Neuroendoscope analogue simulation training method and system

Abstract

The invention relates to the technical field of neuroendoscopy analogue simulation training, and discloses a neuroendoscopy analogue simulation training method and a neuroendoscopy analogue simulation training system, through an operation unit, an instrument unit and a display unit, submillimeter-level motion precision and high-frequency and high-precision torque feedback of an operation platform are simulated, force sense perception accurate to milliNewton level can be realized, and the training efficiency is improved. A micro sensor network in the simulation instrument not only can monitor conventional parameters such as bending angle and stretching length in real time, but also can accurately capture detailed information such as opening and closing force and stress distribution of the tip end of the instrument, and feeds back the detailed information to the system in real time; the operation environment and the operation visual field are personally experienced through the VR head-mounted display, a trainee can accurately avoid a key structure and reduce the operation risk in operation planning and operation in the aspect of displaying an intracranial microstructure through the arranged anatomical model construction unit, and real-time updating is performed according to real physiological and pathological changes of a human body through dynamic simulation.

Description

Technical Field

[0001] The present invention relates to the technical field of neuroendoscopic simulation training, and in particular to a neuroendoscopic simulation training method and system. Background Art

[0002] In the modern field of neurosurgery, neuroendoscopic surgery has become a key means for treating many neurological diseases due to its significant advantages such as minimal trauma and rapid postoperative recovery, and is gradually occupying a core position in clinical practice. However, it cannot be ignored that this surgical operation needs to penetrate into the most complex, delicate and fragile intracranial region of the human body. The extremely complex and high-risk surgical environment poses almost harsh requirements on the doctor's operation proficiency, accuracy and emergency handling ability.

[0003] When facing this emerging and highly difficult surgical technique, the traditional neurosurgeon training mode exposes many limitations. Theoretical learning can indeed enable doctors to master basic anatomical knowledge and surgical principles, but it lacks the intuitive experience of actual operation; although cadaver dissection provides certain practical opportunities, cadaver resources are scarce and costly, making it difficult to meet the training needs of a large number of doctors, and the physical properties of cadaver tissues are different from those of living bodies, so the real surgical scene cannot be fully restored; in clinical practice under the guidance of a teaching teacher, due to the diversity and complexity of patient conditions and the consideration of medical safety, novice doctors are difficult to obtain enough and step-by-step operation opportunities, and may also bring additional risks to patients due to unskilled operation.

[0004] Some existing surgical simulation training technologies, although attempting to fill this training gap, mostly focus on general surgical simulation and do not fully consider the uniqueness of neuroendoscopic surgery. In terms of anatomical structure restoration, only the general outline can be presented, and key details such as tiny blood vessel branches and nerve fiber directions cannot be refined; the simulation of instrument operation is simple and crude, and the special feel of neuroendoscopic instruments cannot be accurately simulated, such as the flexible rotation of the endoscope lens and the fine clamping feedback of the biopsy forceps; the surgical scene simulation is single and stereotyped, lacking the real restoration of intraoperative emergencies such as acute bleeding and instrument failures, making it difficult for doctors to obtain comprehensive and combat-like training.

[0005] To solve the above problems, a neuroendoscopic simulation training method and system are proposed in this application. Summary of the Invention

[0006] The purpose of the present invention is to provide a neuroendoscopic simulation training method and system to solve the problems in the prior art that the real surgical scene cannot be fully restored and the intuitive experience of actual operation is lacking as mentioned in the above background art.

