A method for dynamically correcting a target point under respiration

Abstract

The present application relates to medical intervention surgery navigation technology, and discloses a kind of dynamic correction method of puncture target under breathing dynamics, the method first obtains real-time pose and dynamic needle insertion point of puncture needle, analyzes needle tip motion state and effective puncture depth;Second, in combination with the mapping inside and outside tissue and viscoelastic phase lag characteristic, utilize surface substitute signal to predict dynamic target coordinates under no puncture interference and analyze feedforward speed;Subsequently, based on delay feedback blocking algebraic ring, tissue deformation is decoupled into radial volume row and axially frictional drag activated by speed smooth activation, and the deformation offset vector is obtained by physical limiting amplitude;Then, the predicted coordinates and deformation offset are superimposed to generate correction coordinates, and spatial tracking error is extracted;Finally, the speed command is generated by fusing feedforward tracking, position closed-loop error elimination and active impedance dissipation component.The present application improves nonlinear deformation compensation and suppresses actuator chattering, which helps to improve dynamic puncture accuracy.

Description

Technical Field

[0001] This invention pertains to medical interventional surgical navigation technology, and particularly relates to a method for dynamic correction of puncture target points under respiratory dynamics. Background Technology

[0002] During thoracic and abdominal interventional surgeries, the periodic displacement of the target point caused by respiratory motion severely interferes with puncture accuracy. Existing navigation systems generally rely on preoperative static images or surface respiratory signals for target location prediction, their core logic based on the simplified assumption that the in vivo target point and surface displacement are linearly and synchronously mapped. However, the viscoelastic nature of human soft tissue leads to a significant difference between the in vivo target trajectory and the surface signal: on the one hand, the spatial distribution of the main respiratory direction and the secondary deformation direction is inconsistent, making it impossible to accurately represent the target displacement vector by linear scaling in a single direction; on the other hand, the inherent phase lag characteristic of tissue causes a time delay in the movement of the in vivo target point relative to the surface signal, and existing open-loop prediction methods continuously accumulate positioning errors because they ignore this dynamic characteristic. More importantly, after the puncture needle penetrates the skin and enters the tissue, the interaction between the needle and the soft tissue triggers two types of nonlinear deformation effects—the volume displacement effect caused by radial compression and the dragging effect caused by axial friction. Both of these effects cause the target position to deviate from the respiratory trajectory without puncture interference. Current technical solutions treat the puncture needle simply as a passive tracker, failing to establish a decoupled model of needle-tissue contact mechanics. They neither distinguish the independent action mechanisms of radial pushing and axial friction, nor do they have the ability to dynamically compensate for the causal activation characteristics of needle insertion speed. This results in a continuous increase in spatial error, sluggish system response, and abnormal oscillation of the actuator during the needle tip approaching the target point, ultimately restricting the accuracy and clinical safety of the puncture operation. Summary of the Invention

[0003] The purpose of this invention is to provide a method for dynamic correction of puncture target under respiratory dynamics, aiming to solve at least one of the above-mentioned problems.

[0004] This invention is implemented as follows: a method for dynamic correction of puncture target points under respiratory dynamics, comprising the following steps: S10, acquiring real-time three-dimensional pose data of the puncture needle and dynamic needle insertion point coordinates on the skin surface driven by respiratory motion, and analyzing the three-dimensional physical motion velocity of the needle tip, the effective puncture depth, and the instantaneous needle insertion velocity after low-pass filtering; S20, acquiring the surface respiratory substitute signal, filtering out differential noise based on a state estimator, and predicting the dynamic target point coordinates and analyzing the feedforward velocity under no puncture interference by combining the amplitude mapping relationship between soft tissue and the viscoelastic phase hysteresis characteristics; S30, establishing a spatial phase based on the target point real coordinate feedback from the previous control cycle. For the position vector, the soft tissue deformation model is decoupled into radial pushing volume displacement and axial friction drag activated by the instantaneous speed of needle insertion. After physical limiting processing, the tissue deformation offset vector is calculated. S40, the dynamic target point coordinates without puncture interference are superimposed with the tissue deformation offset vector to obtain the corrected coordinates of the target point in the current cycle, and the spatial tracking error vector is extracted by combining the current three-dimensional coordinates of the needle tip. S50, based on the analytical feedforward speed and the spatial tracking error vector, and by introducing an active impedance control term composed of the deviation between the three-dimensional physical motion speed of the needle tip and the analytical feedforward speed, a dynamic correction speed command for the needle tip is generated to drive the actuator to move.

[0005] In a further technical solution, the analytical mechanism for obtaining the effective puncture depth in step S10 includes: based on rigid body kinematics, combining the pose of the needle tail sensor and the physical effective nominal length of the puncture needle, mapping and obtaining the real-time three-dimensional position coordinate vector of the needle tip and the real-time attitude direction unit vector of the needle tip; based on the relative deviation between the real-time three-dimensional position coordinate vector of the needle tip and the dynamic needle insertion point coordinates on the skin surface, and projecting it onto the real-time attitude direction unit vector of the needle tip, extracting the non-negative effective puncture depth.

[0006] In a further technical solution, the analytical mechanism for obtaining the three-dimensional physical motion speed of the needle tip and the instantaneous speed of needle insertion after low-pass filtering in step S10 includes: extracting the effective puncture depth and the real-time three-dimensional position coordinate vector of the needle tip based on the rate of change in the time domain, and introducing state observation or low-pass filtering mechanism to eliminate signal boundary jumps and differential high-frequency noise, thereby outputting the instantaneous speed of needle insertion and the three-dimensional physical motion speed of the needle tip respectively.

[0007] In a further technical solution, the mechanism for predicting the dynamic target coordinates without puncture interference in step S20 includes: superimposing the in vivo and in vivo displacement mapping components in the main respiratory direction on the three-dimensional coordinate vector of the target resting state, and subtracting the phase lag displacement component in the secondary lag direction to synthesize the dynamic target coordinates without puncture interference; wherein, the phase lag displacement component is generated by the instantaneous velocity of the surface respiratory substitution signal being constrained by the tissue lag time constant and the in vivo and in vivo amplitude coupling coefficient.

