AI auxiliary analysis method and system for operation video

Abstract

The invention relates to the technical field of AI auxiliary analysis, and particularly discloses an operation video-oriented AI auxiliary analysis method and system. The method comprises the following steps of S1, acquiring an operation video and performing preprocessing; S2, performing AI auxiliary analysis on an operation illumination condition; S3, performing AI auxiliary analysis on an operation patient index; S4, performing AI auxiliary analysis on an operation risk index; according to the invention, real-time risk early warning can be realized, and doctors are helped to take measures to avoid or reduce surgical risks; compared with a traditional method, manual intervention and errors are remarkably reduced, and the objectivity and reliability of analysis are improved; by introducing the image processing technology and the AI technology, accurate analysis of operation videos is achieved, comprehensive operation auxiliary information is provided for doctors, the safety and efficiency of operations are effectively improved, optimal configuration of medical resources is promoted, and the method has important clinical value and application prospects.

Description

Technical Field

[0001] The present invention relates to the field of AI-assisted analysis technology, and in particular to an AI-assisted analysis method and system for surgical videos. Background Art

[0002] In the field of modern surgery, the core elements to ensure patient health and surgical success lie in surgical safety and risk assessment. With the rapid development of medical technology, the complexity and sophistication of surgery are constantly increasing, which in turn puts more stringent requirements on the surgical environment. During the operation, doctors need to make accurate operational decisions in a very short time, because any small mistake may cause serious complications or adverse postoperative results. In view of this, the recording and analysis of surgical videos is particularly important. It not only helps to improve the success rate of surgery, but is also an indispensable part of ensuring patient safety.

[0003] In recent years, the application of AI technology in the medical field has made significant progress, especially in image recognition, natural language processing, behavior recognition, etc. AI technology can automatically extract key information from surgical videos through learning and analyzing large amounts of data, providing more objective and comprehensive surgical evaluations.

[0004] Traditional surgical video analysis methods rely too much on manual viewing and have significant drawbacks. First, this method is time-consuming and labor-intensive, which greatly limits its widespread application in clinical practice and may delay doctors' timely assessment of surgical results and patient conditions. Secondly, traditional methods cannot achieve real-time risk warning and dynamic adjustment. Doctors often perceive risks only after problems emerge, missing the best time to respond. At the same time, manually marking key events and operation steps in surgical videos is not only time-consuming and error-prone, which undoubtedly affects the accuracy of subsequent analysis. In addition, traditional methods have low analysis efficiency, strong subjectivity, and difficulty in capturing subtle changes during surgery. More importantly, for the monitoring of patients' physiological indicator data, traditional methods mainly rely on medical equipment, cannot be combined with real-time surgical videos for dynamic analysis, and cannot intuitively display indicator data during surgery, thereby limiting doctors' overall control of surgery and accurate decision-making. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides an AI-assisted analysis method and system for surgical videos to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above-mentioned object, the present invention provides the following technical solutions: an AI-assisted analysis method for surgical videos, comprising the following steps: S1: acquiring surgical videos and preprocessing them, S2: AI-assisted analysis of surgical lighting conditions, S3: AI-assisted analysis of surgical patient indicators, S4: AI-assisted analysis of surgical risk indicators, S5: evaluating a quantitative index of surgical risk, and S6: establishing an interactive user interface; S1: Obtaining and preprocessing a surgical video: performing video monitoring on the surgical process, obtaining a first surgical video, and preprocessing the first surgical video to obtain a second surgical video; S2: AI-assisted analysis of surgical lighting conditions: Use image processing technology to extract surgical images in the second surgical video, analyze the surgical light domain adaptation index, and use AI-assisted analysis of surgical lighting conditions. If the analysis result shows that the lighting conditions meet the requirements, further execute S3, otherwise jump out of the step loop; S3: AI-assisted analysis of surgical patient indicators: Extract the patient's physiological indicator data from the second surgical video that meets the surgical lighting conditions, analyze and obtain the physiological homeostasis assessment coefficient, use AI to assist in analyzing the patient's physiological homeostasis, and issue warning information for abnormal results; S4: AI-assisted analysis of surgical risk indicators: Extract the patient's lesion area, bleeding volume, and operation time from the second surgical video that meets the surgical lighting conditions, and analyze and obtain a quantitative surgical risk index based on the physiological homeostasis assessment coefficient; S5: Evaluate the quantitative surgical risk index: Evaluate the quantitative surgical risk index by comparing it with the preset risk threshold, analyze the surgical risk status of the patient through AI assistance, and issue warning information for data with abnormal evaluation results; S6: Establish an interactive user interface: Establish an interactive user interface for real-time display of surgical video images, patient physiological homeostasis, surgical risk assessment results and early warning information, and send data with abnormal judgment and assessment results to doctors and medical staff.

