electrode reconstruction tool to obtain a contact activation area for each contact of the detection target; whole-brain fiber bundle tracking feature extraction is carried out based on a standard brain map and the contact activation area, and structure connection features aiming at the detection target are obtained; and performing feature recognition on the structure connection features based on a feature recognition model, and generating a feature recognition result indicating an optimal contact. According to the invention, the problem of low accuracy of contact selection in the prior art is solved.

Description

Technical Field

[0001] This application relates to the field of brain image processing technology. Specifically, this application relates to a contact selection method, device, electronic device, and storage medium. Background Art

[0002] Parkinson's disease (PD) seriously affects the quality of life of patients. In the prior art, deep brain stimulation (DBS) therapy is an effective means for treating Parkinson's disease, which realizes treatment by determining the optimal contact after surgery for deep brain stimulation. However, in the prior art, currently clinicians usually use a trial-and-error method to select the optimal contact, which is time-consuming and relies on doctor experience. It may also cause side effects due to stimulation, and the accuracy is not high. In addition, the method of identifying the optimal contact by obtaining microelectrode recordings (MER) is susceptible to the quality of the MER signal, and the adaptability of this method is limited, and the selection accuracy of contact selection is low.

[0003] As can be seen from the above, the problem of how to improve the effect of contact selection still needs to be solved. Summary of the Invention

[0004] This application provides a contact selection method, device, electronic device, and storage medium respectively, which can solve the problem of low accuracy of contact selection in the related art. The technical solutions are as follows:

[0005] According to one aspect of this application, a contact selection method, characterized in that it is applied to endoscopic images and includes:

[0006] Performing multi-modal image acquisition on a detection target to obtain a whole-brain fiber tractography image;

[0007] Estimating the activated tissue volume of the whole-brain fiber tractography image based on an electrode reconstruction tool to obtain a contact activation region for each contact of the detection target;

[0008] Extracting structural features from the whole-brain fiber tractography image based on a standard brain atlas and the contact activation region to obtain a structural connection matrix for the detection target;

[0009] Performing feature recognition on the structural connection matrix based on a feature recognition model to generate a feature recognition result indicating the optimal contact.

[0010] According to one aspect of this application, a contact selection device, characterized in that it includes:

[0011] An image acquisition module, configured to perform multi-modal image acquisition on a detection target to obtain a whole-brain fiber tractography image;

[0012] A region module, configured to estimate the activated tissue volume of the whole-brain fiber tract imaging image based on an electrode reconstruction tool, and obtain the contact activation regions of each contact for the detection target;

[0013] A feature extraction module, configured to extract brain fiber tract tracking features based on a standard brain atlas and the contact activation regions, and obtain the structural connection features for the detection target;

[0014] An identification module, configured to perform feature identification on the structural connection features based on a feature identification model, and generate a feature identification result indicating the optimal contact.

[0015] In an exemplary embodiment, the image acquisition module includes:

[0016] A first imaging module, configured to perform magnetic resonance imaging on the detection target to obtain a diffusion tensor imaging image and a preoperative weighted image;

[0017] A second imaging module, configured to perform helical system imaging on the detection target to obtain a postoperative computed tomography image;

[0018] A registration module, configured to register the preoperative diffusion tensor imaging image, the postoperative computed tomography image, and the preoperative weighted image to obtain a whole-brain fiber tract imaging image.

[0019] In an exemplary embodiment, the whole-brain fiber imaging image includes a first whole-brain fiber imaging image and a second whole-brain fiber imaging image; the registration module includes:

[0020] A first linear registration unit, configured to linearly register the postoperative computed tomography image based on the preoperative weighted image to obtain a first registered image;

[0021] A first non-linear registration unit, configured to non-linearly register the first registered image to obtain a first whole-brain fiber imaging image indicating the neuroanatomical atlas of the detection target;

[0022] A second linear registration unit, configured to linearly register the preoperative diffusion tensor imaging image based on the preoperative weighted image to obtain a second registered image;

[0023] A second non-linear registration unit, configured to non-linearly register the second registered image to obtain a second whole-brain fiber imaging image indicating the connection strength information in different brain regions of the detection target.

[0024] In an exemplary embodiment, the region module includes:

[0025] A correction unit, configured to perform brain shift correction processing on the first whole-brain fiber tract imaging image to generate a corrected image;

[0026] A trajectory detection unit, configured to perform electrode trajectory detection on the corrected image based on an electrode reconstruction tool to generate an electrode trajectory image;

[0027] An estimation unit, configured to perform electric field amplitude value conversion processing on the electrode trajectory image to obtain an estimated value of the activated tissue volume for each contact;

[0028] A region determination unit, configured to determine a contact activation region of each contact of the detection target in the first whole-brain fiber tractography image based on the estimated value of the activated tissue volume for each contact.

[0029] In an exemplary embodiment, the feature extraction module includes:

[0030] A whole-brain fiber tractography unit, configured to perform whole-brain fiber tractography processing on a second whole-brain fiber tractography image to obtain a microstructural image;

[0031] A matrix determination unit, configured to perform region segmentation on the microstructural image based on the contact activation region and a standard brain atlas to determine a structural connection matrix between different regions in the microstructural image.

[0032] In an exemplary embodiment, the matrix determination unit includes:

[0033] A region import sub-unit, configured to import the contact activation region into the microstructural image;

[0034] A region definition sub-unit, configured to define a region of interest in the microstructural image based on a standard brain atlas to determine the region of interest in the microstructural image;

[0035] A measurement sub-unit, configured to perform structural connectivity measurement on the microstructural image to obtain a structural connection matrix indicating the connection between the contact activation region and the region of interest.

