Image recognition system for resectoscope

Abstract

The invention provides an image recognition system for a resectoscope, and aims to improve the accuracy and safety of a prostate electric resection operation. The system comprises an image acquisition unit, a tissue characteristic module, a depth estimation module and an intelligent control module. The image acquisition unit acquires an operation video in real time, the tissue feature module analyzes image data, the spectral analysis unit extracts reflection spectral features of different tissues, and the tissue layering unit identifies a gland layer, an interstitial layer and an envelope layer of a prostate based on the spectral data. The network tracking unit identifies capillary distribution through spectral absorption characteristics of hemoglobin. And the depth estimation module performs depth measurement by using the layered identification result and the tissue deformation data, and realizes depth correction through compensation adjustment. And the intelligent control module integrates capillary distribution, corrected depth data and tissue stress information, generates a resection track of the electric resection ring through a position positioning unit and a track generation unit, and optimizes a path by using a stress analysis unit.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to an image recognition system for a resectoscope. Background Art

[0002] In the prior art, image-assisted systems for electrosurgical resection have been used in clinical operations, usually using image acquisition and processing modules to acquire and analyze real-time video images of the surgical area. These systems help surgeons identify different types of tissue structures through image segmentation and basic tissue analysis techniques, thereby achieving more accurate resection operations.

[0003] However, existing electroresectoscope-assisted systems still have significant limitations in terms of accuracy and safety. Most systems cannot effectively distinguish between different layers of prostate tissue and lack the ability to identify the glandular layer, interstitial layer, and capsule layer in detail. At the same time, the existing systems are not capable of compensating and correcting tissue deformation, vascular distribution, and depth measurement errors that occur during surgery, which can easily lead to accidental injury to microvessels and deviations in depth estimation. In addition, these systems usually lack real-time tissue stress analysis, making it difficult to prevent tissue tearing during surgical resection.

[0004] Therefore, it is necessary to develop an improved image recognition system for resectoscope. Summary of the invention

[0005] The present application provides an image recognition system for an electrosurgical resectoscope to improve the accuracy and safety of electrosurgical resection of the prostate.

[0006] The present application provides an image recognition system for an electroresectoscope, comprising:

[0007] An image acquisition unit, used for acquiring video images during the electrosurgical resection procedure in real time;

[0008] A tissue feature module is used to receive the video image from the image acquisition unit and perform feature analysis processing, and the tissue feature module includes a spectral analysis unit, a tissue stratification unit and a network tracking unit; the spectral analysis unit is used to extract the reflectance spectral features of different tissues from the video image; the tissue stratification unit is used to hierarchically divide the prostate tissue based on the spectral feature data provided by the spectral analysis unit, realize hierarchical identification of the glandular layer, the interstitial layer and the capsule layer, and obtain a hierarchical identification result; the network tracking unit is used to receive the spectral feature data output by the spectral analysis unit, identify the microvessel distribution by analyzing the spectral absorption characteristics of hemoglobin, and obtain microvessel distribution information;

[0009] A depth estimation module, used for performing depth estimation based on the layered identification result provided by the tissue feature module, the depth estimation module includes a deformation monitoring unit, an optical path depth unit and a compensation adjustment unit; the deformation monitoring unit is used to collect tissue deformation data during the electroresection process; the optical path depth unit is used to generate a depth measurement result of the prostate tissue based on the tissue deformation data provided by the deformation monitoring unit and the layered identification result provided by the tissue layering unit, using the ratio of the reflected light intensity at wavelengths of 470nm and 630nm; the compensation adjustment unit is used to perform real-time correction on the depth measurement result according to the received depth measurement result provided by the optical path depth unit and the tissue deformation data provided by the deformation monitoring unit, and generate corrected depth data;

[0010] An intelligent control module is used to integrate the data output by each module to control the electrosurgical resection operation process, and the intelligent control module includes a position positioning unit, a trajectory generating unit and a stress analyzing unit; the position positioning unit is used to identify the three-dimensional spatial relationship data of the electrosurgical resection ring relative to the prostate tissue and obtain the position information based on the microvessel distribution information provided by the network tracking unit and the corrected depth data provided by the compensation adjusting unit; the trajectory generating unit is used to generate a predicted resection trajectory of the electrosurgical resection ring based on the received position information provided by the position positioning unit and the corrected depth data provided by the compensation adjusting unit; the stress analyzing unit is used to analyze the tissue stress distribution during the resection process based on the predicted resection trajectory to prevent tissue tearing, and feed back the analysis results to the trajectory generating unit to optimize the resection path.

[0011] Furthermore, the spectrum analysis unit is specifically used for:

[0012] Receiving a video image acquired in real time, and dividing the video image into a plurality of regions; each region represents a tissue sample;

[0013] For each region, a specific spectral filtering process is applied to obtain the reflected light intensity information at different wavelengths; based on the reflected light intensity information, a reflection spectrum curve at different wavelengths is generated;

[0014] Normalize the acquired reflectance spectrum curve to remove noise and deviation caused by brightness and angle changes in the video image;

[0015] The normalized spectral curve is subjected to feature extraction to identify the characteristic wavelengths and corresponding reflection intensities of different tissues; the reflection characteristics at the characteristic wavelengths are analyzed to extract the spectral feature data of each tissue.

[0016] Furthermore, the tissue stratification unit is specifically used for:

[0017] receiving spectral characteristic data provided by the spectral analysis unit, and preliminarily identifying different tissue regions according to the reflection characteristics of each tissue type at a specific wavelength;

[0018] The initially identified tissue regions are further classified by comparing the spectral features of each region with the typical spectral patterns in the pre-stored tissue feature database to classify each region as the glandular layer, interstitial layer or capsule layer;

[0019] Generate and output hierarchical recognition results, and mark the hierarchical attributes of each area to form a complete organizational hierarchical diagram.

[0020] Furthermore, the network tracking unit is specifically used for:

[0021] receiving the spectral characteristic data output by the spectral analysis unit, and screening out the characteristic wavelength range related to the absorption characteristics of hemoglobin;

[0022] Conduct in-depth analysis of the filtered spectral data to identify regions with significant absorption characteristics within the characteristic wavelength range of hemoglobin; generate a preliminary microvascular distribution map by aggregating regions with significant absorption characteristics;

[0023] The preliminary microvessel distribution map is further processed to optimize and confirm the specific location of the microvessels and generate the final microvessel distribution information.

[0024] Furthermore, the deformation monitoring unit is specifically used for:

[0025] Real-time capture of prostate tissue images during electroresection and marking of multiple key reference points;

[0026] Analyze the position changes of key reference points during the electrosurgical cutting process, calculate the displacement and relative deformation degree of each reference point; generate preliminary deformation data by comparing the position differences of the reference points in the current image and the reference image;

[0027] The displacement data of each reference point is integrated to form a complete tissue deformation map, and the deformation intensity and distribution of the tissue at different positions are recorded.

[0028] Furthermore, the optical path depth unit calculates the depth D of the prostate tissue using the following formula (1):

[0029]

[0030] Where I1 is the reflected light intensity at a wavelength of 470 nm; I2 is the reflected light intensity at a wavelength of 630 nm; R1 and R2 represent the standard reflectances of the reference prostate tissue at wavelengths of 470 nm and 630 nm, respectively; K is a correction coefficient related to the prostate tissue type; f(λ1, λ2, t) is a dual-wavelength optical path difference correction function, which is used to correct the deviation caused by the optical path difference of different wavelengths of light in the tissue and temperature changes; C(n) is an interlayer interface correction function; wherein the dual-wavelength optical path difference correction function f(λ1, λ2, t) adopts the following formula (2):

[0031]

[0032] Where t represents the real-time temperature of the prostate tissue; α and β are experimental calibration coefficients; μ1 and μ2 are tissue scattering coefficients corresponding to wavelengths of 470nm and 630nm, respectively;

[0033] The interlayer interface correction function C(n) adopts the following formula (3):

[0034]

[0035] Wherein, n is the number of tissue layers, including glandular layer, interstitial layer and capsule layer; γ i is the weight coefficient of the i-th layer; Δn i is the refractive index difference between the i-th layer and the adjacent layer; d i represents the distance from the current measurement position to the interface of the i-th layer; σ i is the attenuation factor, which is determined by the absorption characteristics of the i-th layer of tissue.