[0007] Preferably, the specific method of obtaining the surgical video and performing preprocessing is: The first surgical video is preprocessed by adjusting the video screen color, brightness and contrast and removing noise to obtain a second surgical video.

[0008] Preferably, the specific method for analyzing the surgical light field adaptation index is: S201: Obtaining the brightness correlation coefficient between the standard image and the image to be tested NCC , substituting it into the formula Get the surgical light field adaptation index ODAI ,in, represents the hyperbolic tangent function, k represents the correction factor; S202: Analyze surgical lighting conditions: extract the surgical light field adaptation index and compare it with the preset adaptation index threshold. If the surgical light field adaptation index is greater than or equal to the preset adaptation index threshold, it is determined that the surgery meets the standard lighting conditions, and further execute S3. If the surgical light field adaptation index is less than the preset adaptation index threshold, it is determined that the surgery does not meet the standard lighting conditions, and the step loop is jumped.

[0009] Preferably, the brightness correlation coefficient of the standard image and the image to be tested is obtained NCC The specific content is: The first step is to use a light meter to calibrate the surgical lamps one by one in a standard environment, adjust the height and angle of the surgical lamps, and use a high-definition camera to capture images from different surgical perspectives, including but not limited to the surgeon's perspective, assistant's perspective, and bird's-eye perspective, and mark them as standard images; The second step is to obtain the surgical video image and mark it as the image to be tested; pre-process the standard image and the image to be tested, wherein the sizes of the standard image and the image to be tested are ; Extract the brightness value of the pixel located at row x and column y in the standard image , substituting it into the formula Get the average brightness value of the standard image , extract the brightness value of the pixel located in the xth row and yth column in the image to be tested , substituting it into the formula Get the average brightness value of the image to be tested , where M represents the number of pixels in the height direction of the standard image and the surgical video image, i.e., the number of rows, and M represents the number of pixels in the width direction of the standard image and the surgical video image, i.e., the number of columns, x=1, 2, …, M, y=1, 2, …, N; The third step is to calculate the correlation coefficient of the brightness values ​​of the standard image and the image to be tested. NCC , the calculation model is: .

[0010] Preferably, the specific method of AI-assisted analysis of surgical patient indicators is: The first step is to locate the patient's body temperature detection area ROI through the facial feature recognition algorithm, and obtain the pixel value of the green light G channel in the xth row and yth column in the detection area. , substituting it into the formula Get the average intensity of the green channel pixels Ia ,in, It is represented by the total number of pixels in the ROI to be detected, and the average intensity of the green channel pixels is filtered and marked as If, using the peak detection algorithm to determine the heartbeat peak, when If If the heart rate exceeds the preset threshold, it is determined as a heartbeat peak and the time interval between adjacent peaks is recorded. , substituting it into the formula The patient's heart rate HR is obtained, where n represents the total number of heartbeat peaks in the area to be detected, and i represents the number of the heartbeat peak in the area to be detected; The second step is to obtain the elapsed time interval Heart rate change after , body temperature change and respiratory rate changes , calculate the heart rate change rate , body temperature change rate , respiratory rate change rate ; The third step is to calculate the physiological homeostasis assessment coefficient. The specific calculation model is: ,in, SEC represents the patient's physiological homeostasis assessment coefficient, rRR 0 means the preset maximum breathing rate change rate, e represents a natural constant; The fourth step is to compare the physiological homeostasis assessment coefficient with the preset steady-state assessment coefficient threshold to analyze the patient's physiological homeostasis condition. If the physiological homeostasis assessment coefficient is greater than or equal to the preset steady-state assessment coefficient threshold, the patient's physiological indicators are judged to be normal. If the physiological homeostasis assessment coefficient is less than the preset steady-state assessment coefficient threshold, the patient's physiological indicators are judged to be abnormal, and a warning signal is issued for the abnormal result and sent to the doctor.