[0036] In an exemplary embodiment, the recognition module includes:

[0037] A bilateral feature extraction unit, configured to extract bilateral features of each contact activation region relative to other regions based on the structural connection matrix;

[0038] A determination unit, configured to input the bilateral features corresponding to each contact activation region into a feature recognition model to perform priority determination on each contact activation region and generate a feature recognition result indicating the contact activation region with the highest priority among them.

[0039] According to one aspect of the present application, an electronic device includes at least one processor and at least one memory, wherein computer-readable instructions are stored on the memory; the computer-readable instructions are loaded and executed by the processor, so that the electronic device implements the contact selection method as described above.

[0040] According to one aspect of the present application, a storage medium stores computer-readable instructions, and the computer-readable instructions are loaded and executed by a processor to implement the contact selection method as described above.

[0041] According to one aspect of the present application, a computer program product includes computer-readable instructions. The computer-readable instructions are stored in a storage medium, and a processor of an electronic device reads the computer-readable instructions from the storage medium, loads and executes the computer-readable instructions, so that the electronic device implements the contact selection method as described above.

[0042] The beneficial effects brought by the technical solution provided by the present application are:

[0043] In the above technical solution, by determining the contact activation area for the whole-brain fiber tract imaging image based on the electrode reconstruction tool, the visualization of the stimulation position and range is realized. Integrating the standard brain atlas and the contact activation area into the structural connection matrix of the whole-brain fiber tract imaging image improves the dimension and efficiency of obtaining contact-related information, improves the recognition accuracy of the optimal contact, and provides an intelligent decision for postoperative planning. Using the feature recognition model to determine the optimal contact improves the accuracy and efficiency of finding the optimal contact. Thus, it can effectively solve the problem of low accuracy of contact selection existing in the related technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments of the present application.

[0045] Figure 1 is a schematic diagram of the implementation environment related to the present application;

[0046] Figure 2 is a flowchart of a contact selection method shown according to an exemplary embodiment;

[0047] Figure 3 is a schematic diagram of the determination process of the contact activation area shown according to an exemplary embodiment;

[0048] Figure 4 is a schematic diagram of the generation of the structural connection matrix shown according to an exemplary embodiment;

[0049] Figure 5It is a schematic diagram of a specific implementation of a contact selection method in an application scenario;

[0050] Figure 6 It is a structural block diagram of a contact selection device shown according to an exemplary embodiment;

[0051] Figure 7 It is a hardware structure diagram of a server shown according to an exemplary embodiment;

[0052] Figure 8 It is a structural block diagram of an electronic device shown according to an exemplary embodiment. Detailed implementation manners

[0053] The embodiments of the present application will be described in detail below. The examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be construed as a limitation to the present application.

[0054] Those skilled in the art of the present technology can understand that unless specifically stated otherwise, the singular forms "a", "an", "the" and "said" used herein may also include the plural forms. It should be further understood that the term "including" used in the specification of the present application means the presence of the described features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or their groups. It should be understood that when we say that an element is "connected" or "coupled" to another element, it can be directly connected or coupled to other elements, or there may also be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The phrase "and / or" used herein includes all or any unit and all combinations of one or more related listed items.

[0055] As mentioned above, in the prior art, currently clinicians usually use the trial-and-error method to select the optimal contact, which is time-consuming and relies on the doctor's experience. It may also cause side effects due to stimulation, and the accuracy rate is not high. In addition, the method of identifying the optimal contact by obtaining microelectrode recordings (MER) is easily affected by the quality of the MER signal. The adaptability of this method is limited, and the accuracy rate of contact selection is low.

[0056] As can be seen from the above, there are still defects in the low accuracy rate of contact selection in the related art.

[0057] Therefore, the contact selection method provided in this application can effectively improve the contact selection effect of contact selection. Correspondingly, this contact selection method is applicable to a contact selection device, which can be deployed in an electronic device. The electronic device can be a computer device configured with a von Neumann architecture. For example, the computer device includes a desktop computer, a laptop computer, a server, etc.

[0058] To make the objectives, technical solutions, and advantages of this application clearer, the following will further describe the embodiments of this application in detail with reference to the accompanying drawings.

[0059] Figure 1 It is a schematic diagram of an implementation environment related to a contact selection method. It should be noted that this implementation environment is only an example adapted to the present invention and should not be considered as providing any limitation to the scope of use of the present invention.

[0060] This implementation environment includes a collection end 110 and a server end 130.

[0061] Specifically, the collection end 110, which can also be regarded as a whole-brain fiber tractography image acquisition device, includes but is not limited to electronic devices with whole-brain fiber tractography image acquisition functions such as a magnetic resonance imaging system and a spiral system. For example, the collection end 110 is a 3-T magnetic resonance imaging system.

[0062] The server end 130 can be an electronic device such as a desktop computer, a laptop computer, a server, etc., or a computer cluster composed of multiple servers, or even a cloud computing center composed of multiple servers. Among them, the server end 130 is used to provide background services. For example, the background services include but are not limited to contact selection services, etc.

[0063] A network communication connection is pre-established between the server end 130 and the collection end 110 by means of wire or wireless, etc., and data transmission between the server end 130 and the collection end 110 is realized through this network communication connection. The transmitted data includes but is not limited to: whole-brain fiber tractography images, etc.

[0064] In an application scenario, through the interaction between the collection end 110 and the server end 130, the collection end 110 acquires a whole-brain fiber tractography image and uploads the whole-brain fiber tractography image to the server end 130 to request the server end 130 to provide a contact selection service.

[0065] For the server 130, after receiving the whole-brain fiber tract imaging image uploaded by the acquisition terminal 110, it calls the contact selection service to analyze the whole-brain fiber tract imaging image, fully obtain the feature information in the whole-brain fiber tract imaging image, determine the contact activation regions of each contact of the detection target, generate a structural connection matrix for the detection target, and perform feature recognition on the structural connection matrix based on the feature recognition model to determine the optimal contact among each contact, so as to solve the problem of low accuracy in determining the optimal contact in the related art.