[0036] Furthermore, the compensation adjustment unit uses the correction formula provided by the following formula (4) for correction:

[0037]

[0038] Wherein, D' is the corrected depth value, which is the depth value finally output to the intelligent control module; D0 is the initial depth value, which is provided by the optical path depth unit; is the depth gradient correction term, where is the depth gradient value, which indicates the rate of change of the depth value around the measurement point; k1 is the correction coefficient related to tissue characteristics; k2·κ is the depth curvature correction term, where κ is the depth curvature, which indicates the second-order derivative of the depth value with the spatial position, reflecting the curvature characteristics of the depth distribution; k2 is the curvature correction coefficient, which is used to adjust the influence of the depth curvature in the correction process; d max is the maximum deformation provided by the deformation monitoring unit, which is the deformation value of the position with the largest deformation in the resection area; T1 is the deformation threshold; f(d max) is the maximum deformation compensation function, which is implemented using the following formula (5):

[0039]

[0040] Among them, γ1 is the compensation gain coefficient; α1 is the nonlinear index used to adjust the growth rate of the compensation function.

[0041] Furthermore, the stress analysis unit is specifically used for:

[0042] The discrete tangent point coordinate sequence P (x, y, z) on the predicted resection trajectory output by the acquisition trajectory generation unit;

[0043] For each discrete cutting point, obtain the tissue elastic coefficient E(θ) and Poisson's ratio v(θ) at different angles in the circumferential direction, as well as the tissue thickness h and the resection radius r of the electric resection ring at the corresponding position;

[0044] According to the following formulas (6) and (7), the stress distribution model at the tangent point is established to calculate the radial stress σ r and shear stress τ rθ :

[0045]

[0046] Among them, F r and F θ are the radial force and circumferential force during the resection process; K1 and K2 are correction coefficients related to tissue type; E(θ) and v(θ) are the elastic coefficient and Poisson's ratio corresponding to each angle, respectively;

[0047] Obtain the ultimate tensile strength σ of the target tissue c and the ultimate urgent strength τ c , and based on the Mohr strength theory, it is determined whether there is a tearing risk. When any of the conditions provided by the following formulas (8) to (10) are met, a trajectory correction signal is generated:

[0048] σ r >0.8σ c (8)

[0049] |τ rθ |>0.8τ c (9)

[0050] (σ r -τ rθ )>0.9σ c (10)

[0051] Among them, the speed trajectory correction signal includes the azimuth information of the high stress area and the recommended resection speed adjustment coefficient to adjust the movement speed and resection path of the electrosurgical resection ring to avoid the risk of tissue tearing.

[0052] Furthermore, the position positioning unit is specifically used for:

[0053] constructing a microvessel distribution map according to the microvessel distribution information received from the network tracking unit;

[0054] Receiving the corrected depth data provided by the compensation adjustment unit, and combining it with the microvascular distribution map to construct a depth distribution model of the prostate tissue;

[0055] Based on the depth distribution model, the optimal position of the electrosurgical ring in the prostate tissue is calculated;

[0056] Generating three-dimensional spatial relationship data of the electric cutting ring relative to the prostate tissue according to the calculated position of the electric cutting ring, wherein the three-dimensional spatial relationship data includes the distance between the electric cutting ring and the microvessels and the relative position in the depth direction;

[0057] The obtained three-dimensional spatial relationship data is output and provided to the resection trajectory generation unit as position information.

[0058] Furthermore, the trajectory generating unit is specifically used for:

[0059] Receive the position information provided by the position positioning unit, and analyze the spatial position relationship of the electrosurgical resection ring relative to the prostate tissue, including the distance information from the electrosurgical resection ring to the microvessels and key tissue structures, and the relative position data in different depth directions;

[0060] Receiving the corrected depth data provided by the compensation adjustment unit to obtain the accurate position of the prostate tissue at different cutting depths;

[0061] Determine the preliminary resection path of the electrosurgical resection loop based on the position information and the corrected depth data;

[0062] Optimize the initial resection path and dynamically adjust the movement direction and speed of the electrosurgical resection loop;

[0063] According to the optimized path, the final predicted resection trajectory is generated and used as the operation instruction of the electrosurgical resection loop.

[0064] This application has the following beneficial technical effects:

[0065] (1) The system uses the tissue stratification unit to identify the prostate tissue in layers based on the spectral feature data provided by the spectral analysis unit, which can effectively distinguish the glandular layer, interstitial layer and capsule layer, and provide more detailed layer information. This precise layered identification improves the targetedness of surgical resection and avoids accidental injury to non-target tissues. (2) The system analyzes the spectral absorption characteristics of hemoglobin through the network tracking unit and identifies the distribution of microvessels in real time. This function helps avoid microvessels in resection path planning, reduces the risk of damage to microvessels, and significantly improves the safety of surgery. (3) The system uses the network tracking unit to analyze the spectral absorption characteristics of hemoglobin and identifies the distribution of microvessels in real time. This function helps avoid microvessels in resection path planning, reduces the risk of damage to microvessels, and significantly improves the safety of surgery. (4) The intelligent control module combines the position positioning unit, trajectory generation unit and stress analysis unit to generate a resection path based on real-time data and monitor tissue stress to prevent tissue tearing during surgery. Path optimization and stress feedback further ensure the accuracy and stability of the surgery and effectively reduce the possibility of tissue damage during resection. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 It is a schematic diagram of an image recognition system for an electroresectoscope provided in the first embodiment of the present application. DETAILED DESCRIPTION

[0067] Many specific details are described in the following description to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present application, so the present application is not limited by the specific implementation disclosed below.

[0068] The first embodiment of the present application provides an image recognition system for a resectoscope. Figure 1 , which is a schematic diagram of the first embodiment of the present application. Figure 1 The first embodiment of the present application provides an image recognition system for a resectoscope and is described in detail.

[0069] The image recognition system for a resectoscope includes an image acquisition unit 101 , a tissue feature module 102 , a depth estimation module 103 and an intelligent control module 104 .

[0070] The image acquisition unit 101 is used to acquire video images during the electrosurgical resection procedure in real time.

[0071] In the image recognition system provided in this embodiment, the image acquisition unit 101 is designed as a key component that can capture and transmit video images during electrosurgical resection in real time, ensuring that the system can continuously and efficiently collect the required image data during the operation. The unit 101 is connected to a high-resolution camera device to capture the real-time video stream of the surgical site and convert it into a digital signal for subsequent processing. The camera device can obtain the subtle structural details of the tissue through an optical sensor to accurately reflect the actual state of the surgical area. The image acquisition unit can not only capture the brightness, contrast and color information of the image, but also provide clear image data under complex lighting conditions to ensure the reliability and accuracy of the image information.

[0072] In order to adapt to the rapid movements that may occur during the electrosurgical resection, the image acquisition unit 101 usually has a high frame rate, at least 30 frames per second, to avoid image blur caused by the rapid movement of surgical instruments. This can ensure that the image information at each moment is accurate, which is convenient for the subsequent modules to analyze the tissue in real time. In addition, the image acquisition unit can also integrate spectral filtering functions. Through filters of multiple specific wavelengths, the system can focus on the reflection of light at specific wavelengths, especially the spectral response of prostate tissue at different wavelengths. The capture of these spectral information lays the foundation for the subsequent spectral analysis unit to extract the spectral characteristics of tissue reflection.

[0073] The connection between the image acquisition unit 101 and the subsequent modules is usually carried out through a high-speed data bus to ensure that the collected video images can be quickly transmitted to the tissue feature module 102 for processing. The transmission process can use USB3.0, HDMI or other high-bandwidth interfaces to reduce data loss or processing delays caused by transmission delays. In addition, the image acquisition unit 101 can also integrate automatic white balance and automatic exposure control functions to adapt to the brightness changes of different tissue structures, thereby ensuring the stability of image quality. In order to enhance the convenience of operation during surgery, the unit also supports autofocus function to ensure that clear images of tissues can be captured at different focal lengths.

[0074] Through the above characteristics, the image acquisition unit 101 can provide high-quality, low-latency real-time video image data for the entire electroresectoscope image recognition system, so that subsequent tissue feature analysis, depth estimation and intelligent control modules can perform complex recognition, analysis and control operations based on accurate image information, thereby ensuring the accuracy and safety of the surgical process.

[0075] The tissue feature module 102 is used to receive the video image from the image acquisition unit and perform feature analysis processing. The tissue feature module includes a spectral analysis unit, a tissue stratification unit and a network tracking unit; the spectral analysis unit is used to extract the reflectance spectral characteristics of different tissues from the video image; the tissue stratification unit is used to hierarchically divide the prostate tissue based on the spectral feature data provided by the spectral analysis unit, realize hierarchical identification of the glandular layer, the interstitial layer and the capsule layer, and obtain a hierarchical identification result; the network tracking unit is used to receive the spectral feature data output by the spectral analysis unit, identify the microvessel distribution by analyzing the spectral absorption characteristics of hemoglobin, and obtain microvessel distribution information.