[0007] To achieve the above object, the present invention provides the following technical solution: a neuroendoscopic simulation training system, which includes an operation unit, a display unit, an instrument unit, an anatomical model construction unit, a surgical scene simulation unit, and a training evaluation unit;

[0008] The operation unit includes an operation platform and a motion control system. The main body of the operation platform is constructed of an alloy material with high rigidity and low friction coefficient, and has a robotic arm structure with multiple degrees of freedom. Through the motion control system, its motion accuracy is stably controlled at the sub-millimeter level, simulating every subtle movement of the neuroendoscope in the complex intracranial environment;

[0009] The display unit includes a VR headset. The VR headset selects a display with a 4K resolution, a 120Hz refresh rate, and a latency of 1ms. Combining a high-performance GPU and professional graphics rendering software can present a three-dimensional intracranial surgical field to the doctor in real time. The display of the VR headset is built-in with an eye tracking module, which can capture the doctor's line of sight focus and automatically optimize the display of picture details according to the viewing direction;

[0010] The instrument unit includes a simulated neuroendoscope and simulated supporting instruments, which are replicated one-to-one according to the size, shape, and material characteristics of real neuroendoscopes and their supporting instruments;

[0011] The anatomical model construction unit includes an image database, a three-dimensional cranial anatomical model, and a dynamic simulation module. It collects a large amount of human cranial CT and MRI image data of different ages, genders, races, and those suffering from various neurological diseases, constructs a huge and comprehensive image database, and uses a deep learning algorithm. This deep learning algorithm is based on 3D reconstruction technology of convolutional neural networks, combined with region growing algorithms and level set methods to deeply mine and finely process these image data, and constructs a three-dimensional cranial anatomical model. The dynamic simulation module simulates the dynamic changes in the intracranial cavity under different physiological states based on human physiology principles and clinical big data,

[0012] The surgical scene simulation unit includes a surgical case library and a random event generation algorithm. The surgical case library includes preoperative diagnosis materials, surgical indication analysis, multiple sets of surgical plan planning, and complete postoperative evaluation reports for each surgical case. This case library also includes image materials and animation demonstrations. Using the random event generation algorithm and combining surgical cases, various intraoperative emergencies are simulated in real time and randomly during the surgical process;

[0013] The training evaluation unit includes a data acquisition module, an operation database, and a comparison module. The operations in the surgical cases are imported as the operation database. The data acquisition module collects various operation data of the doctor during the training process. Through the comparison module, the data collected by the data acquisition module is compared with the operation database to analyze and evaluate the surgical operations.

[0014] The joints of the robotic arm are equipped with high-precision, high-sensitivity torque sensors, which can keenly capture extremely subtle force changes and feed back signals to the motion control system. The motion control system immediately instructs the robotic arm to sharply increase the resistance of feedback, so as to timely adjust the operating method.

[0015] The surface of the operating platform is covered with a layer of non-slip, anti-static and easy-to-clean flexible material.

[0016] A shock-absorbing device based on the combination of electromagnetic suspension shock-absorbing technology and high-performance damping materials is provided under the operating platform. The vibration sensing element and intelligent adjustment algorithm in the shock-absorbing device will weaken the vibration amplitude transmitted to the operating platform to the micron level.

[0017] The eye tracking module can capture the doctor's eye focus through an infrared camera and an intelligent image recognition algorithm, and automatically optimize the display of image details based on the gaze direction.

[0018] An ambient light sensor is installed on the display of the VR head display to sense in real time the changes in surgical lighting and the reflective lighting of the intracranial tissue itself in the simulated environment.

[0019] The simulated neural endoscope has a built-in micro camera, which uses a CMOS sensor and can transmit the captured images to the VR headset in real time;

[0020] The supporting instruments include biopsy forceps, aspirators, bipolar electrocoagulation, tumor removal forceps, microscissors and microforceps;

[0021] The instrument unit also includes sensors corresponding to the supporting instruments, which track the bending dynamics of the instrument through the cooperation of the built-in high-precision gyroscope and angle encoder, use laser ranging sensors to provide real-time feedback on the extension state of the instrument, and use strain gauge force sensors to sense the force conditions during instrument operation.

[0022] The data acquisition module collects various operation data of doctors during the training process, including the operation trajectory of the neuroendoscope and supporting instruments, the magnitude and duration of the operation force, the completion time of the operation, the operation sequence and time consumption of key steps, and the timeliness and effectiveness of the response to emergencies during the operation, through high-precision sensors arranged on the instruments and operating platforms, and the tracking of the changes in the viewing angle by the VR head display. The collected data is deeply compared and analyzed with the operation database;

[0023] The comparison module includes a machine learning algorithm based on a neural network classification and regression model, adopts a multi-layer perceptron architecture, and optimizes model parameters through a large number of sample training. The comparison module accurately quantifies the doctor's operating skills.