[0066] Please refer to Figure 2 , this embodiment of the present application provides a contact selection method, which is applicable to an electronic device, and the electronic device may be Figure 1 the server 130 in the shown implementation environment, or may also be a desktop computer, a laptop computer, a server, etc.

[0067] In the following method embodiments, for the sake of description, it is illustrated by taking the execution subject of each step of the method as an electronic device as an example, but this is not a specific limitation thereto.

[0068] As Figure 2 shown, the method may include the following steps:

[0069] Step 210, perform multi-modal image acquisition on the detection target to obtain a whole-brain fiber tract imaging image.

[0070] Among them, the detection target refers to any object with contacts to be detected, and it can also be understood as any object that needs to perform deep brain stimulation of the subthalamic nucleus in the multi-modal image, and this object may be a person, an animal, etc. to be analyzed, and no limitation is made here. Through multi-modal image acquisition, a variety of imaging technologies are used to comprehensively obtain different levels and dimensions of information related to contact selection of the same target.

[0071] In a possible implementation manner, multi-modal image acquisition is performed on PD patients who have received bilateral STN-DBS surgery.

[0072] Step 230, estimate the activated tissue volume of the whole-brain fiber tract imaging image based on an electrode reconstruction tool to obtain the contact activation regions of each contact of the detection target.

[0073] Among them, the DBS electrode for acquiring the whole-brain fiber tract imaging image is reconstructed through the electrode reconstruction tool, and the stimulation range of the DBS electrode is evaluated, and the contact activation regions of each contact are estimated according to the stimulation range of the DBS electrode.

[0074] In a possible implementation manner, the Lead-DBS toolbox is used to reconstruct the electrode and estimate the contact activation regions of each contact of the detection target.

[0075] Step 250: Based on the standard brain atlas and the contact activation regions, extract the structural features from the whole-brain tractography images to obtain a structural connection matrix for the detection target.

[0076] It should be noted that the inventors have realized that due to the existence of various types of nerve fibers in the brain, electrical stimulation may lead to different effects, including activation, inhibition, or alteration of the neuronal activity pattern. Therefore, the effect of deep brain stimulation specifically depends on factors such as the type, direction, and integrity of the stimulated nerve fibers. That is to say, the stimulation effect is affected by the electrode position, stimulation intensity, and connection strength.

[0077] Therefore, in this application, by extracting the structural connection features from the whole-brain tractography images, a feature matrix that can display the structural connection features of the contact activation regions in the detection target is constructed. The feature matrix is used to guide the search for the optimal contacts, obtain the connection strength information in different brain regions, and ensure the accuracy of contact recognition.

[0078] Step 270: Based on the feature recognition model, perform feature recognition on the structural connection matrix to generate a feature recognition result indicating the optimal contacts.

[0079] Among them, the feature recognition model is a convolutional attention network with priority calculation based on the input bilateral features. It predicts the effect of each contact on deep brain stimulation through the bilateral features of each contact activation region and quantifies it as the priority of each contact. It can be understood that the higher the priority of a contact, the better the effect of deep brain stimulation, and the contact with the highest priority is the optimal contact.

[0080] Through the above process, by applying multimodal imaging technology, the limitations of a single imaging method can be overcome. It is possible to observe more comprehensively the arrangement, orientation, and connection status of nerve fibers in the brain. By combining structural information with functional information, researchers can more clearly understand the interactions between different brain regions and their changes in different neurological diseases. Based on the electrode reconstruction tool, the contact activation regions are determined from the whole-brain tractography images to realize the visualization of the stimulation position and range. Integrating the standard brain atlas and the contact activation regions into the structural connection matrix of the whole-brain tractography images can improve the dimension and efficiency of obtaining contact-related information, realize the prediction of the effect of deep deep brain stimulation on the contacts, improve the recognition accuracy of the optimal contacts, and provide intelligent decision-making for postoperative planning. Using the feature recognition model to determine the optimal contacts can improve the accuracy and efficiency of finding the optimal contacts.

[0081] In an exemplary embodiment, step 210 may include the following steps:

[0082] Step 211: Perform magnetic resonance imaging on the detection target to obtain diffusion tensor imaging images and preoperative weighted images.

[0083] Among them, contact-related information in the detection target is extracted by magnetic resonance imaging to generate a preoperative weighted image. At the same time, magnetic resonance imaging of the detection target in different diffusion weighting directions is performed to obtain multiple diffusion tensor imaging (DTI) images.

[0084] In a possible implementation, a 64-channel head coil is used to obtain preoperative weighted images and DTI images on a 3-T magnetic resonance imaging system. Among them, the resolution of the preoperative weighted image is 1 cubic millimeter, and the repetition time / echo time is 1780 milliseconds / 2.32 milliseconds. The resolution of the DTI image is 2 cubic millimeters, and the repetition time / echo time is 8000 / 64 milliseconds.

[0085] Step 213: Perform helical system imaging on the detection target to obtain a postoperative computed tomography (CT) image.

[0086] Among them, helical system imaging is used to obtain electrode position information of the detection target to generate a postoperative CT image to avoid the interference of pneumocephalus on the contact selection process.

[0087] In a possible implementation, a SOMATOM Emotion 16 / Definition AS / DefinitionFlash helical system is used to obtain a postoperative CT image with a slice thickness of 1 millimeter.

[0088] Step 215: Register the preoperative DTI images, postoperative CT images, and preoperative weighted images to obtain a whole-brain fiber tract imaging image.

[0089] Among them, by registering and fusing relevant information in the preoperative DTI images, postoperative CT images, and preoperative weighted images, comprehensive extraction of contact selection-related information is achieved.

[0090] In a possible implementation, the whole-brain fiber imaging image includes a first whole-brain fiber imaging image and a second whole-brain fiber imaging image.