[0008] In a further technical solution, the mechanism for predicting the analytical feedforward velocity in step S20 includes: performing time-domain analytical derivation of the mechanism for synthesizing dynamic target coordinates, generating a feedforward velocity component in the main respiratory direction using the instantaneous velocity of the surface respiratory substitute signal, and generating a feedforward velocity component in the secondary lag direction using the instantaneous acceleration of the surface respiratory substitute signal, and superimposing the two to generate the analytical feedforward velocity.

[0009] In a further technical solution, the basic decoupling mechanism for calculating the tissue deformation offset vector in step S30 includes: using the correction coordinates with a single sampling period delay as the target point's true coordinate feedback in the previous control period, subtracting them from the current real-time three-dimensional position coordinate vector of the needle tip, and constructing a spatial relative position vector containing time delay memory, thereby blocking the algebraic loop within the same sampling period; orthogonally decomposing the spatial relative position vector relative to the unit vector of the real-time attitude direction of the needle tip to obtain an axial projection scalar and a radial perpendicular vector; the initial tissue deformation offset vector is constructed as a vector sum of a radial pushing volume displacement term and an axial friction drag term, wherein: the action direction of the radial pushing volume displacement term is parallel to the radial perpendicular vector, and the action direction of the axial friction drag term is parallel to the unit vector of the real-time attitude direction of the needle tip.

[0010] A further technical solution includes the following constraint rules and amplitude limiting mechanisms for the radial pushing volume displacement term and the axial friction drag term: the amplitude of the radial pushing volume displacement term decreases nonlinearly with the increase of the magnitude of the radial perpendicular vector, and the physical geometric scale of the puncture needle is introduced as a regularization constraint in the denominator of the amplitude decay calculation to eliminate the computational singularity when the radial distance approaches zero; the amplitude of the axial friction drag term is constrained by the soft tissue friction coupling coefficient and the effective puncture depth and decreases with the relative spatial distance; in addition, the axial friction drag term adopts a smooth state activation mechanism with the instantaneous speed of needle insertion as the independent variable for direction and amplitude modulation to ensure that the drag effect conforms to the causal mechanical direction of contact friction; the generated initial tissue deformation offset vector is subjected to amplitude saturation limiting operation based on the maximum yield limit of the tissue to cut off the discrete recursive positive feedback oscillation caused by time delay memory, and the final tissue deformation offset vector is output.

[0011] In a further technical solution, in step S40, the corrected coordinates of the target point within the current cycle are obtained, and the spatial tracking error vector is extracted by combining the current three-dimensional coordinates of the needle tip. The extraction mechanism includes: performing spatial vector addition and superposition of the dynamic target point coordinates under no puncture interference and the tissue deformation offset vector, and using the superposition result as the corrected coordinates of the target point under the current sampling cycle, and using it as the historical state input for the next control cycle for iteration; performing spatial vector subtraction and difference between the corrected coordinates and the current real-time three-dimensional position coordinate vector of the needle tip to obtain the spatial tracking error vector that the system needs to eliminate.

[0012] In a further technical solution, step S50, the mechanism for generating the needle tip dynamic correction speed command includes: the needle tip dynamic correction speed command is synthesized by a feedforward tracking component, an error elimination component, and an active impedance dissipation component; wherein: the feedforward tracking component is directly mapped to the analytical feedforward speed, used to actively match respiratory dynamics; the error elimination component is obtained by multiplying the spatial tracking error vector with the proportional position stiffness gain of the system; the active impedance dissipation component is obtained by multiplying the difference vector between the three-dimensional physical motion speed of the needle tip and the analytical feedforward speed with the active relative damping gain of the system, and this component is introduced into the synthesis operation in the form of negative feedback to provide directional damping and smooth out mechanical chattering during the puncture process.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention breaks through the limitation of existing navigation technology that simply regards target displacement as linear synchronous motion. It innovatively decouples tissue deformation into radial pushing and pushing and axial friction dragging that is smoothly activated by the needle insertion speed. By introducing the delayed feedback of the previous control cycle, it completely blocks the algebraic loop in digital control, making the model's prediction of the target correction coordinates more rigorous and robust in a physical sense.

[0014] 2. In the calibration speed command generation stage, in addition to introducing feedforward tracking and position closed loop, this scheme innovatively constructs an active impedance dissipation mechanism with the deviation between the three-dimensional physical motion speed of the needle tip and the analytical feedforward speed as variables. This directional damping mechanism can smoothly respond to sudden tissue interaction torques, which is beneficial to suppressing overshoot and high-frequency oscillations generated by the actuator when approaching the target point at high speed or encountering non-uniform tissue interfaces.

[0015] 3. When predicting interference-free dynamic target points, this method not only considers the amplitude mapping of the main respiratory direction, but also introduces the tissue hysteresis time constant and the displacement component of the secondary hysteresis direction to conduct a prospective mathematical model of the viscoelastic hysteresis behavior of soft tissue, thereby eliminating the dynamic positioning error that accumulates continuously due to signal time delay in the traditional open-loop tracking system. Attached Figure Description

[0016] Figure 1 A flowchart of a method for dynamic correction of puncture target under respiratory dynamic conditions provided by the present invention; Figure 2 The structural diagram of the tissue deformation offset vector decoupling and limiting mechanism provided by the present invention is shown. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0018] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0019] like Figure 1 As shown, an embodiment of the present invention provides a method for dynamic correction of puncture target points under respiratory dynamics, comprising the following steps: S10: Acquire real-time three-dimensional pose data of the puncture needle and dynamic needle insertion point coordinates on the skin surface driven by respiratory motion. Analyze this data to obtain the three-dimensional physical velocity of the needle tip, the effective puncture depth, and the instantaneous insertion velocity after low-pass filtering. The real-time three-dimensional pose data of the puncture needle represents its position and orientation information in three-dimensional space, typically acquired in real-time via an external tracking device. This data is fundamental for determining the needle tip position and orientation. The dynamic needle insertion point coordinates on the skin surface refer to the position where the puncture needle enters the patient's skin surface; this position changes periodically due to the patient's respiratory motion. The effective puncture depth represents the actual distance the needle tip penetrates into the soft tissue. This depth is an important parameter for assessing the degree of needle-tissue interaction.