[0024] Among them, a method for endoscopic simulation training is also proposed. The method includes the following steps:

[0025] Case selection and import: After the user logs in to the system, the system scientifically divides the massive built-in cases into three different levels: primary, intermediate, and advanced according to multi-dimensional criteria such as surgical difficulty, disease type, and anatomical complexity. The user accurately selects the most suitable endoscopic surgery case based on their current skill level, training objectives, and weak links. If the user encounters special and rare actual cases in clinical work and needs targeted training, they only need to upload the complete CT and MRI image data of the patient to the server. The intelligent image recognition and processing engine built into the system will be quickly activated, and deep learning algorithms will be used to automatically identify, segment, and three-dimensional reconstruct the images, generating a high-precision three-dimensional anatomical model and corresponding surgical scenarios dedicated to the training of this case;

[0026] Preoperative simulation planning: After selecting a case and importing the relevant models and scenarios, in the VR environment, the user holds a virtual endoscope and interactive tools to comprehensively examine the diseased tissue and its intricate surrounding anatomical structures. The marking tools provided by the system clearly mark the surgical approach, key anatomical landmarks, and structures that need to be protected. The system records every step and detail planned by the user in real time and compares it with the information in the surgical case library. The system provides the user with detailed prompt information and optimization suggestions;

[0027] Surgical operation training: The user holds the simulation endoscope and performs surgical operations on the simulation operation platform. The system drives the VR headset to present a realistic surgical vision feedback by collecting the operation data of the trainee in real time. At the same time, biopsy forceps, aspirators, bipolar electrocoagulation, tumor forceps, micro scissors, and micro forceps are flexibly used to complete tissue sampling, tumor resection, and hemostasis operations. During the operation, the system randomly simulates various unexpected situations during the surgery, and the user makes temporary responses. If they don't know how to respond, press the pause button, and the system will immediately pop up a detailed operation demonstration video recorded by senior experts. The video comprehensively demonstrates how to solve the current problem from the correct operation techniques, instrument selection, and response strategies. After the user trainee watches and learns, they continue to advance the surgery until the entire surgical process is successfully completed;

[0028] Postoperative evaluation and summary: After the user completes the surgery, the training evaluation unit evaluates the entire operation process of the user, displays the quantitative scores of the trainee's surgical operations in the form of charts, covering key indicators such as surgical time, operation accuracy, and scores for dealing with unexpected events, and compares them with expert data. Through the playback of the operation trajectory, the error points and deficiencies of the trainee are accurately pointed out.

[0029] Compared with the prior art, the beneficial effects of the present invention are:

[0030] 1. The present invention simulates the submillimeter motion accuracy and high-frequency, high-precision torque feedback of the operating platform through the operation unit, instrument unit and display unit. When dealing with delicate operations such as those near intracranial micro-blood vessels, compared with the defect that the traditional simulation system can only roughly simulate the movements and cannot accurately perceive the subtle changes in the operating force, the present invention can accurately perceive the force to the millinewton level, allowing trainees to clearly feel the subtle interaction between the instrument and the tissue. The micro-sensor network inside the simulation instrument can not only monitor conventional parameters such as bending angle and stretching length in real time, but also accurately capture the opening and closing force, the force distribution at the tip of the instrument and other detailed information, and immediately feed it back to the system, and the surgical environment and surgical field can be immersively experienced through the VR headset.

[0031] 2. The present invention provides trainees with unprecedented depth of anatomical cognition in terms of displaying the subtle intracranial structures through the anatomical model construction unit, so that they can accurately avoid key structures in surgical planning and operation, reduce surgical risks, and through dynamic simulation, update in real time according to the real physiological and pathological changes of the human body, simulate dynamic scenes such as brain tissue pulsation caused by heartbeat, morphological and texture changes during tumor growth, etc. This is in sharp contrast to most existing static or simple dynamic models, allowing trainees to fully adapt to the uncertainty in real surgery, improve their ability to respond, and combine massive expert cases with random event algorithms to make surgical scene simulation richer, truly honing trainees' emergency decision-making and processing capabilities.

