Artificial intelligence system of endoscope for minimally invasive surgery

Abstract

The invention belongs to the field of surgical instruments, and provides a surgical minimally invasive surgery endoscope artificial intelligence system which comprises a control box, a mechanical arm, a visual sensor, an inner needle assembly and an air bag. The air bag is used for being placed in an operation opening, a minimally invasive hole or a natural hole of a human body in a closed state and expanding by transmitting air pressure to support and open organs or tissues; the visual sensor comprises a CMOS sensor, a light path assembly, a light filter, a light field and a prism assembly, the visual sensor is combined with the endoscope or the endoscope, and the visual sensor is driven by the mechanical arm to enter a human body along a surgical opening, a minimally invasive hole or a natural hole of the human body so as to capture images of organ tissues in the body; the control box comprises a feature processing unit, a mechanical arm driving control unit, an inner needle control unit and an air bag control unit, the feature processing unit is used for segmenting, recognizing and classifying captured images, the mechanical arm driving control unit is used for driving the mechanical arm to move, and the inner needle control unit is used for controlling the position, angle and depth of an inner needle assembly or a catheter; the airbag control unit is used for controlling air pressure of airbags; according to the method, the time of characteristic analysis of clinical pathology can be shortened, and the reliability is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of surgical instruments, in particular to a surgical minimally invasive surgery endoscope artificial intelligence system. BACKGROUND

[0002] With the development of clinical medicine, the application range of artificial intelligence and optics in the medical field is gradually entering, in order to improve the precise data analysis and surgical operation knowledge base of clinical medicine, thereby improving the level of surgical medicine, At present, the endoscope system uses an optical lens to observe the internal situation, without feature processing, and then analyzes and classifies the existing clinical lesions and pathological conditions to obtain accurate medical clinical data information for reference and diagnosis by doctors, and the clinical pathological time is long and the reliability is low. SUMMARY

[0003] In view of the above technical problems, the present application provides a surgical minimally invasive surgery endoscope artificial intelligence system, which can obtain dynamic images of lesions or tissues and organs in real time and process and analyze image features, and also can compare and learn big data and be referenced by doctors, thereby reducing the time of feature analysis of clinical pathology and improving the reliability.

[0004] Other features and advantages of the present disclosure will become apparent from the following detailed description, or will be learned by practice of the present disclosure.

[0005] The present application discloses a surgical minimally invasive surgery endoscope artificial intelligence system, which comprises a control box, a mechanical arm, a visual sensor, an inner needle assembly and an air bag. The inner needle assembly is used to perform puncture action under the driving of the mechanical arm; The air bag is used to be placed in a surgical opening, a minimally invasive hole or a natural hole of the human body in a closed state, and is inflated by transmitting air pressure to support the organ or tissue, thereby forming a space for surgical operation or observation; The visual sensor comprises a CMOS sensor, an optical path assembly, a filter, an optical field and a rhombus assembly, and is combined with an endoscope or a laparoscope, and is used to enter the human body along the surgical opening, the minimally invasive hole or the natural hole of the human body under the driving of the mechanical arm to capture the image of the internal organ tissue; The control box comprises a feature processing unit, a mechanical arm driving control unit, an inner needle control unit and an air bag control unit, the feature processing unit is used to segment, identify and classify the captured image, the mechanical arm driving control unit is used to drive the movement of the mechanical arm, the inner needle control unit is used to control the position, angle and depth of the inner needle assembly or catheter, and the air bag control unit is used to control the air pressure of the air bag.

[0006] Further, the visual sensor has a multi-angle rotating function.

[0007] Further, the feature processing unit is specifically configured to: quantize the image features into data, including the size, shape, color, thickness, and other substances of the organs; and perform real-time analysis on the pathological benchmark data in the medical database through an algorithm, extract and compare the states of the organ tissue, lesions, and adjacent tissue organs, and form a medical data and analysis report for reference by an on-site doctor or a remote doctor.