[0020] S20 acquires the surface respiratory substitute signal and filters out differential noise based on the state estimator. Combining the amplitude mapping relationship between soft tissue and intra- and extra-soft tissue and the viscoelastic phase hysteresis characteristic, it predicts the dynamic target coordinates and analytical feedforward velocity without puncture interference. The surface respiratory substitute signal is a signal acquired through surface sensors (such as breathing belts, infrared sensors, etc.) that indirectly reflects the patient's respiratory motion state. This signal is used to predict the motion of the target point within the body. The state estimator is a signal processing module used to process the surface respiratory substitute signal, such as filtering out measurement noise and estimating parameters like displacement, velocity, and acceleration of respiratory motion, thereby providing more stable and accurate respiratory motion information. The amplitude mapping relationship between soft tissue and intra- and extra-soft tissue and the viscoelastic phase hysteresis characteristic describe the complex nonlinear relationship between surface respiratory motion and intra-body target motion. The amplitude mapping relationship reflects the proportional relationship between the amplitudes of surface and intra-body motion, while the viscoelastic phase hysteresis characteristic characterizes the time delay of the intra-body target motion relative to the surface signal. The dynamic target coordinates without puncture interference represent the real-time three-dimensional position of the target point predicted solely by respiratory motion, without the puncture needle entering the tissue. The analytical feedforward velocity is the pre-calculated velocity of the target point based on the predicted dynamic target trajectory without puncture interference. This velocity is used to proactively guide the movement of the puncture needle.

[0021] S30, based on the target point's true coordinate feedback from the previous control cycle, establish a spatial relative position vector, decouple the soft tissue deformation model into radial pushing volume displacement and axial friction drag activated by the instantaneous speed of the needle insertion, and calculate the tissue deformation offset vector after physical limiting processing; the tissue deformation offset vector represents the additional displacement of the target point position relative to the undisturbed state caused by the interaction between the needle body and the surrounding soft tissue, such as radial compression and axial friction drag, when the puncture needle moves inside the tissue.

[0022] S40, superimpose the dynamic target coordinates under no puncture interference with the tissue deformation offset vector to obtain the corrected coordinates of the target in the current cycle, and extract the spatial tracking error vector by combining the current three-dimensional coordinates of the needle tip. The S50, based on analytical feedforward velocity and spatial tracking error vector, introduces an active impedance control term consisting of the deviation between the three-dimensional physical motion velocity of the needle tip and the analytical feedforward velocity. This generates a dynamic correction velocity command for the needle tip to drive the actuator. The active impedance control term is a control strategy that, by simulating mechanical impedance characteristics, generates an additional control force or velocity component based on the deviation between the actual movement velocity of the needle tip and the target feedforward velocity. This adjusts the interaction between the needle tip and the tissue, improving the stability and safety of tracking. The dynamic correction velocity command for the needle tip is the final real-time velocity command generated by the system to drive the actuator (e.g., a robotic arm) to accurately track the dynamic target point with the puncture needle tip.

[0023] This embodiment provides a method for dynamic correction of puncture target points under dynamic respiratory conditions, the specific implementation of which is as follows: First, real-time three-dimensional pose data of the puncture needle and dynamic needle insertion point coordinates on the skin surface driven by respiratory motion are acquired. The three-dimensional physical motion velocity of the needle tip, the effective puncture depth, and the instantaneous insertion velocity after low-pass filtering are then analyzed. Specifically, the real-time three-dimensional pose data of the puncture needle can be obtained by attaching optical markers to the puncture needle and using an external optical tracking system for real-time capture and calculation. The dynamic needle insertion point coordinates on the skin surface can be obtained by attaching a traceable marker to the needle insertion area on the patient's skin and monitoring it in real-time using the same optical tracking system. The three-dimensional physical motion velocity of the needle tip can be obtained by performing time differentiation on the real-time acquired needle tip pose data and smoothing it using an appropriate digital filter. The effective puncture depth can be obtained by calculating the projection of the relative distance between the real-time position of the needle tip and the dynamic needle insertion point on the skin surface onto the needle axis. The instantaneous insertion velocity can be obtained by performing time differentiation on the effective puncture depth and then applying a one-dimensional low-pass filter to eliminate high-frequency noise.

[0024] Secondly, the surface respiratory substitution signal is acquired, and differential noise is filtered out based on a state estimator. Combining the amplitude mapping relationship between soft tissue interior and exterior and the viscoelastic phase hysteresis characteristics, the dynamic target coordinates and analytical feedforward velocity under puncture-free interference are predicted. Specifically, the surface respiratory substitution signal can be obtained by measuring the circumference change of a breathing belt sensor worn on the patient's chest and abdomen. This signal is then input to a state estimator, such as a Kalman filter, to estimate the displacement, velocity, and acceleration of respiratory motion, effectively filtering out differential noise generated during the measurement process. Based on this, a pre-established respiratory motion model reflecting the amplitude mapping relationship between soft tissue interior and exterior and the viscoelastic phase hysteresis characteristics is used to predict the dynamic coordinates of the target under puncture-free interference. This model can be personalized based on preoperative imaging data (such as 4DCT). The analytical feedforward velocity is obtained by calculating the time derivative of the predicted dynamic target coordinates under puncture-free interference.

[0025] Furthermore, a spatial relative position vector is established based on the target's true coordinate feedback from the previous control cycle. This decouples the soft tissue deformation model into radial pushing volume displacement and axial friction drag, causally activated by the needle insertion velocity. After physical limiting, the tissue deformation offset vector is calculated. Specifically, the spatial relative position vector is established by vector subtraction of the target's corrected coordinates obtained in the previous control cycle and the real-time three-dimensional position coordinates of the needle tip at the current moment. The soft tissue deformation model is decoupled into two main components: a radial pushing volume displacement component, which simulates the radial compression effect of the puncture needle on the surrounding tissue, related to the needle radius and radial relative displacement; and an axial friction drag component, which simulates the friction between the needle and tissue during axial movement, influenced by the needle insertion velocity and puncture depth. The calculated original tissue deformation offset vector is then physically limited to ensure its amplitude is within a reasonable physical range, preventing unrealistically large deformation estimates.

