Real-time data visualization system in neurosurgery

Abstract

The invention relates to the technical field of neurosurgery operation equipment, and discloses a neurosurgery operation real-time data visualization system which comprises a real-time data acquisition module, a feature construction module, a visualization processing module, a decision output module, a core processing unit, a data interface module, an instrument linkage module and the like. The real-time data acquisition module acquires multi-source physiological data and divides the multi-source physiological data into a structured data stream and a dynamic image stream; the feature construction module generates a first feature set and a second feature set respectively; the visual processing module fuses the features and then maps the features to a display mode; the decision output module calls a target visualization scheme; and the core processing unit performs logic verification to generate a final display instruction. The system can also realize the functions of data interface communication docking, instrument linkage control, instruction distribution, scene storage and the like. According to the system, real-time acquisition, intelligent processing and dynamic visualization of multi-source data are realized, and the accuracy, the safety and the efficiency of a neurosurgery operation are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of neurosurgical operation equipment, and particularly to a real-time data visualization system during neurosurgery. Background Art

[0002] In the field of neurosurgery, precision and real-time nature are the key factors to ensure the success of the operation. During traditional operations, doctors mainly rely on preoperative imaging data (such as CT, MRI) and limited intraoperative real-time monitoring data (such as electrophysiological signals) for decision-making. However, such data often has the following significant defects: The data source is single and lacks integration. Preoperative imaging data cannot dynamically reflect the real-time deformation, blood flow changes and other dynamic information of the brain tissue during the operation. And multi-source data such as intraoperative electrophysiological signals and microscope images are usually presented independently. Doctors need to switch between different devices to view, and it is difficult to quickly establish data associations, resulting in low decision-making efficiency and easy misjudgment. For example, in glioma resection surgery, it is difficult to fuse the imaging features of the tumor boundary with the electrophysiological signals of the surrounding nerve fibers in real time by traditional methods, which may cause tumor residue or nerve injury.

[0003] The visualization means is lagging and lacks dynamic adaptability. Existing visualization systems mostly adopt static display modes and cannot dynamically adjust display parameters according to the surgical process. For example, in deep brain stimulation electrode implantation surgery, as the electrode gradually penetrates into the brain tissue, the electrophysiological signals and ultrasonic images of different depths of the nerve nucleus need to be matched in real time. However, traditional systems cannot automatically optimize the display level, and doctors need to manually adjust the parameters, which may miss the best operation opportunity.

[0004] The linkage between surgical instruments and the data system is insufficient. During traditional operations, the operations of instruments (such as the activation of a laser scalpel and an ultrasonic aspirator) and data feedback are independent of each other. Doctors need to coordinate the two based on experience and it is difficult to achieve precise closed-loop control. For example, in vascular malformation resection surgery, it is impossible to automatically trigger the pre-activation of hemostatic instruments according to real-time blood flow parameters, which may increase the risk of intraoperative bleeding.

[0005] The data processing and verification mechanism is imperfect. Multi-source data is prone to delay, loss or parameter conflict during transmission and fusion. Traditional systems lack an effective real-time verification mechanism, which may lead to distorted display results. For example, the time synchronization deviation between electrophysiological signals and imaging data may cause functional area positioning errors and affect the safety of the operation.

[0006] With the development of neurosurgery towards precision and minimally invasive directions, the need for intraoperative real-time integration of multi-dimensional data and intelligent visualization is becoming increasingly urgent. In the existing technologies, although some studies have attempted to integrate multi-modal images (such as intraoperative MRI and fluorescence imaging), a full-process closed-loop system covering physiological signals, image data, and instrument control has not been formed, and significant technical bottlenecks still exist in key technical links such as data hierarchical processing, dynamic rendering, and instrument linkage. Therefore, there is an urgent need to develop a system that can realize real-time acquisition, intelligent processing, dynamic visualization, and instrument collaborative control of multi-source physiological data to improve the safety and effectiveness of neurosurgical operations. Summary of the Invention

[0007] The purpose of the present invention is to provide a real-time data visualization system for neurosurgery during operation to solve the problems proposed in the above background technology.

[0008] To achieve the above purpose, the present invention provides the following technical solution: A real-time data visualization system for neurosurgery during operation, the system includes: A real-time data acquisition module, configured to acquire multi-source physiological data during the operation process, and divide the multi-source physiological data into a structured data stream and a dynamic image stream based on a preset data hierarchical rule; A feature construction module, configured to perform neural signal analysis on the structured data stream to generate a first feature set, and perform image segmentation processing on the dynamic image stream to generate a second feature set; A visualization processing module, configured to map the first feature set and the second feature set to corresponding display modes after feature fusion according to a preset visualization parameter matrix, and use the display mode as the rendering parameter of the current scene; A decision output module, configured to call a target visualization scheme in a preset scheme database based on the rendering parameter, and use the target visualization scheme as the output configuration of the current scene; A core processing unit, configured to send the multi-source physiological data to the feature construction module, send the first feature set and the second feature set to the visualization processing module, and further configured to perform logical verification on the rendering parameter and the output configuration to generate a final display instruction.

[0009] Preferably, the feature construction module performing image segmentation processing on the dynamic image stream includes: Dividing a continuous frame sequence in the dynamic image stream into a tissue region group and an instrument region group, and performing boundary recognition calculation on the tissue region group based on a preset segmentation model to generate a contour feature set; Performing spatial coordinate conversion processing on the depth imaging data in the dynamic image stream, extracting the geometric distribution features of each coordinate region, and constructing a spatial atlas; Perform position - associated fusion of the contour feature set and the spatial atlas to generate the second feature set.

[0010] Preferably, the preset data layering rule includes a basic acquisition set and an enhanced acquisition set; the basic acquisition set includes electrophysiological identifiers, imaging identifiers, and position - sensing identifiers; the enhanced acquisition set includes blood flow parameter identifiers and metabolic feature identifiers, and each identifier corresponds to an independent data conversion channel.

[0011] Preferably, the system further includes a data interface module, which is used to realize the communication docking between the real - time data acquisition module, the feature construction module, the visualization processing module, and the decision output module and the surgical equipment respectively; The real - time data acquisition module divides multi - source physiological data based on the preset data layering rule, including: Receive a mixed data packet in real - time from the surgical equipment through the data interface module, and match the header mark of the mixed data packet according to the identifiers in the basic acquisition set to separate the basic data segment; Traverse the extended mark of the mixed data packet according to the identifiers in the enhanced acquisition set to extract the enhanced data segment; Align the basic data segment and the enhanced data segment in time series and write them into the structured data storage area and the dynamic image buffer respectively.

[0012] Preferably, when the preset visualization parameter matrix adopts a hierarchical mapping model, the display mode is the hierarchical mapping result of the composite fusion value of the first feature set and the second feature set; When the preset visualization parameter matrix adopts a dynamic adjustment model, the display mode is a set of continuous variables obtained by dynamically calibrating the joint analysis result of the first feature set and the second feature set through an optimization algorithm.