[0008] Further, the feature processing unit is further configured to convert the image features and the medical data and analysis report into a knowledge base.

[0009] Further, the system further comprises a display device configured to display the real-time image of the visual sensor and the processing result of the feature processing unit.

[0010] Further, the system further comprises a manual control device, which comprises a monitoring focus knob for controlling the visual sensor and a control handle for controlling the mechanical arm.

[0011] Further, the control box further comprises a physiological signal acquisition and perception evaluation unit, which comprises: a non-invasive electroencephalogram acquisition interface configured to acquire electroencephalogram signals through access electrodes attached to the scalp of a patient; an inquiry information receiving module configured to acquire inquiry data related to the patient's physical perception and past medical history; an electroencephalogram signal processing module configured to filter, artifact suppress, and feature extract the electroencephalogram signals, and perform fusion analysis on the features based on the inquiry data to obtain perception feedback parameters related to pain, swelling, or numbness and the like, and provide the perception feedback parameters for the feature processing unit and / or the mechanical arm driving control unit to call.

[0012] The technical solution of the present disclosure has the following beneficial effects: Precise diagnosis and real-time analysis: the system can analyze and compare pathological data in the medical database in real time by combining artificial intelligence technology to segment, identify, and classify captured images, and provide doctors with precise medical data, analysis reports, and diagnostic references. This helps to improve the accuracy and efficiency of diagnosis.

[0013] Automation and precise control: the coordinated work of the inner needle assembly, the mechanical arm, and the air bag can achieve precise operation control in minimally invasive surgery. The mechanical arm driving control unit can accurately control the movement of the mechanical arm, and the inner needle control unit can accurately adjust the angle, position, and depth of the inner needle, thereby improving the accuracy and safety of the surgery.

[0014] Multi-angle observation and depth image capture: The visual sensor has a multi-angle rotation function, allowing for more comprehensive and three-dimensional image capture of internal organs and tissues during surgery, providing clearer and more detailed views to help surgeons make more accurate judgments.

[0015] Enhanced surgical operating space: The air bag provides support for surgical openings or minimally invasive holes by inflating, forming an operating space. This can effectively expand the operating field of view, improve the flexibility and operability of surgery, especially in small internal spaces.

[0016] Knowledge base and data accumulation: The feature processing unit not only analyzes image data in real time, but also converts it into a medical knowledge base for future learning and reference, promoting the accumulation of medical data and experience and improving the technical level of the medical industry.

[0017] Remote support and cooperation: The system can transmit medical data and analysis reports to remote doctors or experts, supporting remote assistance and improving the accuracy and safety of surgery, especially in resource-limited areas.

[0018] Convenient and flexible operation: The system is equipped with manual control devices such as monitoring focus knobs and control handles, enhancing flexibility during surgery, making it easy for doctors to fine-tune and operate, ensuring the accuracy and safety of surgery. BRIEF DESCRIPTION OF DRAWINGS

[0019] Fig. 1 A schematic diagram of a surgical minimally invasive endoscopic artificial intelligence system according to an embodiment of the present specification; Fig. 2 A structural block diagram of a surgical minimally invasive endoscopic artificial intelligence system according to an embodiment of the present specification. DETAILED DESCRIPTION

[0020] Example implementations will now be described more fully with reference to the accompanying drawings. Example implementations may, however, be implemented in many different forms and should not be construed as limited to the examples set forth herein; rather, these implementations are provided so that this disclosure will be thorough and complete, and will fully convey the scope of example implementations to those skilled in the art. The described features, structures, or characteristics can be combined in one or more implementations. In the following description, numerous specific details are provided to give a thorough understanding of implementations of the disclosure. One skilled in the relevant art will recognize, however, that the implementations of the disclosure can be practiced without one or more of the specific details, or with other methods, components, devices, steps, etc. In other instances, well-known structures have not been described in detail so as not to obscure aspects of the disclosure.