[0011] Preferably, the specific method of AI-assisted analysis of surgical risk indicators is: The first step is to select video frame images from the second surgical video, convert the selected video frame images from the RGB color space to the HSV color space, use Gaussian filtering or median filtering preprocessing methods to remove noise interference in the image, extract the patient's lesion area from the video frame image based on the convolutional neural network segmentation algorithm, and obtain the number of pixels in the lesion area pn , get the spatial resolution r , substituting it into the formula Get the patient's lesion area A ; The second step is to install a spectrum acquisition device on the surgical video recording device. During the operation, the spectrum data of the surgical area is collected at a time interval of once per second, and the collected spectrum data is marked as , where λ represents the wavelength, tIndicates the time, a pure blood sample is extracted from the management database, placed in a cuvette, and spectral measurement is performed under the same equipment and conditions as the spectrum acquisition in the surgical area to obtain the spectral data of the blood sample , by analyzing the spectrum of the blood sample, determine the characteristic wavelength λ of the blood and substitute it into the formula Get the absorbance of blood ,in, Indicates the reference light intensity, that is, the light intensity measurement value at the same wavelength when there is no blood absorption; obtains the optical path length , according to the Lambert-Beer law formula Get the blood concentration at each characteristic wavelength λ and time point t ,in, represents the molar absorption coefficient, obtain the volume V of the surgical area, and substitute it into the formula Get the patient's surgical bleeding BL ; The third step is to read the patient's lesion area A , Bleeding BL and duration of surgery OT , and standardize them respectively to calculate the quantitative index of surgical risk ORQI , the calculation model is as follows: ,in, SEC represents the patient's physiological homeostasis assessment coefficient, S Indicates the area of ​​the patient's surgical area, BL 0 represents the maximum allowable bleeding volume preset for the corresponding surgery type. OT 0 means the preset maximum operation time for the corresponding operation type. Represent the weight factors of lesion area, bleeding volume and operation time, respectively, and .

[0012] Preferably, the specific method for evaluating the quantitative index of surgical risk is: The surgical risk quantification index is extracted and compared with the preset risk threshold. If the surgical risk quantification index is less than or equal to the preset risk threshold, the patient's surgery is judged to have no abnormal risk. If the surgical risk quantification index is greater than the preset risk threshold, the patient's surgical risk is judged to be abnormal. The abnormal data will be promptly warned to the doctor by means of alarm sounds and pop-up images.

[0013] To achieve the above object, the present invention provides the following technical solution: an AI-assisted analysis system for surgical videos, implementing the above-mentioned AI-assisted analysis method for surgical videos, comprising: A surgical video acquisition and processing module: used for performing video monitoring of the surgical process, acquiring a first surgical video, and preprocessing the first surgical video to obtain a second surgical video; AI-assisted analysis module for surgical lighting conditions: This module uses image processing technology to extract surgical images from the second surgical video, analyzes the surgical light domain adaptation index, and uses AI to assist in analyzing surgical lighting conditions. AI-assisted analysis module for surgical patient indicators: used to extract the patient's physiological indicator data from the second surgical video that meets the surgical lighting conditions, analyze and obtain the physiological homeostasis assessment coefficient, analyze the patient's physiological homeostasis through AI assistance, and issue early warning information for abnormal results; Surgical risk index AI-assisted analysis module: used to extract the patient's lesion area, bleeding volume and operation time from the second surgical video that meets the surgical lighting conditions, and analyze and obtain the quantitative surgical risk index based on the physiological steady-state assessment coefficient; Module for evaluating the quantitative index of surgical risk: used to evaluate the quantitative index of surgical risk by comparing it with the preset risk threshold, analyze the surgical risk status of the patient through AI assistance, and issue early warning information for data with abnormal evaluation results; Establish an interactive user interface module: Establish an interactive user interface to display the surgical video screen, the patient's physiological steady-state condition, the surgical risk assessment results and early warning information in real time, and send the data with abnormal judgment and assessment results to doctors and medical staff.

[0014] Technical effects and advantages of the present invention: 1. By introducing advanced image processing and artificial intelligence technologies, the present invention can automatically and quickly analyze surgical videos and accurately extract key information, thereby greatly improving the efficiency and accuracy of surgical video analysis; 2. The present invention can monitor indicator data during the operation in real time, realize real-time risk warning, and help doctors take measures before or in the early stages of problems, effectively avoiding or reducing surgical risks; compared with traditional methods, it significantly reduces human intervention and errors, and improves the objectivity and reliability of analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0016] Figure 1 The figure is a flow chart of an AI-assisted analysis method for surgical videos of the present invention.