[0013] Preferably, the system further includes an instrument linkage module connected to the core processing unit, and the instrument linkage module is connected to the surgical instrument database through the data interface module; The instrument linkage module is used to screen the available device list from the surgical instrument database according to the instrument control requirements in the final display instruction, and generate an instrument activation sequence to optimize the output response process.

[0014] Preferably, the instrument activation sequence generated by the instrument linkage module includes: Load the surgical scene grid model, and locate the real - time position nodes of each device in the available device list in the grid model; Calculate the optimal response path from the deployment location of each device to the target display area based on the path optimization algorithm, and prioritize the list of available devices according to the response timeliness; Integrate the optimal response path and the priority ranking into the grid model to generate a visual instrument activation sequence.

[0015] Preferably, when the core processing unit performs logical verification on the rendering parameters and output configuration, a dual audit mode of an integrity verification mechanism and a consistency comparison mechanism is adopted. The integrity verification mechanism is used to confirm the completeness of data elements, and the consistency comparison mechanism is used to solve the spatial conflicts between parameters.

[0016] Preferably, the system further includes an instruction distribution module connected to the core processing unit. The instruction distribution module is used to convert the final display instruction into a device control code, and send the device control code to the specified display device through the data interface module to start the output program.

[0017] Preferably, the system further includes a scene storage module connected to the core processing unit. The scene storage module is used to archive the multi-source physiological data, the first feature set, the second feature set, the display mode, and the final display instruction, and generate a panoramic scene record chain according to the time distribution.

[0018] Compared with the prior art, the beneficial effects of the present invention are: In terms of multi-source data integration and hierarchical processing, the real-time data acquisition module divides the multi-source physiological data into structured data streams (such as electrophysiological markers, blood flow parameters) and dynamic image streams (such as microscope videos, ultrasonic imaging) based on preset data stratification rules, and realizes efficient communication docking with surgical equipment through the data interface module. This stratification mechanism not only ensures the independent acquisition and synchronous processing of basic physiological indicators (such as electroencephalogram signals) and enhanced parameters (such as regional cerebral blood flow), but also guarantees the spatio-temporal consistency of data through time series alignment technology, solving the problems of data mixing and poor synchronization in traditional systems. For example, in epilepsy focus resection surgery, electrocorticogram (structured data) and intraoperative fluorescence images (dynamic image stream) can be accurately collected simultaneously to provide multi-dimensional basis for focus localization.

[0019] The feature construction module extracts key features from structured data streams and dynamic image streams respectively through neural signal analysis and image segmentation techniques. For the dynamic image stream, by dividing the tissue region group and the instrument region group, combining with a preset segmentation model for boundary recognition, and constructing a spatial atlas to achieve the fusion of contour features and geometric distributions, the morphological and spatial position features of targets such as brain tissue, tumors, and surgical instruments can be accurately extracted. This processing method breaks through the limitation of traditional images that only provide macroscopic structures and can refine to tissue boundary features at the sub-millimeter level. For example, in fiber tractography surgery, a three-dimensional contour feature set of white matter fiber tracts can be generated in real time to assist doctors in avoiding important functional areas.

[0020] The visualization processing module fuses multi-dimensional features and maps them to a dynamic display mode based on a preset visualization parameter matrix (layered mapping model or dynamic adjustment model). The layered mapping model can achieve the overlay display of data at different levels (such as macroscopic anatomical structures and microscopic electrophysiological activities), while the dynamic adjustment model dynamically calibrates feature parameters through an optimization algorithm, enabling the display interface to automatically switch the key information as the surgical process progresses. For example, in aneurysm clipping surgery, the system can dynamically adjust the color coding and transparency of the blood vessel image according to real-time blood flow parameters to visually display hemodynamic changes and help doctors evaluate the clipping effect.

[0021] The logical verification mechanism of the decision output module and the core processing unit ensures the reliability of the visualization scheme. The integrity verification mechanism confirms the completeness of data elements to avoid misjudgment caused by data loss; the consistency comparison mechanism solves the spatial conflicts between parameters. For example, when fusing MRI and ultrasound images, the coordinate system is automatically calibrated to eliminate image misalignment caused by equipment errors. This dual review mode significantly improves the stability of the system and reduces the surgical risks caused by data processing errors.

[0022] The introduction of the instrument linkage module and the instruction distribution module realizes the deep cooperation between the data visualization system and surgical instruments. The instrument linkage module generates an instrument activation sequence according to the final display instruction, calculates the optimal response path through a path optimization algorithm and performs priority sorting. For example, when emergency hemostasis is required, the ultrasonic hemostat is automatically activated with priority and the shortest operation path is planned to shorten the instrument response time. The instruction distribution module converts the display instruction into a device control code to achieve seamless docking with devices such as microscopes and neuronavigation systems, forming a closed loop of "data acquisition - analysis - visualization - instrument control" to improve the coherence and accuracy of surgical operations.

[0023] The scene storage module archives multi-source data and processing results to generate a panoramic scene record chain by time, providing complete digital materials for postoperative review, surgical plan optimization, and teaching. Doctors can analyze the data changes and operation effects at different surgical stages by tracing back the record chain to promote the iterative improvement of surgical techniques. Brief Description of the Drawings

[0024] Figure 1 This is the working principle diagram of the real-time data visualization system in neurosurgery according to the present invention; Figure 2 This is the working flow chart of the image segmentation process of the feature construction module; Figure 3 This is the working flow chart of the data interface module; Figure 4 This is the design diagram of the visualization parameter matrix mapping mode. Detailed Description of the Preferred Embodiments

[0025] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0026] Please refer to Figures 1-4 , a real-time data visualization system in neurosurgery according to the present invention, which system includes a real-time data acquisition module, a feature construction module, a visualization processing module, a decision output module and a core processing unit. The specific implementation is as follows: The real-time data acquisition module acquires multi-source physiological data during the operation, and divides the data into a structured data stream and a dynamic image stream based on a preset data layering rule. Among them, the multi-source physiological data covers various physiological indicators and image information of the patient during the operation process.

[0027] The feature construction module performs neural signal analysis on the structured data stream to generate a first feature set reflecting neural activity characteristics; at the same time, it performs image segmentation processing on the dynamic image stream to generate a second feature set containing image structure characteristics.

[0028] The visualization processing module performs feature fusion on the first feature set and the second feature set according to a preset visualization parameter matrix, and the fused features are mapped to the corresponding display mode, which is used as the rendering parameter for the current scene.

[0029] The decision output module calls a matching target visualization scheme from a preset scheme database based on the rendering parameter, and this scheme is used as the output configuration for the current scene.

[0030] As the control center of the system, the core processing unit is responsible for transmitting multi-source physiological data to the feature construction module and transmitting the first feature set and the second feature set to the visualization processing module. At the same time, it performs logical verification on the rendering parameters and output configurations, generates the final display instruction, and ensures the accuracy and reliability of the system output.

