CN112132805B

CN112132805B - Ultrasonic robot state normalization method and system based on human body characteristics

Info

Publication number: CN112132805B
Application number: CN202011003002.8A
Authority: CN
Inventors: 黄彦玮; 孙正隆; 肖维; 刘恒利
Original assignee: Chinese University of Hong Kong Shenzhen; Shenzhen Institute of Artificial Intelligence and Robotics
Current assignee: Chinese University of Hong Kong Shenzhen; Shenzhen Institute of Artificial Intelligence and Robotics
Priority date: 2020-09-22
Filing date: 2020-09-22
Publication date: 2024-02-09
Anticipated expiration: 2040-09-22
Also published as: CN112132805A

Abstract

The invention discloses a method and a system for normalizing the state of an ultrasonic robot based on human body characteristics, wherein the method comprises the following steps: initializing a preset human body digital twin model based on a human body parameterized vector according to a scanning result of the physical sign of the person to be detected by the ultrasonic robot; calibrating and updating the human body digital twin model based on an ultrasonic image standard image; and carrying out normalized representation on the mechanical arm track of the human body digital twin model based on the chessboard grid so as to realize parameterization of human body characteristics. The invention can realize vector and normalized representation of human body characteristic parameters, thereby realizing human body characteristic parameterization so as to analyze human body characteristics.

Description

Ultrasonic robot state normalization method and system based on human body characteristics

Technical Field

The invention relates to the technical field of artificial intelligence, in particular to an ultrasonic robot state normalization method and system based on human body characteristics.

Background

Ultrasound technology is a very important technology in clinical analysis, and ultrasound detection is also increasingly important in human detection. In conventional ultrasound scanning, there are a number of unavoidable limitations to manual freehand manipulation.

Robotic-assisted ultrasound scanning systems have evolved from the 90 s of the last century to date. The existing ultrasonic robot technology can be divided into two categories of semi-autonomous ultrasonic and fully autonomous ultrasonic according to the degree of automation. The semi-autonomous ultrasound aims at combining the respective advantages of a robot and a robot, and the doctor can concentrate attention on a high-level decision-making level by using the machine to enhance the operation capability of the doctor and reduce the operation difficulty of the doctor; whereas "fully autonomous ultrasound" is intended to replace to some extent the work of the sonographer with a robotic and computer-aided system. The ultrasonic robot can realize autonomous movement, and a doctor only needs to finally audit the operation normalization and the accuracy of the robot. However, there are many limitations in the technology when it is performed on the scanned human body characteristics, which seriously affect the analysis result.

Accordingly, there is a need for improvement and development in the art.

Disclosure of Invention

The invention aims to solve the technical problems that the state normalization method and system of the ultrasonic robot based on human body characteristics are provided for overcoming the defects in the prior art, and aims to solve the problems that the prior art has a plurality of limitations when scanning human body characteristics are carried out, and the analysis result is seriously influenced.

The technical scheme adopted by the invention for solving the problems is as follows:

in a first aspect, an embodiment of the present invention provides a method for normalizing a state of an ultrasound robot based on a human feature, where the method includes:

initializing a preset human body digital twin model based on a human body parameterized vector according to a scanning result of the physical sign of the person to be detected by the ultrasonic robot;

calibrating and updating the human body digital twin model based on an ultrasonic image standard image;

and carrying out normalized representation on the mechanical arm track of the human body digital twin model based on the chessboard grid so as to realize parameterization of human body characteristics.

In one embodiment, the method further comprises:

and dynamically updating the human body digital twin model in real time according to an explicit positioning point in the human body ultrasonic image, wherein the explicit positioning point is used for reflecting the human body surface characteristics in the human body ultrasonic image.

In one embodiment, the initializing the preset digital twin model of the human body based on the parameterized vector of the human body according to the scan result of the ultrasonic robot to the sign of the human body to be measured includes:

according to the scanning result of the physical sign of the person to be detected by the ultrasonic robot, identifying the explicit positioning point of the body surface of the person to be detected;

Based on the relative relation between the explicit positioning points, obtaining the human body parameterized vector;

initializing the human body digital twin model based on the human body parameterized vector.

In one embodiment, the method further comprises:

and establishing a state working space of the ultrasonic robot based on the human body digital twin model.

In one embodiment, the method further comprises:

real-time adaptive control is achieved by using the artifact information of the force/moment sensor or the ultrasound image as feedback.

In one embodiment, the system comprises:

the model initialization unit is used for initializing a preset human body digital twin model based on a human body parameterized vector according to the scanning result of the physical sign of the person to be detected by the ultrasonic robot;

the calibration updating unit is used for carrying out calibration updating on the human body digital twin model based on the ultrasonic image standard image;

and the parameterization unit is used for carrying out normalized representation on the mechanical arm track of the human body digital twin model based on the chessboard grid so as to realize parameterization on human body characteristics.

In one embodiment, the system further comprises:

the dynamic updating unit is used for dynamically updating the human body digital twin model in real time according to the explicit positioning points in the human body ultrasonic image, wherein the explicit positioning points are used for reflecting the human body surface characteristics in the human body ultrasonic image.

In one embodiment, the model initializing unit includes:

the identification subunit is used for identifying the explicit positioning points on the body surface of the person to be detected according to the scanning result of the sign of the person to be detected by the ultrasonic robot;

a parameter vectorization subunit, configured to obtain the human body parameterized vector based on a relative relationship between the explicit positioning points;

and the initialization subunit is used for initializing the human body digital twin model based on the human body parameterized vector.

In one embodiment, the system further comprises:

and the working space establishing unit is used for establishing the state working space of the ultrasonic robot based on the human body digital twin model.

In one embodiment, the system further comprises:

and the self-adaptive control unit is used for realizing real-time self-adaptive control by taking artifact information of the force/moment sensor or the ultrasonic image as feedback.

The invention has the beneficial effects that: according to the scanning result of the physical sign of the person to be detected by the ultrasonic robot, a preset human body digital twin model is initialized based on a human body parameterized vector; calibrating and updating the human body digital twin model based on an ultrasonic image standard image; and carrying out normalized representation on the mechanical arm track of the human body digital twin model based on the chessboard grid so as to realize parameterization of human body characteristics. The invention can realize vector and normalized representation of human body characteristic parameters, thereby realizing human body characteristic parameterization so as to analyze human body characteristics.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments described in the present invention, and other drawings may be obtained according to the drawings without inventive effort to those skilled in the art.

Fig. 1 is a schematic flow chart of an ultrasonic robot state normalization method based on human body characteristics according to an embodiment of the present invention.

Fig. 2 is a flowchart of an overall idea of the ultrasonic robot state normalization method based on human body characteristics according to the embodiment of the present invention when implemented.

Fig. 3 is a schematic diagram of human body parameterization in a human body feature-based ultrasonic robot state normalization method according to an embodiment of the present invention.