[0032] 3. The present invention provides a training evaluation unit through which preoperative simulation planning can deeply collide with expert wisdom to provide trainees with real-time and accurate optimization suggestions and guide them to make scientific decisions. Postoperative evaluation uses cutting-edge machine learning algorithms and massive expert data to give quantitative scores and detailed improvement suggestions, accurately locate trainees' shortcomings, and help them improve quickly. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 A schematic diagram of the architecture of a neuroendoscopy simulation training system of the present invention;

[0034] Figure 2 It is a schematic diagram of the overall assembly of a neuroendoscopy simulation training system of the present invention;

[0035] Figure 3 A schematic diagram of the workflow of a neuroendoscopy simulation training system of the present invention;

[0036] Figure 4 This is a flow chart of a neuroendoscopy simulation training system of the present invention;

[0037] In the figure:

[0038] 1. Operation unit; 11. Operation platform;

[0039] 2. Display unit;

[0040] 3. Instrument unit;

[0041] 4. Anatomical model construction unit; 41. Image database; 42. Three-dimensional cranial anatomical model; 43. Dynamic simulation module;

[0042] 5. Surgical scene simulation unit; 51. Surgical case library; 52. Random event generation algorithm;

[0043] 6. Training evaluation unit; 61. Data acquisition module; 62. Operation database; 63. Comparison module. Detailed implementation manners

[0044] 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.

[0045] Embodiment 1

[0046] Please refer to Figures 1-4 , the present invention provides a technical solution: a neuroendoscopic simulation training system, which includes an operation unit 1, a display unit 2, an instrument unit 3, an anatomical model construction unit 4, a surgical scene simulation unit 5, and a training evaluation unit 6;

[0047] The operation unit 1 below the operation platform 11 includes an operation platform 11 and a motion control system. The main body of the operation platform 11 below the operation platform 11 is constructed of an alloy material with high rigidity and low friction coefficient, and has a robotic arm structure with multiple degrees of freedom, generally three degrees of freedom, namely the degrees of freedom in the X, Y, and Z axis directions. Through the motion control system, its motion accuracy is stably controlled at the sub-millimeter level, simulating every subtle movement of the neuroendoscope in the complex intracranial environment;

[0048] The display unit 2 below the operation platform 11 includes a VR headset. The VR headset below the operation platform 11 selects a display with a 4K resolution, a 120Hz refresh rate, and a latency of 1ms. Combining a high-performance GPU and professional graphics rendering software can present a three-dimensional intracranial surgical field to the doctor in real time. The display of the VR headset below the operation platform 11 is built-in with an eye tracking module, which can accurately capture the doctor's line of sight focus and automatically optimize the detailed display of the picture according to the viewing direction;

[0049] The instrument unit 3 under the operation platform 11 includes a simulated neuroendoscope and simulated supporting instruments, which are replicated one-to-one according to the size, shape, and material characteristics of real neuroendoscopes and their supporting instruments. The outer surface of the instruments is carefully processed by special techniques and coated with a coating that simulates the touch of real instruments. The physical parameters such as the friction coefficient and elastic modulus of this coating are highly consistent with those of actual instruments, enabling doctors to truly feel the same sense of weight, friction, and instrument feedback force as in real surgeries when holding and operating, and thus quickly forming accurate muscle memory, laying a solid foundation for actual surgical operations.

[0050] The anatomical model construction unit 4 under the operation platform 11 includes an image database 41, a three-dimensional cranial anatomical model 42, and a dynamic simulation module 43. It collects a large amount of CT and MRI image data of the human cranial brain of different ages, genders, races, and those suffering from various neurological diseases, constructing a huge and comprehensive image database 41. Using a deep learning algorithm based on 3D reconstruction technology of convolutional neural networks, combined with region growing algorithms and level set methods, it deeply mines and finely processes these image data to construct a three-dimensional cranial anatomical model 42. The dynamic simulation module 43 simulates the dynamic changes in the cranial cavity under different physiological states based on human physiology principles and clinical big data.