[0021] Further, the accompanying drawings are only schematic and are non-limiting. Identical components have been designated with the same reference numerals throughout the drawings, wherein repeating description of these components has been omitted for clarity. Some of the blocks in the drawings are functional blocks, which can be implemented in software, hardware or a combination thereof. These functional blocks can be implemented in one or more hardware modules or integrated circuits, which are designed to perform the functions described in the drawings.

[0022] As shown in the drawings, Figs. 1-2 The surgical minimally invasive surgery endoscope artificial intelligence system provided by the embodiments of the present disclosure comprises a control box 110, a mechanical arm 120, a visual sensor 150, an inner needle assembly 180, and an air bag. Wherein: The inner needle assembly 180 is used to perform a puncture action under the driving of the mechanical arm 120; The air bag is used to be placed in a surgical opening, a minimally invasive hole or a natural hole of a human body in a closed state, and is inflated by transmitting air pressure to support an organ or tissue, thereby forming a space for surgical operation or observation; The visual sensor 150 comprises a CMOS sensor, an optical path assembly, a filter, a light field, and a rhombus assembly. The visual sensor 150 is combined with an endoscope or a laparoscope, is installed at the end of the mechanical arm 130, and is used to enter the human body along a surgical opening, a minimally invasive hole or a natural hole of a human body under the driving of the mechanical arm 120 to capture images of organs and tissues in the body; The control box 110 comprises a feature processing unit 111, a mechanical arm driving control unit 112, an inner needle control unit 113, and an air bag control unit 114. The feature processing unit 111 is used to segment, identify and classify the captured images. The mechanical arm driving control unit 112 is used to drive the mechanical arm 120 to move. The inner needle control unit 113 is used to control the position, angle and depth of the inner needle assembly 180 or a catheter. The air bag control unit 114 is used to control the air pressure of the air bag.

[0023] The present disclosure aims to provide more accurate and efficient minimally invasive surgical operations by integrating advanced optical technology, mechanical control, air bag support and artificial intelligence analysis. The system comprises a control box 110, a mechanical arm 120, a visual sensor 150, an inner needle assembly 180 and an air bag, and is used to perform surgical operations on a human body located on an operating table 140.

[0024] The inner needle assembly 180 is one of the key components in the system, which completes the puncture action by the mechanical arm 120. This means that the mechanical arm 120 can accurately position the inner needle to the required surgical site under control, reducing human operation errors, thereby improving the accuracy and safety of the operation. The inner needle assembly 180 can be used for puncture, sampling, injection or other operations in minimally invasive surgery, especially suitable for surgical operations that require delicate operations.

[0025] The role of the air bag is to provide additional space in minimally invasive surgery. It can be placed in the surgical opening, minimally invasive hole or natural hole of the human body in the closed state, and inflated by air pressure to form a stable space. This inflation effect can support and separate human organs or tissues, thereby providing more operating space for surgery, and also providing a clearer view for the visual sensor 150. The control unit of the air bag can adjust the air pressure of the air bag to ensure the stability and adaptability of the inflation process, reduce the pressure and damage to the surrounding tissues.

[0026] The visual sensor 150 is an important part of the system, which combines CMOS sensor, optical path assembly, optical filter, light field and rhombus component, and can enter the human body for image capture under the control of the mechanical arm 120. The visual sensor 150 not only can provide high-definition endoscopic images, but also has real-time imaging capability for internal organs and tissues. By combining the visual sensor 150 with an endoscope or a laparoscope, it can observe the internal situation from multiple angles, helping surgeons obtain more information from different angles, thereby making more accurate judgments.

[0027] The control box 110 is the brain of the entire system, integrating multiple control units. The feature processing unit 111 is responsible for intelligent processing of captured images, including image segmentation, recognition and classification. Through these processes, the system can extract useful medical data and analyze the differences between lesions and normal tissues in real time, forming corresponding medical reports. These analysis results can provide reliable reference for surgeons to help them make decisions during surgery. The mechanical arm drive control unit 112 is responsible for controlling the movement of the mechanical arm 120 to ensure accurate positioning of the inner needle and visual sensor 150. The mechanical arm 120 can move accurately to the required position according to the doctor's instructions. The inner needle control unit 113 can accurately control the angle, position and depth of the inner needle to ensure that the inner needle accurately reaches the designated position without damaging the surrounding tissues. The air bag control unit 114 adjusts the air pressure to ensure that the air bag maintains the appropriate inflation state to provide an ideal operating space for the surgery.