Fig. 4 is a schematic diagram of a chest scanning embodiment in a state normalization method of an ultrasound robot based on human body features according to an embodiment of the present invention.

Fig. 5 is a schematic diagram of ultrasonic robot operation space establishment in the ultrasonic robot state normalization method based on human body characteristics according to the embodiment of the present invention.

Fig. 6 is a semantic analysis diagram of a renal ultrasound image in a human feature-based ultrasound robot state normalization method according to an embodiment of the present invention.

Fig. 7 is a flowchart of PPO algorithm training in a human feature-based ultrasound robot state normalization method according to an embodiment of the present invention.

Fig. 8 is a schematic diagram of a large data platform in the human body feature-based ultrasonic robot state normalization method according to the embodiment of the present invention.

Fig. 9 is a running thought diagram of a data platform in the ultrasonic robot state normalization method based on human body characteristics according to the embodiment of the invention.

Fig. 10 is a functional schematic diagram of an ultrasonic robot scanning control system according to an embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention clearer and more obvious, the present invention will be described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.

It should be noted that, if directional indications (such as up, down, left, right, front, and rear … …) are included in the embodiments of the present invention, the directional indications are merely used to explain the relative positional relationship, movement conditions, etc. between the components in a specific posture (as shown in the drawings), and if the specific posture is changed, the directional indications are correspondingly changed.

Ultrasound technology is a very important technology in clinical analysis, and ultrasound detection is also increasingly important in human detection. In conventional ultrasound scanning, there are a number of unavoidable limitations to the freehand operation of the physician. The disadvantages are mainly the following: 1. long term wrist joint forces cause harm to the physical health of the operator (e.g., doctor). The prevalence of repetitive strain injury, carpal tunnel syndrome and musculoskeletal disease are high in the sonographer population; 2. it is difficult for a doctor to avoid an error in operation due to, for example, shake of hands, distraction, etc.; 3. the handheld ultrasonic scanning operation requires that the doctor directly contacts with the patient, so that the possibility of infection of the doctor is greatly improved, and the life safety of the doctor can be threatened when the doctor is serious. Fourth, traditional hand-held ultrasound analysis has obvious regional limitations. In order to solve the problems, the ultrasonic scanning system assisted by the robot has profound significance and is also a development trend of future ultrasonic robots.

Robotic-assisted ultrasound scanning systems have evolved from the 90 s of the last century to date. The existing ultrasonic robot technology can be divided into two categories of semi-autonomous ultrasonic and fully autonomous ultrasonic according to the degree of automation. The semi-autonomous ultrasound aims at combining the respective advantages of a robot and a robot, and the doctor can concentrate attention on a high-level decision-making level by using the machine to enhance the operation capability of the doctor and reduce the operation difficulty of the doctor; whereas "fully autonomous ultrasound" is intended to replace to some extent the work of the sonographer with a robotic and computer-aided system. The ultrasonic robot can realize autonomous movement, and a doctor only needs to finally audit the operation standardization and accuracy of the ultrasonic robot.

The semi-autonomous ultrasound mainly focuses on the research of a semi-autonomous algorithm for cooperatively controlling a robot and a person, and a representative of a commercial system at home and abroad is an MELODY expert system produced by AdEchoTech company. The system consists of a parallel robot slave hand end on the side of a person to be detected and a master hand end on the side of an operator. With the help of the system, an imaging expert can control a plurality of degrees of freedom of the slave hand of the person to be tested through the master hand of the remote control ultrasonic robot, and finally complete the remote scanning work of the person to be tested. Domestic similar systems have a remote ultrasound robotic system manufactured by Hua Dazhi, which also uses a similar architecture to achieve remote real-time control. In the above two examples, autonomy is mainly represented by the fact that the operator needs to control part of the degrees of freedom of the ultrasound robot, while the remaining degrees of freedom are autonomously controlled by the ultrasound machine. Through real-time feedback of the sensor, the ultrasonic robot shows certain adaptability to the rigidity and shape change of the body surface of the person to be tested.

The development of "fully autonomous ultrasound" systems is still relatively primitive. The only system capable of realizing commercial full-automatic scanning and analysis is a Siemens full-emulsion ultrasonic scanning system at present. The system can acquire an ultrasonic image sequence of a group of breasts of a patient through autonomous scanning, then reconstruct the image sequence in three dimensions to extract a coronal section view of the breasts, and finally analyze the section view by utilizing ultrasonic analysis to obtain an analysis result. Other studies on automated control methods have focused mainly on laboratory level, but their definition of "autonomous" varies, and can be roughly divided into the following two main categories: 1. and the self-adaption of the ultrasonic robot to the environment is realized aiming at the feedback of the sensor. The aim of this study is to construct a mapping relationship between the real-time feedback of the sensor to the motion state of the ultrasound robot, including but not limited to the following classes: a. the self-adaptive control based on the feedback of ultrasonic robot force, b. The compensation of the ultrasonic robot to the biological movement (such as pulse, breath, heart beat); c. visual servo control of the ultrasonic robot on specific features in the ultrasonic image, and the like. And secondly, autonomous scanning and analysis based on a mechanical arm. The purpose of the research is to use a mechanical arm to replace a human hand to scan specific areas such as skin points, lines, faces, bodies and the like of a person to be tested. By utilizing the characteristics of high precision and high response speed of the robot, the robot can obtain a serialized image with higher quality than human hand scanning. After 3D reconstruction is carried out on the serialized images, a 3D model of a specific part of a person to be detected can be obtained, and then through operations such as slicing and the like, a computer-aided analysis technology can obtain an analysis result of the disease and finally generate an evaluation report. The two methods are also different in evaluation indexes, the former method (aiming at realizing the self-adaption of the robot to the environment by the sensor feedback) focuses on exploring the improvement of response capability of the feedback to the robot, and simultaneously emphasizing theoretical innovation, and the latter method (based on autonomous scanning and analysis of the mechanical arm) focuses on combining with clinical medicine, so that indexes such as accuracy and specificity of disease analysis are emphasized.

However, existing "semi-autonomous ultrasound" and "fully autonomous ultrasound" suffer from a number of drawbacks. The development of "semi-self-help ultrasound" at the hardware and software level is relatively mature. Master-slave architecture, wired and wireless medium-and-remote communication modes have gradually become industry consensus and industry standard. However, the disadvantage of this solution is also quite obvious, i.e. the degree of autonomy is low. Although the regional limitation of ultrasonic scanning is solved to a certain extent and the difficulty of ultrasonic scanning is reduced, the problem of labor cost is not fundamentally solved: one expert can only manage one machine at a time, and the efficiency of the operation is not really improved.