[0026] Subsequently, the dynamic target coordinates under puncture-free interference are superimposed with the tissue deformation offset vector to obtain the corrected coordinates of the target within the current cycle. The spatial tracking error vector is then extracted by combining this with the current three-dimensional coordinates of the needle tip. Specifically, the corrected coordinates of the target at the current moment are obtained by vector superimposing the predicted dynamic target coordinates under puncture-free interference with the calculated tissue deformation offset vector. These coordinates comprehensively consider respiratory motion and needle-tissue interaction. The spatial tracking error vector is obtained by vector subtracting these corrected coordinates from the current three-dimensional coordinates of the puncture needle tip, representing the real-time spatial deviation between the needle tip and the target.

[0027] Finally, based on the analytical feedforward velocity and the spatial tracking error vector, and by introducing an active impedance control term consisting of the deviation between the three-dimensional physical motion velocity of the needle tip and the analytical feedforward velocity, a dynamic correction velocity command for the needle tip is generated to drive the actuator motion. Specifically, the generation of the dynamic correction velocity command integrates feedforward control and feedback control. The analytical feedforward velocity provides predictive guidance for the future motion of the target point, while the spatial tracking error vector provides correction for the current deviation. The active impedance control term dynamically adjusts the control output according to the deviation between the actual physical motion velocity of the needle tip and the analytical feedforward velocity to simulate a virtual mechanical impedance. This allows for a compliant response and suppression of unnecessary vibrations or shocks when the needle tip interacts with the tissue, improving the stability and safety of the puncture. The generated command is then sent to the robotic actuator to precisely control the real-time movement of the puncture needle.

[0028] This method accurately predicts the dynamic position of the target point under respiratory motion by acquiring real-time data on the puncture needle posture and skin insertion point, combined with state estimation of surface respiratory signals and soft tissue viscoelastic properties. Simultaneously, this method decouples and incorporates corrections for radial pushing and axial frictional deformation caused by needle-tissue interaction, effectively compensating for additional deformation during puncture. By introducing active impedance control, the stability and safety of needle tip tracking are further improved, significantly reducing systematic errors, feedback lag, and mechanical chattering problems present in traditional methods, thereby enhancing the puncture accuracy of interventional surgery under respiratory dynamics.

[0029] This application further proposes an analytical mechanism in step S10 for analyzing the three-dimensional physical motion velocity of the needle tip, the effective puncture depth, and the instantaneous needle insertion velocity after low-pass filtering. This mechanism includes: based on rigid body kinematics, combining the pose of the needle tail sensor and the effective nominal length of the puncture needle, mapping and obtaining the real-time three-dimensional position coordinate vector of the needle tip and the real-time attitude direction unit vector of the needle tip; based on the relative deviation between the real-time three-dimensional position coordinate vector of the needle tip and the dynamic needle insertion point coordinates on the skin surface, and projecting it onto the real-time attitude direction unit vector of the needle tip, extracting the non-negative effective puncture depth; extracting the effective puncture depth and the real-time three-dimensional position coordinate vector of the needle tip based on the rate of change in the time domain, and introducing state observation or low-pass filtering mechanism to eliminate signal boundary jumps and differential high-frequency noise, thereby outputting the instantaneous needle insertion velocity and the three-dimensional physical motion velocity of the needle tip respectively.

[0030] The three-dimensional physical velocity of the needle tip, the effective puncture depth, and the instantaneous needle insertion velocity after low-pass filtering are obtained from the analysis, specifically calculated according to the following formula:

[0031]

[0032]

[0033]

[0034]

[0035] In the above formula, This represents the real-time three-dimensional position coordinate vector of the needle tip at time t; This represents the real-time three-dimensional position coordinate vector of the needle tail sensor; This represents the real-time three-dimensional rotation matrix of the needle tail sensor; Indicates the physical effective nominal length of the puncture needle; This represents a unit vector indicating the real-time attitude direction of the needle tip. This represents the real-time three-dimensional coordinate vector of the needle insertion point on the skin surface at time t, driven by respiratory motion. Indicates the actual effective puncture depth; Indicates one-dimensional low-pass filtering The instantaneous effective needle insertion scalar speed after insertion; Indicates three-dimensional low-pass filtering The processed three-dimensional real physical motion velocity vector of the needle tip.

[0036] To avoid differential jumps in the max function when the needle tip just enters or exits the skin, you can... Replace with a smooth activation function, for example Alternatively, a sigmoid weighted function with a transition region can be used, making the effective puncture depth continuously differentiable at the boundary, thus improving the smoothness of subsequent differential calculations.

[0037] Specifically, the real-time three-dimensional position coordinate vector of the needle tip The calculation is performed by using the real-time three-dimensional position coordinate vector of the needle tail sensor. With real-time three-dimensional rotation matrix based on needle tail sensor and the physical effective nominal length of the puncture needle This is achieved by superimposing the calculated relative positions of the needle tips. The effective nominal length of the puncture needle is determined by... This represents a fixed distance from the needle tip sensor mounting position to the needle tip. By rotating this length along the needle body axis (usually the Z-axis in the sensor coordinate system) and adding it to the needle tip sensor position, the real-time three-dimensional position of the needle tip in the global coordinate system can be accurately calculated. This calculation method effectively overcomes the limitation that the sensor cannot be directly mounted on the needle tip, providing precise spatial positioning of the needle tip.

[0038] Real-time attitude direction unit vector of needle tip The calculation is based on the real-time three-dimensional rotation matrix of the needle tail sensor. This is obtained by transforming the needle axis direction (usually a unit vector along the Z-axis) in the local coordinate system to the global coordinate system. This unit vector represents the spatial orientation of the needle tip at the current moment and is crucial for subsequent determination of the puncture direction, calculation of the effective puncture depth, and tissue deformation modeling. By acquiring and updating this vector in real time, the attitude changes of the puncture needle in three-dimensional space can be accurately reflected.

[0039] Actual effective puncture depth The calculation first uses the real-time three-dimensional position coordinate vector of the needle tip. Subtract the time t driven by respiratory movement, and the real-time three-dimensional coordinate vector of the dynamic needle insertion point on the skin surface. This yields the displacement vector of the needle tip relative to the insertion point. Subsequently, this displacement vector is compared with the unit vector of the real-time attitude direction of the needle tip. Perform a dot product operation to obtain the projection of the displacement along the needle tip direction. Finally, through... The function ensures that the effective puncture depth is always a non-negative value; that is, it is calculated as a positive value only when the needle tip enters the skin, and zero otherwise. This calculation method can accurately quantify the actual depth to which the needle tip enters the tissue, providing crucial input for subsequent tissue deformation models and control strategies.