[0017] Figure 2 This is a structural schematic diagram of an AI-assisted analysis system for surgical videos of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] Example 1 See also Figure 1 As shown, the present invention provides an AI-assisted analysis method for surgical videos, comprising the following steps: S1: acquiring surgical videos and performing preprocessing, S2: AI-assisted analysis of surgical lighting conditions, S3: AI-assisted analysis of surgical patient indicators, S4: AI-assisted analysis of surgical risk indicators, S5: evaluating a quantitative index of surgical risk, and S6: establishing an interactive user interface; S1: Obtaining and preprocessing a surgical video: performing video monitoring on the surgical process, obtaining a first surgical video, and preprocessing the first surgical video to obtain a second surgical video; In this embodiment, it should be specifically explained that the specific method of obtaining the surgical video and performing preprocessing is: Preprocessing the first surgical video by adjusting the video screen color, brightness and contrast and removing noise to obtain a second surgical video; The adjustment of the video screen color is based on the white or gray reference in the first surgical video, and the screen color is corrected according to its RGB value; the brightness and contrast adjustment adopts the gamma correction formula to change the brightness of the image according to the gamma value; the histogram stretching method is used to stretch the distribution range of the image pixel value to the entire grayscale range to adjust the contrast; the noise removal is performed on the first surgical video by using the mean filtering or median filtering method.

[0020] S2: AI-assisted analysis of surgical lighting conditions: Use image processing technology to extract surgical images in the second surgical video, analyze the surgical light domain adaptation index, and use AI-assisted analysis of surgical lighting conditions. If the analysis result shows that the lighting conditions meet the requirements, further execute S3, otherwise jump out of the step loop; In this embodiment, it should be specifically explained that the specific method of AI-assisted analysis of surgical lighting conditions is: The first step is to use a light meter to calibrate the surgical lamps one by one in a standard environment, adjust the height and angle of the surgical lamps, and use a high-definition camera to capture images from different surgical perspectives, including but not limited to the surgeon's perspective, assistant's perspective, and bird's-eye perspective, and mark them as standard images; In this embodiment, it should be specifically explained that the standard environment refers to an environment with known light intensity, uniform light, and in accordance with the current type of surgery; the standard image is obtained in a strictly controlled professional light measurement environment, where not only the light intensity is precisely calibrated to meet the ideal surgical lighting requirements, but also the uniformity meets the best practice standards, ensuring that there is no obvious difference in light and shade in the entire surgical area. For example, a professional light meter is used to finely adjust the surgical shadowless lamp at different heights and angles, and a series of images with detailed light intensity numerical labels are taken in conjunction with a specially designed simulated surgical scene.

[0021] The second step is to obtain the surgical video image and mark it as the image to be tested; pre-process the standard image and the image to be tested, wherein the sizes of the standard image and the image to be tested are ; Extract the brightness value of the pixel located at row x and column y in the standard image , substituting it into the formula Get the average brightness value of the standard image , extract the brightness value of the pixel located in the xth row and yth column in the image to be tested , substituting it into the formula Get the average brightness value of the image to be tested , where M represents the number of pixels in the height direction of the standard image and the surgical video image, i.e., the number of rows, and M represents the number of pixels in the width direction of the standard image and the surgical video image, i.e., the number of columns, x=1, 2, …, M, y=1, 2, …, N; The third step is to calculate the correlation coefficient of the brightness values ​​of the standard image and the image to be tested. NCC , the calculation model is: , where the value range of NCC is between -1 and 1. NCC close to 1 indicates that the brightness distribution of the two images is similar, that is, the lighting condition of the surgical video image is close to the lighting condition of the standard image, which meets the surgical lighting conditions; NCC close to -1 indicates that the difference in brightness distribution of the two images is greater, and the relationship is inverse. NCC close to 0 indicates that there is almost no linear brightness correlation between the two images, which does not meet the surgical lighting conditions.

[0022] In this embodiment, it should be specifically explained that the specific method for analyzing the surgical light domain adaptation index is: S201: Obtaining the brightness correlation coefficient of the standard image and the image to be tested NCC , substituting it into the formula Get the surgical light field adaptation index ODAI ,in, represents the hyperbolic tangent function, k represents the correction factor; S202: Analyze surgical lighting conditions: extract the surgical light field adaptation index and compare it with the preset adaptation index threshold. If the surgical light field adaptation index is greater than or equal to the preset adaptation index threshold, it is determined that the surgery meets the standard lighting conditions, and further execute S3. If the surgical light field adaptation index is less than the preset adaptation index threshold, it is determined that the surgery does not meet the standard lighting conditions, and the step loop is jumped.