In contrast, "autonomous ultrasound" is currently in a rapidly evolving stage. The whole milk ultrasound scanning system can realize a scanning integrated medical system, but the high efficiency is realized by sacrificing the universality of equipment. The device can only scan hemispherical anterior convex organs such as the breast, which greatly limits the usability of the device and reduces the cost performance of the device. There are significant limitations to both of the above-described broad categories of methods in terms of functionality. The limitation of the method one (the realization of the self-adaptation of the robot to the environment aiming at the feedback of the sensor) is that only local information (the feedback information of the sensor) is used as the state quantity of the robot, and the correlation of the anatomical structure of the human body and the disease type and the human body structure is not considered. The second method (autonomous scanning and analysis based on a mechanical arm) adopts the thought of industrial part processing, and can only perform standardized scanning on one or more specific organs of a human body. The scanning path generated by the control architecture has single track and poor generalization potential, and is difficult to meet the market demand of generalization of the medical robot.

It is emphasized that clinical sonographers do not only perform constant-speed standardized scans as do ultrasound robots when scanning patients, but often do individual judgment paths (which are commonly referred to in the medical arts as "handling skills" of sonographers) based on their own knowledge of anatomy and pathology, in combination with information from current ultrasound images. In contrast, the shortcomings of current "autonomous ultrasound" are mainly in the following three aspects:

1. the normalization method of the body surface coordinates of the human body is lacking. In the prior art, the self-adaptive technology is basically based on the local feedback control developed by the control theory, and an intelligent normalization method combining the global information of the human body is absent. The defect severely limits the development of autonomy of the ultrasonic robot, on one hand, the ultrasonic robot cannot generate self-adaptive scanning tracks for the to-be-tested personnel with different body heights, and on the other hand, the tracks generated by scanning of the ultrasonic robot cannot be effectively expressed in a normalized space.

2. Ultrasonic robots cannot achieve "step decisions". At present, the ultrasonic robot still uses a control method of decision in advance, namely 'one-time decision, permanent execution', but cannot realize step decision in the scanning process like a doctor. Specifically, taking ultrasonic lung scanning as an example, the current ultrasonic robot cannot adjust analysis and decision thinking in real time according to abnormal point conditions of the lung of a patient judged in the current ultrasonic image in the scanning process;

3. A system framework with the ability to learn the physician's "manipulation" is lacking. Because an algorithm framework capable of migrating the cognition of doctors to diseases is still lacking at present, the clinical analysis experience of doctors cannot be well applied to the clinical use and analysis of ultrasonic robots; on the other hand, while semi-autonomous ultrasound robots are capable of providing large amounts of clinical data, the large amounts of clinical scan data (from the semi-autonomous ultrasound control method) are not fully utilized due to the lack of an efficient network of training data.

In order to solve the defects in the prior art, the application provides an ultrasonic robot state normalization method and system based on human body characteristics. Specifically, as shown in fig. 1, the method for normalizing the state of the ultrasound robot based on the human body characteristics in the present embodiment includes the following steps:

step S100, initializing a preset human body digital twin model based on a human body parameterized vector according to a scanning result of the physical sign of the person to be detected by the ultrasonic robot;

step 200, calibrating and updating the human body digital twin model based on an ultrasonic image standard image;

and step 300, carrying out normalized representation on the mechanical arm track of the human body digital twin model based on the chessboard grid so as to realize parameterization of human body characteristics.

In specific implementation, the embodiment is based on the ultrasonic robot to scan the person to be tested to obtain a scanning result, and then parameterizing the human body features according to the scanning result. Specifically, the present embodiment first requires the scanning result to be obtained by the ultrasonic robot. The ultrasonic robot used in this embodiment may be a semi-autonomous ultrasonic robot or a fully autonomous ultrasonic robot. According to the embodiment, the human body can be scanned by the autonomous ultrasonic robot or the semi-autonomous ultrasonic robot according to the preset scanning skills to obtain the human body ultrasonic image and the scanning result corresponding to the human body ultrasonic image. Then scanning a human body through an autonomous ultrasonic robot or a semi-autonomous ultrasonic robot according to a preset scanning skill to obtain a human body ultrasonic image and a scanning result corresponding to the human body ultrasonic image; and finally, inputting the scanning result into a preset learning frame for training to obtain a scanning thought decision maker and the video auxiliary analysis system.

Suppose that several doctors (doctors 1,2,3, … …) operate the semi-autonomous ultrasonic robot to scan several persons under test, and each person under test is divided into several scan partitions (partitions a, B, C … …) from the body surface. Taking doctor 1 as an example, the doctor first scans from partition a by the ultrasound robot according to the scanning procedure. In the process that a doctor scans the subarea A by using the ultrasonic robot, the doctor firstly finds implicit ultrasonic positioning points (provided that the implicit ultrasonic positioning points are a and b in the subarea A) on the body surface of the human body by referring to an ultrasonic scanning example provided in a medical manual, and the doctor scans the areas near the a and the b. The present embodiment assumes that the doctor only finds the outlier c near a and b, and that doctor will mark the location of c. Here, the scan of the partition a by the doctor has been completed, the doctor will judge the next partition (here, B is assumed) based on the position where the abnormal locating point appears and the characteristics of the abnormal locating point, and then the doctor will repeat the above process using the ultrasonic robot until the scan process ends. Only all doctors will scan the same part of the human body.

In one implementation manner, in this embodiment, the video acquired by the ultrasonic probe in the scanning result and the scanning track of the ultrasonic probe may be acquired, and the learning frame may be trained based on the video and the scanning track. And carrying out data recording on the scanning process based on the human body digital twin model, and carrying out reinforcement learning training on the learning framework by using the data to obtain a scanning thought decision maker and the video auxiliary analysis system. The data in this embodiment includes the position of the abnormal feature, the abnormal feature in the next partition, the ultrasound video of the abnormal feature, and the determination result of the abnormal feature. When the abnormal characteristics in the next partition and the determination results of the abnormal characteristics are input into a preset learning frame for training, the scanning thought decision maker is obtained, and the scanning thought decision period is used for judging the abnormal characteristics and the positions of the abnormal characteristics; and inputting the positions of the abnormal features and the determination results of the abnormal features into a preset learning frame for training to obtain the video auxiliary analysis system.

In practice, the following data and results will be generated after the scan of doctor 1 is completed: 1. the position of an abnormal ultrasonic locating point c; 2. analyzing results of the abnormal ultrasonic locating point c; 3. an ultrasonic video at an abnormal ultrasonic locating point c; 4. the next partition B is selected (notably, the locations of anchor points a, B and the found abnormal ultrasound anchor point c will be normalized into the digital twin model of the human body). The model to be trained is a scanning thought decision maker and a medical video auxiliary analysis system, wherein the medical video auxiliary analysis system learns the analysis experience of a doctor on a medical video mainly by analyzing the analysis result of the doctor and the abnormal ultrasonic locating point position. The input of the method is a marked video and the position of an abnormal ultrasonic locating point, and the output is the judgment of the current abnormal ultrasonic locating point; and the scanning thought decision maker combines the judgment of the doctor on the position of the abnormal ultrasonic locating point and the current abnormal ultrasonic locating point to learn the selection of the doctor on the next analysis partition. In addition, as the concerned position in the scanning process of each doctor has a great relation with the abnormality of the person to be tested, if c is an abnormal multiple position, the occurrence of c is subject to a certain distribution function, and the experience can be obtained through a statistical method. In this embodiment, the explicit positioning point is used to reflect the body surface characteristics of the human body in the human body ultrasound image, and the implicit positioning point is used to reflect the abnormal positioning point existing in the human body ultrasound image.