[0076] In the image recognition system for resectoscope provided in this embodiment, the tissue feature module 102 is a key component for receiving the video image from the image acquisition unit 101 and performing deep feature analysis and processing, ensuring that various features of the prostate tissue can be accurately identified and processed. The tissue feature module 102 includes three main units: a spectral analysis unit, a tissue stratification unit, and a network tracking unit, each of which has a specific function in the system to achieve detailed analysis and accurate identification of the tissue structure.

[0077] The spectral analysis unit extracts the reflective spectral characteristics of different tissues from the video images received by the image acquisition unit. Specifically, the spectral analysis unit is able to decompose the video images into spectral data of different wavelengths to identify the light reflection intensity characteristics of different tissues at specific wavelengths. The unit separates the image into layers of multiple wavelength ranges by applying spectral filtering technology to obtain the unique reflection characteristics of each tissue at different wavelengths. For example, the spectral analysis unit can analyze the reflection intensity of wavelengths of 470nm and 630nm, and these wavelength data provide a basis for subsequent tissue identification. By identifying the specific spectral reflection patterns of each tissue, the spectral analysis unit can help distinguish different tissue types, laying the foundation for the system's stratification and tracking functions.

[0078] The tissue stratification unit divides the prostate tissue into layers based on the spectral feature data provided by the spectral analysis unit, thereby realizing the hierarchical identification of the glandular layer, interstitial layer and capsule layer. The hierarchical identification process determines the optical characteristics of different layers in the prostate tissue through a comprehensive analysis of the spectral feature data. The tissue stratification unit will first receive the spectral feature data output by the spectral analysis unit, and compare it with the typical spectral features of each layer of prostate tissue preset in the system, so as to identify the specific location of the glandular layer, interstitial layer and capsule layer. Each layer of tissue has a different spectral response. The tissue stratification unit can automatically identify and mark the position and boundary of each layer based on these spectral differences, and finally form a hierarchical identification result, which is output to the subsequent depth estimation module for further processing.

[0079] The network tracking unit is used to identify and mark the microvessel distribution information in the prostate tissue, providing a reference for path planning and avoidance strategies during surgery. The unit identifies the distribution of microvessels by receiving the spectral feature data provided by the spectral analysis unit, especially the spectral absorption characteristics of hemoglobin. The network tracking unit compares the spectral feature data with the typical absorption pattern of hemoglobin and marks the areas in the image that meet the hemoglobin absorption characteristics. These areas are considered to be areas that may contain microvessels, and the network tracking unit generates microvessel distribution information by analyzing the distribution of these areas. This information not only helps the system avoid microvessels in resection path planning, thereby reducing the risk of bleeding, but also provides an important reference for the subsequent intelligent control module to ensure that the resection path is carried out within a safe range.

[0080] The tissue feature module 102 can accurately analyze and identify the spectral characteristics, layered structure and microvascular distribution of prostate tissue through the synergistic effect of the spectral analysis unit, tissue stratification unit and network tracking unit, thereby providing reliable basic data for the depth estimation and intelligent control of the system. The design of this module ensures that during the electrosurgical resection, various tissue feature information can be extracted and processed in real time and accurately to improve the accuracy and safety of the operation.

[0081] Furthermore, the spectrum analysis unit is specifically used for:

[0082] Receiving a video image acquired in real time, and dividing the video image into a plurality of regions; each region represents a tissue sample;

[0083] For each region, a specific spectral filtering process is applied to obtain the reflected light intensity information at different wavelengths; based on the reflected light intensity information, a reflection spectrum curve at different wavelengths is generated;

[0084] Normalize the acquired reflectance spectrum curve to remove noise and deviation caused by brightness and angle changes in the video image;

[0085] The normalized spectral curve is subjected to feature extraction to identify the characteristic wavelengths and corresponding reflection intensities of different tissues; the reflection characteristics at the characteristic wavelengths are analyzed to extract the spectral feature data of each tissue.

[0086] The spectral analysis unit provided in this embodiment is used to accurately extract spectral feature data of different tissues from real-time electrosurgical surgery video images, providing key support for subsequent tissue stratification and microvascular distribution analysis. The working process of this unit includes steps such as image segmentation, spectral data extraction, normalization processing and feature extraction, and each step is interrelated to ensure the scientificity and accuracy of the entire process.

[0087] The spectral analysis unit first receives the surgical video images transmitted in real time from the image acquisition unit, and divides these video images into multiple small areas, each of which corresponds to a tissue sample. This segmentation process is based on a preset image segmentation algorithm, which can divide the entire surgical field into several uniform and relatively independent small blocks, thereby reducing background interference in spectral analysis. The pixel data in each segmented area will be regarded as an independent analysis unit, providing clear tissue positioning for subsequent spectral processing.

[0088] For each segmented area, the spectral analysis unit uses a specific spectral filtering technology to process the video data in the area and extract the reflected light intensity information at different wavelengths. This process separates the spectral data into several different wavelength bands by using narrow-band filters or spectrometers, each of which represents the reflective characteristics of the tissue in the area at a specific wavelength. By scanning key wavelength ranges such as 470nm and 630nm, the main optical properties of prostate tissue can be obtained. The output of this process is the reflected light intensity data of each segmented area at different wavelengths, which is used to further generate a reflective spectrum curve.

[0089] The generated reflectance spectral curves may be affected by changes in ambient light or camera angle during video acquisition, so the spectral analysis unit normalizes these curves to remove unnecessary noise and deviations. The normalization process uses the brightness information in the image and the reference reflectance data to adjust the spectral data of each area to a uniform intensity range, ensuring that the spectral characteristics reflect the true physical properties of the tissue rather than interference caused by the environment or equipment.

[0090] The normalized spectral curve is further used for feature extraction. The spectral analysis unit identifies the characteristic wavelengths of different tissues and their corresponding reflection intensities by analyzing the curves. The characteristic wavelengths are key data points that reflect tissue characteristics. These data can be compared with a preset standard spectral database to accurately identify the unique spectral characteristics of each tissue. For example, the glandular layer may show a higher reflection intensity within a specific wavelength range, while the capsule layer exhibits a different spectral distribution pattern.

[0091] The spectrum analysis unit finally extracts the spectrum characteristic data of each tissue based on the reflection characteristics at the characteristic wavelength. These data contain the summary of the spectrum characteristics of each segmented area, such as the reflection intensity at a specific wavelength, the shape characteristics of the reflection curve, etc., providing accurate spectrum analysis results for the subsequent tissue stratification unit and network tracking unit.

[0092] Through the above series of processing, the spectral analysis unit can extract the spectral characteristics of different tissues in electrocautery surgery in real time and efficiently, providing a solid data foundation for the overall accuracy and safety of the system.

[0093] Furthermore, the tissue stratification unit is specifically used for:

[0094] receiving spectral characteristic data provided by the spectral analysis unit, and preliminarily identifying different tissue regions according to the reflection characteristics of each tissue type at a specific wavelength;

[0095] The initially identified tissue regions are further classified by comparing the spectral features of each region with the typical spectral patterns in the pre-stored tissue feature database to classify each region as the glandular layer, interstitial layer or capsule layer;

[0096] Generate and output hierarchical recognition results, and mark the hierarchical attributes of each area to form a complete organizational hierarchical diagram.

[0097] The tissue stratification unit provided in this embodiment is specially designed for accurate stratification and identification of prostate tissue during electroresection surgery. It can divide the tissue in the surgical area into layers based on the spectral feature data provided by the spectral analysis unit and generate detailed stratification results.

[0098] The tissue stratification unit first receives the spectral feature data from the spectral analysis unit. The spectral feature data contains the reflection intensity information of each tissue area at different wavelengths, which can directly reflect the optical properties of the tissue. By analyzing the reflection intensity of specific wavelengths (such as 470nm and 630nm), the unit can preliminarily identify different tissue areas in the surgical field of view. This preliminary identification process is based on the typical reflection characteristics exhibited by each tissue at a specific wavelength. For example, the glandular layer usually has a higher reflectivity, while the interstitial layer and the capsule layer may exhibit different spectral characteristics. By analyzing these spectral data, the system can preliminarily divide the tissues in the surgical area.