[0051] It should be understood that this three-dimensional cranial anatomical model 42 fully displays every fine structure of the skull (from the outer cortical bone with its dense and clearly distinguishable texture to the inner diploe structure at a glance), different partitions of the brain tissue (such as various functional areas of the cerebral cortex, with clear functional divisions such as the motor area, sensory area, and language area), the fine branching directions of cerebral blood vessels (the minimum resolvable blood vessel diameter reaches 0.1 mm, and even tiny perforating blood vessels can be presented, with their slender lumen and running paths clearly visible), and the intricate arrangement relationship of nerve fiber bundles (through diffusion tensor imaging DTI data reconstruction, visualizing the direction and connection of nerve fibers to display cranial nerve conduction). Moreover, by fusing and learning the image data of different individuals, the model can also intelligently present common anatomical variations, such as abnormal cerebral blood vessel courses in some populations (such as developmental variations of the anterior cerebral artery, showing tortuosity, abnormal branching, etc., and incomplete forms of the Willis circle, such as the absence or hypoplasia of a certain section of blood vessel).

[0052] The surgical scenario simulation unit 5 under the operation platform 11 includes a surgical case library 51 and a random event generation algorithm 52. The surgical case library 51 includes preoperative diagnosis data, surgical indication analysis, multiple sets of surgical plan planning, and complete postoperative evaluation reports for each surgical case. This case library also includes image materials and animation demonstrations. Using the random event generation algorithm 52 and combining surgical cases, it simulates various intraoperative emergencies triggered in real time and randomly during the surgical process.

[0053] Below the operation platform 11, the training and evaluation unit 6 includes a data acquisition module 61, an operation database 62, and a comparison module 63. The operations in the imported surgical cases are used as the operation database 62. The data acquisition module 61 collects various operation data of the doctor during the training process. The comparison module 63 compares the data collected by the data acquisition module 61 with the operation database 62 to analyze and evaluate the surgical operations.

[0054] It should be noted that: The surgical case library 51 includes hundreds of surgical types such as common pituitary tumor resection, endoscopic ventriculostomy for hydrocephalus, as well as rare and complex intracranial deep tumor removal and skull base complex lesion surgery.

[0055] A high-precision and high-sensitivity torque sensor is equipped at the robotic arm joint below the operation platform 11, which can keenly capture extremely subtle force changes and feedback the signals to the motion control system. The motion control system immediately commands the resistance fed back by the robotic arm to increase sharply for timely adjustment of the operation technique.

[0056] The surface of the operation platform 11 is covered with a flexible material with anti-slip, anti-static, and easy-to-clean properties.

[0057] Below the operation platform 11, a shock absorption device combining electromagnetic suspension shock absorption technology and high-performance damping materials is provided. The vibration sensing element and intelligent adjustment algorithm in the shock absorption device weaken the vibration amplitude transmitted to the operation platform 11 to the micron level.

[0058] Below the operation platform 11, the eye tracking module can capture the doctor's eye focus through an infrared camera and intelligent image recognition algorithm, and automatically optimize the detailed display of the picture according to the gaze direction.

[0059] An ambient light sensor is installed on the display of the VR headset below the operation platform 11 to real-time sense the changes in surgical lighting and the reflected light of the intracranial tissues in the simulated environment. For example, when the simulated endoscope approaches the highly reflective skull area, the display will automatically reduce the brightness of this area instantly, and at the same time highlight the details of the surrounding soft tissues to prevent the doctor from being unable to see the key parts due to reflection, greatly improving the comfort and accuracy of the visual experience.