In one embodiment, the autonomous scanning process of the ultrasound robot is achieved in the following manner. First, the global camera will provide the coordinates of explicit anchor points on the body surface of the person under test (explicit anchor points can be implemented by externally applied markers) which will be used to dynamically update the human body digital twin model. Then according to the human body correlation information provided by the human body digital twin model, the robot obtains the coordinates of the implicit ultrasonic positioning points a and b. Then by combining the standard human body characteristics provided in the medical tool book by adopting a visual servo method, the ultrasonic robot can find implicit ultrasonic locating points a and b and further update a human body digital twin body model through the anatomical result of the human body (the closer process is helpful to reduce the locating point searching time and the scanning of abnormal locating points can have higher precision). The ultrasonic robot will then scan the points of positions c and d of the possible abnormal anchor points according to the correlation information between the anchor points and the abnormal anchor points provided by the human body digital twin body model. The ultrasound robot will then produce a marked ultrasound video, and this piece of video will be analyzed by the medical video-aided analysis system to produce analysis results at locations c and d of the outlier. Finally, the robot adjusts the analysis thought according to the analysis result, and the process changes the previous one-time decision process in the traditional medical ultrasound into a step-by-step decision process.

Further, the basis of autonomous ultrasonic analysis is the construction of a human body digital twin model, and the core of the construction of the human body digital twin model is the definition of a digital twin model which can meet the ultrasonic analysis requirement (the concept of digital twin technology is based on the basic state of a physical entity, and the established model and collected data are subjected to highly realistic analysis in a dynamic real-time mode for monitoring, predicting and optimizing the physical entity). In combination with practical clinical application, the human body digital twin model is required to provide information related to human body structures for ultrasonic positioning on one hand, and normalize execution tracks of an ultrasonic robot on different human bodies (namely the gender, height and body shape of a person to be tested) on the other hand, so that an algorithm can conveniently learn an analysis technique of doctors of the same kind of diseases. Based on this, the clinical requirements put some constraints on the design of the digital twin model of the human body. On one hand, as the real human body is different from the ultrasonic prosthesis and can keep still all the time, the disturbance formed by subconscious movements of respiration and body brings certain requirements to the dynamics of the digital twin model of the human body; on the other hand, the real-time requirement of the ultrasonic robot control brings a certain constraint to the complexity of the digital twin model of the human body.

In this embodiment, the purpose of the human body digital twin model is to fully express "correlation information existing between human body special structures that perform positioning function for a specific ultrasonic scanning task" by designing a compact topology structure. In addition, the digital twin body model has the following characteristics: 1. information is expressed fully rather than redundantly: for a specific ultrasonic scanning task, "sufficient" means that the vertex information of the topological structure can provide a rough positioning function for completing a complete ultrasonic scanning process, and meanwhile, the general area of scanning is explicitly defined, and "non-redundant" means that the information (non-complete human body) of other structures of the human body which are not related to the specific scanning task is not contained; 2. dynamics and real-time: "dynamic" means that the model can be synchronized with the motion of the body of the person to be tested (including following the breath of the person to be tested, following the body posture transformation, etc.). And 'real-time' means that the calculated amount of model update is small by using a simple expression mode based on topology, so that real-time update according to sensor feedback is possible to realize; 3. normalization of the probe trajectory can be achieved: on the one hand, for a specific scanning track, each scanning point in the track can be uniquely expressed in the human body digital twin model; on the other hand, the scanning path of the human body in the human body digital twin body can be mapped into an effective execution path in the robot operation space.

When the human body parameterization is carried out and the human body digital twin model is constructed, the method mainly comprises the following steps: calibrating and updating the human body digital twin model based on an ultrasonic image standard image; then, carrying out normalized representation on the mechanical arm track of the human body digital twin model based on the chessboard grid; and finally, dynamically updating the human body digital twin model in real time according to the explicit positioning point in the human body ultrasonic image. Specifically, as shown in fig. 3, in order to establish a human body structure with real-time dynamic characteristics, the embodiment designs a positioning method for positioning an implicit ultrasonic positioning point of a human body through an explicit positioning point and further positioning an implicit part where an abnormal point may exist through the implicit positioning point, and realizes an initialization method and a dynamic updating method of a human body digital twin model based on the explicit positioning point and an updating method of the human body digital twin model based on the implicit ultrasonic positioning point. Specifically, the present embodiment will realize the following steps: firstly, identifying an explicit positioning point on the body surface of a human body. By using the body surface chessboard, the embodiment can quickly extract the absolute coordinates of the explicit positioning points of the human body by using the camera, and by connecting adjacent angular point coordinates, a three-dimensional grid which can approximate the outline of the body surface is obtained, each angular point of the grid has three-dimensional space coordinate information of the body surface.2. Based on the parameterization method of the body of the explicit positioning points of the human body, based on the explicit positioning point coordinates of the human body surface in step 1, the embodiment can measure the relative relation of the explicit positioning points of the body surface of the human body (as shown in figure 3, FIG. 3 is a diagram showing relevant human signs obtained by a state normalization method of a human-based ultrasonic robot commonly used in anthropometry, wherein 1-24 refer to 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, shoulder widths, respectively, chest half girth, waist height, ankle half girth, thigh half girth, 11, thigh half girth, waist half girth, obtaining a parameterized vector of the human body by measuring the physical sign of the human body; 3. initialization of human digital twins based on parameterized vectors of the human body. The digital twin model of the human body is defined based on the parameterized vector in the previous step. Specifically, for a particular ultrasound scanning task, the present embodiment uses topological lines to represent relative positional information between the bones and organs of the human body. In particular, the present embodiment focuses on the position on the topological line from which a particular ultrasound cross section of the human body can be extracted. When initializing, the embodiment first roughly locates the topological line according to the prior human body information, and then represents the topological line as the position coordinate relative to the chessboard angular point.