[0023] S3: AI-assisted analysis of surgical patient indicators: Extract the patient's physiological indicator data from the second surgical video that meets the surgical lighting conditions, analyze and obtain the physiological homeostasis assessment coefficient, use AI to assist in analyzing the patient's physiological homeostasis, and issue warning information for abnormal results; In this embodiment, it should be specifically explained that the specific method of AI-assisted analysis of surgical patient indicators is: The patient's physiological index data includes the patient's heart rate, body temperature and respiratory rate; The first step is to locate the patient's body temperature detection area ROI through the facial feature recognition algorithm, and obtain the pixel value of the green light G channel in the xth row and yth column in the detection area. , substituting it into the formula Get the average intensity of the green channel pixels Ia ,in, It is represented by the total number of pixels in the ROI to be detected, and the average intensity of the green channel pixels is filtered and marked as If , using the peak detection algorithm to determine the heartbeat peak, when If If the heart rate exceeds the preset threshold, it is determined as a heartbeat peak and the time interval between adjacent peaks is recorded. , substituting it into the formula The patient's heart rate HR is obtained, where n represents the total number of heartbeat peaks in the area to be detected, and i represents the number of the heartbeat peak in the area to be detected; The second step is to obtain the elapsed time interval Heart rate change after , body temperature change and respiratory rate changes , calculate the heart rate change rate , body temperature change rate , respiratory rate change rate ; The third step is to calculate the physiological homeostasis assessment coefficient. The specific calculation model is: ,in, SEC represents the patient's physiological homeostasis assessment coefficient, rRR 0 means the preset maximum breathing rate change rate, e represents a natural constant; The fourth step is to compare the physiological homeostasis assessment coefficient with the preset steady-state assessment coefficient threshold to analyze the patient's physiological homeostasis condition. If the physiological homeostasis assessment coefficient is greater than or equal to the preset steady-state assessment coefficient threshold, the patient's physiological indicators are judged to be normal. If the physiological homeostasis assessment coefficient is less than the preset steady-state assessment coefficient threshold, the patient's physiological indicators are judged to be abnormal, and a warning signal is issued for the abnormal result and sent to the doctor.

[0024] S4: AI-assisted analysis of surgical risk indicators: Extract the patient's lesion area, bleeding volume, and operation time from the second surgical video that meets the surgical lighting conditions, and analyze and obtain a quantitative surgical risk index based on the physiological homeostasis assessment coefficient; In this embodiment, it should be specifically explained that the specific method of AI-assisted analysis of surgical risk indicators is: The first step is to select video frame images from the second surgical video, convert the selected video frame images from the RGB color space to the HSV color space, use Gaussian filtering or median filtering preprocessing methods to remove noise interference in the image, extract the patient's lesion area from the video frame image based on the convolutional neural network segmentation algorithm, and obtain the number of pixels in the lesion area pn , get the spatial resolution r , substituting it into the formula Get the patient's lesion area A ; In this embodiment, it should be specifically explained that the convolutional neural network segmentation algorithm can refer to the following references: Research on image segmentation algorithm based on convolutional neural network, Lesion area segmentation technology based on medical image, Research on dermoscopic image lesion segmentation based on deep convolutional neural network, Research on image segmentation method based on total variation spectral transform and deep convolutional neural network.

[0025] The second step is to install a spectrum acquisition device on the surgical video recording device. During the operation, the spectrum data of the surgical area is collected at a time interval of once per second, and the collected spectrum data is marked as , where λ represents the wavelength, t Indicates the time, a pure blood sample is extracted from the management database, placed in a cuvette, and spectral measurement is performed under the same equipment and conditions as the spectrum acquisition in the surgical area to obtain the spectral data of the blood sample , by analyzing the spectrum of the blood sample, determine the characteristic wavelength λ of the blood and substitute it into the formula Get the absorbance of blood ,in, Indicates the reference light intensity, that is, the light intensity measurement value at the same wavelength when there is no blood absorption; obtains the optical path length , according to the Lambert-Beer law formula Get the blood concentration at each characteristic wavelength λ and time point t ,in, represents the molar absorption coefficient, obtain the volume V of the surgical area, and substitute it into the formula Get the patient's surgical bleeding BL ; The third step is to read the patient's lesion area A , Bleeding BL and duration of surgery OT , and standardize them respectively to calculate the quantitative index of surgical risk ORQI , the calculation model is as follows: ,in, SEC represents the patient's physiological homeostasis assessment coefficient, S Indicates the area of ​​the patient's surgical area, BL 0 represents the maximum allowable bleeding volume preset for the corresponding surgery type. OT 0 means the preset maximum operation time for the corresponding operation type. Represent the weight factors of lesion area, bleeding volume and operation time, respectively, and .