[0031] The present invention will be further described below in conjunction with Embodiments 1 to 5: Embodiment 1:

[0032] The implementation manner of the feature construction module for performing image segmentation processing on the dynamic image stream in this embodiment specifically includes steps such as continuous frame sequence region division, tissue region boundary recognition calculation, depth imaging data spatial coordinate conversion and geometric feature extraction, and contour feature set and spatial atlas position association fusion. Each step realizes the generation of the second feature set through specific processing logics and data interactions, as follows: During the operation of the system, the feature construction module receives the dynamic image stream output by the real-time data acquisition module. The dynamic image stream is a continuous image sequence collected in real time during the operation through imaging devices (such as intraoperative MRI, CT, or neuro-navigation system), containing dynamic visual information such as the tissue morphology, instrument position, and spatial structure in the surgical area. The feature construction module first preprocesses the continuous frame sequence of the dynamic image stream, and improves the image quality through image preprocessing algorithms (such as noise reduction and contrast enhancement) to lay a foundation for subsequent segmentation processing.

[0033] The preprocessed continuous frame sequence enters the region division link. The feature construction module divides each image in the continuous frame sequence into a tissue region group and an instrument region group based on the semantic features of the image content. Among them, the tissue region group covers biological tissue regions such as nerve tissue, blood vessels, and brain parenchyma in the surgical field of view; the instrument region group includes medical instrument regions such as scalpels, forceps, and electrodes used during the operation. The region division can be realized through an image semantic segmentation model based on deep learning. For example, classic segmentation networks such as U-Net are used. By training on the labeled surgical image data set, the model is enabled to automatically distinguish between tissue and instrument regions. During the division process, the module classifies the pixel points of each frame of image and assigns a label of "tissue" or "instrument" to each pixel point, thereby generating a corresponding region mask and dividing the entire image into two independent region groups.

[0034] After the region division is completed, the feature construction module performs boundary recognition calculations on the tissue region group. Specifically, the module processes each region within the tissue region group based on a preset segmentation model. The preset segmentation model can be a traditional image processing model based on edge detection algorithms (such as the Canny operator) or a contour extraction model based on deep learning. Taking the traditional algorithm as an example, the module first grayscales the image of the tissue region group, then applies the Canny operator to detect the edge pixels in the image, and filters out the effective edge contours by setting high and low thresholds. Then, the edge pixels are connected through a contour tracking algorithm (such as the Suzuki algorithm) to form a continuous contour curve, thereby generating a contour feature set that can accurately describe the tissue boundary. The contour feature set contains geometric feature parameters such as the contour coordinate sequence, contour perimeter, area, and curvature of each tissue region, which are used to characterize the morphological and structural boundaries of the tissue.

[0035] Meanwhile, the feature construction module performs spatial coordinate transformation processing on the depth imaging data in the dynamic image stream. The depth imaging data is usually provided by imaging devices with depth perception capabilities (such as stereo vision cameras, laser scanning devices) and contains the three-dimensional coordinate information (X, Y, Z) of each point in the scene. The module first performs coordinate system transformation on the depth imaging data, converting the coordinate values in the device coordinate system to the coordinate values in a unified world coordinate system to ensure the spatial consistency of different frame data and different modality data. During the transformation process, the internal parameters of the device (such as focal length, optical center) and external parameters (such as rotation matrix, translation vector) need to be considered, and the coordinate system is unified through coordinate transformation formulas (such as rotation and translation transformations).

[0036] After the coordinate system transformation is completed, the feature construction module performs spatial block processing on the depth data in the world coordinate system, dividing the entire surgical scene into multiple regular coordinate regions (such as three-dimensional grid cells). For each coordinate region, the module extracts the geometric distribution features therein, including the spatial position distribution, density, volume ratio, and connection relationship with adjacent regions of the tissue or instruments in the region. By statistically analyzing the geometric feature parameters in each coordinate region, a spatial atlas reflecting the spatial structure of the surgical scene is constructed. The spatial atlas is stored in the form of a three-dimensional matrix or graph structure, and each element corresponds to the geometric feature vector of a coordinate region. The vector contains the position coordinates of the region and various geometric feature values.

[0037] After generating the contour feature set and the spatial atlas, the feature construction module needs to perform position correlation and fusion on the two. The specific steps are as follows: First, map the coordinate points of each tissue contour in the contour feature set to the world coordinate system to obtain the three-dimensional spatial coordinates of each contour point; then, determine the coordinate region to which each contour point belongs according to the coordinate block rule of the spatial atlas; next, correlate and fuse the geometric feature parameters (such as contour coordinates, perimeter, area, etc.) in the contour feature set with the geometric distribution features in the corresponding coordinate region in the spatial atlas. For example, for a certain coordinate region, combine the perimeter and area of the tissue contour in this region with features such as its spatial density and volume ratio to form a composite feature vector for this region. Through this position correlation and fusion, the contour feature set can be matched with the spatial structure information in the spatial atlas to generate a second feature set containing tissue morphological features and spatial position features.

[0038] The specific manifestation form of the second feature set is a set of multi-dimensional feature vectors. Each feature vector corresponds to a specific region or object in the surgical scene and contains the contour geometric features, spatial position features, and association features with the surrounding environment of this region or object. For example, for a certain section of nerve tissue, its feature vector may contain parameters such as the contour coordinate sequence, area, curvature, coordinates of the spatial region where it is located, tissue density of this region, and distance from adjacent blood vessels. These feature parameters provide a rich data source for subsequent visualization processing, enabling the visualization processing module to flexibly fuse and map the features according to different display requirements to generate intuitive and accurate visualization results.

[0039] During the whole processing process, the feature construction module realizes the conversion from the original image data to the structured feature data through multi-level processing of the dynamic image stream. The regional division of the continuous frame sequence provides a clear target region for subsequent feature extraction. The boundary recognition calculation of the tissue region accurately captures the morphological features of the tissue. The spatial coordinate conversion and geometric feature extraction of the depth data construct the spatial structure model of the scene. And the position correlation and fusion of the contour features and the spatial atlas organically combine the morphological features and spatial features, and finally generate a comprehensive and accurate second feature set. This processing flow not only makes full use of the visual information in the dynamic image stream, but also endows the feature data with semantic information in the spatial dimension through spatial coordinate conversion and feature fusion technology, providing a solid feature basis for the visualization of real-time data during neurosurgery, helping doctors to observe and analyze the tissue morphology and spatial structure of the surgical area more clearly and accurately during the operation, thereby improving the accuracy and safety of the operation.