[0028] In more detail, the system can greatly improve the precision and safety of minimally invasive surgery through precise control of the mechanical arm 120, high-definition imaging of the visual sensor 150, and intelligent analysis of artificial intelligence. It not only reduces surgical trauma, but also provides more valuable medical data to doctors based on real-time image analysis and feature recognition, improving decision-making ability during surgery In an embodiment, the visual sensor 150 has a multi-angle rotation function. The visual sensor 150 includes a CMOS sensor, an optical path assembly, a filter, a light field, and a rhombus assembly, which, in combination with an endoscope or a laparoscope, can capture high-definition images of internal organs and tissues. These components provide high-quality image support through precise optical design and electronic control. For example, the multi-angle rotation function can be achieved through the following structural design: Rotary structure design: The visual sensor 150 is installed on a multi-degree-of-freedom rotary base, which includes three rotation axes (e.g., horizontal, vertical, and tilt axes), each driven by a micro servo motor. This allows the visual sensor 150 to rotate flexibly in three-dimensional space to meet the observation needs of different surgical scenarios. For example, when the surgery needs to observe a specific angle of the lesion tissue, the doctor can adjust the rotation angle of the sensor through the control system without frequently changing the insertion angle of the endoscope.

[0029] Angle control unit: The control box 110 integrates a dedicated visual sensor 150 angle control unit. This unit receives instructions from the doctor through the manual control device 160 (such as a control handle) or a preset program, and adjusts the rotation angle and position of the visual sensor 150 in real time. For example, during a stomach examination, the visual sensor 150 can achieve horizontal rotation from 0° to 360° or vertical flipping from -90° to +90° through the angle control unit to fully observe the internal wall structure of the stomach cavity.

[0030] Transmission mechanism: The multi-angle rotation function of the visual sensor 150 relies on the internal gear transmission mechanism. The gear system is made of high-precision materials and combined with a micro servo motor to achieve smooth rotation without jamming. For example, using a planetary gear set can improve transmission efficiency and stability, ensuring image stability and clarity during high-speed rotation or fine adjustment.

[0031] Flexible connection design: To reduce the stress limitation of the visual sensor 150 during rotation, the connection part is coated with flexible materials (such as high-strength polymers). This design not only absorbs the impact force during rotation, but also improves the durability of the visual sensor 150, adapting to complex stress conditions in the surgical environment.

[0032] Through the multi-angle rotation function, the visual sensor 150 can flexibly capture images of different angles in a narrow minimally invasive surgery space, providing more comprehensive observation data for the doctor. This function is particularly suitable for the examination and treatment of complex lesions, such as in hepatobiliary surgery, the visual sensor 150 can bypass the blood vessels of the liver and capture images of the lesion area that is difficult to observe directly, thereby assisting the doctor to complete more accurate operations.

[0033] Overall, the multi-angle rotation function of the visual sensor 150 combined with the flexible mechanical arm 120 control enables the system to better meet different surgical needs, improve surgical precision and efficiency, and provide comprehensive image support for doctors, significantly reducing surgical risks In an embodiment, the feature processing unit 111 is specifically configured to: quantize image features into data, including the size, shape, color, thickness, and other substances of organs; and perform real-time analysis by comparing the data with pathological reference data in a medical database through an algorithm, extract and compare the state of organ tissues, lesions, and adjacent organs, and form a medical data and analysis report for reference by an on-site doctor or a remote doctor.