Further, since the human body has a similar structure and also has a difference (there is a difference in sex, height, and body shape). Thus, the present embodiment can use topological lines to represent relative position information between bones and organs of a human body based on similarity to achieve general guidance of the motion of an ultrasonic probe; on the other hand, due to the variability, the specific structure inside the human body needs to be precisely positioned after the ultrasonic probe contacts the human body. Thus, at the beginning of the scan, the robotic arm determines the approximate scan location on the body surface (notably, this coarse positioning is done automatically under guidance of the chessboard and the digital twin phantom of the body) with the help of the medical prior knowledge provided by the digital twin of the body. Whereas after the probe touches the human body, the robotic arm will obtain posterior information about the "anchor point" location (which is exact). The mechanical arm will update the relative coordinates of the human digital twin body model relative to the body surface chessboard using this posterior information. When the relative coordinates of a certain locating point are updated, the model uses the correlation information of the human body model to update and optimize the position information of the rest locating points.

Further, since the position of the mechanical arm track in the chessboard of the human body can be uniquely represented, any discrete point in the chessboard surface path during the scanning process of the mechanical arm can be represented as a linear combination of four angular points close to the discrete point in the chessboard, and thus, any motion track of the mechanical arm can be uniquely represented as a linear combination of the angular point coordinates of the chessboard. In addition, as the topological lines can be uniquely represented by the chessboard, the relative position relation between the motion trail of the mechanical arm and the topological lines of the human body can be obtained in the embodiment. The relative positional relationship will be updated synchronously when the human body topological line is updated in the previous step. It is worth mentioning that this relative position is information of real interest to the doctor when scanning, since it describes the relative positional relationship of the scanning trajectory generated according to the doctor's scanning concept with respect to the "anchor point". Because the relative position relation between the human body digital twin model and the camera can be dynamically changed due to the respiration, subconscious motion of the patient and the like, the embodiment designs a method for dynamically updating the digital twin body in real time based on the chessboard. Since the positions of the corner points in the chessboard can be conveniently extracted by an external camera, the computer can update the human body digital twin body model by utilizing the positions of the corner points in a very large way.

The present embodiment will be described with reference to the scanning of the chest cavity of a human body as an example. The aim of this example is to achieve automatic tracking of the robot's human rib aperture under the guidance of the human digital twin. As shown in fig. 4 (b), in order to achieve scanning of the chest, the present embodiment first needs to obtain coordinates of explicit positioning points on the body surface of the human body. The nipple of the person to be measured is used as positioning information, the relative distance between the nipples of the human body is measured, the parameterized vector of the human body is obtained, and finally, the initialization of the digital twin model of the human body is realized based on the parameterized vector. The present embodiment generates topological lines (yellow solid lines and green solid lines) of the rib slit shape on both sides of the human body based on a priori medical knowledge of the rib slit, and represents the topological lines as position coordinates with respect to the chessboard angular points; furthermore, the embodiment realizes the real-time update of the topological line position based on the ultrasonic image characteristics of the locating point. As shown in (a) of fig. 4, if a rib nip enters an acoustic window of an ultrasonic probe, a batwing-like batwing-shaped batwing image will be able to be detected on an ultrasonic image. Based on the position where the bat sign image can be seen, the present embodiment will correct the exact position of the rib slit, while the position of the right rib slit will also be updated according to the structural correlation of the human body. At the same time, the relative position of the robot track and the human body topological line is updated (as shown in (c) of fig. 4). Finally, the embodiment updates the dynamic model of the human body digital twin model based on the chessboard grid in real time. For example, when the human body rotates, the positions of the corner points of the chessboard grid on the surface of the human body are updated, and the positions of topological lines are synchronously updated.

Further, the present embodiment further defines a working space of the ultrasonic robot based on the human body digital twin model. The robot's workspace is a vector space that is spanned by all possible states of the robot for describing the possible states of the robot's end effector. From the perspective of empirical migration, the creation of a state space is a process that parameterizes the operating experience of a physician. Since the imitation capability of the robot to the human is closely related to the dimension of the working space of the ultrasonic robot, whether the state space can be reasonably defined determines whether the ultrasonic robot can reproduce the operation of the operator well or not. Generally, the higher the dimension, the more mimicking an ultrasound robot is to a person, but the negative impact is that the difficulty of data processing and training will increase. In addition, decoupling of different dimension variables in the robot operation space is beneficial to simplifying the complexity of the model, reducing the calculated amount and improving the dynamic property and the real-time property of the model.

In combination with the clinical requirements of ultrasound analysis, this embodiment defines the state space of a fully autonomous ultrasound robot. As shown in fig. 5, it is assumed that the robot operates the ultrasonic probe to contact the curved surface S1-Fo-S2 of the human skin, and the contact point thereof is F0. The four marked points on the plane S1-Fo-S2 that are adjacent to the contact point Fo are denoted V1, V2, V3, V4, respectively. With the intersection point of two diagonal lines of V1-V3 and V2-V4 as the center, a non-rectangular coordinate system xyz-o is established, and then the projection of the contact point Fo onto the plane x-o-y can be uniquely expressed as a linear combination of four vertexes V1, V2, V3 and V4. Because the points are on a plane, determining a point on the plane requires a four-dimensional vector v= {1,2,3,4} consisting of vertex numbers, and a 2-dimensional coordinate o= { x, y } representation on the plane x-o-y. If it is also desired to indicate the positive pressure of the doctor pressing against the skin surface and the orientation of the probe, it is also necessary to add two dimensions of data F2, To represent the component of the force/moment sensing probe's contact force with the skin perpendicular to the skin surface and the positive angle of rotation of the probe about axis Fz, respectively.

In addition, in combination with the operation of the doctor during scanning, the present embodiment will also increase the data of the two dimensions of the frequency H of the ultrasound and the depth of focus D of the ultrasound. The frequency H of ultrasound, among other things, affects the depth range of ultrasound imaging and the clarity of imaging. In particular, lower frequency ultrasound energy presents deep organs farther from the body surface, but at the same time also blurs the imaging darkness; whereas high-frequency imaging can see clear textures of shallow surface layers, deep organs of a human body cannot be observed due to the fact that the imaging depth is shallow. The focusing depth D is another effective index affecting the imaging quality, and the recognition rate of the image can be effectively improved due to the fact that the signal-to-noise ratio of the ultrasonic image is low and the focusing depth is reasonably selected.

In summary, the state space vector of the ultrasonic robot in this patent is defined as follows:

，Fz/>。

in addition, the ultrasonic robot has other active degrees of freedom based on local feedback, and the control mode of the degrees of freedom is the same as that of the semi-autonomous ultrasonic robot. I.e. real-time adaptive control is achieved by means of the force/moment sensor or the artifact information of the ultrasound image as feedback. The purpose of realizing self-adaption is to enable the ultrasonic probe to always keep fit with the skin, so that the imaging effect of the algorithm is stable. The degrees of freedom controlled in this patent using such control means are the ultrasound probe divide The other two degrees of freedom are used for counteracting the deflection moment possibly applied by the ultrasonic probe in each direction.