[0060] A micro camera is built into the simulated neurosurgical endoscope below the operation platform 11. The micro camera below the operation platform 11 uses a CMOS sensor. The error of the simulated neurosurgical endoscope is strictly controlled within 0.5 degrees, and the captured images can be transmitted to the VR headset in real time;

[0061] The supporting instruments 11 include biopsy forceps, aspirator, bipolar coagulator, tumor forceps, micro scissors, and micro forceps;

[0062] The instrument unit 3 under the operation platform 11 also includes sensors corresponding to the supporting instruments. It is realized by the cooperation of the built-in high-precision gyroscope and the angle encoder to track the bending dynamics of the instrument. A laser ranging sensor is used to provide real-time feedback on the extension state of the instrument. A strain gauge force sensor is used to sense the force during instrument operation. In the case of biopsy forceps, the error of the feedback force when the simulated biopsy forceps and the real biopsy forceps clamp tissues is accurately controlled within 0.1 Newton, enabling trainees to accurately grasp the operation force with ease, and the operation feel is consistent with that of real surgery.

[0063] The data acquisition module 61 under the operation platform 11 collects various operation data of the doctor during the training process through high-precision sensors arranged on the instruments and the operation platform 11, as well as the tracking of the perspective change by the VR headset, including the operation trajectories of the neuroendoscope and the supporting instruments, the magnitude and duration of the operation force, the operation completion time, the operation sequence and time-consuming of the key steps, and the timeliness and effectiveness of the response to intraoperative emergencies. The collected data is deeply compared and analyzed with the operation database 62.

[0064] The comparison module 63 includes a machine learning algorithm based on a classification and regression model of neural networks, adopting a multi-layer perceptron architecture. The model parameters are optimized through a large number of sample trainings. The comparison module 63 accurately quantifies and scores the doctor's operation skills.

[0065] Please refer to Figures 1-4 , corresponding to the above-mentioned Embodiment 1, this embodiment also proposes a large-space multi-person VR interactive experience method, which includes:

[0066] Case selection and import. After the user logs in to the system, the system scientifically divides the built-in massive cases into three different levels: primary, intermediate, and advanced according to multi-dimensional criteria such as surgical difficulty, disease type, and anatomical complexity. The user accurately selects the most suitable neuroendoscopic surgery case according to their current skill level, training objectives, and weak links. If the user encounters special and rare actual cases in clinical work and needs targeted training, they only need to upload the complete CT and MRI image data of the patient to the server, and the built-in intelligent image recognition and processing engine of the system will be quickly activated. Using deep learning algorithms, the images are automatically recognized, segmented, and three-dimensionally reconstructed to generate a high-precision three-dimensional anatomical model and the corresponding surgical scene dedicated to the training of this case.

[0067] Preoperative simulation and planning. After selecting a case and importing relevant models and scenarios, in the VR environment, the user holds a virtual neuroendoscope and interactive tools to comprehensively examine the diseased tissue and the intricate anatomical structures around it. The marking tools provided by the system clearly mark the surgical approach, key anatomical landmarks, and structures that need to be protected with emphasis. The system records every step and detail of the user's plan in real time and compares it with the information in the surgical case library 51. The system provides the user with detailed prompt information and optimization suggestions;

[0068] Surgical operation training. The user holds a simulated neuroendoscope and performs surgical operations on the simulation operation platform 11. The system collects the operation data of the trainee in real time and drives the VR headset to present a realistic surgical field feedback. At the same time, the user flexibly uses biopsy forceps, aspirator, bipolar coagulation, tumor forceps, micro scissors, and micro forceps to complete tissue sampling, tumor resection, and hemostasis operations. During the operation, the system randomly simulates various unexpected situations during the surgery, and the user makes temporary responses. If the user doesn't know how to respond, press the pause button, and the system immediately pops up a detailed operation demonstration video recorded by senior experts. The video comprehensively demonstrates how to resolve the current problem from the correct operation techniques, instrument selection, and response strategies. After the user trainee watches and learns, continue to advance the surgery until the entire surgical process is successfully completed;

[0069] Postoperative evaluation and summary. After the user completes the surgery, the training evaluation unit 6 evaluates the entire operation process of the user, displays the quantitative score of the user's surgical operation in the form of a chart, covering key indicators such as operation time, operation accuracy, and scores for dealing with unexpected events, and compares it with the expert data. Through the playback of the operation trajectory, accurately point out the error points and deficiencies of the trainee.