[0091] In an exemplary embodiment, step 215 may include the following steps:

[0092] Step 2151: Perform linear registration of the postoperative CT image based on the preoperative weighted image to obtain a first registered image.

[0093] Step 2153: Perform non-linear registration on the first registered image to obtain a first whole-brain fiber imaging image indicating the neuroanatomical atlas of the detection target.

[0094] Among them, the registered images are normalized through a non-linear registration algorithm to standardize the image format. For example, the registered images are non-linearly registered to the MNI ICBM 2009b NLIN asymmetric space.

[0095] In a possible implementation, the postoperative computed tomography images are linearly registered to the preoperative weighted images through the Advanced Normalization Tools (ANT).

[0096] Step 2155: Linearly register the preoperative diffusion tensor imaging images based on the preoperative weighted images to obtain a second registered image.

[0097] Step 2157: Non-linearly register the second registered image to obtain a second whole-brain fiber imaging image indicating the connection strength information in different brain regions of the detection target.

[0098] In a possible implementation, the preoperative diffusion tensor imaging images and the preoperative weighted images are registered to the MNI space through SPM12.

[0099] Through the above process, through multiple image acquisitions, the dimension and breadth of obtaining contact-related information are expanded, ensuring the accuracy of subsequent optimal contact recognition. By linearly registering the comprehensive image information, the information comprehensiveness is improved, and through non-linear registration for normalization, the images are standardized, facilitating subsequent analysis of data using neuroanatomical atlases and improving the image processing efficiency.

[0100] In an exemplary embodiment, step 230 may include the following steps:

[0101] Step 231: Perform brain shift correction processing on the first whole-brain fiber tract imaging image to generate a corrected image.

[0102] Among them, the image deviation in the first whole-brain fiber tract imaging image is eliminated through the brain shift correction processing, thereby improving the consistency and comparability of the images. By eliminating the shift and deformation, the brain features can be extracted more accurately and statistical analysis can be performed, improving the reliability of the research results.

[0103] Step 233: Detect the electrode trajectories on the corrected image based on the electrode reconstruction tool to generate an electrode trajectory image.

[0104] In a possible implementation, the PaCER toolbox is used to manually refine and detect the electrode trajectories from the corrected image.

[0105] Step 235: Perform electric field amplitude value conversion processing on the electrode trajectory image to obtain the estimated value of the activated tissue volume for each contact.

[0106] In a possible implementation, the SimBio / FieldTrip toolbox is used to convert the electric field amplitude values by setting the average stimulus voltage to 1.6 V and applying a heuristic threshold of 0.2 V / mm, generating estimated values of the activated tissue volume corresponding to each contact point.

[0107] Step 237: Based on the estimated values of the activated tissue volume for each contact point, determine the contact activation regions of the detection target at each contact point in the first whole-brain fiber tractography image.

[0108] Specifically, as Figure 3 shown, Figure 3 illustrates the determination process of the contact activation region. Among them, the postoperative CT is the postoperative computed tomography image, and the preoperative T1 is the preoperative weighted image. In parts a and b, co-registration is achieved through linear registration of the postoperative computed tomography image and the preoperative weighted image, and normalized to the MNI space. The image accuracy is improved by correcting brain shift in part c. The PaCER toolbox is used to manually refine the detection of electrode trajectories from the postoperative computed tomography image, reconstruct and manually adjust the electrodes. Finally, the contact activation regions of each contact point determined according to the reconstructed electrode information are as shown by the arrows in part e.

[0109] Through the above process, the integration of multi-modal data in the first whole-brain fiber tractography image is promoted, enabling the information obtained from different imaging sources in the first whole-brain fiber tractography image to be better integrated together to support comprehensive analysis. By detecting the electrode trajectories in the first whole-brain fiber tractography image, the stimulation intensity of deep brain stimulation is visualized, improving the recognition accuracy of each contact point of the target. By determining the contact activation regions of each contact point of the detection target, the optimal contact recognition efficiency is improved.

[0110] In an exemplary embodiment, step 250 may include the following steps:

[0111] Step 251: Perform whole-brain fiber tractography processing on the second whole-brain fiber tractography image to obtain a microstructure image.

[0112] Among them, by performing whole-brain fiber tracking on the diffusion tensor imaging components in the second whole-brain fiber tractography image to visualize the microstructure of the detection target, the microstructure of the detection target is modeled to generate a microstructure image.

[0113] In a possible implementation, whole-brain fiber tractography is performed based on the voxel-based morphometry (VBM) interface, and SPM is used to image brain regions such as gray matter, white matter, and cerebrospinal fluid to obtain a microstructure image.

[0114] In a possible implementation, whole-brain fiber tractography processing is performed on the second whole-brain fiber tractography image based on the generalized Q-sampling imaging algorithm.

[0115] Step 253: Based on the contact activation region and the standard brain atlas, perform regional segmentation on the microstructure image to determine the structural connection matrix between different regions in the microstructure image.

[0116] In an exemplary embodiment, step 253 may include the following steps:

[0117] Step 2531: Import the contact activation region into the microstructure image.

[0118] It can be understood that by determining the specific coordinates of the contact activation region in the microstructure image, the microstructure image can identify each contact activation region, thereby completing the import of the contact activation region.

[0119] Step 2533: Define the regions of interest in the microstructure image based on the standard brain atlas to determine the regions of interest in the microstructure image.

[0120] Among them, the standard brain atlas contains the definitions of different regions of interest in the detection target. By defining the regions of interest in the microstructure image through the standard brain atlas and warping the image back to a single native space, the specific coordinates corresponding to each region of interest in the microstructure image can be determined.