[0040] Example 2: The implementation manner of the preset data layering rule and the data interface module described in this embodiment specifically includes the structural composition of the preset data layering rule, the communication docking function of the data interface module, the layering division process of multi-source physiological data, and data storage management, etc. Each part realizes the orderly acquisition and classification of multi-source physiological data through a specific logical architecture and data processing process, as follows: The preset data layering rule in the system serves as the basis for dividing multi-source physiological data, and its architecture is divided into two levels: the basic acquisition set and the enhanced acquisition set. The basic acquisition set includes three types of core basic identifiers: electrophysiological identifier, imaging identifier, and position sensing identifier. Among them, the electrophysiological identifier corresponds to electrophysiological index data such as nerve electrical signals and electroencephalogram signals collected by electrodes or sensors during the operation, which is used to reflect the functional activity state of nerve tissues; the imaging identifier corresponds to two-dimensional or three-dimensional image data collected by intraoperative imaging devices such as MRI, CT, and ultrasound, which is used to present the anatomical structure of the operation area; the position sensing identifier corresponds to the spatial position coordinate data of surgical instruments and patient anatomical landmark points obtained through a positioning system (such as a neuronavigation system), which is used to determine the relative position relationship between the instrument and the tissue. These three types of identifiers constitute the acquisition framework of the basic physiological information of the operation, covering the most basic electrophysiology, imaging, and spatial position data dimensions in neurosurgery.

[0041] On the basis of the basic acquisition set, the enhanced acquisition set further expands two types of advanced identifiers: blood flow parameter identifier and metabolic feature identifier. The blood flow parameter identifier is used to collect hemodynamic parameters such as blood flow velocity, blood flow volume, and vascular resistance in the operation area, which can be obtained through devices such as laser Doppler flowmetry and reflects the blood perfusion situation of tissues; the metabolic feature identifier is used to collect metabolic indicators such as tissue oxygen metabolism rate and glucose metabolism rate, which can be obtained through means such as near-infrared spectroscopy and is used to evaluate the metabolic state and functional activity of tissues. The setting of the enhanced acquisition set enables the system to obtain more in-depth physiological function data and provides more comprehensive information support for surgical decision-making. It should be noted that each identifier corresponds to an independent data conversion channel, that is, data of different identifiers are respectively converted and transmitted through independent signal processing links during the acquisition process, avoiding interference between different types of data and ensuring the accuracy and reliability of data acquisition.

[0042] The data interface module of the system settings is the data interaction hub between the hardware layer and the software layer. Through standardized communication protocols (such as TCP / IP, USB, RS-232, etc.), it realizes the bidirectional communication docking between the real-time data acquisition module, the feature construction module, the visualization processing module, the decision output module and surgical devices (such as anesthetic machines, monitors, imaging devices, neuronavigation systems, etc.). Specifically, the data interface module assigns a unique communication address to each connected surgical device, establishes a data transmission channel between the device and each module of the system, and performs format conversion and protocol parsing on the transmitted data, enabling surgical devices of different manufacturers and models to be compatible with the system and realizing the unified acquisition and processing of multi-source data. For example, for the DICOM format image data output by the imaging device, the data interface module can convert it into a general image format recognizable by the system; for the analog signal output by the electrophysiological device, the data interface module converts it into a digital signal through an analog-to-digital conversion circuit and encapsulates it according to the preset data protocol.

[0043] When the real-time data acquisition module divides multi-source physiological data based on the preset data layering rules, it first receives mixed data packets from surgical devices in real time through the data interface module. The mixed data packet is a comprehensive data packet containing multiple types of data output by the surgical device at the same time node. For example, it may contain electrophysiological signal data, image slice data, instrument position coordinate data, blood flow parameter data, etc. Since different types of data are stored in different positions and formats in the data packet, it is necessary to separate and extract them through the preset data layering rules.

[0044] The specific processing flow is as follows: The real-time data acquisition module matches the header mark of the mixed data packet according to the identifier in the basic acquisition set. The header mark is a fixed field in the data packet used to identify the data type and basic attributes. For example, the header mark corresponding to the electrophysiological identifier may contain the prefix "EP_", and the header mark corresponding to the imaging identifier may contain the prefix "IM_", etc. By identifying the header mark, the module can quickly locate the start position and length of the basic data segment in the data packet, so as to separate the basic data segment, including electrophysiological data, imaging data and position sensing data. The separated basic data segment is transmitted to the subsequent processing link for structured processing.

[0045] The module traverses the extended tags of the mixed data packets according to the identifiers in the enhanced acquisition set. The extended tags are optional fields in the data packets used to identify advanced data types, usually located after the basic data segment. For example, the extended tag corresponding to the blood flow parameter identifier may be "BF_", and the extended tag corresponding to the metabolic feature identifier may be "ME_". By scanning the extended tags one by one, the module identifies the position and content of the enhanced data segment, and extracts the blood flow parameter data and metabolic feature data. This hierarchical identification mechanism ensures the orderly separation of the basic data and the enhanced data, avoiding processing errors caused by data mixing.

[0046] After separating the basic data segment and the enhanced data segment, the real-time data acquisition module needs to perform time series alignment processing on the two types of data. Due to possible differences in the sampling frequency and clock synchronization of surgical devices, there may be deviations in the timestamps of different types of data. The module extracts the timestamp information from each data segment and calibrates the data based on the system global clock, adjusting the basic data segment and the enhanced data segment to the same time axis to ensure that the data at the same time node can be accurately corresponded. Time series alignment can be achieved through algorithms such as linear interpolation and synchronous sampling. For example, for image data with a higher sampling frequency and electrophysiological data with a lower sampling frequency, interpolation algorithms are used to generate equally spaced synchronous data points on the time axis.

[0047] The time-aligned basic data segment and enhanced data segment are respectively written into the structured data storage area and the dynamic image buffer. The structured data storage area is a storage module built based on a relational database (such as MySQL) or a time series database (such as InfluxDB), which is used to store structured data streams, such as the numerical sequence of electrophysiological signals and the coordinate sequence of instrument positions, and supports fast query and retrieval according to keywords such as timestamps and data types; the dynamic image buffer is a storage module built based on cache technology (such as memory cache, solid-state disk cache), which is used to store dynamic image streams, such as continuous image slice data, depth imaging data, etc., to meet the requirements of real-time processing for data read and write speeds. By storing different types of data in the corresponding areas, the system realizes the classified management of multi-source physiological data, providing a structured and time-series consistent data source for the subsequent feature construction module.

[0048] The preset data layering rule realizes the scientific classification of multi-source physiological data through the layering architecture of the basic acquisition set and the enhanced acquisition set; the data interface module ensures the compatible acquisition of data from different devices through standardized communication docking and data format conversion; the layering division process of multi-source physiological data realizes the precise separation and orderly storage of mixed data packets through steps such as header tag matching, extended tag traversal, and time series alignment.