[0099] After the initial division is completed, the tissue stratification unit further classifies each area. This classification process is completed by comparing the spectral characteristics of each area with the corresponding standard spectral patterns in the pre-stored tissue feature database. The database contains typical spectral data of the prostate glandular layer, interstitial layer, and capsule layer, covering the reflection intensity and distribution characteristics of these tissues at various specific wavelengths. By matching the spectral characteristics of each area with the data in the database one by one, the system can accurately classify each area as a glandular layer, interstitial layer, or capsule layer. This comparison method is based on verified standard feature patterns and can significantly improve the accuracy of classification and ensure the reliability of stratification results.

[0100] After the classification is completed, the tissue stratification unit generates the final stratification recognition result. The stratification result includes the hierarchical attributes of each area, such as indicating whether a certain area is a glandular layer, an interstitial layer, or a capsule layer. These stratification attributes are further integrated into a complete tissue stratification map, which clearly shows the distribution of each tissue layer in the entire surgical field of view. The tissue stratification map can not only provide a reliable basis for the resection path planning in electrosurgical resection, but also help the surgeon intuitively understand the tissue composition of the surgical area, thereby improving the accuracy and safety of the operation.

[0101] Through this implementation, the tissue stratification unit can complete the stratification identification task of the surgical area with high accuracy and efficiency, and the formed tissue stratification map provides important reference data for other modules of the system.

[0102] Furthermore, the network tracking unit is specifically used for:

[0103] receiving the spectral characteristic data output by the spectral analysis unit, and screening out the characteristic wavelength range related to the absorption characteristics of hemoglobin;

[0104] Conduct in-depth analysis of the filtered spectral data to identify regions with significant absorption characteristics within the characteristic wavelength range of hemoglobin; generate a preliminary microvascular distribution map by aggregating regions with significant absorption characteristics;

[0105] The preliminary microvessel distribution map is further processed to optimize and confirm the specific location of the microvessels and generate the final microvessel distribution information.

[0106] The network tracking unit provided in this embodiment is used to accurately identify the microvascular distribution information in the prostate tissue during surgery, ensuring that the microvascular area is effectively avoided during the electroresection operation, thereby reducing the risk of bleeding and improving the safety and accuracy of the surgery. The unit performs multi-level processing on the data provided by the spectral analysis unit to ultimately generate detailed microvascular distribution information.

[0107] After receiving the spectral characteristic data output by the spectral analysis unit, the network tracking unit first screens the data to extract the characteristic wavelength range related to the absorption characteristics of hemoglobin. This process uses the significant spectral absorption characteristics of hemoglobin in a specific wavelength range (such as 470nm and 630nm) to screen out key data points that can reflect the presence of blood. These characteristic wavelength ranges are used for subsequent analysis to ensure that the processing focuses on areas related to microvascular distribution, thereby excluding spectral information unrelated to other tissues.

[0108] After screening out the characteristic wavelength range, the network tracking unit conducts an in-depth analysis of the screened spectral data to identify areas that show significant absorption characteristics within these wavelength ranges. These areas usually correspond to blood-containing areas in the tissue, especially where microvessels are densely distributed. The analysis process gradually determines which areas have high absorption characteristics associated with blood by comparing the absorption intensity of each tissue area at the characteristic wavelength. Each area with significant absorption characteristics will be marked, laying the foundation for generating a preliminary microvascular distribution map.

[0109] The network tracking unit then generates a preliminary microvessel distribution map by aggregating all regions with significant absorption characteristics. This distribution map shows all possible locations of microvessels in the surgical field. The aggregation process ensures that the distribution map accurately reflects the actual location of microvessels by merging adjacent regions and excluding isolated non-relevant regions, while avoiding interference from noise and misjudgment.

[0110] After generating the preliminary distribution map, the network tracking unit further processes it to optimize and confirm the specific location of the microvessels. This optimization process combines further analysis of the spectral feature data with the spatial relationship of the tissue structure, and ultimately generates detailed microvascular distribution information by adjusting boundaries, eliminating artifacts, and enhancing microvascular coherence. This distribution information includes not only the specific location of the microvessels, but may also include important features such as their direction and density distribution, providing key support for subsequent resection path planning and safe avoidance.

[0111] Through this layer-by-layer processing, the network tracking unit can complete the task of identifying microvascular distribution with high accuracy and high reliability.

[0112] The depth estimation module 103 is used to perform depth estimation based on the layered identification result provided by the tissue feature module, and the depth estimation module includes a deformation monitoring unit, an optical path depth unit and a compensation adjustment unit; the deformation monitoring unit is used to collect tissue deformation data during the electroresection process; the optical path depth unit is used to generate a depth measurement result of the prostate tissue based on the tissue deformation data provided by the deformation monitoring unit and the layered identification result provided by the tissue layering unit, using the ratio of the reflected light intensities at wavelengths of 470nm and 630nm; the compensation adjustment unit is used to perform real-time correction on the depth measurement result according to the received depth measurement result provided by the optical path depth unit and the tissue deformation data provided by the deformation monitoring unit, so as to generate corrected depth data.

[0113] In the image recognition system for resectoscope provided in this embodiment, the depth estimation module 103 is used to accurately measure and correct the depth information of the prostate tissue during the operation to support accurate resection operation. The depth estimation module 103 is based on the hierarchical recognition result provided by the tissue feature module 102, and realizes high-precision depth estimation and dynamic correction by integrating the deformation monitoring unit, the optical path depth unit and the compensation adjustment unit.

[0114] The deformation monitoring unit is responsible for collecting tissue deformation data during the electrosurgical resection process in real time to ensure that any deformation that occurs during tissue force or resection can be detected in a timely manner. Since the physical structure of the tissue may deform during electrosurgical resection, these deformations will affect the accuracy of depth measurement. Therefore, the unit uses multi-point tracking technology to capture tissue deformation data by analyzing the displacement of key reference points or marker points in the tissue. By comparing the currently collected deformation data with the initial morphological data, the deformation monitoring unit can generate accurate deformation information, which will serve as important input data for depth measurement and correction, ensuring that the system can adapt to the changing state of the tissue during surgery.

[0115] The optical path depth unit is used to perform preliminary depth measurement based on the tissue deformation data provided by the deformation monitoring unit and the layered identification results provided by the tissue feature module. In order to improve the accuracy of depth measurement, the optical path depth unit uses the ratio of reflected light intensity at a specific wavelength to infer depth information, especially by analyzing the difference in reflection intensity of light at 470nm and 630nm wavelengths in the tissue to generate depth measurement results of the prostate tissue. After receiving the layered identification data of the tissue, the optical path depth unit can accurately identify the depth changes of different tissue layers, which is particularly important for layered resection of prostate tissue. By combining the intensity of reflected light of different wavelengths with the layered information of the tissue, the optical path depth unit provides a basic depth measurement value for subsequent compensation adjustments.

[0116] The compensation adjustment unit is responsible for further real-time correction of the depth measurement results of the optical path depth unit to ensure the accuracy of the depth data. Due to tissue deformation and possible measurement errors during surgery, the compensation adjustment unit will use the preliminary depth results generated by the optical path depth unit and the real-time deformation data provided by the deformation monitoring unit to correct the depth results. The compensation adjustment unit adapts different correction modes according to different deformation conditions to ensure that the measured depth data can adapt to the dynamic changes of the tissue in real time. This correction mechanism can accurately adjust the measurement results to generate corrected depth data, provide reliable depth information for the intelligent control module 104, and support further resection trajectory planning and surgical path optimization.

[0117] Through the synergistic effect of the three steps of deformation monitoring, preliminary depth measurement, and compensation correction in the depth estimation module 103, the system can always provide accurate tissue depth data during the operation, adapt to tissue deformation, and dynamically correct errors. This module design ensures the accuracy of the electrosurgical process and provides an accurate depth information basis for the subsequent operation of the intelligent control module, thereby greatly improving the safety and effectiveness of the operation.

[0118] Furthermore, the deformation monitoring unit is specifically used for:

[0119] Real-time capture of prostate tissue images during electroresection and marking of multiple key reference points;

[0120] Analyze the position changes of key reference points during the electrosurgical cutting process, calculate the displacement and relative deformation degree of each reference point; generate preliminary deformation data by comparing the position differences of the reference points in the current image and the reference image;

[0121] The displacement data of each reference point is integrated to form a complete tissue deformation map, and the deformation intensity and distribution of the tissue at different positions are recorded.

[0122] The deformation monitoring unit provided in this embodiment is intended to monitor the deformation of prostate tissue in real time during electrosurgical resection to ensure accurate assessment and adjustment of dynamically changing tissue states during the surgical operation. By continuously capturing and analyzing tissue images, the deformation monitoring unit can generate comprehensive tissue deformation data, providing key support for the system's depth estimation and path optimization.