[0121] In a possible implementation, the standard brain atlas is determined according to the standard space where the second whole-brain tractography image is located. For example, in the standard space, the Automated Anatomical Labeling 3 (AAL3) atlas and the DISTAL medium atlas are selected as the standard brain atlases. Among them, the AAL3 atlas contains 166 gray matter regions of interest for capturing cortical and subcortical structures, and the DISTAL medium atlas contains 38 sub-regions of interest for obtaining deep brain stimulation-related structures.

[0122] Step 2535: Measure the structural connectivity of the microstructure image to obtain a structural connection matrix indicating the connection between the contact activation region and the regions of interest.

[0123] It should be noted that the stimulating effect of deep brain stimulation of the subthalamic nucleus is related to specific brain regions that affect clinical outcomes. For example, the supplementary motor area (SMA), primary motor cortex (PMC), superior frontal gyrus (SFG), thalamus in the brain, and brain regions related to side effects, such as the dentate mammillary tract (DRTT) and internal capsule. By dividing the microstructure image into different regions, and then analyzing the structural connection relationship based on the regions to determine the interaction between the contact activation region and the region of interest, various influencing factors and their interactions can be considered to provide a more comprehensive analysis of the system state, effectively handle the complex effect of deep brain stimulation in the contact activation region, and improve the adaptability to deep brain stimulation. Based on the overall brain connectivity between deep brain stimulation and symptom-related brain regions, it provides a scientific basis for decision-making and can improve the accuracy and comprehensiveness of optimal contact identification.

[0124] In a possible implementation, the structural connection matrix uses a typical structural connectivity measurement method, and the connectivity strength and microstructure diffusion characteristics between the contact activation region and the region of interest are represented by fractional anisotropy (FA) and the number of streamlines (NOS) respectively.

[0125] Specifically, as Figure 4 shown, the generalized Q-sampling imaging algorithm is used to process the second whole-brain fiber tractography image data for whole-brain fiber tracking to establish a microstructure image. After importing the contact activation region (VTA), the regions of interest defined by the brain atlas AAL3, and the regions of interest defined by the DISTAL medium atlas into the microstructure image, a structural connection matrix between the contact activation region (VTA) and the regions of interest is obtained.

[0126] In a possible implementation, the software of DSI Studio is used for whole-brain fiber tracking to calculate the structural connection matrix.

[0127] Through the above process, it is realized to estimate the stimulation range using the contact activation region, extract the structural connectivity characteristics of the contact activation region and the whole-brain region, increase the information dimension for determining the optimal contact, and improve the accuracy of determining the optimal contact.

[0128] In an exemplary embodiment, step 270 may include the following steps:

[0129] Step 271, extracting the bilateral features of each contact activation region relative to other regions based on the structural connection matrix.

[0130] Among them, the bilateral feature is the structural connection label between each contact activation region and other brain regions. The structural connection relationship of each contact in the brain is accurately quantified through the bilateral features of each contact activation region, ensuring the efficiency and accuracy of the feature analysis of each contact.

[0131] For example, when the AAL3 atlas in the microstructure image contains 166 gray matter regions of interest for capturing cortical and subcortical structures, the DISTAL medium atlas contains 38 sub-regions of interest for obtaining deep brain stimulation-related structures, and there are 8 contact activation regions corresponding to 8 contacts, significant features between the bilateral 8 contacts and other brain regions are extracted, and 1×424-dimensional bilateral features corresponding to each contact are obtained. The obtained bilateral features are normalized by z-scores and imported into the classifier of the feature recognition model to achieve the classification of the optimal contact points.

[0132] Step 273: Input the bilateral features corresponding to each contact activation region into the feature recognition model to determine the priority of each contact activation region, and generate a feature recognition result indicating the contact activation region with the highest priority among them.

[0133] Specifically, by inputting the bilateral features corresponding to each contact activation region, the feature recognition model extracts the significant features between each contact activation region corresponding to each contact and other brain regions, and obtains multi-dimensional features for each contact. The features of each obtained contact are normalized by z-scores and imported into the classifier to generate the priority for each contact.

[0134] Among them, the feature recognition model includes seven stages, each stage includes at least one convolutional layer and one classifier module, and after each convolutional layer, there is a batch normalization layer and a rectified linear unit. Among them, at least three intermediate stages (such as from the fourth stage to the sixth stage) end with a max pooling layer. The last stage also includes a fully connected layer and a function output layer.

[0135] In a possible implementation, the CBAM module is selected as the classifier module to save parameters and computing resources, and at the same time ensure that the classifier module is seamlessly integrated into the existing network architecture as a plug-and-play module.

[0136] Through the above process, the structural connectivity between each contact activation region and each region is identified through bilateral features, which helps to analyze the superiority of each contact, can help select the optimal contact, and predict and optimize the efficacy of deep brain stimulation for each contact. By generating priorities through the feature recognition model, the accurate recognition of each contact is improved, and at the same time, the speed of contact recognition is relatively fast, and the efficiency of determining the optimal contact is improved.

[0137] Figure 5It is a schematic diagram of the specific implementation of a contact selection method in an application scenario. In this application scenario, a retrospective analysis was conducted on 100 PD patients (age range, 40 - 80 years old; average disease course, 8 years) who underwent bilateral STN-DBS surgery in the hospital, and the inclusion and exclusion criteria were listed in detail. Among them, the DBS surgery was performed using a stereotactic surgical robot under general anesthesia, such as robotic stereotactic assistance (ROSA, MedTech Surgical, Inc.), and two DBS electrodes targeting bilateral STN were implanted.