Furthermore, the embodiment can also be based on human body parameterized modeling and robot state space establishment, and a medical video auxiliary analysis system and a scanning thought decision-making device are designed, so that the robot can learn the ultrasonic scanning thought of a doctor. The core problem solved by this system is to enable the robot to judge where to start and where to end the scan like a doctor. The embodiment adopts a reinforcement learning mode, and uses the video acquired by the ultrasonic probe in the scanning process of the doctor and the scanning track of the ultrasonic probe to train the robot. And recording data in the scanning process during the analysis of the ultrasonic doctor on the modeling-processed human body data model, and using the data to train the reinforcement learning of the robot. Through the reinforcement learning method, the robot can make dynamic decisions during scanning, and self-adaptive autonomous control is realized by combining force sensor information, so that ultrasonic scanning of a human body can be intelligently completed, and the robot has stronger generalization capability for different diseases.

In particular, the robot state space established by the present embodiment is dense, the scanning process is continuous, and the dense space and continuous scanning process can cause a lot of difficulties in the training of the ultrasound image, because the present embodiment requires training an evaluator that can generate evaluation results for all positions and images of the human body to realize real-time feedback based on the analysis results. To reduce the difficulty of training, the present embodiment desirably uses a discretization method to compress the feasible region of the possible operation space of the robot as much as possible. Since the doctor will mark the position of the found abnormality, this embodiment will design a marking tool to help the doctor record the position of the person who finds the abnormality and record the image of the abnormality at that position. Clinically, a doctor scans different patients to generate a large number of positioning results and clinical images of abnormal conditions, and the embodiment clusters the different positioning results to separate a main scanning position and a secondary scanning position in the body surface of the human body, wherein the main scanning position approximately accords with a focus area recorded in a medical tool book, and the secondary scanning position is also important (because the secondary scanning position records some special disease positions).

Sufficient sample data to be acquired by an autonomous ultrasound robot is provided to the robot for training. The information acquired by the ultrasonic probe can be used by the robot after being processed in advance. In the case of an autonomous ultrasound robot, although the robot learns the analysis skills of the operator, the robot still has a great deal of difference from the processing of the acquired image information, and the present embodiment analyzes the ultrasound image so that the information can be used by the robot. The present embodiment uses FCN (fully convolutional networks) networks for semantic segmentation of ultrasound images. Semantic segmentation is to classify the types of pixels in each part of the judgment image into different types of pixels. The difference is that the semantic meaning is represented by different colors, as shown in fig. 6, fig. 6 is a semantic analysis of a renal ultrasound image.

The FCN is characterized in that both the input and the output are two-dimensional images and the input and the output have corresponding spatial structures, in which case the present embodiment can treat the output of the FCN as a map of heat, with the heat indicating the location and coverage of the object to be detected. A higher heat is displayed in the area where the object is located and a lower heat is displayed in the background area, which can also be seen as classifying each pixel point on the image, whether this point is located on the object to be detected. The first step of FCN is to convolve the image in a full sense to preserve spatial information on the image to the maximum extent, while the picture at output is still a two-dimensional picture. The FCN network pair then performs an up-sampling operation, where the image is scaled down several times after the full convolution operation, and restored to its original size after the up-sampling operation. Finally, the image is optimized using the neglected connection structure, since the result is very coarse if the result after the full convolution is directly up-sampled. The image semantic analysis process is shown in fig. 6.

The embodiment adopts the mapping relation between the image result and the analysis thought of the gradient PPO algorithm, wherein the PPO algorithm is an enhanced learning algorithm, can obtain specific probability distribution of (state, action) relation pairs, and can process continuous action problems. The PPO network uses a model of normal distribution, and the probability of various reactions that occur in a continuous motion is represented by the mean μ and the variance σ. Taking the normalized human body digital model structure as an environment, taking the robot as a main body, marking body surface sign positioning points selected by doctors as benefits, taking the state of the robot as input, taking actions taken by the robot as output, and establishing a learning model of the robot. After training, when the robot encounters each condition, a probability density function reaching the next state can be obtained, so that an autonomous decision is realized, and the method of an operator is like to flexibly change according to the information of an ultrasonic image.

In order to realize reinforcement learning of the robot, when the embodiment uses a PPO (Proximal Policy Optimization, deep reinforcement learning) algorithm, an 'actor commentator' algorithm system is required to be introduced, in reality, the commentators comment on the actors according to the performances of the actors, and the actors improve the performances of the actors according to the comments of the commentators, so that the algorithm system used in the embodiment is the same. Specifically, as shown in fig. 7, the present embodiment inputs environmental information into an "actor" algorithm network to obtain two values, wherein the two values are respectively a normal distribution mean and variance representing a next reaction of the robot, a "reaction" is obtained through the constructed normal distribution sample, then the normal distribution sample is input into the environment to obtain "rewards" and a next "state", then the previously obtained "states", "actions", "rewards" are stored as a set of data, then the "states" are input into a new "actor" network, and the steps of inputting environmental information and storing a set of data are circulated until a certain amount of "states", "actions", "rewards" data sets are stored, and the "new actor network" network is not updated in the process. Inputting the status obtained after the circulation in the process into a criticism network, and calculating discount rewards; all stored combinations of states are input into a criticism network to obtain V values of all states, and then the criticism network is updated by back propagation of the two values. All stored state combinations are input into an original 'actor' network and a new 'actor' network to obtain a normal distribution, all stored 'actions' are combined into an 'action' set to be input into the normal distribution, and the distribution corresponding to each 'action' is obtained. Based on this distribution, actions are selected, acting on the environment, the environment feeding back the next state, etc. And then back-propagating, updating the 'actor' network, and cycling through the whole PPO algorithm steps for a plurality of times until training is finished.

Further, the embodiment further uploads the scanning result and the data output by the video auxiliary analysis system and the scanning idea decision maker to a cloud database, so as to package the scanning result and the data output by the video auxiliary analysis system and the scanning idea decision maker, and analyze the target portion.

Specifically, the present embodiment provides a big data platform, whose architecture is shown in fig. 8. According to the embodiment, a plurality of single ultrasonic analysis devices are connected into the cloud platform, so that data standardization, data archiving storage and real-time sharing are realized. The remote ultrasonic system is used for realizing a series of applications of remote ultrasonic, including remote analysis and remote teaching, and supporting the real-time online analysis of experts.