[0123] The deformation monitoring unit first captures images of prostate tissue during the electroresection process in real time. These images are acquired through high-resolution cameras and can clearly present the surface structure and local features of the tissue. In each frame, the system automatically marks multiple key reference points, which are usually selected at the boundaries, junctions or locations with significant features of the tissue in order to track its changes during the electroresection process. The selection criteria for reference points are based on their stability and repeatability, ensuring that the system can accurately re-identify these points in consecutive images.

[0124] After acquiring the real-time image, the deformation monitoring unit analyzes the position change of each key reference point and calculates its displacement and relative deformation degree. This analysis measures the movement distance and direction in space by comparing the coordinates of the reference point in the current frame image with the initial coordinates of the reference point in the baseline image. The calculation of the deformation degree is based on the displacement relationship between the reference point and the surrounding area, which can accurately reflect the dynamic changes of local tissues. This process not only captures the changes of a single point, but also takes into account the overall change trend between reference points, laying the foundation for the subsequent generation of deformation maps.

[0125] Subsequently, the deformation monitoring unit generates a complete tissue deformation map by integrating the displacement data of all key reference points. This deformation map visually displays the deformation intensity and distribution of the tissue at different locations, clearly marking the areas with greater deformation and relatively stable areas. The deformation map also includes the dynamic change trend of each area, providing reference data support for other modules of the system. For example, in the optimization of the electrosurgical resection path, the high deformation area information in the deformation map can be used to adjust the resection path to avoid depth errors or damage risks caused by tissue deformation.

[0126] Through real-time image capture, key reference point analysis and deformation map generation, the deformation monitoring unit can accurately monitor the dynamic changes of tissues during surgery.

[0127] Furthermore, the optical path depth unit calculates the depth D of the prostate tissue using the following formula (1):

[0128]

[0129] Wherein, I1 is the reflected light intensity of wavelength 470nm; I2 is the reflected light intensity of wavelength 630nm; R1 and R2 represent the standard reflectance of wavelength 470nm and 630nm on the reference prostate tissue respectively; K is the correction coefficient related to the prostate tissue type; f(λ1,λ2,t) is the dual-wavelength optical path difference correction function, which is used to correct the deviation caused by the optical path difference of different wavelengths of light in the tissue and temperature change; C(n) is the interlayer interface correction function;

[0130] In formula 1, I1 represents the reflected light intensity at a wavelength of 470nm, and I2 represents the reflected light intensity at a wavelength of 630nm. These values ​​are obtained by collecting light intensity information when the light beam analysis unit collects real-time video images, reflecting the reflective characteristics of tissues at different wavelengths. For example, if the reflected intensity I1 at 470nm is measured to be 0.6 (normalized unit), and the reflected intensity I2 at 630nm is measured to be 0.4, then I1 / I2=1.5.

[0131] R1 and R2 are the standard reflectances of reference prostate tissue at wavelengths of 470nm and 630nm, respectively. These reference values ​​are obtained through experimental calibration and are usually stored in the database of the device. For example, experimental calibration shows that the standard reflectance R1 at 470nm is 0.7, and the standard reflectance R2 at 630nm is 0.5. These values ​​are used as benchmarks in measurements.

[0132] K is a correction factor related to the prostate tissue type, reflecting the effect of the physical and chemical properties of the tissue on light propagation. The value of K is experimentally calibrated and adjusted according to the tissue type. For example, K = 1.2 may be used for the glandular layer, while K = 1.0 may be used for the interstitial layer.

[0133] The dual-wavelength optical path difference correction function f(λ1,λ2,t) adopts the following formula (2):

[0134]

[0135] Where t represents the real-time temperature of the prostate tissue; α and β are experimental calibration coefficients;

[0136] μ1 and μ2 are the tissue scattering coefficients at wavelengths of 470 nm and 630 nm, respectively. These values ​​are obtained experimentally. For example, if μ1 = 0.8 and μ2 = 0.6, then

[0137] The interlayer interface correction function C(n) adopts the following formula (3):

[0138]

[0139] Where n is the number of layers of the tissue, including the glandular layer, the interstitial layer and the capsule layer; for example, if there are 3 layers, then n=3.

[0140] γ i is the weight coefficient of the i-th layer, indicating the importance of the layer on light propagation, and can be set experimentally, for example, γ1 of the glandular layer is 0.8.

[0141] Δn i is the refractive index difference between the i-th layer and the adjacent layer; measured by optical experiments, for example, Δn1=0.1 represents the refractive index difference between the glandular layer and the interstitial layer.

[0142] d i Indicates the distance from the current measurement position to the interface of the i-th layer; determined by image layering and depth calculation, for example, d1 = 0.5 mm.

[0143] σ i is the attenuation factor, which is determined by the absorption characteristics of the i-th layer of tissue. For example, σ1 of the glandular layer is 0.3.

[0144] Furthermore, the compensation adjustment unit uses the correction formula provided by the following formula (4) for correction:

[0145]

[0146] Wherein, D' is the corrected depth value, which is the depth value finally output to the intelligent control module; D0 is the initial depth value, which is provided by the optical path depth unit; is the depth gradient correction term, where is the depth gradient value, which indicates the rate of change of the depth value around the measurement point; k1 is the correction coefficient related to tissue characteristics; k2·κ is the depth curvature correction term, where κ is the depth curvature, which indicates the second-order derivative of the depth value with the spatial position, reflecting the curvature characteristics of the depth distribution; k2 is the curvature correction coefficient, which is used to adjust the influence of the depth curvature in the correction process; d max The maximum deformation provided by the deformation monitoring unit is the deformation value of the position with the largest deformation in the resection area; T1 is the deformation threshold;

[0147] D0 is the initial depth value provided by the optical path depth unit, calculated by the ratio of the reflected light intensity at wavelengths of 470nm and 630nm. This value is the original depth measurement result without any deformation or gradient correction. For example, in one measurement, D0 is 2.5mm.

[0148] is the depth gradient value, which indicates the rate of change of depth around the measurement point, and is used to reflect the local change of depth in space. The gradient is obtained by performing a derivative calculation on the depth values ​​of the surrounding measurement points. For example, at a certain measurement point, the depth value changes from 2.5mm to 2.8mm, then the gradient where Δx is the distance between the measurement points.

[0149] k1 is a correction factor related to tissue characteristics, which is used to adjust the influence of depth gradient on the correction result. Its value is calibrated by actual risk. For example, k1 = 0.8 can be taken for the glandular layer, and k1 = 0.6 may be taken for the capsule layer.

[0150] κ is the depth curvature, which represents the second-order derivative of the depth value with the spatial position, reflecting the curvature characteristics of the depth distribution. The curvature is calculated by the gradient The derivative is taken again to describe the curvature of the depth change. For example, around the measurement point, the depth gradient changes from 0.3 to 0.1, then the curvature κ = (0.1-0.3) / Δx.

[0151] k2 is the curvature correction coefficient, which controls the influence weight of curvature in the correction process. For example, if the experiment shows that curvature has a greater influence on the depth error, k2 can take a higher value, such as 1.2; if the influence is small, it can take 0.8.

[0152] d max The maximum deformation provided by the deformation monitoring unit indicates the displacement value of the point with the largest deformation in the resection area. max =0.5mm, indicating that the tissue has undergone significant deformation due to external force during the resection process.

[0153] T1 is the deformation threshold, which is used to determine whether the deformation requires additional compensation. For example, if T1 = 0.4 mm, when d max When this value is exceeded, the system will trigger nonlinear compensation.

[0154] f(d max ) is the maximum deformation compensation function, which is implemented using the following formula (5):

[0155]

[0156] Among them, γ1 is the compensation gain coefficient, which can be obtained through experimental data or directly set according to expert knowledge; α1 is the nonlinear index, which can be obtained through experimental data or directly set according to expert knowledge, for example, α1=2, α1 is used to adjust the growth rate of the compensation function.

[0157] The intelligent control module 104 is used to integrate the data output by each module to control the electrosurgical resection operation process. The intelligent control module includes a position positioning unit, a trajectory generation unit and a stress analysis unit; the position positioning unit is used to identify the three-dimensional spatial relationship data of the electrosurgical resection ring relative to the prostate tissue and obtain the position information based on the microvessel distribution information provided by the network tracking unit and the corrected depth data provided by the compensation adjustment unit; the trajectory generation unit is used to generate a predicted resection trajectory of the electrosurgical resection ring based on the received position information provided by the position positioning unit and the corrected depth data provided by the compensation adjustment unit; the stress analysis unit is used to analyze the tissue stress distribution during the resection process based on the predicted resection trajectory to prevent tissue tearing, and feed the analysis results back to the trajectory generation unit to optimize the resection path.