[0138] Considering the micro-lesion effect, DBS programming was performed 1 month after the surgery. Senior neurologists used a monopolar review strategy to stimulate and test each contact to observe the effects on the patient's motor improvement and side effects. The electrodes were selected in constant voltage mode and monopolar mode. The initial DBS parameters were set as follows: the pulse width was 60 ms, the frequency was 130 Hz, the stimulation voltage gradually increased from 0.7 V to 2.0 V, and the average voltage was 1.6 V. The optimal contact was selected based on the improvement of electro-stimulated symptoms and the absence of side effects, and the average clinical programming time was 3 hours. The contact method was not changed within 3 months after programming due to dissatisfaction with symptom improvement or side effects. This study was approved by the local hospital's Human Research Ethics Committee, and all participants provided written informed consent.

[0139] For 100 PD patients, linear registration and non-linear registration were performed on the postoperative computed tomography images (CT), preoperative weighted images (T1), and preoperative diffusion tensor imaging images (DTI), and they were normalized to the MNI space. Then, the DBS electrodes were reconstructed, and the contact activation area (VTA) of each contact was estimated using the Lead-DBS toolbox. The stimulation range was estimated using the activated tissue volume of each contact, and the structural connectivity features of the activation area and the whole brain area, including FA and NOS, were extracted. The stimulation position and range were visualized to obtain the contact activation area of each contact. After that, based on the standard brain atlases AAL3 and DISTAL, the contact activation area and the region of interest were divided, and the structural connection features of the whole brain were extracted by the tractography method to generate a feature matrix. Finally, a convolutional attention network was used to identify the structural connection features of the feature matrix to determine the optimal contact.

[0140] Among them, the feature recognition model is the CNN+CBAM network. During the training process of the feature recognition model, in order to make the feature recognition accuracy of the feature recognition model as high as possible, a training set and a test set were generated from the feature matrix to be labeled, and the feature recognition model was trained to improve the judgment ability of the feature recognition model. By continuously updating the model parameters of the feature recognition model, the result of the feature recognition model was trained towards identifying the optimal contact.

[0141] In this application scenario, by combining DBS electrode reconstruction with VTA estimation results, the visualization of the stimulation location and range is achieved. By integrating the brain atlas and electrode contact VTA into the structural connection features of the whole brain, and using the Convolutional Block Attention Module (CBAM) to integrate the Convolutional Neural Network (CNN) (i.e., the convolutional attention network) to identify the structural connection features, the purpose of improving the efficiency of personalized optimal contact selection is realized.

[0142] In the present invention, experiments are carried out on the data of 800 contacts from 100 PD patients. The data are randomly assigned, 80% for training and 20% for testing, and 5-fold cross-validation is adopted. The results show that the average accuracy of the method for selecting the best starting contacts is 97.63%, the average precision is 94.50%, the average recall rate is 94.46%, and the average specificity is 98.18%, which proves the feasibility and effectiveness of the method. Therefore, the results of the present invention can provide useful guiding value for neurosurgeons. It can provide a reference for postoperative programming, thereby promoting the rapid and accurate programming of Deep Brain Stimulation (DBS) therapy, and may further promote the development and application of new technologies in deep brain stimulation surgery.

[0143] The following is an embodiment of the device of the present application, which can be used to execute the contact selection method involved in the present application. For the details not disclosed in the embodiment of the device of the present application, please refer to the method embodiment of the contact selection method involved in the present application.

[0144] Please refer to Figure 6 , in the embodiment of the present application, a contact selection device 1000 is provided, including but not limited to: an image acquisition module 1010, a region module 1030, a feature extraction module 1050, and an identification module 1070.

[0145] Among them, the image acquisition module 1010 is used to perform multi-modal image acquisition on the detection target to obtain a whole-brain fiber tract imaging image;

[0146] The region module 1030 is used to estimate the activated tissue volume of the whole-brain fiber tract imaging image based on an electrode reconstruction tool to obtain a contact activation region for each contact of the detection target;

[0147] The feature extraction module 1050 is used to extract brain fiber tract tracking features based on a standard brain atlas and the contact activation region to obtain structural connection features for the detection target;

[0148] The identification module 1070 is used to perform feature identification on the structural connection features based on a feature identification model to generate a feature identification result indicating the optimal contact.

[0149] It should be noted that when the contact selection device provided in the above embodiments performs contact selection, only the division of the above functional modules is used for illustration. In actual applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the contact selection device will be divided into different functional modules to complete all or part of the functions described above.

[0150] In addition, the contact selection device and the contact selection method provided in the above embodiments belong to the same concept. The specific ways in which each module performs operations have been described in detail in the method embodiments, and will not be elaborated here.

[0151] Figure 7 is a schematic structural diagram of a server shown according to an exemplary embodiment. This server is applicable to Figure 1 the server 130 in the shown implementation environment.

[0152] It should be noted that this server is only an example adapted to the present application and cannot be considered as providing any limitation to the scope of use of the present application. This server cannot be interpreted as requiring dependence on or necessarily having Figure 7 one or more components in the shown exemplary server 2000.

[0153] The hardware structure of the server 2000 may vary greatly due to different configurations or performances. For example, Figure 7 as shown, the server 2000 includes: a power supply 210, an interface 230, at least one memory 250, and at least one central processing unit (CPU) 270.

[0154] Specifically, the power supply 210 is used to provide working voltages for each hardware device on the server 2000.

[0155] The interface 230 includes at least one wired or wireless network interface 231 for interacting with external devices. For example, for Figure 1 the interaction between the acquisition end 110 and the server end 130 in the shown implementation environment.

[0156] Of course, in other examples adapted to the present application, the interface 230 may further include at least one serial-parallel conversion interface 233, at least one input-output interface 235, and at least one USB interface 237, etc. As Figure 7 shown, this is not a specific limitation here.

[0157] The memory 250, as the carrier for resource storage, can be a read-only memory, a random access memory, a magnetic disk, an optical disc, etc. The resources stored thereon include an operating system 251, application programs 253, data 255, etc. The storage method can be transient storage or permanent storage.