[0049] Example 3: The implementation methods of the preset visualization parameter matrix and display mode described in this embodiment specifically include the feature fusion logic, display mode generation mechanism, and application scenarios in two modes: the hierarchical mapping model and the dynamic adjustment model. Combining the specific data types and visualization requirements in neurosurgery, the details are as follows: The preset visualization parameter matrix in the system is the core rule set that controls feature fusion and display mode generation. Its essence is a parameter set that defines the mapping relationship between feature inputs and visual outputs. According to different requirements for data display during surgery, the preset visualization parameter matrix can be configured as a hierarchical mapping model or a dynamic adjustment model, which are respectively applicable to the hierarchical display of structured features and the real-time presentation of dynamic features.

[0050] Implementation method of the hierarchical mapping model: When the preset visualization parameter matrix adopts the hierarchical mapping model, the processing logic of the system focuses on hierarchically mapping the composite fusion value of the first feature set (neuro-signal analysis result) and the second feature set (image segmentation feature). Taking the localization of epileptic foci in epilepsy surgery as an example: Feature fusion process: The first feature set includes neuro-electrophysiological features such as the spike frequency, amplitude, and spatial distribution of electroencephalogram (EEG) signals, and the second feature set includes image features such as the contour coordinates, volume, and spatial distance from surrounding blood vessels of abnormal brain sulci and gyri in magnetic resonance imaging (MRI) images. The visualization processing module first standardizes the two types of features, unifying the microvolt-level amplitude data of EEG signals and the millimeter-level spatial data of image coordinates into the normalization interval of [0,1] to avoid fusion deviation caused by dimensional differences.

[0051] Definition of hierarchical rules: The hierarchical mapping model divides the composite fusion value into three levels: the anatomical layer, the functional layer, and the risk layer. The anatomical layer corresponds to the display level dominated by image features, showing the anatomical structure of the surgical area (such as the cerebral cortex, ventricles), mapping tissue density through grayscale values, and outlining tissue boundaries with contour lines; the functional layer corresponds to the display level dominated by neuro-electrophysiological features, superimposed on the anatomical layer in the form of a heat map, and the color gradient (such as from blue to red) represents the degree of EEG signal abnormality (such as the high or low spike frequency); the risk layer is the comprehensive evaluation result of the first two layers, calculating the spatial overlap area of the anatomical structure and functional abnormality through an algorithm, and marking the high-risk areas of surgical operations (such as the boundary of epileptic foci near the motor cortex) with grid lines.

[0052] Display mode generation: The calculation of the composite fusion value is based on the feature weighted summation rule. For example, the weight of the image contour feature in the anatomical layer accounts for 70%, and the weight of the neural electrical signal feature in the functional layer accounts for 60%. The risk layer dynamically allocates weights according to the spatial overlap degree of the anatomical and functional features. Each layer corresponds to an independent display channel, and the display transparency or hidden state of the layer can be controlled through interface controls (such as sliders, checkboxes), facilitating doctors to selectively view according to the surgical process. For example, before opening the dura mater, doctors can focus on observing the tissue structure of the anatomical layer; during cortical electrical stimulation, the electroencephalogram activity heat map of the functional layer is superimposed and displayed; when designing the resection range, refer to the high-risk area markings in the risk layer.

[0053] Implementation method of the dynamic adjustment model: When the preset visualization parameter matrix adopts the dynamic adjustment model, the system performs real-time calibration on the joint analysis results of the first feature set and the second feature set through an optimization algorithm, generating a continuously changing display mode. Taking the hemodynamic monitoring in aneurysm clipping surgery as an example: Feature joint analysis: The first feature set includes real-time blood flow parameters such as blood flow velocity and pulsatility index measured by transcranial Doppler ultrasound (TCD), and the second feature set includes the morphological features of the aneurysm (such as aneurysm neck width, aneurysm body volume) and the position coordinates of the clipping instrument in the intraoperative angiography (DSA) image. The visualization processing module establishes an association model between the blood flow parameters and the image features. For example, the correlation threshold between abnormal blood flow velocity and aneurysm morphological changes is determined through regression analysis.

[0054] Application of the optimization algorithm: The optimization algorithm adopts a sliding window mechanism to perform real-time analysis on the feature data within the last 5 seconds. When it detects a sudden increase in blood flow velocity (such as exceeding 120% of the baseline value), the algorithm automatically triggers the dynamic calibration process: First, mark the real-time deformation area of the aneurysm body in the DSA image (by comparing the previous and current frames through image registration technology); then, jointly calculate the blood flow velocity change value and the volume change rate of the deformation area to generate a continuous variable reflecting the blood flow-structure coupling effect (such as "blood flow-deformation index"); finally, dynamically adjust the color coding and graphic markings of the display mode according to the numerical range of this index (such as 0-100) - when the index is below 40, it is displayed in green and the aneurysm body contour line is solid; when the index is between 40 and 70, it is displayed in yellow and the contour line becomes dashed; when the index exceeds 70, it is displayed in red, triggering a flashing alarm and marking the risk area.

[0055] Dynamic update of display mode: A set of continuous variables is synchronously displayed in the visualization interface in the form of a time-series curve. The horizontal axis represents time (unit: second), and the vertical axis represents the normalized values of each characteristic parameter. For example, curve A represents the real-time change in blood flow velocity, and curve B represents the change rate of aneurysm volume. The two are distinguished by color and superimposed on the angiography image. Doctors can view the details of any time interval through the interface zoom tool. At the same time, the system automatically saves the dynamic data of the most recent 30 minutes for retrospective analysis. When the surgical instrument (such as an aneurysm clip) approaches the aneurysm neck, the dynamic adjustment model will, according to the association between the instrument position coordinates (from the position sensing marker) and the blood flow parameters, adjust the spatial calibration parameters of the display mode in real time to ensure the spatial alignment accuracy between the instrument image and the blood flow characteristics.

[0056] Cooperative application scenarios of the two models: In complex neurosurgical operations (such as resection of gliomas in the functional area), the preset visualization parameter matrix can automatically switch models or use the two models in combination according to the surgical stage. For example: Preoperative planning stage: The hierarchical mapping model is adopted to statically display the anatomical layer of the MRI image (showing the spatial relationship between the tumor and the white matter fiber bundle), the functional layer of the DTI fiber tractography (showing the direction of the nerve conduction tract), and the risk layer of the intraoperative electrophysiological monitoring (showing the localization result of the cortical functional area) in layers to help doctors formulate the resection path.

[0057] Intraoperative resection stage: Switch to the dynamic adjustment model, and fuse in real time the cell morphological characteristics (the second feature set) of the laser confocal microscope and the abnormal discharge characteristics (the first feature set) of the electrocorticogram (ECoG). Through an optimization algorithm, a "tumor resection boundary probability map" is dynamically generated and superimposed on the surgical field image in the form of a semi-transparent heat map. The boundary line moves smoothly with the real-time data, prompting the doctor whether the current resection range is close to the tumor edge or the functional area.

[0058] Stage before closing the incision: Re-enable the hierarchical mapping model to display the image layer of the hemostasis effect (such as using ultrasound to detect whether there is a hematoma), the functional layer of the electroencephalogram signal (showing whether the cortical electrical activity has returned to normal after the operation), and the confirmation layer of the instrument position (showing whether all surgical instruments have been removed), and ensure the safety of the surgical closure through multi-level verification.