[0158] In the image recognition system for resectoscope provided in this embodiment, the intelligent control module 104 acts as the core of data integration and operation control during the operation, and is responsible for receiving and integrating data from multiple modules in real time to ensure the safety and accuracy of the resection operation. The module includes a position positioning unit, a trajectory generation unit, and a stress analysis unit, which work closely together in data processing and path optimization.

[0159] The function of the position positioning unit is to identify the three-dimensional spatial relationship of the electrosurgical resection ring relative to the prostate tissue based on the microvascular distribution information provided by the network tracking unit and the corrected depth data provided by the compensation adjustment unit. First, through the microvascular distribution data obtained by the network tracking unit, the unit can determine the vascular distribution area in the resection path to avoid the resection path entering the high-risk area. Then, based on the corrected depth data provided by the compensation adjustment unit, the position positioning unit further accurately identifies the depth and position relationship of the electrosurgical resection ring in the surgical area, including the distance relative to the microvessels and different tissue layers, and finally generates complete three-dimensional spatial position information. This information provides a detailed electrosurgical resection path basis for the trajectory generation unit to ensure the safety and accuracy of the path design.

[0160] After obtaining the position information, the trajectory generation unit uses this data to formulate the predicted resection trajectory of the electrosurgical resection ring. The trajectory generation unit dynamically generates the optimal resection path based on the microvascular distribution and corrected depth data in the position information. Specifically, the trajectory generation unit can use the microvascular avoidance data in the position information to design a path that can both efficiently resect the target tissue and avoid microvessels. In addition, the unit also uses depth data to ensure that the resection depth meets the requirements of different tissue layers, thereby ensuring the resection effect while avoiding over-resection or under-resection. The output of the trajectory generation unit is an optimized resection trajectory, which provides clear directional guidance for the electrosurgical resection ring to perform the resection operation along the safest and most efficient path during surgery.

[0161] The stress analysis unit analyzes the stress distribution of the tissue during the electrosurgical resection based on the predicted resection trajectory of the trajectory generation unit to prevent tissue tearing or damage that may occur during the resection process. Specifically, the stress analysis unit evaluates the tissue stress distribution on the resection path and predicts the stress concentration during the surgical operation. By analyzing the stress data, the stress analysis unit can identify high-stress areas that may cause tissue tearing and feed this information back to the trajectory generation unit. The trajectory generation unit can optimize the resection path accordingly to ensure that the resection trajectory operates in an area with less tissue stress, thereby reducing the risk of tissue damage. Through the coordination of the three, the intelligent control module 104 can perform real-time path planning and adjustment throughout the operation, improve the safety and accuracy of the operation, and enable the system to accurately avoid key tissues and ensure the resection effect when operating in different tissue layers and different positions.

[0162] Furthermore, the stress analysis unit is specifically used for:

[0163] The discrete tangent point coordinate sequence P(z,y,z) on the predicted resection trajectory output by the acquisition trajectory generation unit;

[0164] For each discrete cutting point, obtain the tissue elastic coefficient E(θ) and Poisson's ratio v(θ) at different angles in the circumferential direction, as well as the tissue thickness h and the resection radius r of the electric resection ring at the corresponding position;

[0165] The stress analysis unit first receives the predicted resection trajectory output by the trajectory generation unit. The trajectory consists of a series of discrete tangent point coordinates P (x, y, z). These coordinates are obtained by discretizing the motion path of the electrosurgical resection ring and represent the cutting position of the electrosurgical resection ring in three-dimensional space. For each tangent point, the system further obtains the tissue elastic coefficient E (θ), Poisson's ratio v (θ), tissue thickness h at the corresponding position and the resection radius r of the electrosurgical resection ring at different circumferential angles. For example, at a specific tangent point, the tissue thickness h = 0.5 mm, the resection radius r = 2.0 mm, and the elastic coefficients in the 0°, 90° and 180° directions are E (0°) = 0.8, E (90°) = 0.9 and E (180°) = 0.85 respectively.

[0166] The system then establishes the stress distribution model at the tangent point according to the following formulas (6) and (7) to calculate the radial stress σ r and shear stress τ rθ :

[0167]

[0168] Among them, F r and F θ are the radial force and circumferential force during the resection process; K1 and K2 are correction coefficients related to tissue type; E(θ) and v(θ) are the elastic coefficient and Poisson's ratio corresponding to each angle, respectively.

[0169] F r and F θ They are radial force and circumferential force, respectively, which can be measured in real time by the force sensor in the surgical equipment. For example, during a certain cutting process, the radial force F r =1.2N, hoop force F θ =0.8N.

[0170] ρ is the radial distance from the tangent point to the center of the electrocautery ring, which can be calculated from the three-dimensional coordinates of the tangent point, for example, ρ = 1.5 mm.

[0171] K1 and K2 are correction coefficients related to tissue types, which are calibrated through experiments to reflect the response characteristics of different tissues to stress. For example, for the glandular layer, K1 = 1.1 and K2 = 1.2 may be taken.

[0172] Based on formulas (6) and (7), the system can calculate the radial stress and shear stress at each tangent point. For example, under the above parameters, if E(θ) = 0.8, v(θ) = 0.4, h = 0.5 mm, r = 2.0 mm, it is calculated as:

[0173]

[0174] Obtain the ultimate tensile strength σ of the target tissue c and the ultimate urgent strength τ c , and based on the Mohr strength theory, it is determined whether there is a tearing risk. When any of the conditions provided by the following formulas (8) to (10) are met, a trajectory correction signal is generated:

[0175] σ r >0.8σ c (8)

[0176] |τ rθ |>0.8τ c (9)

[0177] (σ r -τ rθ )>0.9σ c (10)

[0178] The velocity trajectory correction signal includes the azimuth information of the high stress area and the recommended resection speed adjustment coefficient to adjust the movement speed and resection path of the electrosurgical resection loop to avoid the risk of tissue tearing. c and the ultimate urgent strength τ c Can be determined by experiment.

[0179] Next, the system calculates the ultimate tensile strength σ of the target tissue based on the c and the ultimate shear strength τ c Determine whether there is a risk of tearing. These parameters are determined experimentally, such as

[0180] σ c =5.0MPa, τ c =3.0MPa.

[0181] According to formulas (8), (9) and (10), the judgment condition of tearing risk is:

[0182] σ r >0.8·σ c

[0183] |τ rθ |>0.8·τ c

[0184] (σ r -τrθ )>0.9·σ c

[0185] In the above calculation:

[0186] σ r =3.52MPa,0.8·σ c =4.0MPa, condition (8) is not satisfied;

[0187] τ rθ =2.08MPa,0.8·τ c =2.4MPa, condition (9) is not satisfied;

[0188] σ r -τ rθ =1.44MPa,0.9·σ c =4.5MPa, condition (10) is not satisfied.

[0189] If any of the conditions are met, the system will generate a trajectory positive signal. This signal includes the azimuth information of the high stress area and the recommended resection speed adjustment factor. For example, if condition (8) is met, the system may recommend adjusting the resection speed from 2 mm / s to 1 mm / s to reduce tissue stress and avoid high stress areas by adjusting the resection path.

[0190] Through the above steps, the stress analysis unit can evaluate the tissue stress distribution during the electrosurgical cutting process in real time, dynamically generate correction signals and adjust cutting parameters.

[0191] Furthermore, the position positioning unit is specifically used for:

[0192] constructing a microvessel distribution map according to the microvessel distribution information received from the network tracking unit;

[0193] Receiving the corrected depth data provided by the compensation adjustment unit, and combining it with the microvascular distribution map to construct a depth distribution model of the prostate tissue;

[0194] Based on the depth distribution model, the optimal position of the electrosurgical ring in the prostate tissue is calculated;

[0195] Generating three-dimensional spatial relationship data of the electric cutting ring relative to the prostate tissue according to the calculated position of the electric cutting ring, wherein the three-dimensional spatial relationship data includes the distance between the electric cutting ring and the microvessels and the relative position in the depth direction;

[0196] The obtained three-dimensional spatial relationship data is output and provided to the resection trajectory generation unit as position information.

[0197] The position positioning unit in the image recognition system for the resectoscope provided in this embodiment accurately locates the spatial position of the resection ring in the prostate tissue by receiving data from the network tracking unit and the compensation adjustment unit. The unit combines the microvascular distribution information with the corrected depth data to form comprehensive three-dimensional spatial relationship data, providing a scientific basis for the subsequent generation of the resection trajectory.