[0158] Among them, the operating system 251 is used to manage and control each hardware device and application program 253 on the server 2000, so as to realize the operation and processing of the massive data 255 in the memory 250 by the central processing unit 270. It can be Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSD TM, etc.

[0159] The application program 253 is a computer-readable instruction that completes at least one specific task based on the operating system 251. It can include at least one module ( Figure 7 (not shown), and each module can respectively contain computer-readable instructions for the server 2000. For example, the contact selection device can be regarded as an application program 253 deployed on the server 2000.

[0160] The data 255 can be photos, pictures, etc. stored in the magnetic disk, and can also be various training image sets, images to be processed, etc., which are stored in the memory 250.

[0161] The central processing unit 270 can include one or more than one processors, and is set to communicate with the memory 250 through at least one communication bus, so as to read the computer-readable instructions stored in the memory 250, and then realize the operation and processing of the massive data 255 in the memory 250. For example, the contact selection method is completed in the form of reading a series of computer-readable instructions stored in the memory 250 by the central processing unit 270.

[0162] In addition, the present application can also be implemented by a hardware circuit or a combination of a hardware circuit and software. Therefore, the implementation of the present application is not limited to any specific hardware circuit, software, and the combination of both.

[0163] Please refer to Figure 8 , in the embodiments of the present application, an electronic device 4000 is provided. The electronic device 400 can include: a desktop computer, a laptop computer, a server, etc.

[0164] In Figure 8 , the electronic device 4000 includes at least one processor 4001 and at least one memory 4003.

[0165] Among them, the data interaction between the processor 4001 and the memory 4003 can be realized through at least one communication bus 4002. The communication bus 4002 may include a path for transmitting data between the processor 4001 and the memory 4003. The communication bus 4002 can be a PCI (Peripheral Component Interconnect) bus, an EISA (Extended Industry Standard Architecture) bus, or the like. The communication bus 4002 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 8 only a thick line is used to represent it in Figure 8 , but it does not mean that there is only one bus or one type of bus.

[0166] Optionally, the electronic device 4000 may further include a transceiver 4004, and the transceiver 4004 can be used for data interaction between the electronic device and other electronic devices, such as data sending and / or data receiving, etc. It should be noted that in practical applications, the transceiver 4004 is not limited to one, and the structure of the electronic device 4000 does not constitute a limitation to the embodiments of the present application.

[0167] The processor 4001 can be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute various exemplary logic blocks, modules, and circuits described in combination with the disclosure of the present application. The processor 4001 can also be a combination that realizes computing functions, such as a combination including one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0168] The memory 4003 can be a ROM (Read Only Memory), or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory), or other types of dynamic storage devices that can store information and instructions. It can also be an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory), or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage media, or other magnetic storage devices, or any other medium that can be used to carry or store the desired computer-readable instructions in the form of instructions or data structures and can be accessed by the electronic device 400, but is not limited thereto.

[0169] Computer-readable instructions are stored on the memory 4003, and the processor 4001 can read the computer-readable instructions stored in the memory 4003 through the communication bus 4002.

[0170] When the computer-readable instructions are executed by the processor 4001, the contact selection method in the above embodiments is implemented.

[0171] In addition, an embodiment of the present application provides a storage medium on which computer-readable instructions are stored. The computer-readable instructions are loaded and executed by a processor to implement the contact selection method as described above.

[0172] An embodiment of the present application provides a computer program product. The computer program product includes computer-readable instructions. The computer-readable instructions are stored in a storage medium, and the processor of the electronic device reads the computer-readable instructions from the storage medium, loads and executes the computer-readable instructions, so that the electronic device implements the contact selection method as described above.

[0173] Compared with related technologies, the present application applies multimodal imaging technology, which can overcome the limitations of single imaging methods. It can observe the arrangement, orientation, and connection status of nerve fibers in the brain more comprehensively. By combining structural information with functional information, researchers can better understand the interactions between different brain regions and their changes in different neurological diseases. Based on the electrode reconstruction tool, the contact activation area is determined for the whole-brain tractography image, realizing the visualization of the stimulation position and range. The standard brain atlas and the contact activation area are integrated into the structural connection matrix of the whole-brain tractography image, improving the dimension and efficiency of obtaining contact-related information, realizing the effect prediction of deep brain stimulation for the contacts, improving the recognition accuracy of the optimal contacts, and providing intelligent decision-making for postoperative planning. The feature recognition model is used to determine the optimal contacts, improving the accuracy and efficiency of finding the optimal contacts.

[0174] Through multiple image acquisitions, the dimension and breadth of obtaining contact-related information are expanded, ensuring the accuracy of subsequent optimal contact recognition. By linearly registering comprehensive image information, the degree of information integration is improved, and normalization is performed through non-linear registration to standardize the images, facilitating subsequent data analysis using the neuroanatomical atlas and improving the image processing efficiency. Promote the integration of multimodal data in the first whole-brain tractography image, enabling better integration of information obtained from different imaging sources in the first whole-brain tractography image and supporting comprehensive analysis. Through the detection of electrode trajectories in the first whole-brain tractography image, the stimulation intensity of deep brain stimulation is visualized, improving the recognition accuracy of each target contact. By determining the contact activation area of each detected target contact, the efficiency of optimal contact recognition is improved. By dividing the microstructure image into different regions, and then analyzing the structural connection relationship based on the regions, the interaction between the contact activation area and the region of interest is determined. Multiple influencing factors and their interactions can be considered to provide a more comprehensive system state analysis, effectively handling the complex effects of deep brain stimulation in the contact activation area and improving the adaptability to deep brain stimulation. Based on the overall brain connectivity between deep brain stimulation and symptom-related brain regions, a scientific decision-making basis is provided, which can improve the accuracy and comprehensiveness of optimal contact recognition. Realize the estimation of the stimulation range using the contact activation area, extract the structural connectivity features of the contact activation area and the whole-brain area, increase the dimension of optimal contact determination information, and improve the accuracy of optimal contact determination. By bilaterally identifying the structural connectivity between each contact activation area and each region, it helps to analyze the superiority of each contact, can assist in selecting the optimal contact, and predict and optimize the efficacy of deep brain stimulation for each contact. Through the generation of priorities by the feature recognition model, the accurate recognition of each contact is improved, and at the same time, the speed of contact recognition is relatively fast, and the efficiency of optimal contact determination is improved.