[0040] Instantaneous effective needle insertion scalar speed The calculation first involves determining the actual effective puncture depth. The instantaneous rate of change is obtained by performing a time derivative. Since the original derivative signal may contain high-frequency noise, the derivative result is subjected to a one-dimensional low-pass filter to obtain a smooth and accurate needle insertion speed. Processing. One-dimensional low-pass filtering. It can effectively filter out measurement noise and high-frequency jitter, thereby providing a stable and reliable instantaneous needle feed speed and ensuring the smoothness of subsequent control commands.

[0041] The three-dimensional real physical velocity vector of the needle tip The calculation is based on the real-time three-dimensional position coordinate vector of the needle tip. By performing time differentiation, the instantaneous velocity vector of the needle tip in three-dimensional space is obtained. To eliminate the influence of sensor noise and motion jitter on the velocity calculation, this velocity vector is subjected to a three-dimensional low-pass filter. Processing. Three-dimensional low-pass filtering. This ensures that the obtained three-dimensional real physical motion velocity vector of the needle tip is accurate. It is smooth and accurate, providing precise feedback information for the generation of subsequent motion control and impedance control terms.

[0042] Through the above technical solution, this application provides a method for accurately and robustly analyzing the motion state of a puncture needle. By introducing needle tip position and attitude calculation based on needle tail sensor data and puncture needle physical parameters, the real-time position and orientation of the needle tip in three-dimensional space can be accurately obtained, overcoming the difficulty of directly measuring the needle tip. Furthermore, by projecting and limiting the needle tip position and dynamic insertion point coordinates, the actual effective puncture depth is accurately quantified, avoiding depth misjudgment caused by respiratory movements or needle body shaking. One-dimensional and three-dimensional low-pass filtering of the time derivative results of depth and position effectively suppresses measurement noise and high-frequency jitter, ensuring the stability and accuracy of the instantaneous effective insertion scalar velocity and the three-dimensional real physical motion velocity vector of the needle tip. These accurate and filtered kinematic parameters provide reliable basic data for the generation of target point dynamic correction instructions in subsequent steps, significantly improving the accuracy and stability of the entire puncture target point dynamic correction method, thereby ensuring the safety and effectiveness of the puncture operation.

[0043] This application further proposes a prediction mechanism in step S20 for predicting the dynamic target coordinates and analytical feedforward velocity under non-puncture interference, including: based on the three-dimensional coordinate vector of the target at rest, superimposing the in vivo and external displacement mapping components in the main respiratory direction, and subtracting the phase lag displacement component in the secondary lag direction to synthesize the dynamic target coordinates under non-puncture interference; wherein, the phase lag displacement component is generated by the instantaneous velocity of the surface respiratory substitute signal constrained by the tissue lag time constant and the in vivo and external amplitude coupling coefficient; the mechanism for synthesizing the dynamic target coordinates is analyzed in the time domain, using the instantaneous velocity of the surface respiratory substitute signal to generate the feedforward velocity component in the main respiratory direction, and using the instantaneous acceleration of the surface respiratory substitute signal to generate the feedforward velocity component in the secondary lag direction, and the two are superimposed to generate the analytical feedforward velocity. The prediction of the dynamic target coordinates and analytical feedforward velocity under non-puncture interference is specifically calculated according to the following formula:

[0044]

[0045] In the above formula, This represents the three-dimensional coordinate vector of the dynamic target point at time t without interference. Represents the three-dimensional coordinate vector of the target point in its resting state; and These represent the in vivo and in vitro amplitude coupling scalar coefficients in the main respiratory direction and the secondary hysteresis direction, respectively; , and These represent the real-time displacement, velocity, and acceleration scalars of the body surface substitution signal output by state estimation at time t, respectively. and These represent the unit vectors of the primary respiratory displacement direction and the secondary motion direction, respectively; The formula represents the tissue phase lag time constant, and the minus sign in the formula is used to characterize the physical lag effect. This represents the analytical feedforward target velocity vector.

[0046] To ensure the accuracy and feasibility of dynamic target coordinate prediction without puncture interference, the unit vectors of the primary respiratory displacement direction and the secondary motion direction mentioned in the above formula are... and In vivo and in vitro amplitude coupling scalar coefficients and and the tissue phase lag time constant The specific parameter identification methods include: during the preoperative planning stage, acquiring 4DCT image sequences of the patient's complete respiratory cycle, and using image registration algorithms to extract the three-dimensional centroid coordinate sequences of the target point at each respiratory phase; performing principal component analysis (PCA) on the three-dimensional centroid coordinate sequences, and extracting the first principal component eigenvector with the largest variance contribution rate as the unit vector of the main respiratory displacement direction. Extract the second principal component feature vector as the secondary motion direction unit vector. After directional decoupling, the body surface respiratory substitution signal recordings from the CT scan are acquired synchronously. The displacement is then measured in real-time using the body surface signal. As the independent variable, the three-dimensional centroid coordinate sequence of the target point is used in... The projected distance in the direction of respiration is the dependent variable. The in vivo and in vitro amplitude coupling scalar coefficients in the main respiratory direction are obtained by least squares fitting. Similarly, using body surface velocity signals As the independent variable, the target point is With the projected displacement in the direction as the dependent variable, a nonlinear curve fitting algorithm incorporating an exponential term is used to simultaneously identify the in-body and out-of-body amplitude coupling scalar coefficients in the secondary lag direction. and tissue phase lag time constant .when and When coupling in a product form within a single dataset, further acquisition of surface signals and target motion data under multiple conditions of different respiratory rates and / or amplitudes can be performed for joint fitting, or a priori constraint of a first-order hysteresis model of tissue viscoelasticity can be introduced to achieve [further optimization]. and Separation and identification.