[0198] The position positioning unit first receives the microvessel distribution information provided by the network tracking unit and constructs a microvessel distribution map. The microvessel distribution map represents the spatial distribution of microvessels in the tissue in the form of an image, including information such as the shape, position, and direction of the microvessels. For example, in an area of ​​​​electrical resection surgery, the microvessel distribution map may show several branched microvessel structures, and these microvessels are concentrated in certain specific areas of the tissue. The construction of the microvessel distribution map depends on the recognition results of the hemoglobin spectral characteristics by the network tracking unit to ensure that the reflected distribution information is accurate and dynamically updated.

[0199] The positioning unit also receives the corrected depth data provided by the compensation adjustment unit. The corrected depth data is the depth measurement value corrected for tissue deformation and other errors, and can accurately reflect the true depth of the prostate tissue at different locations. After combining these depth data with the microvascular distribution map, the positioning unit constructs a depth distribution model of the prostate tissue. The depth distribution model is represented in the form of a three-dimensional grid, showing the distribution of the depth value of the tissue in space, and overlaps with the microvascular distribution map to form a comprehensive prostate tissue characteristic map. For example, in a certain resection area, the depth distribution model may show that the thickness of the glandular layer is 1.5mm, the thickness of the interstitial layer is 2.0mm, and a dense microvascular network is distributed within these layers.

[0200] Based on the constructed depth distribution model, the positioning unit further calculates the optimal position of the electrosurgical ring in the prostate tissue. This process comprehensively considers the avoidance requirements of microvessels and the characteristics of tissue depth to ensure that the electrosurgical ring can effectively remove the target tissue while avoiding damage to the microvessels during the resection path. For example, by analyzing the depth distribution model, the positioning unit may conclude that the optimal position is in an area 0.5 mm away from the microvascular network, and adjust the movement path of the electrosurgical ring to this area.

[0201] Subsequently, the positioning unit generates three-dimensional spatial relationship data of the electrosurgical resection ring relative to the prostate tissue based on the calculated optimal position. The three-dimensional spatial relationship data specifically includes information such as the distance between the electrosurgical resection ring and the microvessels and the relative position in the depth direction. For example, the three-dimensional relationship data may indicate that the distance between the electrosurgical resection ring and the nearest microvessel is 0.7 mm, and the distance to the boundary of the glandular layer is 0.2 mm. These data intuitively reflect the precise positioning of the electrosurgical resection ring in the surgical area, providing the necessary input for the subsequent resection trajectory planning.

[0202] Finally, the position positioning unit outputs the generated three-dimensional spatial relationship data as position information to the resection trajectory generation unit. The resection trajectory generation unit uses this data to formulate the optimal resection path to achieve safe and accurate electrosurgical resection.

[0203] Furthermore, the trajectory generating unit is specifically used for:

[0204] Receive the position information provided by the position positioning unit, and analyze the spatial position relationship of the electrosurgical resection ring relative to the prostate tissue, including the distance information from the electrosurgical resection ring to the microvessels and key tissue structures, and the relative position data in different depth directions;

[0205] Receiving the corrected depth data provided by the compensation adjustment unit to obtain the accurate position of the prostate tissue at different cutting depths;

[0206] Determine the preliminary resection path of the electrosurgical resection loop based on the position information and the corrected depth data;

[0207] Optimize the initial resection path and dynamically adjust the movement direction and speed of the electrosurgical resection loop;

[0208] According to the optimized path, the final predicted resection trajectory is generated and used as the operation instruction of the electrosurgical resection loop.

[0209] The trajectory generation unit provided in this embodiment plays a core role in the image recognition system for the resectoscope, and is specifically used to generate the predicted resection trajectory of the resection loop based on the position information and depth data, and optimize it to ensure the accuracy and safety of the operation. Through multi-stage data processing and path planning, the unit can generate dynamically adjusted resection instructions to adapt to the real-time changes of tissues and environment during the operation.

[0210] The trajectory generation unit first receives the position information from the position positioning unit and analyzes the spatial position relationship of the electrosurgical resection ring relative to the prostate tissue. This position information includes the distance from the electrosurgical resection ring to the microvascular network and other key tissue structures, as well as the relative position data of the electrosurgical resection ring in different depth directions. For example, at a specific cut point, the position information may show that the distance between the electrosurgical resection ring and the nearest microvessel is 0.6mm, and the vertical distance from the glandular layer boundary is 1.2mm. These data provide a detailed spatial reference for the planning of the resection path, ensuring that the electrosurgical resection ring can avoid key tissues.

[0211] At the same time, the trajectory generation unit receives the corrected depth data provided by the compensation adjustment unit. These depth data are deformed and error corrected to accurately reflect the true depth distribution of prostate tissue at different locations. For example, in a certain surgical area, the depth data may indicate that the thickness of the glandular layer is 1.5mm, while the thickness of the interstitial layer is 2.0mm. By combining the depth data and position information, the trajectory generation unit can fully grasp the three-dimensional characteristics of the surgical area.

[0212] Based on the above data, the trajectory generation unit first determines the preliminary resection path of the electrosurgical resection ring. The planning of the preliminary path aims to resect the target tissue while avoiding microvessels and other sensitive areas as much as possible. For example, the system may calculate a path to ensure that the electrosurgical resection ring always runs within a safe range of at least 0.5 mm from the microvascular network. This preliminary path reflects the spatial layout of the target area for resection, but does not yet take into account the need for dynamic adjustment.

[0213] After determining the preliminary path, the trajectory generation unit optimizes it. The optimization process dynamically adjusts the movement direction and speed of the electrosurgical ring based on real-time position information and depth data. For example, when the path passes through an area with significant tissue deformation, the system may reduce the speed of the electrosurgical ring to reduce the risk of tissue tearing, or adjust the direction to avoid high-stress areas. The optimization algorithm takes into account safety and efficiency, ensuring that the resection operation meets the target requirements while minimizing tissue damage.

[0214] Finally, the trajectory generation unit generates a predicted resection trajectory based on the optimized path and outputs it as an operation instruction for the electrosurgical resection ring. The predicted trajectory is represented by a series of continuous coordinate points, which clearly indicate the movement path and speed of the electrosurgical resection ring in three-dimensional space. For example, the system may output a trajectory that instructs the electrosurgical resection ring to move in a specific direction at a speed of 2 mm per second, while slowing down to 1 mm per second at key locations. Such operation instructions can be used directly to control the movement of the electrosurgical resection ring, thereby achieving precise tissue resection.

[0215] Through multi-stage processing of the trajectory generation unit, the system is able to generate and optimize the electrocautery path in real time, dynamically adapting to tissue characteristics and surgical requirements.

[0216] Although the present application is disclosed as above in the form of a preferred embodiment, it is not intended to limit the present application. Any technical personnel in this field may make possible changes and modifications without departing from the spirit and scope of the present application. Therefore, the scope of protection of the present application shall be based on the scope defined by the claims of the present application.

Claims

1. An image recognition system for a resectoscope, characterized in that: include: An image acquisition unit, used for acquiring video images during the electrosurgical resection procedure in real time; A tissue feature module is used to receive the video image from the image acquisition unit and perform feature analysis processing, and the tissue feature module includes a spectral analysis unit, a tissue stratification unit and a network tracking unit; the spectral analysis unit is used to extract the reflectance spectral features of different tissues from the video image; the tissue stratification unit is used to hierarchically divide the prostate tissue based on the spectral feature data provided by the spectral analysis unit, realize hierarchical identification of the glandular layer, the interstitial layer and the capsule layer, and obtain a hierarchical identification result; the network tracking unit is used to receive the spectral feature data output by the spectral analysis unit, identify the microvessel distribution by analyzing the spectral absorption characteristics of hemoglobin, and obtain microvessel distribution information; A depth estimation module, used for performing depth estimation based on the hierarchical recognition result provided by the tissue feature module, the depth estimation module includes a deformation monitoring unit, an optical path depth unit and a compensation adjustment unit; the deformation monitoring unit is used for collecting tissue deformation data during the electroresection process; The optical path depth unit is used to generate a depth measurement result of the prostate tissue based on the tissue deformation data provided by the deformation monitoring unit and the stratification identification result provided by the tissue stratification unit, using the reflected light intensity ratio of 470nm and 630nm wavelengths; The compensation adjustment unit is used to perform real-time correction on the depth measurement result according to the received depth measurement result provided by the optical path depth unit and the tissue deformation data provided by the deformation monitoring unit, so as to generate corrected depth data; An intelligent control module, used to integrate the data output by each module and control the electrosurgical cutting operation process, the intelligent control module includes a position positioning unit, a trajectory generation unit and a stress analysis unit; The position positioning unit is used to identify the three-dimensional spatial relationship data of the electrosurgical resection ring relative to the prostate tissue and obtain the position information according to the microvessel distribution information provided by the network tracking unit and the corrected depth data provided by the compensation adjustment unit; The trajectory generating unit is used to generate a predicted resection trajectory of the electrosurgical resection ring according to the received position information provided by the position positioning unit and the corrected depth data provided by the compensation adjustment unit; The stress analysis unit is used to analyze the tissue stress distribution during the resection process based on the predicted resection trajectory to prevent tissue tearing, and feed the analysis result back to the trajectory generation unit to optimize the resection path.