In order to achieve the data sharing concept, the invention provides a solution of communicating and interconnecting a plurality of ultrasonic scanning equipment terminals, and the single-machine stored data is instantly stored and recalled in different places and is archived and analyzed through a communication technology and a cloud server. The ultrasonic scanning image stores data in a general format so as to facilitate scanning quantitative information extraction off-line or off-site. In order to facilitate later-stage big data analysis, the platform system needs to perform standardization processing on the data. And carrying out ultrasonic analysis on a patient by using an ultrasonic system connected with a platform, carrying out ultrasonic scanning strictly according to an international standardization ultrasonic scanning flow in the analysis process, recording scanning tracks on the body of a person to be tested and stay time of each body area by various sensors on an ultrasonic probe in real time, uniformly uploading the data to a cloud platform, and carrying out standardization processing on the data by the cloud platform. The autonomous ultrasonic scanning robot trained by the autonomous scanning algorithm of the robot performs scanning analysis on a person to be detected, and the ultrasonic probe, various force sensors, depth cameras and other devices are arranged on the end effector of the robot, so that the state information of the robot in the scanning process can be reflected in real time. By connecting the autonomous ultrasonic robot with the cloud platform, the executed scanning track is saved and uploaded to the cloud platform each time the scanning task is completed, as shown in fig. 9.

In specific implementation, after the traditional ultrasonic scanning is used by an operator, the acquired picture needs to be manually calibrated, the calibration process is slightly complicated, the cloud platform data input of the platform does not need the operator to calibrate, the cloud platform performs centralized processing on the scanned data, the whole ultrasonic analysis process is uploaded in a video form, and in the scanning process, the operation of the operator can be divided into a plurality of sections, and the carotid ultrasonic analysis is taken as an example. Firstly, a doctor needs to place a probe near an carotid artery region on the neck of a person to be examined; and secondly, the doctor further adjusts the image acquired by the prior knowledge and the image in the ultrasonic probe until an image which can be analyzed appears, and if the doctor can analyze the image only by needing more ultrasonic image information, the doctor needs to further adjust the angle of the probe until the image which can be analyzed appears. In the traditional ultrasonic analysis, a doctor needs to manually mark on an ultrasonic image to determine what each part is in the image, and in the method provided by the invention, the corresponding position can be found by only looking at different paragraphs of a video, so that the complicated step of marking is omitted.

When the ultrasonic analysis method is specifically applied, a doctor performs ultrasonic analysis by using an ultrasonic system connected with the cloud platform, the scanning flow of the doctor scans according to the standardized ultrasonic scanning flow, and the whole scanning process is recorded by the ultrasonic system. The medical analysis information acquired by the ultrasonic probe is uploaded to a cloud database, the data which is uploaded are packaged by the database in the cloud database, the data are changed into useful medical information, for example, the scanning track and the scanning duration of each doctor in ultrasonic scanning are matched to a human body digital model, and in the human body digital model matched with the medical information, the sequence of the whole analysis and the focus of the patient can be intuitively seen, and the analysis focus of the doctor is on the focus of the patient. The whole scanning flow of the doctor is stored and uploaded in a video form, video data intuitively show how the doctor scans a certain position, an area for imagining to place the probe is found according to medical knowledge, the area can be defined as a positioning area, after the positioning area is found, the doctor further adjusts according to ultrasonic images until judging images appear, the whole process is stored in a video form, and the doctor can easily adjust a progress bar of the video to view the images in the ultrasonic probe at corresponding time. The packaged medical information is stored in a cloud database of a central expert system, and each group of stored medical information can be called out at any time. Medical data in a central expert system can be read from an ultrasonic robot platform in each place, medical information in the central expert system can be used for training medical staff just contacted with ultrasonic analysis or providing an intuitive analysis template for the medical staff to learn, an expert doctor can train the skills of ultrasonic scanning for the medical staff positioned in different areas by means of the system, data generated in the training process can also be transmitted back to a cloud database, the re-enrichment of the information in the database is realized, the medical information in the database can be transmitted to each ultrasonic robot connected with the cloud platform, the life learning of the ultrasonic robot is realized by utilizing the transmitted data to realize and strengthen the learning method, and the ultrasonic robot can truly perform super analysis like a doctor and can also continuously update the analysis mode of the ultrasonic robot.

Further, in the present embodiment, three-dimensional reconstruction and slicing of medical images are important medical aids. Three-dimensional reconstruction can provide more intuitive three-dimensional information to the physician, while slicing can provide information about the organ at a particular viewing angle. Therefore, three-dimensional image reconstruction and slicing are very important contents. Based on the requirement of clinical reconstruction and segmentation of abnormal tissues, the invention researches how to segment, reconstruct and reconstruct the lesion tissue of the focal zone imaging specialty, and then slice the reconstructed model, and further performs ultrasonic diagnosis by combining with the imaging characteristics of the focal zone.

Ultrasound image lesion tissue segmentation is a challenging task in three-dimensional reconstruction procedures. Because of the blurring of image edges and the weak boundary diffusion effect between adjacent surfaces caused by the imaging principle of ultrasonic images, high-precision tissue segmentation is difficult to achieve by a traditional threshold-type-based method. For this case, an adjacent voxel segmentation method is used to process the image. Firstly, tracking of pathological tissue surface voxels is achieved through iterative self-adaptive reclassification based on ultrasonic medical images, a preliminary segmented pathological tissue interface is obtained, then the normal direction of the preliminary interface is optimized by combining Gaussian standard deviation, tracking in the three-dimensional direction is achieved, estimation of the boundary normal direction is improved, therefore robustness and segmentation accuracy of an overall algorithm are improved, segmentation and extraction of pathological tissue are completed, and an optimization basis is provided for reconstruction of pulmonary pathological tissue.

Since the three-dimensional reconstruction accuracy of the lesion tissue has a great influence on the diagnosis and evaluation of diseases by the robot, the selection of the reconstruction algorithm is critical to the diagnosis of the robot. For this case we use a moving cube algorithm to reconstruct the tissue. The basic idea of three-dimensional reconstruction is to resample a data structure describing the three-dimensional space according to the relative position between two-dimensional ultrasound pictures and interpolate the blank positions in the data structure. The basic idea of the algorithm is as follows: firstly, an algorithm establishes a three-dimensional space coordinate system and plans a cuboid enveloping frame capable of accommodating a target organ in space; secondly, the algorithm cuts the cuboid enveloping frame into a plurality of volume pixels with equal volume according to the resolution of the target image, and finds the volume pixels corresponding to the pixel points in each ultrasonic image according to the relative position information of the ultrasonic fault picture. Furthermore, the algorithm judges whether 8 vertexes of each small cube are respectively positioned in the target organ according to the segmentation result obtained in the adjacent voxel tracking method, and generates internal isosurfaces of the body pixels according to the vertexes (256 different situations are all provided); finally, the algorithm combines the voxels into the target organ whose isosurface will constitute the surface of the target organ.

Taking pulmonary ultrasound as an example, a video recorded by a scanning ultrasonic probe of a doctor is derived, and an ultrasonic image of each frame is segmented into abnormal tissues by using an adjacent voxel tracking method. Tracking of abnormal tissue surface voxels is achieved through iterative self-adaptive reclassification, a preliminarily segmented lesion tissue interface is obtained, then the normal direction of the preliminary interface is optimized by combining with Gaussian standard deviation, tracking of the three-dimensional direction is achieved, estimation of the boundary normal direction is improved, and finally abnormal region extraction of an ultrasonic image is achieved. And then carrying out three-dimensional reconstruction on the three-dimensional reconstruction by a moving cube algorithm.