[0047] In one embodiment, a nonlinear least squares method with prior regularization constraints is employed. and The separation and identification process involves several steps. First, the actual projected displacement sequence of the target point in the secondary motion direction is extracted using preoperative 4DCT image sequences and used as the fitting target. Then, according to the prediction model in this scheme, this projected displacement can theoretically be characterized by multiplying the instantaneous velocity of the surface respiratory substitution signal by a combination coefficient, where the combination coefficient is... and The product of the two factors. To eliminate the non-unique separability problem caused by the product, an additional regularization penalty term is introduced into the objective function, besides the sum of squared fitting errors, to constrain the product. The degree of deviation from the preset prior value. This prior value can be determined by referring to the range of viscoelastic hysteresis time constants of similar soft tissues in existing literature, such as approximately 0.1 to 0.5 seconds for liver parenchyma and approximately 0.05 to 0.2 seconds for lung tissue; alternatively, it can be estimated by analyzing the phase difference between the surface signal and the target movement at different respiratory rates through preoperative experiments. The initial value range. When solving this nonlinear least squares problem with regularization constraints, the Levenberg-Marquardt algorithm or gradient descent method can be used, while simultaneously optimizing... and The separated parameter values ​​are obtained.

[0048] As an alternative or supplement, frequency domain identification can also be used. At multiple different respiratory frequencies, the surface respiratory substitution signal is used as input, and the displacement of the target point in the secondary motion direction is used as output. The amplitude ratio and phase difference at each frequency point are calculated. Utilizing the characteristic that the amplitude ratio of a first-order hysteresis system decreases with frequency while the phase difference increases with frequency, two independent equations are established, which can be uniquely solved by simultaneous application. and This method requires no additional assumptions and can directly utilize data from different respiratory rates to eliminate product coupling.

[0049] Specifically, the three-dimensional coordinate vector of the dynamic target point without interference This represents the real-time three-dimensional spatial position of the target point under the influence of respiratory motion, without interaction between the puncture needle and tissue. The calculation comprehensively considers the target point's resting position, displacement caused by respiratory motion, and secondary hysteresis due to the viscoelastic properties of the tissue. In this way, the system can pre-determine the expected trajectory of the target point, providing a basis for subsequent puncture correction. (Target point resting state three-dimensional coordinate vector) This represents the initial or average three-dimensional spatial position of the target point when the patient is at rest or in a state of respiratory arrest. It is usually obtained through preoperative imaging data (such as CT, MRI) and serves as a reference point for the dynamic movement of the target point. The in vivo and in vitro amplitude coupling scalar coefficients in the main respiratory direction and secondary hysteresis direction are also considered. and It is used to quantify the amplitude mapping relationship between surface respiratory alternative signals (such as chest and abdominal movements) and the actual movement of targets in the body. This reflects the coupling strength in the main respiratory direction, while This reflects the coupling strength in the secondary motion direction, particularly the hysteresis motion related to tissue viscoelasticity. These coefficients can be determined by synchronous measurement and modeling of surface signals and target motion before or during surgery, ensuring that surface signals accurately reflect the true motion amplitude of the target within the body. The real-time displacement, velocity, and acceleration scalars of the surface substitute signal output by state estimation are... , and It originates from surface respiratory substitute signals (e.g., chest and abdominal motion data acquired through external sensors such as breathing belts or optical tracking systems) and is processed by a state estimator. The state estimator filters the raw signal, particularly removing differential noise, thus providing smooth and accurate real-time displacement, velocity, and acceleration information. These signals are the core inputs driving the target dynamic prediction model. The primary respiratory displacement direction and the secondary motion direction are represented by unit vectors. and The main motion patterns of the target in three-dimensional space are defined. It indicates the direction of the main displacement of the target point during the respiratory cycle, which is usually consistent with the direction of the main axis of respiratory motion. This indicates the direction of secondary motion orthogonal to or at a certain angle to the main respiratory direction. Motion in this direction is often related to the viscoelastic properties and phase lag effect of the tissue. Tissue phase lag time constant. This is used to characterize the phase delay between the internal motion of soft tissue and the surface signal or driving force when driven by respiratory movements. Due to the viscoelastic properties of soft tissue, its response is not instantaneous but exhibits a certain hysteresis. This hysteresis effect is quantified, enabling the prediction model to more accurately reflect the actual dynamic behavior of the target. The minus sign in the formula explicitly indicates this physical hysteresis effect. Analyzing the feedforward target velocity vector. This represents the expected velocity of the target at time t under conditions of no puncture interference. It is calculated based on the velocity and acceleration information of the surface respiratory substitute signal, combined with the amplitude coupling coefficient and the phase lag time constant. This feedforward velocity vector provides important forward-looking information for subsequent needle tip dynamic correction, enabling the actuator to respond to the target's movement in advance, thereby reducing tracking errors.

[0050] Through the above technical solution, this application can accurately predict the dynamic target coordinates and analytical feedforward velocity under conditions without puncture interference. Specifically, this is achieved by introducing a three-dimensional coordinate vector of the target at rest. As a benchmark, and combined with the in vivo and in vitro amplitude coupling scalar coefficients in the main respiratory direction and the secondary hysteresis direction. and And the real-time displacement, velocity, and acceleration scalars of the body surface substitute signal after processing by the state estimator. , and This allows for accurate mapping of surface respiratory movements to the actual movement of targets within the body. Furthermore, it introduces unit vectors for the primary respiratory displacement direction and the secondary movement direction. and and the tissue phase lag time constant This application fully considers the viscoelastic phase hysteresis characteristics of soft tissue, thereby more realistically simulating the complex dynamic behavior of the target. This refined predictive model enables the system to obtain high-precision dynamic position and movement trend information of the target before the puncture needle makes significant contact with the target, providing a reliable feedforward control signal for subsequent dynamic correction of the puncture needle, significantly improving the predictability and accuracy of the puncture process, and effectively reducing puncture deviation caused by respiratory movements.