2. The image recognition system for a resectoscope according to claim 1, characterized in that: The spectrum analysis unit is specifically used for: Receiving a video image acquired in real time, and dividing the video image into a plurality of regions; each region represents a tissue sample; For each region, a specific spectral filtering process is applied to obtain the reflected light intensity information at different wavelengths; based on the reflected light intensity information, a reflection spectrum curve at different wavelengths is generated; Normalize the acquired reflectance spectrum curve to remove noise and deviation caused by brightness and angle changes in the video image; The normalized spectral curve is subjected to feature extraction to identify the characteristic wavelengths and corresponding reflection intensities of different tissues; the reflection characteristics at the characteristic wavelengths are analyzed to extract the spectral feature data of each tissue.

3. The image recognition system for a resectoscope according to claim 1, characterized in that: The tissue stratification unit is specifically used for: receiving spectral characteristic data provided by the spectral analysis unit, and preliminarily identifying different tissue regions according to the reflection characteristics of each tissue type at a specific wavelength; The initially identified tissue regions are further classified by comparing the spectral features of each region with the typical spectral patterns in the pre-stored tissue feature database to classify each region as the glandular layer, interstitial layer or capsule layer; Generate and output hierarchical recognition results, and mark the hierarchical attributes of each area to form a complete organizational hierarchical diagram.

4. The image recognition system for a resectoscope according to claim 1, characterized in that: The network tracking unit is specifically used for: receiving the spectral characteristic data output by the spectral analysis unit, and screening out the characteristic wavelength range related to the absorption characteristics of hemoglobin; Conduct in-depth analysis of the filtered spectral data to identify regions with significant absorption characteristics within the characteristic wavelength range of hemoglobin; A preliminary microvascular distribution map was generated by aggregating regions with significant absorption properties; The preliminary microvessel distribution map is further processed to optimize and confirm the specific location of the microvessels and generate the final microvessel distribution information.

5. The image recognition system for a resectoscope according to claim 1, characterized in that: The deformation monitoring unit is specifically used for: Real-time capture of prostate tissue images during electroresection and marking of multiple key reference points; Analyze the position changes of key reference points during the electrosurgical cutting process and calculate the displacement and relative deformation degree of each reference point; By comparing the position difference of the reference points in the current image and the reference image, preliminary deformation data is generated; The displacement data of each reference point is integrated to form a complete tissue deformation map, and the deformation intensity and distribution of the tissue at different positions are recorded.

6. The image recognition system for a resectoscope according to claim 1, characterized in that: The optical path depth unit calculates the depth D of the prostate tissue using the following formula (1): Where I1 is the reflected light intensity at a wavelength of 470 nm; I2 is the reflected light intensity at a wavelength of 630 nm; R1 and R2 represent the standard reflectances of the reference prostate tissue at wavelengths of 470 nm and 630 nm, respectively; K is a correction coefficient related to the prostate tissue type; f(λ1, λ2, t) is a dual-wavelength optical path difference correction function, which is used to correct the deviation caused by the optical path difference of different wavelengths of light in the tissue and temperature changes; C(n) is an interlayer interface correction function; wherein the dual-wavelength optical path difference correction function f(λ1, λ2, t) adopts the following formula (2): Where t represents the real-time temperature of the prostate tissue; α and β are experimental calibration coefficients; μ1 and μ2 are tissue scattering coefficients corresponding to wavelengths of 470nm and 630nm, respectively; The interlayer interface correction function C(n) adopts the following formula (3): Wherein, n is the number of tissue layers, including glandular layer, interstitial layer and capsule layer; γ i is the weight coefficient of the i-th layer; Δn i is the refractive index difference between the i-th layer and the adjacent layer; d i represents the distance from the current measurement position to the interface of the i-th layer; σ i is the attenuation factor, which is determined by the absorption characteristics of the i-th layer of tissue.

7. The image recognition system for a resectoscope according to claim 1, characterized in that: The compensation adjustment unit uses the correction formula provided by the following formula (4) for correction: Wherein, D' is the corrected depth value, which is the depth value finally output to the intelligent control module; D0 is the initial depth value, which is provided by the optical path depth unit; is the depth gradient correction term, where is the depth gradient value, which indicates the rate of change of the depth value around the measurement point; k1 is the correction coefficient related to tissue characteristics; k2·κ is the depth curvature correction term, where k is the depth curvature, which indicates the second-order derivative of the depth value with the spatial position, reflecting the curvature characteristics of the depth distribution; k2 is the curvature correction coefficient, which is used to adjust the influence of the depth curvature in the correction process; d max is the maximum deformation provided by the deformation monitoring unit, which is the deformation value of the position with the largest deformation in the resection area; T1 is the deformation threshold; f(d max ) is the maximum deformation compensation function, which is implemented using the following formula (5): Among them, γ1 is the compensation gain coefficient; α1 is the nonlinear index used to adjust the growth rate of the compensation function.

8. The image recognition system for a resectoscope according to claim 1, characterized in that: The stress analysis unit is specifically used for: The discrete tangent point coordinate sequence P (x, y, z) on the predicted resection trajectory output by the acquisition trajectory generation unit; For each discrete cutting point, obtain the tissue elastic coefficient E(θ) and Poisson's ratio v(θ) at different angles in the circumferential direction, as well as the tissue thickness h and the resection radius r of the electric resection ring at the corresponding position; According to the following formulas (6) and (7), the stress distribution model at the tangent point is established to calculate the radial stress σ r and shear stress τ rθ : Among them, F r and F θ are the radial force and circumferential force during the resection process; K1 and K2 are correction coefficients related to tissue type; E(θ) and v(θ) are the elastic coefficient and Poisson's ratio corresponding to each angle, respectively; Obtain the ultimate tensile strength σ of the target tissue c and the ultimate urgent strength τ c , and based on the Mohr strength theory, it is determined whether there is a tearing risk. When any of the conditions provided by the following formulas (8) to (10) are met, a trajectory correction signal is generated: s r >0.8σ c (8) |t rθ |>0.8t c (9) (s r -t rθ )>0.9σ c (10) Among them, the speed trajectory correction signal includes the azimuth information of the high stress area and the recommended resection speed adjustment coefficient to adjust the movement speed and resection path of the electrosurgical resection ring to avoid the risk of tissue tearing.

9. The image recognition system for a resectoscope according to claim 1, characterized in that: The position positioning unit is specifically used for: constructing a microvessel distribution map according to the microvessel distribution information received from the network tracking unit; Receiving the corrected depth data provided by the compensation adjustment unit, and combining it with the microvascular distribution map to construct a depth distribution model of the prostate tissue; Based on the depth distribution model, the optimal position of the electrosurgical ring in the prostate tissue is calculated; Generating three-dimensional spatial relationship data of the electric cutting ring relative to the prostate tissue according to the calculated position of the electric cutting ring, wherein the three-dimensional spatial relationship data includes the distance between the electric cutting ring and the microvessels and the relative position in the depth direction; The obtained three-dimensional spatial relationship data is output and provided to the resection trajectory generation unit as position information.

10. The image recognition system for a resectoscope according to claim 1, characterized in that: The trajectory generating unit is specifically used for: Receive the position information provided by the position positioning unit, and analyze the spatial position relationship of the electrosurgical resection ring relative to the prostate tissue, including the distance information from the electrosurgical resection ring to the microvessels and key tissue structures, and the relative position data in different depth directions; Receiving the corrected depth data provided by the compensation adjustment unit to obtain the accurate position of the prostate tissue at different cutting depths; Determine the preliminary resection path of the electrosurgical resection loop based on the position information and the corrected depth data; Optimize the initial resection path and dynamically adjust the movement direction and speed of the electrosurgical resection loop; According to the optimized path, the final predicted resection trajectory is generated and used as the operation instruction of the electrosurgical resection loop.

Images