[0034] Among them, the feature processing unit 111 is mainly responsible for in-depth analysis of the images captured by the visual sensor 150, converting them into useful medical data, and comparing and analyzing them with pathological reference data in an advanced algorithm and medical database, and finally generating medical data and analysis reports. The specific workflow and structure design are as follows: Image feature quantization: The feature processing unit 111 first digitizes the captured images, extracts the basic features of organs and tissues through image segmentation, edge detection, and other techniques. These features include but are not limited to the size, shape, color, thickness, texture, structural density of organs, and other substances (such as the boundary of a tumor, color changes in the lesion area, etc.). For example, using image recognition algorithms (such as convolutional neural networks CNN) can accurately extract the contours of different organs or tissues from the image and quantify their size and shape. These quantified data can provide objective basis for subsequent diagnosis.

[0035] Medical database comparison and real-time analysis: After image feature quantization, the feature processing unit 111 compares these data with pathological reference data in a medical database through artificial intelligence algorithms (such as deep learning or machine learning models). For example, the system can compare the size, shape, thickness, etc. of an organ with the organ features of normal people in the database to quickly detect abnormalities. For lesion analysis, the system can identify the differences between the lesion area and normal tissue, such as the shape, color, and edge features of a tumor, and compare them with the annotated pathological data in the database in real time. This comparison can be achieved through dynamic algorithms to ensure the accuracy and timeliness of the analysis process.

[0036] Tissue and lesion state extraction and comparison: The feature processing unit 111 can further extract state information of organ tissues and lesions. For example, in the detection of tumors, the feature processing unit 111 can analyze the growth rate, location, morphology, and other characteristics of the tumor, and compare them with standard pathological data to assess the nature and risk level of the lesion. In addition, the system can also analyze the relationship between the lesion and the surrounding healthy tissue, such as whether the tumor has invaded the surrounding blood vessels, lymph nodes, or other organs. Through these detailed comparisons, the system can provide real-time state reports of the lesion, helping doctors judge the expansion of the lesion and the feasibility of surgical resection.

[0037] Medical data and analysis report generation: Based on the image features and comparison results, the feature processing unit 111 finally generates medical data and analysis reports. These reports contain detailed parameters of organs (such as size, shape, color, etc.), detailed state of lesions, comparison analysis with normal tissues, and possible diagnostic results. These reports can be displayed in the form of charts, text, images, or a combination of them, making it easy for doctors to quickly understand and use. The system can generate these reports in real time for on-site doctors to reference, and also transmit the reports to remote doctors through a secure network, supporting remote consultation and diagnosis.

[0038] Multi-level algorithm support: The feature processing unit 111 uses multi-level algorithms for processing. The basic layer algorithm is responsible for image preprocessing and preliminary feature extraction, the deep learning algorithm further quantizes and recognizes features, and the machine learning algorithm performs disease prediction, analysis, and comparison to form the final medical report. The combination of these algorithms can ensure the accuracy of the processing and the efficiency of the system.

[0039] Through the above functions, the feature processing unit 111 not only provides in-depth analysis of medical images, but also forms detailed and accurate medical data and analysis reports through real-time comparison with medical databases, and converts the image features and medical data, analysis reports into a knowledge base. These data and reports can provide scientific basis for doctors, help on-site doctors make timely and accurate decisions during surgery, and also support remote doctors to participate in diagnosis and surgical decision-making, improving the accuracy of diagnosis and the success rate of surgery The system also includes a display device 170 for displaying the real-time image of the visual sensor 150 and the processing results of the feature processing unit 111.

[0040] The system also includes a manual control device 160, which includes a monitoring focus knob 161 for controlling the visual sensor 150 and a control handle 162 for controlling the mechanical arm 120.

[0041] In an embodiment, the control box further comprises a physiological signal acquisition and perception evaluation unit, which comprises: a non-invasive electroencephalogram acquisition interface for acquiring electroencephalogram signals through access electrodes attached to the patient's scalp; an inquiry information receiving module for obtaining inquiry data related to the patient's physical perception and medical history; and an electroencephalogram signal processing module for filtering, artifact suppression and feature extraction of the electroencephalogram signals, and fusion analysis of the features based on the inquiry data to obtain perception feedback parameters related to pain, swelling or numbness and other physical perceptions and provide the feature processing unit and / or the mechanical arm driving control unit with the perception feedback parameters.