[0026] S5: Evaluate the quantitative surgical risk index: Evaluate the quantitative surgical risk index by comparing it with the preset risk threshold, analyze the surgical risk status of the patient through AI assistance, and issue warning information for data with abnormal evaluation results; In this embodiment, it should be specifically explained that the specific method for evaluating the quantitative index of surgical risk is: The surgical risk quantification index is extracted and compared with the preset risk threshold. If the surgical risk quantification index is less than or equal to the preset risk threshold, the patient's surgery is judged to have no abnormal risk. If the surgical risk quantification index is greater than the preset risk threshold, the patient's surgical risk is judged to be abnormal. The abnormal data will be promptly warned to the doctor by means of alarm sounds and pop-up images.

[0027] S6: Establish an interactive user interface: Establish an interactive user interface for real-time display of surgical video images, patient physiological homeostasis, surgical risk assessment results and early warning information, and send data with abnormal judgment and assessment results to doctors and medical staff.

[0028] Example 2 See also Figure 2 As shown, the present invention provides an AI-assisted analysis system for surgical videos, including a surgical video acquisition and processing module, a surgical lighting condition AI-assisted analysis module, a surgical patient index AI-assisted analysis module, a surgical risk index AI-assisted analysis module, a surgical risk quantitative index evaluation module, and an interactive user interface establishment module; A surgical video acquisition and processing module: used for performing video monitoring of the surgical process, acquiring a first surgical video, and preprocessing the first surgical video to obtain a second surgical video; AI-assisted analysis module for surgical lighting conditions: This module uses image processing technology to extract surgical images from the second surgical video, analyzes the surgical light domain adaptation index, and uses AI to assist in analyzing surgical lighting conditions. AI-assisted analysis module for surgical patient indicators: used to extract the patient's physiological indicator data from the second surgical video that meets the surgical lighting conditions, analyze and obtain the physiological homeostasis assessment coefficient, analyze the patient's physiological homeostasis through AI assistance, and issue early warning information for abnormal results; Surgical risk index AI-assisted analysis module: used to extract the patient's lesion area, bleeding volume and operation time from the second surgical video that meets the surgical lighting conditions, and analyze and obtain the quantitative surgical risk index based on the physiological steady-state assessment coefficient; Module for evaluating the quantitative index of surgical risk: used to evaluate the quantitative index of surgical risk by comparing it with the preset risk threshold, analyze the surgical risk status of the patient through AI assistance, and issue early warning information for data with abnormal evaluation results; Establish an interactive user interface module: Establish an interactive user interface to display the surgical video screen, the patient's physiological steady-state condition, the surgical risk assessment results and early warning information in real time, and send the data with abnormal judgment and assessment results to doctors and medical staff.

[0029] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

[0030] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. An AI-assisted analysis method for surgical videos, characterized in that: The following steps are involved: S1: Obtaining and preprocessing a surgical video: performing video monitoring on the surgical process, obtaining a first surgical video, and preprocessing the first surgical video to obtain a second surgical video; S2: AI-assisted analysis of surgical lighting conditions: Use image processing technology to extract surgical images in the second surgical video, analyze the surgical light domain adaptation index, and use AI-assisted analysis of surgical lighting conditions. If the analysis result shows that the lighting conditions meet the requirements, further execute S3, otherwise jump out of the step loop; S3: AI-assisted analysis of surgical patient indicators: Extract the patient's physiological indicator data from the second surgical video that meets the surgical lighting conditions, analyze and obtain the physiological homeostasis assessment coefficient, use AI to assist in analyzing the patient's physiological homeostasis, and issue warning information for abnormal results; S4: AI-assisted analysis of surgical risk indicators: Extract the patient's lesion area, bleeding volume, and operation time from the second surgical video that meets the surgical lighting conditions, and analyze and obtain a quantitative surgical risk index based on the physiological homeostasis assessment coefficient; S5: Evaluate the quantitative surgical risk index: Evaluate the quantitative surgical risk index by comparing it with the preset risk threshold, analyze the surgical risk status of the patient through AI assistance, and issue warning information for data with abnormal evaluation results; S6: Establish an interactive user interface: Establish an interactive user interface for real-time display of surgical video images, patient physiological homeostasis, surgical risk assessment results and early warning information, and send data with abnormal judgment and assessment results to doctors and medical staff.