[0070] 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 scope of protection of the present invention.

Claims

1. A neuroendoscopy simulation training system, characterized by: The system comprises an operation unit (1), a display unit (2), an instrument unit (3), an anatomical model construction unit (4), a surgical scene simulation unit (5), and a training evaluation unit (6). The operating unit (1) comprises an operating platform (11) and a motion control system. The operating platform (11) is constructed of a high-rigidity, low-friction alloy material and has a multi-degree-of-freedom mechanical arm structure. Through the motion control system, the motion accuracy is stably controlled at the sub-millimeter level, simulating every subtle movement of the neuroendoscope in the complex intracranial environment. The display unit (2) comprises a VR head display, which uses a display with a 4K resolution, a 120 Hz refresh rate and a delay of 1 ms, and is capable of presenting a three-dimensional intracranial surgical field of view to a doctor in real time in combination with a high-performance GPU and professional graphics rendering software. The display of the VR head display has an eye tracking module built in, which can capture the doctor's line of sight and automatically optimize the display of picture details according to the gaze direction; The instrument unit (3) includes a simulated neuroendoscope and simulated supporting instruments, which are reproduced one-to-one according to the size, shape, and material characteristics of a real neuroendoscope and its supporting instruments; The anatomical model construction unit (4) includes an image database (41), a three-dimensional cranial anatomical model (42) and a dynamic simulation module (43), which collects a large amount of human cranial CT and MRI image data of different age groups, genders, races and people with various neurological diseases, and constructs a large and comprehensive image database (41). A deep learning algorithm is used. The deep learning algorithm is based on the 3D reconstruction technology of convolutional neural networks, combined with the regional growing algorithm and the level set method to deeply mine and finely process these image data to construct a three-dimensional cranial anatomical model (42). The dynamic simulation module (43) simulates the dynamic changes in the skull under different physiological states based on the principles of human physiology and clinical big data. The surgical scene simulation unit (5) includes a surgical case library (51) and a random event generation algorithm (52). The surgical case library (51) includes preoperative diagnosis data, surgical indication analysis, multiple surgical plan plans, and a complete postoperative evaluation report for each surgical case. The case library also includes image data and animation demonstrations. The random event generation algorithm (52) is used in combination with the surgical cases to simulate the real-time and random triggering of various intraoperative emergencies during the surgical process. The training evaluation unit (6) includes a data acquisition module (61), an operation database (62) and a comparison module (63). The operations in the surgical case are imported as the operation database (62). The data acquisition module (61) collects various operation data of the doctor during the training process. The data collected by the data acquisition module (61) is compared with the operation database (62) through the comparison module (63) to analyze and evaluate the surgical operation.

2. A neuroendoscopy simulation training system according to claim 1, characterized in that: The joints of the robotic arm are equipped with high-precision, high-sensitivity torque sensors, which can keenly capture extremely subtle force changes and feed back signals to the motion control system. The motion control system immediately instructs the robotic arm to sharply increase the resistance of feedback, so as to timely adjust the operating method.

3. A neuroendoscopy simulation training system according to claim 1, characterized in that: The surface of the operating platform (11) is covered with a layer of non-slip, anti-static and easy-to-clean flexible material.

4. A neuroendoscopy simulation training system according to claim 1, characterized in that: A shock absorbing device based on a combination of electromagnetic suspension shock absorbing technology and high-performance damping material is provided below the operating platform (11); the vibration sensing element and intelligent adjustment algorithm in the shock absorbing device weaken the vibration amplitude transmitted to the operating platform (11) to the micron level.

5. A neuroendoscopy simulation training system according to claim 1, characterized in that: The eye tracking module can accurately capture the doctor's eye focus through an infrared camera and an intelligent image recognition algorithm, and automatically optimize the display of image details based on the gaze direction.