[0042] Exemplarily, the non-invasive electroencephalogram acquisition interface uses scalp electrodes for electroencephalogram (EEG) acquisition, and can select 8 channels of distribution points (such as Fp1, Fp2, F3, F4, C3, C4, Pz, Oz according to the international 10-20 system), and the reference and ground electrodes are arranged near the mastoid or earlobe; the front-end specifications include a sampling rate of 200-500 Hz (such as 250 Hz), 24-bit A / D, input noise not higher than 1 µV_rms, inter-channel crosstalk not higher than -60 dB, and electrode-skin impedance detection function with a threshold not higher than 10 kΩ (acceptable not higher than 20 kΩ). The interface is connected to the control box through a shielded wire or a low-latency wireless link, and if a wireless method is used, the link latency is preferably not more than 20 ms. The inquiry information receiving module acquires structured inquiry data through a touch screen or a voice form, which at least includes chief complaint (pain, swelling, numbness, etc.), site, degree (0-10), duration, trigger, accompanying symptoms, medical history (such as neuropathy, migraine, epilepsy, diabetic peripheral neuropathy, etc.), and current medication. The above physiological signals and inquiry data are shared with the endoscopic image and the mechanical arm state after entering the control box to obtain a unified time stamp for subsequent fusion.

[0043] After power-on, the system first performs impedance self-test to determine the contact quality of each channel: when the impedance of a channel is greater than 10 kΩ, the operator is guided to increase the conductive paste or adjust the electrode; if it does not meet the standard for 5 s continuously, it is allowed to continue in a "degraded operation" mode, but the weight of this channel in subsequent processing is set to zero. After the impedance is qualified, resting baseline data is collected, with 30 s of closed eyes and 30 s of open eyes, to estimate the relative power mean and standard deviation of the individual in the δ (0.5-4 Hz), θ (4-7 Hz), α (8-13 Hz), β (13-30 Hz), γ (30-45 Hz) frequency bands and record the self-evaluation discomfort degree of the three 0-10 scales, which are used as individual benchmarks for subsequent feature normalization and threshold setting. The inquiry information receiving module writes the structured inquiry data into the case context and binds it with the baseline features in this stage, thereby forming prior information for subsequent fusion.

[0044] The collected electroencephalogram signals are processed in a sliding window, with a window length of 2.0 s and an overlap of 0.5 s with the previous window to update every 0.5 s; the signal is first filtered by zero-phase band-pass filtering (0.5-45 Hz) and power frequency notch filtering (50 Hz, or 60 Hz according to the region) to suppress low-frequency drift and power interference. Then, artifact suppression is performed: when the instantaneous amplitude exceeds ±100 µV, linear interpolation repair is performed; when the high-frequency energy ratio E_>30Hz / E_total exceeds 0.35, it is determined that there is electromyographic / movement artifact, and the window is down-weighted or discarded; eye movement / blinking artifact uses fast independent component analysis or reference EOG channel for component rejection. The system calculates the window-level quality score Q (range 0-1) based on impedance, saturation rate and artifact proportion; when Q is lower than 0.6, the window data is down-weighted or discarded to reduce false positives.

[0045] Based on the estimation of power spectral density by the Welch method, five segments of relative power, θ / α ratio, (β+γ) / (α+θ) ratio and spectral entropy are extracted, combined with the Hjorth activity, migration, complexity, sample entropy and zero-crossing rate in the time domain; to enhance the sensitivity to the positioning of the operation stimulus, the left-right hemisphere difference (such as |F3-F4|, |C3-C4|) and the frontal-central ratio ((Fp1+Fp2) / (C3+C4)) are calculated. All features are z-score normalized with the mean and standard deviation of the individual baseline to obtain the standardized feature vector for classification and discrimination, thereby reducing the influence of individual differences on the subsequent threshold criterion.