2. The AI-assisted analysis method for surgical videos according to claim 1, characterized in that: The specific method of obtaining the surgical video and performing preprocessing is: The first surgical video is preprocessed by adjusting the video screen color, brightness and contrast and removing noise to obtain a second surgical video.

3. The AI-assisted analysis method for surgical videos according to claim 1, characterized in that: The specific method for analyzing the surgical light field adaptation index is: S201: Obtaining the brightness correlation coefficient between the standard image and the image to be tested NCC , substituting it into the formula Get the surgical light field adaptation index ODAI ,in, represents the hyperbolic tangent function, k represents the correction factor; S202: Analyze surgical lighting conditions: extract the surgical light field adaptation index and compare it with the preset adaptation index threshold. If the surgical light field adaptation index is greater than or equal to the preset adaptation index threshold, it is determined that the surgery meets the standard lighting conditions, and further execute S3. If the surgical light field adaptation index is less than the preset adaptation index threshold, it is determined that the surgery does not meet the standard lighting conditions, and the step loop is jumped.

4. The AI-assisted analysis method for surgical videos according to claim 3, characterized in that: The brightness value correlation coefficient of the standard image and the image to be tested is obtained NCC The specific content is: The first step is to use a light meter to calibrate the surgical lamps one by one in a standard environment, adjust the height and angle of the surgical lamps, and use a high-definition camera to capture images from different surgical perspectives, including but not limited to the surgeon's perspective, assistant's perspective, and bird's-eye perspective, and mark them as standard images; The second step is to obtain the surgical video image and mark it as the image to be tested; pre-process the standard image and the image to be tested, wherein the sizes of the standard image and the image to be tested are ; Extract the brightness value of the pixel located at row x and column y in the standard image , substituting it into the formula Get the average brightness value of the standard image , extract the brightness value of the pixel located in the xth row and yth column in the image to be tested , substituting it into the formula Get the average brightness value of the image to be tested , where M represents the number of pixels in the height direction of the standard image and the surgical video image, i.e., the number of rows, and M represents the number of pixels in the width direction of the standard image and the surgical video image, i.e., the number of columns, x=1, 2, …, M, y=1, 2, …, N; The third step is to calculate the correlation coefficient of the brightness values ​​of the standard image and the image to be tested. NCC , the calculation model is: .

5. The AI-assisted analysis method for surgical videos according to claim 1, characterized in that: The specific method of AI-assisted analysis of surgical patient indicators is as follows: The first step is to locate the patient's body temperature detection area ROI through the facial feature recognition algorithm, and obtain the pixel value of the green light G channel in the xth row and yth column in the detection area. , substituting it into the formula Get the average intensity of the green channel pixels Ia ,in, It is represented by the total number of pixels in the ROI to be detected, and the average intensity of the green channel pixels is filtered and marked as If , using the peak detection algorithm to determine the heartbeat peak, when If If the heart rate exceeds the preset threshold, it is determined as a heartbeat peak and the time interval between adjacent peaks is recorded. , substituting it into the formula The patient's heart rate HR is obtained, where n represents the total number of heartbeat peaks in the area to be detected, and i represents the number of the heartbeat peak in the area to be detected; The second step is to obtain the elapsed time interval Heart rate change after , body temperature change and respiratory rate changes , calculate the heart rate change rate , body temperature change rate , respiratory rate change rate ; The third step is to calculate the physiological homeostasis assessment coefficient. The specific calculation model is: ,in, SEC represents the patient's physiological homeostasis assessment coefficient, R 0 means the preset maximum breathing rate change rate, e represents a natural constant; The fourth step is to compare the physiological homeostasis assessment coefficient with the preset steady-state assessment coefficient threshold to analyze the patient's physiological homeostasis condition. If the physiological homeostasis assessment coefficient is greater than or equal to the preset steady-state assessment coefficient threshold, the patient's physiological indicators are judged to be normal. If the physiological homeostasis assessment coefficient is less than the preset steady-state assessment coefficient threshold, the patient's physiological indicators are judged to be abnormal, and a warning signal is issued for the abnormal result and sent to the doctor.