[0059] Throughout the implementation process, the preset visualization parameter matrix realizes the flexible visualization of multi-source data during neurosurgery by demonstrating the hierarchical features of the hierarchical mapping model and calibrating the real-time parameters of the dynamic adjustment model. The hierarchical mapping model is suitable for scenarios that require clear differentiation of data dimensions, helping doctors establish a three-dimensional understanding of "anatomy-function-risk"; the dynamic adjustment model is good at handling feature associations that change over time and providing dynamic feedback for real-time surgical decision-making. The combination of the two models enables the system to dynamically switch data display strategies according to the diverse needs of the surgical process, enhancing the clinical practicality of the visualization results, assisting doctors in quickly obtaining key information in complex surgical environments, and optimizing the surgical operation process.

[0060] Example 4: The connection and working process between the instrument linkage module and the core processing unit described in this embodiment specifically include module hardware architecture, loading of the surgical scene grid model, positioning of device position nodes, calculation of the optimal response path, priority sorting logic, and activation sequence generation mechanism, etc. In combination with the actual requirements of instrument collaborative operation in neurosurgery, it is described in detail as follows: In the system, the instrument linkage module establishes a high-speed data connection with the core processing unit through a dedicated communication bus (such as PCIe or USB4.0), and at the same time realizes cross-platform communication with the surgical instrument database through the standardized protocol of the data interface module (such as OPCUA or DICOM). The surgical instrument database stores the information of commonly used instruments in neurosurgery, including instrument types (such as suction devices, electrocoagulation forceps, neuronavigation probes), physical parameters (such as length, diameter, working angle), spatial positioning identifiers (such as the coordinates of the Marker point at the end of the instrument), and control protocols (such as the format of motor drive instructions). The hardware carrier of the instrument linkage module is an embedded computing unit (such as an FPGA or GPU acceleration board), which has the ability to process real-time data and run path planning algorithms.

[0061] When the core processing unit generates the final display instruction, the instrument control requirements in the instruction (such as "activate the neuronavigation probe for target positioning") are parsed and transmitted to the instrument linkage module. The module first performs the operation of loading the surgical scene grid model: based on the preoperative CT / MRI image data and the intraoperative real-time image stream, a grid-based spatial model of the surgical area is generated through a three-dimensional reconstruction algorithm. The model is represented in the form of triangular grids or voxel grids, and the spatial resolution can reach the sub-millimeter level (such as 0.5mm×0.5mm×0.5mm). The grid model covers the patient's head anatomical structure (such as the skull, cerebral sulci and gyri), the surgical instruments already placed (such as retractors), and the boundaries of the surgical field of view, providing a spatial reference framework for instrument position positioning and path planning.

[0062] When locating the real-time position nodes of each device in the list of available devices in the grid model, the module obtains the spatial positioning data of the instrument through the data interface module. The positioning data can come from an optical tracking system (such as NDI Polaris) or an electromagnetic positioning system, and the output format is six-degree-of-freedom coordinates (X, Y, Z, Rx, Ry, Rz) in the world coordinate system, where (X, Y, Z) are the three-dimensional coordinates of the Marker point at the end of the instrument, and (Rx, Ry, Rz) are the Euler angle attitude parameters of the instrument. The module maps the coordinate data of each device to the surgical scene grid model to generate corresponding position nodes - for rod-shaped instruments (such as electrocoagulation forceps), the position node is the end point of the instrument; for sheet-shaped instruments (such as brain retractors), the position node is the geometric center of the contact surface of the instrument. The position nodes are displayed as colored sphere icons in the visualization interface, and the color corresponds to the instrument type (e.g., red represents electrocoagulation type, and blue represents aspiration type) for easy identification by the doctor.

[0063] When calculating the optimal response path based on the path optimization algorithm, the module first determines the target display area. The target display area is determined by the surgical operation requirements in the final display instruction. For example, in the tumor resection scenario, the target display area is an annular area 5 mm outside the tumor boundary; in the vascular anastomosis scenario, the target display area is a local area 2 mm around the anastomosis site. The path optimization algorithm uses an improved A* algorithm. "The A* algorithm is a heuristic path search algorithm that finds the shortest path through the evaluation function f(n) = g(n) + h(n), where g(n) is the actual cost from the starting point to the current node, and h(n) is the heuristic estimated cost from the current node to the end point. The improved A algorithm in this system is specifically manifested as: calculating the optimal path through the cost function C by comprehensively considering the geometric distance, travel time, and anatomical conflict cost. The conflict cost is determined based on the obstacle distribution in the surgical scene grid model, and its cost function is defined as:

[0064] Among them, is the total path cost, dimensionless; is the path geometric distance, in millimeters (mm), calculated by the Euclidean distance between nodes in the grid model; is the path travel time, in seconds (s), calculated based on the instrument movement speed parameter (such as the maximum movement speed of electrocoagulation forceps is 10 mm / s) and the path length; is the path conflict cost, dimensionless, with a value range of [0, 1], determined according to the spatial overlap degree between the path and anatomical structures (such as important blood vessels, nerve bundles). The larger the overlapping volume, the The higher the value; for example, in deep brain stimulation electrode implantation surgery, the improved A* algorithm will avoid the lateral ventricle area and plan the shortest conflict-free path from the skull incision to the target point, further enhancing readability. , , are weight coefficients, satisfying , and the default value is , , , and can be manually adjusted according to the type of surgery (such as increasing the weight to avoid risks) in a fine operation scenario.

[0065] During algorithm execution, starting from the current deployment position of the device (i.e., the position node coordinates), search for the shortest path to the center of the target display area in the grid model while avoiding anatomical structure obstacles (such as the brainstem and large blood vessels). The path is represented by a set of grid edges, and each grid node records its parent node and the cumulative cost value, finally generating a zigzag optimal path from the starting point to the ending point. For example, for a suction device located on the right side of the surgical incision, the target display area is the left deep brain tumor cavity, and the algorithm will plan a path that bypasses the motor cortex to avoid damaging important functional areas.

[0066] When prioritizing the available device list according to the response timeliness, the module comprehensively evaluates based on the activation time, moving speed, and total path cost of the instrument. The response timeliness calculation formula is:

[0067] where is the response timeliness, with the unit of seconds (s); is the activation time of the instrument from the standby state to the active state, with the unit of seconds (s). For example, the activation time of a neuro-navigation probe is 2 s, and that of an electrocoagulation forceps is 0.5 s; is the maximum moving speed of the instrument, with the unit of millimeters per second (mm / s), which is preset according to the instrument type.

[0068] The module calculates the value for each instrument in the available device list and generates a priority queue in ascending order. For example, when both a suction device and an electrocoagulation forceps are needed simultaneously, if the of the suction device, and the seconds of the electrocoagulation forceps, then the electrocoagulation forceps has a higher priority than the suction device, and the system preferentially activates the electrocoagulation forceps and displays its optimal path.