[0046] The system constructs a channel-band weight matrix based on the inquiry results to reflect prior attention (for example, increasing the frontal channel and θ / α ratio weight when the chief complaint is "facial acid swelling", and increasing the central β, γ weight when the chief complaint is "limb numbness"), and inputting the standardized features into a lightweight classifier (such as logistic regression or linear support vector machine) to obtain the unsmoothed probability p_raw to represent the confidence of "current significant discomfort". To suppress transient fluctuations, a first-order exponential smoothing is used to obtain p_t, with a smoothing coefficient α preferably 0.2-0.3; and the quality score Q is used to modulate p_t to obtain the final perception feedback parameter P=p_t·Q. The system maps P to a four-level label L to facilitate human-machine interface prompting, where P less than 0.4 is determined as "no obvious", 0.4-0.6 is "mild", 0.6-0.8 is "moderate", and not less than 0.8 is "severe", both the parameter and the label can be called by the feature processing unit and / or the mechanical arm driving control unit.

[0047] As a supplement, to ensure safety during the operation, the system sets two levels of linkage thresholds: when P is not less than T1=0.6 and the duration is not less than 1.5 s, the linear speed of the mechanical arm is reduced in proportion to the original set 30%-70% interval; when P is not less than T2=0.8 and the duration is not less than 1.0 s, the system automatically suspends the advancement of the inner needle and prompts the operator to check the operation path and the balloon pressure. When the "severe" state lasts for not less than 5 s, the system suggests changing the strategy or repositioning. The monitoring interface displays P and L with bar indicators, color prompts, and a trend curve of the last 10 s, while providing a "manual override" button and using a double confirmation mechanism to avoid accidental triggering; the end-to-end delay of acquisition-processing-linkage is preferably not more than 200 ms to ensure the real-time feedback.

[0048] The system writes the timestamp, P and L, channel quality score Q, surgical step label, corresponding endoscope frame identifier, mechanical arm state, and interview summary into the case database and knowledge base for postoperative backtracking and offline model updating; by default, a desensitization strategy is adopted to remove identifiable personal information. When the data quality is insufficient (for example, Q is less than 0.4 for more than 2 s continuously), the system automatically enters "observation mode" and only prompts without linkage with the actuator; when electrode shedding or impedance surge is detected, a channel-level alarm is triggered and the weight distribution is adjusted; for special groups with a history of epilepsy, etc., the linkage threshold can be increased based on the prior interview or changed to a prompt mode only, and the surgeon decides whether to enable linkage.

[0049] Based on the above various embodiments, the beneficial effects of the present application are: Precise diagnosis and real-time analysis: the system can analyze and compare pathological data in the medical database in real time by combining artificial intelligence technology to segment, recognize and classify captured images, providing precise medical data, analysis reports and diagnostic references for doctors. This helps to improve the accuracy and efficiency of diagnosis.

[0050] Automation and precise control: the coordinated work of the inner needle assembly 180, the mechanical arm 120, and the balloon can achieve precise operation control in minimally invasive surgery. The mechanical arm drive control unit 112 can accurately control the movement of the mechanical arm 120, and the inner needle control unit 113 can accurately adjust the angle, position and depth of the inner needle, thereby improving the accuracy and safety of the operation.

[0051] Multi-angle observation and depth image capture: the visual sensor 150 has a multi-angle rotation function, which can capture more comprehensive and three-dimensional images of internal organs and tissues during the operation, providing clearer and more detailed views, which helps surgeons make more accurate judgments.

[0052] Enhanced surgical operating space: The air bag provides support for the surgical opening or minimally invasive hole by inflating, forming an operating space. This can effectively expand the operating field of view, improve the flexibility and operability of the surgery, especially in narrow internal space.