6. The AI-assisted analysis method for surgical videos according to claim 1, characterized in that: The specific method of AI-assisted analysis of surgical risk indicators is: The first step is to select video frame images from the second surgical video, convert the selected video frame images from the RGB color space to the HSV color space, use Gaussian filtering or median filtering preprocessing methods to remove noise interference in the image, extract the patient's lesion area from the video frame image based on the convolutional neural network segmentation algorithm, and obtain the number of pixels in the lesion area pn , get the spatial resolution r , substituting it into the formula Get the patient's lesion area A ; The second step is to install a spectrum acquisition device on the surgical video recording device. During the operation, the spectrum data of the surgical area is collected at a time interval of once per second, and the collected spectrum data is marked as , where λ represents the wavelength, t Indicates the time, a pure blood sample is extracted from the management database, placed in a cuvette, and spectral measurement is performed under the same equipment and conditions as the spectrum acquisition in the surgical area to obtain the spectral data of the blood sample , by analyzing the spectrum of the blood sample, determine the characteristic wavelength λ of the blood and substitute it into the formula Obtain the absorbance of blood ,in, Indicates the reference light intensity, that is, the light intensity measurement value at the same wavelength when there is no blood absorption; obtains the optical path length , according to the Lambert-Beer law formula Get the blood concentration at each characteristic wavelength λ and time point t ,in, represents the molar absorption coefficient, obtain the volume V of the surgical area, and substitute it into the formula Get the patient's surgical bleeding BL ; The third step is to read the patient's lesion area A , Bleeding BL and duration of surgery OT , and standardize them respectively to calculate the quantitative index of surgical risk ORQI , the calculation model is as follows: ,in, SEC represents the patient's physiological homeostasis assessment coefficient, S Indicates the area of ​​the patient's surgical area, BL 0 represents the maximum allowable bleeding volume preset for the corresponding surgery type. OT 0 means the preset maximum operation time for the corresponding operation type. Represent the weight factors of lesion area, bleeding volume and operation time, respectively, and .

7. The AI-assisted analysis method for surgical videos according to claim 1, characterized in that: The specific method for evaluating the quantitative index of surgical risk is: The surgical risk quantification index is extracted and compared with the preset risk threshold. If the surgical risk quantification index is less than or equal to the preset risk threshold, the patient's surgery is judged to have no abnormal risk. If the surgical risk quantification index is greater than the preset risk threshold, the patient's surgical risk is judged to be abnormal. The abnormal data will be promptly warned to the doctor by means of alarm sounds and pop-up images.

8. An AI-assisted analysis system for surgical videos, used to implement the AI-assisted analysis method for surgical videos described in any one of claims 1 to 7, characterized in that: include: A surgical video acquisition and processing module: used for performing video monitoring of the surgical process, acquiring a first surgical video, and preprocessing the first surgical video to obtain a second surgical video; AI-assisted analysis module for surgical lighting conditions: This module uses image processing technology to extract surgical images from the second surgical video, analyzes the surgical light domain adaptation index, and uses AI to assist in analyzing surgical lighting conditions. AI-assisted analysis module for surgical patient indicators: used to extract the patient's physiological indicator data from the second surgical video that meets the surgical lighting conditions, analyze and obtain the physiological homeostasis assessment coefficient, analyze the patient's physiological homeostasis through AI assistance, and issue early warning information for abnormal results; Surgical risk index AI-assisted analysis module: used to extract the patient's lesion area, bleeding volume and operation time from the second surgical video that meets the surgical lighting conditions, and analyze and obtain the quantitative surgical risk index based on the physiological steady-state assessment coefficient; Module for evaluating the quantitative index of surgical risk: used to evaluate the quantitative index of surgical risk by comparing it with the preset risk threshold, analyze the surgical risk status of the patient through AI assistance, and issue early warning information for data with abnormal evaluation results; Establish an interactive user interface module: Establish an interactive user interface to display the surgical video screen, the patient's physiological steady-state condition, the surgical risk assessment results and early warning information in real time, and send the data with abnormal judgment and assessment results to doctors and medical staff.

Images