[0069] When generating the visualization device activation sequence, the module integrates the optimal response path and the priority sorting result into the surgical scene grid model. The specific implementation is as follows: in the visualization interface, the activation animations of the devices are displayed in the order of priority. First, the position node of the device with the highest priority blinks in a highlighted color (such as yellow), and at the same time, a semi-transparent movement trajectory line (such as a blue dashed line) is drawn along the optimal path. The width of the trajectory line gradually narrows as it approaches the target area (from 2 px to 0.5 px), indicating the direction and speed trend of the device movement. When the device reaches the target area, the next-level priority device repeats the above process. The activation sequence also includes the timestamp information of the device control instructions, such as "Electrocoagulation forceps activation time: 00:15:23, estimated arrival time: 00:15:25", which is convenient for the doctor to master the operation rhythm.

[0070] The device linkage module realizes the visualization and optimization of the surgical device activation sequence through the precise construction of the surgical scene grid model, the real-time positioning of the position nodes, the intelligent calculation of the path optimization algorithm, and the quantitative evaluation of the priority sorting. This mechanism not only shortens the device response time but also improves the operation safety by avoiding the anatomical risk areas. For example, in the deep brain stimulation electrode implantation surgery, the device linkage module can calculate the optimal path to avoid the lateral ventricle according to the pre-operative planned electrode trajectory, and preferentially activate the stereotactic frame with real-time position feedback to ensure the accuracy of electrode implantation. By transforming the abstract device control logic into an intuitive visualization activation sequence, the system assists the doctor in efficiently coordinating multi-device operations in a complex surgical environment, reducing human decision-making errors, and improving the standardization and safety of the surgical process.

[0071] Example 5: The implementation methods of the core processing unit logic verification mechanism, instruction distribution, and scene storage module described in this example specifically include the execution process of the dual review mode, the instruction conversion and transmission mechanism, the data archiving logic, and the method for generating the scene record chain, which are described in detail below in combination with the typical application scenarios in neurosurgery. The core processing unit adopts a dual-audit mode of integrity verification mechanism and consistency comparison mechanism for the logical verification of rendering parameters and output configurations. Taking the visualization process of glioma resection surgery as an example: when the visualization processing module generates rendering parameters that include the overlay of neuroelectrophysiological heatmaps and MRI images, and the decision output module calls the preset "functional area protection" visualization scheme as the output configuration, the core processing unit starts the verification process. The integrity verification mechanism first checks whether all necessary data elements are included in the rendering parameters, such as the sampling frequency of neural signal characteristics, the slice thickness of image slices, the threshold range of heatmap color mapping, etc. If it is found that the slice position parameter of a certain layer of MRI image is missing, the system will automatically trigger a data resampling process to retrieve the slice data at the corresponding time point from the dynamic image buffer and insert it into the parameter set. The consistency comparison mechanism then verifies the spatial conflicts between parameters. For example, it checks whether the origin of the coordinate system of the neuroelectrophysiological heatmap is consistent with the origin of the world coordinate system of the MRI image. If the deviation between the two exceeds the preset threshold (such as 0.3 mm), the system calibrates the space of the heatmap through a coordinate transformation algorithm to ensure the precise alignment of the spatial positioning of electrophysiological signals and anatomical structures.

[0072] The function implementation of the instruction distribution module runs through the entire process of system output. Taking the intraoperative real-time navigation scenario as an example: after the core processing unit generates the final display instruction that includes the instrument position, target coordinates, and path planning results, the instruction distribution module first performs protocol parsing on the instruction to identify the display device type (such as the OLED display of the surgical microscope, the holographic projection device of the neuronavigation system). For different devices, the module performs corresponding encoding conversions - for the microscope display, the instruction is converted into a video stream encoding of the HDMI protocol, including parameters such as RGB pixel values and synchronization signals; for the holographic projection device, it is converted into a UDP data packet that supports spatial coordinate mapping, including a three-dimensional coordinate matrix, projection angle parameters, etc. The converted device control encoding is sent to the specified display device through the dedicated communication channel of the data interface module. For example, it is transmitted to the image processing unit of the microscope through an optical fiber link, or transmitted to the holographic projector through Wi-Fi6. After receiving the encoding, the display device starts the corresponding output program: the microscope display renders the anatomical image with dot-shaped instrument markings in real time, and the holographic projection device generates a floating three-dimensional path guidance model above the surgical area, and the doctor can adjust the projection angle through gesture interaction.

[0073] The archiving mechanism of the scenario storage module covers the data throughout the entire life cycle of system operation. During an aneurysm clipping operation, the scenario storage module receives and stores multi-source physiological data in real time, including sequence frames of intraoperative DSA images (each frame contains a grayscale value matrix of 512×512 pixels), blood flow velocity waveform data of transcranial Doppler (sampling frequency 100Hz), and raw electroencephalogram signals of neuroelectrophysiological monitoring (duration about 2 hours, sampling frequency 2000Hz). For the first feature set, store the feature parameters after parsing the neural signals, such as the number of spike waves, average amplitude, and spatial distribution entropy value of the electroencephalogram signals; for the second feature set, store the sequence of contour coordinates after image segmentation (such as the array of edge point coordinates of the aneurysm neck), and the three-dimensional matrix data of the spatial atlas (each voxel contains density values and neighborhood connection relationships). The data related to the display mode stores the configuration information of the preset visualization parameter matrix, such as the hierarchical weights of the hierarchical mapping model and the optimization algorithm parameters of the dynamic adjustment model. The stored content of the final display instruction includes the rendering parameters, output configuration, and verification logs after logical verification (such as the supplementary acquisition records for integrity verification and the calibration parameters for consistency comparison).

[0074] The process of the scenario storage module generating a panoramic scenario record chain according to time distribution is as follows: Taking the start time of the operation as the starting point of the time axis (such as 09:00:00), create a time node every 10 seconds. Each node contains the multi-source physiological data summary, feature set hash value, display mode snapshot, and verification result of the final display instruction at that moment. For example, at the 09:15:30 time node, the record chain stores the 90th frame thumbnail of the DSA image, the real-time average value of the blood flow velocity (120cm / s), the alpha wave power ratio of the electroencephalogram signal (35%), the currently activated layer of the hierarchical mapping model (anatomical layer + functional layer), and the output device status of the instruction distribution module (the microscope display screen is working properly). The record chain uses blockchain technology to ensure data immutability. The hash value of each time node is generated by the SHA-256 algorithm from the hash value of the previous node, the data hash value of the current node, and the timestamp, forming a chained data structure. Doctors can quickly locate to any surgical stage through the time axis sliding control. For example, replay the images and electrophysiological data at the moment of clipping the aneurysm (10:05:12) to check whether the display mode at that time accurately reflects the hemodynamic changes, or trace back whether there are logical loopholes in the parameter configuration during the preoperative planning stage (08:45:00).