[0051] like Figure 2 As shown, this application further proposes a decoupling and limiting mechanism for the tissue deformation offset vector calculated after physical limiting processing in step S30, including: using the corrected coordinates with a single sampling period delay as the target point's true coordinate feedback for the previous control period, subtracting it from the current real-time three-dimensional position coordinate vector of the needle tip, constructing a spatial relative position vector containing time delay memory, thereby blocking the algebraic loop within the same sampling period; orthogonally decomposing the spatial relative position vector relative to the unit vector of the real-time attitude direction of the needle tip to obtain an axial projection scalar and a radial perpendicular vector; the tissue deformation offset vector is constructed as the vector sum of a radial pushing volume displacement term and an axial friction drag term; wherein: the action direction of the radial pushing volume displacement term is parallel to the radial perpendicular vector, and its magnitude varies with the radial perpendicular vector. The modulus length increases but decays nonlinearly, and the physical geometric scale of the puncture needle is introduced as a regularization constraint in the denominator of the decay calculation to eliminate computational singularities when the radial distance approaches zero. The axial friction drag term acts parallel to the unit vector of the real-time needle tip orientation, and its amplitude is constrained by the soft tissue friction coupling coefficient and the effective puncture depth, decaying with relative spatial distance. Furthermore, the axial friction drag term employs a smooth state activation mechanism with the instantaneous needle insertion velocity as the independent variable for direction and amplitude modulation to ensure that the drag effect conforms to the causal mechanical direction of contact friction. The generated initial tissue deformation offset vector undergoes amplitude saturation limiting based on the maximum yield limit of the tissue to truncate the discrete recursive positive feedback oscillation caused by time-delayed memory, outputting the final tissue deformation offset vector. In the presence of anisotropic anatomical constraints, This can be extended to a direction-dependent amplitude limiting threshold, for example, by defining a diagonal matrix. Alternatively, different maximum permissible deformation values ​​can be set according to the main respiratory direction, axial direction, and radial direction to further meet clinical safety requirements. The tissue deformation offset vector is calculated after physical limiting, specifically according to the following formula:

[0052]

[0053]

[0054]

[0055]

[0056] In the above formula, This represents the relative position vector of the needle tip pointing towards the absolute target at time t; Indicates the previous digital sampling period The generated target point correction coordinates; This represents the real-time three-dimensional position coordinate vector of the needle tip at time t; and These represent the scalar projection distance of the relative position vector along the needle insertion axis and the radial perpendicular vector deviating from the needle body axis, respectively. This represents a unit vector indicating the real-time attitude direction of the needle tip. This represents the initial offset vector of tissue deformation before it is limited; Indicates the physical solid radius of the puncture needle; This represents the dimensionless frictional coupling coefficient of soft tissue. Indicates the actual effective puncture depth; This represents the activation sensitivity coefficient for motion states, expressed in seconds per millimeter. Indicates one-dimensional low-pass filtering The instantaneous effective needle insertion scalar speed after insertion; This represents the final tissue deformation offset vector at time t, where 'sat' represents the vector amplitude saturation limiting function. This represents the maximum permissible physical limit threshold for tissue deformation.

[0057] Optionally, to more accurately reflect the influence of the relative position of the needle tip and the target on the axial friction drag direction, the axial friction drag term can be multiplied by the axial relative position sign function. Alternatively, a smooth transition function can be used to reverse the drag direction as the needle tip passes the target point. For example, the axial friction drag term can be modified as follows:

[0058] This symbolic function does not introduce additional dimensions and can further conform to the causal mechanical direction of contact friction.

[0059] To ensure the applicability of tissue deformation offset vector calculation under different anatomical locations and tissue properties, the specific method for obtaining the physical parameters in the above formula is as follows: the maximum permissible tissue deformation physical limit threshold. The setting method is as follows: extract the shortest spatial Euclidean distance from the target point to the adjacent high-risk anatomical structure (such as the edge of a large blood vessel) based on preoperative images, and multiply this distance by... The safety margin coefficient is used to ensure that the needle tip deformation correction command is always within the physiological and anatomical safety range; the dimensionless frictional coupling coefficient of the soft tissue The acquisition method employs online adaptive identification: during the initial uniform velocity phase of the puncture needle insertion into the subcutaneous tissue, the system acquires the measured axial resistance through an external torque sensor. Using an extended Kalman filter (EKF) or recursive least squares (RLS) method, the residual between the measured axial resistance and the theoretical volume displacement force is used as the observation target to estimate and update the data in real time online. The numerical value of the motion state activation sensitivity coefficient; The nonlinear slope used to characterize the transition from static to dynamic friction is obtained through offline bench calibration: that is, in an offline state, the puncture needle is inserted into a standard gelatin phantom at various preset speed trajectories, the velocity-friction hysteresis curve is collected, and the function is fitted using the nonlinear least squares method. The residuals are used to solve for the optimal solution. value.

[0060] Specifically, online identification At that time, Treating this as a time-varying parameter to be estimated, a system state-space model incorporating this parameter is established. The state equations will then... The recursive variation modeling adds process noise to the slowly varying parameters to accommodate the slow parameter drift caused by tissue heterogeneity.

[0061] In the observation equation, the measured axial resistance collected by the external torque sensor is decomposed into two parts: one part is the resistance related to the axial friction drag of the needle body, and the other part is the resistance generated by the axial deformation of the tissue basis. The axial friction drag resistance is directly calculated using the coefficient of the axial friction drag term in the tissue deformation offset vector of this scheme. This coefficient is determined by the real-time effective puncture depth, relative spatial distance, instantaneous needle insertion speed, and... A joint decision. Because... The coefficients appear in linear form, therefore the observation equations are... It is linear and suitable for online estimation using recursive least squares or Kalman filtering.

[0062] The axial component of the tissue's basic deformation resistance can be determined in one of the following ways: During the initial, slow advancement of the puncture needle, before significant frictional drag occurs, the measured axial resistance is used as the baseline resistance, approximating it as the basic deformation resistance; or the basic deformation resistance is used as a slowly varying bias term, and... They are jointly estimated by incorporating the state vector of the extended Kalman filter. When the latter is used, the state vector contains... With two variables, the state transition equation is the identity mapping superposition process noise, and the observation equation is that the measured axial resistance is equal to the sum of the aforementioned axial friction drag resistance and the foundation resistance plus the observation noise.

[0063] Using the state-space model described above, online estimation and updating can be achieved using the standard extended Kalman filter or recursive least squares algorithm. Since all intermediate variables in this model can be calculated in real time using the aforementioned steps, the entire identification process is feasible.