6. A neuroendoscopy simulation training system according to claim 5, characterized in that: An ambient light sensor is installed on the display of the VR head display to sense in real time the changes in surgical lighting and the reflective lighting of the intracranial tissue itself in the simulated environment.

7. A neuroendoscopy simulation training system according to claim 1, characterized in that: The simulated neural endoscope has a built-in micro camera, which uses a CMOS sensor and can transmit the captured images to the VR headset in real time; The supporting instruments include biopsy forceps, aspirators, bipolar electrocoagulation, tumor removal forceps, microscissors and microforceps; The instrument unit (3) also includes a sensor corresponding to the matching instrument, which tracks the bending dynamics of the instrument through the cooperation of a built-in high-precision gyroscope and an angle encoder, uses a laser rangefinder sensor to provide real-time feedback on the extension state of the instrument, and uses a strain gauge force sensor to sense the force applied to the instrument during operation.

8. A neuroendoscopy simulation training system according to claim 1, characterized in that: The data acquisition module (61) collects various operation data of the doctor during the training process, including the operation trajectory of the neuroendoscope and supporting instruments, the magnitude and duration of the operation force, the completion time of the operation, the operation sequence and time consumption of key steps, and the timeliness and effectiveness of the response to emergencies during the operation, through high-precision sensors arranged on the instrument and the operation platform (11) and the tracking of the change of the viewing angle by the VR head display. The collected data is compared and analyzed in depth with the operation database (62); The comparison module (63) includes a machine learning algorithm based on a neural network classification and regression model, adopts a multi-layer perceptron architecture, and optimizes model parameters through a large number of sample training. The comparison module (63) accurately quantifies the doctor's operating skills.

9. A neuroendoscopic simulation training method, characterized in that: The method uses a neuroendoscopy simulation training system according to any one of claims 1 to 8, and the method comprises the following steps: Case selection and import: after the user logs into the system, the system scientifically divides the built-in massive cases into three different levels: elementary, intermediate, and advanced, based on multi-dimensional standards such as surgical difficulty, disease type, and anatomical complexity. Users can accurately select the most suitable neuroendoscopic surgery cases based on their current skill level, training goals, and weak links. If users encounter special and rare actual cases in clinical work and need targeted training, they only need to upload the patient's complete CT and MRI image data to the server. The system's built-in intelligent image recognition and processing engine will start quickly, using deep learning algorithms to automatically identify, segment, and 3D reconstruct images, generating high-precision 3D anatomical models and corresponding surgical scenes specifically for case training. Preoperative simulation planning: After selecting a case and importing relevant models and scenarios, in the VR environment, the user holds a virtual neuroendoscope and interactive tools to examine the lesion tissue and its surrounding complex anatomical structures from all angles. The marking tools provided by the system clearly mark the surgical approach, key anatomical landmarks, and structures that need to be protected. The system records every step and detail of the user's planning in real time and compares it with the information in the surgical case library (51). The system provides the user with detailed prompt information and optimization suggestions. In surgical operation training, the user holds a simulated neuroendoscope and performs surgical operations on a simulated operation platform (11). The system collects the trainee's operation data in real time, drives the VR headset to present realistic surgical field feedback, and flexibly uses biopsy forceps, aspirators, bipolar electrocoagulation, tumor removal forceps, microscissors and microtweezers to complete tissue sampling, tumor resection, and hemostasis operations. During the operation, the system randomly simulates various emergencies during the operation, and the user responds temporarily. If the user does not know how to respond, press the pause button, and the system will immediately pop up a detailed operation demonstration video recorded by a senior expert. The video shows how to resolve the current problem from the perspective of correct operation techniques, instrument selection, and response strategies. After watching and learning, the user trainee continues to advance the operation until the entire surgical process is successfully completed; Postoperative evaluation and summary: After the user completes the operation, the training evaluation unit (6) evaluates the entire operation process of the user, and displays the quantitative score of the trainee's surgical operation in the form of a chart, covering key indicators such as operation time, operation accuracy, and emergency response score, and compares it with expert data. By replaying the operation trajectory, the trainee's mistakes and shortcomings are accurately pointed out.

Images