[0075] In complex surgical scenarios, the dual review mechanism works in tandem with the data storage function. For example, in a functional area tumor resection surgery, when the system simultaneously displays the fluorescence-labeled tumor boundary (the second feature set) and the functional area mapping results of cortical electrical stimulation (the first feature set), the integrity verification mechanism ensures that both the excitation wavelength parameters of the fluorescence image and the frequency parameters of the electrical stimulation are correctly collected, and the consistency comparison mechanism ensures that the spatial overlap area between the functional area coordinates and the tumor boundary is accurately marked. If the verification finds that the fluorescence signal intensity parameters of a certain area are missing, the system automatically marks this area as "data unreliable" and prompts it with a gray shadow on the display interface. The instruction distribution module sends the verified display instructions to both the surgeon's head-mounted display (HMD) and the high-definition monitor at the assistant's position simultaneously. The HMD directly overlays the functional area mapping results on the surgeon's surgical field through augmented reality (AR) technology, while the monitor displays the multi-modal data in a split-screen format. The scenario storage module archives the perspective data of the HMD (such as head rotation angle, gaze point coordinates) in real time, providing data support for post-operative evaluation of the surgeon's operating habits.

[0076] The dual review mode of the core processing unit ensures the accuracy and reliability of the visualization results through a systematic data verification process; the instruction distribution module realizes the compatible drive of multiple types of display devices through a device-independent coding conversion mechanism; the scenario storage module provides a complete data basis for surgical review, technical improvement, and clinical research through full data archiving and time chain management.

[0077] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising 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 device.

[0078] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A real-time data visualization system for neurosurgery, characterized in that, Including: A real-time data acquisition module, configured to obtain multi-source physiological data during a surgical procedure, and divide the multi-source physiological data into a structured data stream and a dynamic image stream based on a preset data layering rule; A feature construction module, configured to perform neural signal analysis on the structured data stream to generate a first feature set, and perform image segmentation processing on the dynamic image stream to generate a second feature set; A visualization processing module, configured to, according to a preset visualization parameter matrix, perform feature fusion on the first feature set and the second feature set and map them to corresponding display modes, and use the display modes as rendering parameters for the current scene; A decision output module, configured to call a target visualization scheme in a preset scheme database based on the rendering parameters, and use the target visualization scheme as the output configuration for the current scene; A core processing unit, configured to send the multi-source physiological data to the feature construction module, send the first feature set and the second feature set to the visualization processing module, and further perform logical verification on the rendering parameters and the output configuration to generate a final display instruction.

2. The real-time data visualization system during neurosurgery according to claim 1, wherein The feature construction module performing image segmentation processing on the dynamic image stream includes: Dividing a continuous frame sequence in the dynamic image stream into a tissue region group and an instrument region group, and performing boundary recognition calculation on the tissue region group based on a preset segmentation model to generate a contour feature set; Performing spatial coordinate conversion processing on depth imaging data in the dynamic image stream, extracting geometric distribution features of each coordinate region, and constructing a spatial atlas; Performing position association fusion on the contour feature set and the spatial atlas to generate the second feature set.

3. A real-time data visualization system during neurosurgery according to claim 1, wherein The preset data layering rule includes a basic acquisition set and an enhanced acquisition set; The basic acquisition set includes electrophysiological identification, imaging identification, and position sensing identification; the enhanced acquisition set includes blood flow parameter identification and metabolic feature identification, and each identification corresponds to an independent data conversion channel.

4. A real-time data visualization system for neurosurgery according to claim 3, wherein, The system further includes a data interface module, and the data interface module is configured to realize communication docking between the real-time data acquisition module, the feature construction module, the visualization processing module, and the decision output module and surgical devices respectively; The real-time data acquisition module dividing the multi-source physiological data based on the preset data layering rule includes: Receiving a mixed data packet from a surgical device in real time through the data interface module, and matching the header mark of the mixed data packet according to the identification in the basic acquisition set to separate a basic data segment; Traversing the extended mark of the mixed data packet according to the identification in the enhanced acquisition set, where the extended mark is a field including a prefix of the enhanced acquisition set identification and a serial number, for example, the extended mark corresponding to the blood flow parameter identification is "BF_001", and the traversing method is to scan the extended header of the data packet in the order of identification types to extract an enhanced data segment; Aligning the basic data segment and the enhanced data segment in time series and writing them into a structured data storage area and a dynamic image buffer respectively.

5. A real-time data visualization system during neurosurgery according to claim 1, characterized in that, When the preset visualization parameter matrix adopts a hierarchical mapping model, the display mode is the hierarchical mapping result of the composite fusion value of the first feature set and the second feature set, where the composite fusion value is the normalized vector obtained by weighted summation of the first feature set and the second feature set according to preset weights (the default weight of neural signal features is 0.6, and the default weight of imaging features is 0.4), and the weights can be dynamically adjusted according to the surgical scenario; When the preset visualization parameter matrix adopts a dynamic adjustment model, the display mode is a set of continuous variables obtained by dynamically calibrating the joint analysis result of the first feature set and the second feature set through an optimization algorithm.

6. The real-time data visualization system during neurosurgery according to claim 1, characterized in that, It further includes an instrument linkage module connected to the core processing unit, and the instrument linkage module is connected to the surgical instrument database through the data interface module; The instrument linkage module is configured to screen the available device list from the surgical instrument database according to the instrument control requirements in the final display instruction, and generate an instrument activation sequence to optimize the output response process.

7. A real-time data visualization system for neurosurgery according to claim 6, characterized in that, The instrument activation sequence generated by the instrument linkage module includes: Loading the surgical scene grid model and locating the real-time position nodes of each device in the available device list in the grid model; Calculating the optimal response path from the deployment position of each device to the target display area based on the path optimization algorithm, and sorting the available device list according to the response timeliness; Integrating the optimal response path and the priority sorting into the grid model to generate a visual instrument activation sequence.

8. A real-time data visualization system for neurosurgery according to claim 1, characterized in that, When the core processing unit performs logical verification on the rendering parameters and output configuration, it adopts a dual audit mode of an integrity verification mechanism and a consistency comparison mechanism. The integrity verification mechanism is used to confirm the completeness of data elements, and the consistency comparison mechanism is used to solve the spatial conflicts between parameters.

9. A real-time data visualization system during neurosurgery according to claim 1, characterized in that, It further includes an instruction distribution module connected to the core processing unit. The instruction distribution module is configured to convert the final display instruction into a device control code, and send the device control code to the specified display device through the data interface module to start the output program.

10. A real-time data visualization system during neurosurgery according to claim 1, wherein It further includes a scene storage module connected to the core processing unit. The scene storage module is configured to archive the multi-source physiological data, the first feature set, the second feature set, the display mode, and the final display instruction, and generate a panoramic scene record chain according to the time distribution.

Images