[0175] It should be understood that although the steps in the flowchart of the accompanying drawings are shown sequentially in the direction of the arrows, these steps are not necessarily executed sequentially in the order indicated by the arrows. Unless there is a clear indication in this document, there is no strict order restriction for the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same moment, but can be executed at different moments, and their execution order is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or sub-steps or stages of other steps.

[0176] The above are only some embodiments of the present application. It should be noted that for those of ordinary skill in the art in this technical field, without departing from the principle of the present application, several improvements and refinements can still be made, and these improvements and refinements should also be regarded as the protection scope of the present application.

Claims

1. A contact selection method, characterized in that, Including: Performing multi-modal image acquisition on a detection target to obtain a whole-brain fiber tractography image; Estimating the activated tissue volume of the whole-brain fiber tractography image based on an electrode reconstruction tool to obtain a contact activation region for each contact of the detection target; Extracting structural features from the whole-brain fiber tractography image based on a standard brain atlas and the contact activation region to obtain a structural connection matrix for the detection target; Performing feature recognition on the structural connection matrix based on a feature recognition model to generate a feature recognition result indicating the optimal contact.

2. The method according to claim 1, wherein The performing multi-modal image acquisition on a detection target to obtain a whole-brain fiber tractography image includes: Performing magnetic resonance imaging on the detection target to obtain a diffusion tensor imaging image and a preoperative weighted image; Performing helical system imaging on the detection target to obtain a postoperative computed tomography image; Registering the preoperative diffusion tensor imaging image, the postoperative computed tomography image, and the preoperative weighted image to obtain a whole-brain fiber tractography image.

3. The method according to claim 2, wherein The whole-brain fiber imaging image includes a first whole-brain fiber imaging image and a second whole-brain fiber imaging image; The registering the preoperative diffusion tensor imaging image, the postoperative computed tomography image, and the preoperative weighted image to obtain a whole-brain fiber tractography image includes: Performing linear registration on the postoperative computed tomography image based on the preoperative weighted image to obtain a first registered image; Performing non-linear registration on the first registered image to obtain a first whole-brain fiber imaging image indicating the neuroanatomical atlas of the detection target; Performing linear registration on the preoperative diffusion tensor imaging image based on the preoperative weighted image to obtain a second registered image; Performing non-linear registration on the second registered image to obtain a second whole-brain fiber imaging image indicating the connection strength information in different brain regions of the detection target.

4. The method according to any one of claims 1 to 3, characterized in that, The estimating the activated tissue volume of the whole-brain fiber tractography image based on an electrode reconstruction tool to obtain a contact activation region for each contact of the detection target includes: Performing brain shift correction processing on the first whole-brain fiber tractography image to generate a corrected image; Detecting electrode trajectories of the corrected image based on an electrode reconstruction tool to generate an electrode trajectory image; Performing electric field amplitude value conversion processing on the electrode trajectory image to obtain an activated tissue volume estimation value for each contact; Based on the activated tissue volume estimation value for each contact, determining the contact activation region for each contact of the detection target in the first whole-brain fiber tractography image.

5. The method according to any one of claims 1 to 3, characterized in that The extracting structural features from the whole-brain fiber tractography image based on a standard brain atlas and the contact activation region to obtain a structural connection matrix for the detection target includes: Performing whole-brain fiber tractography processing on the second whole-brain fiber tractography image to obtain a microstructural image; Performing region segmentation on the microstructural image based on the contact activation region and the standard brain atlas to determine the structural connection matrix between different regions in the microstructural image.

6. The method according to claim 5, wherein The performing region segmentation on the microstructural image based on the contact activation region and the standard brain atlas to determine the structural connection matrix between different regions in the microstructural image includes: Import the contact activation region into the microstructure image; Define regions of interest for the microstructure image based on a standard brain atlas to determine the regions of interest in the microstructure image; Perform a structural connectivity measurement on the microstructure image to obtain a structural connection matrix indicating the connection between the contact activation region and the region of interest.

7. The method according to claim 1, characterized in that, The feature recognition of the structural connection matrix based on the feature recognition model to generate a feature recognition result indicating the optimal contact includes: Extract the bilateral features of each contact activation region relative to other regions based on the structural connection matrix; Input the bilateral features corresponding to each contact activation region into the feature recognition model to determine the priority of each contact activation region and generate a feature recognition result indicating the contact activation region with the highest priority among them.

8. A contact selection device, characterized in that, Including: An image acquisition module for performing multi-modal image acquisition on a detection target to obtain a whole-brain tractography image; A region module for estimating the activated tissue volume of the whole-brain tractography image based on an electrode reconstruction tool to obtain the contact activation region of each contact of the detection target; A feature extraction module for extracting brain tractography feature extraction based on a standard brain atlas and the contact activation region to obtain the structural connection features of the detection target; An identification module for performing feature recognition on the structural connection features based on a feature recognition model to generate a feature recognition result indicating the optimal contact.

9. An electronic device, characterized in that, Including: At least one processor and at least one memory, wherein, The memory stores computer-readable instructions; The computer-readable instructions are loaded and executed by the processor, so that the electronic device implements the contact selection method described in any one of claims 1 to 7.

10. A storage medium having computer-readable instructions stored thereon, characterized in that, The computer-readable instructions are loaded and executed by the processor to implement the contact selection method described in any one of claims 1 to 7.

Images