[0053] Knowledge base and data accumulation: The feature processing unit 111 not only analyzes image data in real time, but also converts it into a medical knowledge base for future learning and reference, promoting the accumulation of medical data and experience, and improving the technical level of the medical industry.

[0054] Remote support and cooperation: The system can transmit medical data and analysis reports to remote doctors or experts, support remote assistance, improve the accuracy and safety of surgery, and is especially suitable for resource-limited areas.

[0055] Convenient and flexible operation: The system is equipped with manual control devices 160, such as monitoring focus knobs 161 and control handles 162, enhancing the flexibility during the operation process, facilitating the doctor to make fine adjustments and operations, and ensuring the accuracy and safety of the operation.

[0056] Other embodiments of the disclosure will be readily apparent to those skilled in the art upon considering the description herein, together with the practices of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the disclosure that follow the general principles of the disclosure and include common knowledge or conventional techniques in the art that are not disclosed by the disclosure. The specification and embodiments are only considered exemplary, and the true scope and spirit of the disclosure are indicated by the claims.

Claims

1. A minimally invasive surgical endoscopic artificial intelligence system, characterized in that, The system includes a control box, a robotic arm, a vision sensor, an inner needle assembly, and an airbag, wherein: The inner needle assembly is used to perform puncture under the drive of the robotic arm; The airbag is used to be placed in a surgical opening, minimally invasive pore, or natural opening of the human body in a closed state. It expands by transmitting air pressure to support the organ or tissue, thereby forming a space for surgical operation or observation. The visual sensor includes a CMOS sensor, an optical path assembly, a filter, a light field, and a prism assembly. The visual sensor, combined with an endoscope or laparoscope, is used to enter the human body through a surgical opening, a minimally invasive orifice, or a natural opening of the human body under the drive of the robotic arm to capture images of internal organs and tissues. The control box includes a feature processing unit, a robotic arm drive control unit, an inner needle control unit, and an airbag control unit. The feature processing unit is used to segment, identify, and classify the captured images. The robotic arm drive control unit is used to drive the robotic arm to move. The inner needle control unit is used to control the position, angle, and depth of the inner needle assembly or cannula. The airbag control unit is used to control the air pressure of the airbag.

2. The artificial intelligence system for minimally invasive surgical endoscopy according to claim 1, characterized in that, The vision sensor has a multi-angle rotation function.

3. The artificial intelligence system for minimally invasive surgical endoscopy according to claim 1, characterized in that, The feature processing unit is specifically used to: quantify image features into data, including the size, shape, color, thickness and other substances of organs; perform real-time analysis by comparing with pathological benchmark data in the medical database through algorithms, extract and compare the state of organ tissues, lesions and adjacent tissues and organs, and form medical data and analysis reports for reference by on-site doctors or remote doctors.

4. The artificial intelligence system for minimally invasive surgical endoscopy according to claim 3, characterized in that, The feature processing unit is also used to convert the acquired image features and the medical data and analysis report into a knowledge base.

5. The artificial intelligence system for minimally invasive surgical endoscopy according to claim 1, characterized in that, The system also includes a display device for displaying the real-time images from the vision sensor and the processing results from the feature processing unit.

6. The artificial intelligence system for minimally invasive surgical endoscopy according to claim 1, characterized in that, The system also includes a manual control device, which includes a monitoring focus knob for controlling the vision sensor and a control handle for controlling the robotic arm.

7. The artificial intelligence system for minimally invasive surgical endoscopy according to claim 1, characterized in that, The control box also includes a physiological signal acquisition and sensing evaluation unit, which includes: A non-invasive EEG acquisition interface for acquiring EEG signals via access electrodes attached to the patient's scalp; The consultation information receiving module is used to acquire consultation data related to the patient's physical sensations and past medical history; The EEG signal processing module is used to filter, suppress artifacts, and extract features from the EEG signals, and to perform fusion analysis on the features based on the consultation data to obtain sensory feedback parameters related to bodily sensations such as pain, soreness, or numbness, which are then called by the feature processing unit and / or the robotic arm drive control unit.

Images