Based on the above embodiment, the present invention further provides an ultrasonic robot state normalization system based on human body characteristics, as shown in fig. 10, the system includes:

the model initializing unit 10 is configured to initialize a preset human body digital twin model based on a human body parameterized vector according to a scanning result of the ultrasonic robot on the sign of the person to be detected;

a calibration updating unit 20, configured to perform calibration updating on the digital twin model of the human body based on an ultrasound image standard image;

and the parameterization unit 30 is used for carrying out normalized representation on the mechanical arm track of the human body digital twin model based on the chessboard grid so as to realize parameterization on human body characteristics.

Further, the system further comprises: the dynamic updating unit is used for dynamically updating the human body digital twin model in real time according to the explicit positioning points in the human body ultrasonic image, wherein the explicit positioning points are used for reflecting the human body surface characteristics in the human body ultrasonic image.

Further, the model initializing unit includes: the identification subunit is used for identifying the explicit positioning points on the body surface of the person to be detected according to the scanning result of the sign of the person to be detected by the ultrasonic robot; a parameter vectorization subunit, configured to obtain the human body parameterized vector based on a relative relationship between the explicit positioning points; and the initialization subunit is used for initializing the human body digital twin model based on the human body parameterized vector.

Further, the system further comprises: and the working space establishing unit is used for establishing the state working space of the ultrasonic robot based on the human body digital twin model.

Further, the system further comprises: and the self-adaptive control unit is used for realizing real-time self-adaptive control by taking artifact information of the force/moment sensor or the ultrasonic image as feedback.

In summary, the invention discloses a human body characteristic-based ultrasonic robot state normalization method and system, wherein the method comprises the following steps: initializing a preset human body digital twin model based on a human body parameterized vector according to a scanning result of the physical sign of the person to be detected by the ultrasonic robot; calibrating and updating the human body digital twin model based on an ultrasonic image standard image; and carrying out normalized representation on the mechanical arm track of the human body digital twin model based on the chessboard grid so as to realize parameterization of human body characteristics. The invention can realize vector and normalized representation of human body characteristic parameters, thereby realizing human body characteristic parameterization so as to analyze human body characteristics.

It is to be understood that the invention is not limited in its application to the examples described above, but is capable of modification and variation in light of the above teachings by those skilled in the art, and that all such modifications and variations are intended to be included within the scope of the appended claims.

Claims

1. An ultrasonic robot state normalization method based on human body characteristics, which is characterized by comprising the following steps:

carrying out normalized representation on the mechanical arm track of the human body digital twin model based on the chessboard grid so as to realize parameterization on human body characteristics;

the initializing the preset human body digital twin model based on the human body parameterized vector according to the scanning result of the physical sign of the person to be detected by the ultrasonic robot comprises the following steps:

according to the scanning result of the physical sign of the person to be detected by the ultrasonic robot, identifying an explicit positioning point of the body surface of the person to be detected, wherein the explicit positioning point is used for reflecting the body surface characteristics of the human body in the human body ultrasonic image;

initializing the human body digital twin model based on the human body parameterized vector;

the calibration updating of the human body digital twin model based on the ultrasonic image standard image comprises the following steps:

extracting absolute coordinates of explicit positioning points of a human body by using a camera, and obtaining a chessboard grid of the body surface contour of the human body by connecting coordinates of adjacent angular points, wherein each angular point of the chessboard grid is provided with three-dimensional space coordinate information of the body surface of the human body;

For a specific ultrasonic scanning task, using topological lines to represent relative position information between bones and organs of a human body;

positioning topological lines according to priori human body information, and expressing the topological lines as position coordinates relative to chessboard angular points;

in the beginning stage of scanning, the mechanical arm determines the scanning position on the body surface of the human body by means of medical priori knowledge provided by a digital twin body of the human body;

after the probe contacts the human body, the robotic arm will obtain posterior information about the location of the anchor point;

the mechanical arm updates the relative coordinates of the human body digital twin body model relative to the body surface chessboard by using posterior information;

when updating the relative coordinates of a certain positioning point, the model uses the correlation information of the human body model to update and optimize the position information of the rest positioning points;

the normalized representation of the manipulator track of the human body digital twin model based on the chessboard grid comprises the following steps:

the positions of the tracks of the mechanical arm in the chessboard of the human body are uniquely represented, any discrete point in the surface path of the chessboard during the scanning process of the mechanical arm is represented as the linear combination of four angular points close to the discrete point in the chessboard, and any motion track of the mechanical arm is uniquely represented as the linear combination of the angular point coordinates of the chessboard.

2. The human feature-based ultrasound robot state normalization method of claim 1, wherein the updating the calibration of the human digital twin model based on the ultrasound image standard image further comprises:

and carrying out real-time dynamic update on the human body digital twin model according to the explicit positioning point in the human body ultrasonic image.

3. The human feature-based ultrasound robot state normalization method of claim 1, wherein the updating the calibration of the human digital twin model based on the ultrasound image standard image further comprises:

4. The human feature-based ultrasound robot state normalization method of claim 1, wherein the normalizing representation of the human digital twin model based on the checkerboard grid further comprises:

the real-time self-adaptive control of the ultrasonic robot is realized by taking artifact information of a force/moment sensor or an ultrasonic image as feedback.

5. An ultrasonic robot state normalization system based on human body characteristics, the system comprising:

the parameterization unit is used for carrying out normalized representation on the mechanical arm track of the human body digital twin model based on the chessboard grid so as to realize parameterization on human body characteristics;

the model initializing unit includes:

the identification subunit is used for identifying explicit positioning points on the body surface of the person to be detected according to the scanning result of the physical sign of the person to be detected by the ultrasonic robot, and the explicit positioning points are used for reflecting the body surface characteristics of the human body in the human body ultrasonic image;

an initialization subunit, configured to initialize the human body digital twin model based on the human body parameterized vector;

the calibration updating unit includes:

the parameterization unit comprises:

6. The human feature based ultrasound robot state normalization system of claim 5, wherein the calibration update unit further comprises:

and the dynamic updating unit is used for dynamically updating the human body digital twin model in real time according to the explicit positioning point in the human body ultrasonic image.

7. The human feature based ultrasound robot state normalization system of claim 5, wherein the calibration update unit further comprises:

8. The human feature based ultrasound robot state normalization system of claim 5, wherein the parameterization unit further comprises:

and the self-adaptive control unit is used for realizing the real-time self-adaptive control of the ultrasonic robot by taking the artifact information of the force/moment sensor or the ultrasonic image as feedback.