[0064] Specifically, the relative position vector of the needle tip pointing to the absolute target point at time t. By the previous digital sampling period Target correction coordinates generated The real-time three-dimensional position coordinate vector of the needle tip at the current time t The vector is obtained through interpolation. This vector accurately represents the spatial positional relationship between the needle tip and the target point at the current moment, providing a foundation for subsequent tissue deformation analysis. By updating this vector in real time, the system can continuously track the dynamic relative position between the needle tip and the target point, which is crucial for maintaining puncture accuracy in complex dynamic environments such as respiratory movements.

[0065] Based on this, the relative position vector is further decomposed. Axial projection distance. It is a relative position vector Unit vector in the real-time attitude direction of the needle tip The projection onto the target represents the distance along the puncture axis between the needle tip and the target point. Radial perpendicular vector. Then by using the relative position vector The distance between the needle tip and the target point, perpendicular to the puncture axis, is obtained by subtracting the axial component. This decomposition method decouples complex tissue deformation into frictional drag along the needle axis and radial pushing perpendicular to the needle axis, making it possible to independently model and analyze deformation mechanisms in different directions, thereby more accurately describing the non-uniform deformation characteristics of soft tissue.

[0066] Subsequently, the initial offset vector of tissue deformation before limiting was calculated. This vector comprehensively considers both the radial pushing volume displacement effect and the axial frictional drag effect. The radial pushing part consists of... This indicates that it reflects the physical solid radius of the puncture needle. and radial perpendicular vector The volumetric displacement effect on the tissue causes it to shift radially outward. The axial frictional drag portion is composed of... It means that, among them It is the dimensionless frictional coupling coefficient of soft tissue, which characterizes the frictional force between the tissue and the needle body; It is the actual effective puncture depth, and the friction increases with the puncture depth; It is an activation function, where It is the sensitivity coefficient for activation during motion. This is the instantaneous effective needle insertion scalar velocity, which activates the frictional drag effect during needle tip movement, and its intensity is non-linearly related to the needle insertion velocity. This dual-mechanism modeling allows the system to more realistically simulate the complex deformation behavior of soft tissue during puncture, thus providing more accurate deformation displacement predictions.

[0067] Finally, at time t, the final deformation offset vector is organized. By analyzing the initial offset vector of the tissue deformation before limiting. Apply the vector amplitude saturation limiting function 'sat' and the maximum permissible tissue deformation physical limiting threshold. The purpose of this is to ensure that the calculated tissue deformation offset does not exceed the physical limit that soft tissue can withstand, and to prevent the model from predicting excessive deformation that would lead to unrealistic results.

[0068] Through the above technical solution, this application can accurately quantify the deformation and displacement of soft tissue during puncture. Specifically, by establishing a spatial relative position vector of the needle tip pointing to the absolute target point, and decoupling it into an axial component along the needle axis and a radial component perpendicular to the needle axis, this application can model the radial pushing volume displacement effect and the axial friction drag effect causally activated by the instantaneous speed of needle insertion, respectively. This refined deformation model considers the physical radius of the puncture needle. Actual puncture depth and needle insertion speed The effect of tissue deformation on the calculated initial offset vector of tissue deformation is... This is more in line with reality. Furthermore, by physically limiting this offset, the final tissue deformation offset vector is ensured. It will not exceed the physical tolerance limit of soft tissue, effectively avoiding correction errors caused by excessive deformation prediction, thereby significantly improving the accuracy and safety of dynamic correction of puncture target under respiratory dynamics, and providing a solid foundation for subsequent target correction coordinate calculation and spatial tracking error extraction.

[0069] This application further proposes that in step S40, the corrected coordinates of the target point within the current cycle are obtained, and the spatial tracking error vector is extracted by combining the current three-dimensional coordinates of the needle tip. The extraction mechanism includes: superimposing the dynamic target point coordinates without puncture interference with the tissue deformation offset vector using spatial vector addition; the superposition result is used as the corrected coordinates of the target point in the current sampling cycle and used as the historical state input for the next control cycle for iteration; the corrected coordinates are subtracted from the current real-time three-dimensional position coordinates of the needle tip to obtain the spatial tracking error vector that the system currently needs to eliminate. The corrected coordinates of the target point within the current cycle are obtained, and the spatial tracking error vector is extracted by combining the current three-dimensional coordinates of the needle tip, specifically calculated according to the following formula:

[0070]

[0071] In the above formula, The target point correction coordinates at time t represent the estimated position of the target point model obtained by superimposing the dynamic target point coordinates without interference and the tissue deformation offset vector. This represents the three-dimensional coordinate vector of the dynamic target point at time t without interference. This represents the final tissue deformation offset vector at time t; This represents the three-dimensional tracking error vector at time t. The vector represents the real-time three-dimensional position coordinates of the needle tip at time t.

[0072] Target point correction coordinates This represents the actual spatial position of the target at the current moment, taking into account the target's own dynamics caused by respiratory motion and the tissue deformation caused by the interaction between the puncture needle and the tissue. The calculation involves superimposing the dynamic target coordinates under no-puncture interference with the tissue deformation offset vector, ensuring a comprehensive and accurate description of the target's position.

[0073] Based on this, the target point coordinates at time t are corrected. The real-time three-dimensional position coordinate vector of the needle tip at time t By performing interpolation, the three-dimensional spatial tracking error vector at time t can be extracted. ,Right now:

[0074] Among them, the three-dimensional spatial tracking error vector This is the spatial difference between the real-time position of the puncture needle tip and the absolute true position of the target point. It intuitively reflects the direction and distance of the puncture needle tip deviating from the target point and is a key input for generating subsequent correction commands. By calculating this error vector in real time, the system can quantify the accuracy of the current puncture and provide a basis for achieving precise dynamic correction.

[0075] Through the above technical solution, the system can accurately obtain the calibration coordinates of the target point during puncture, which comprehensively considers respiratory motion and tissue deformation caused by puncture. More importantly, by calculating the three-dimensional spatial tracking error vector between the real-time position of the needle tip and the coordinates of the absolute true target point, this application can quantify the deviation between the puncture needle tip and the target point. This precise error quantification provides a direct and reliable basis for subsequently generating high-precision needle tip dynamic correction speed commands, thereby significantly improving the accuracy and safety of puncture under respiratory dynamics, ensuring that the puncture needle can track the dynamic target point accurately in real time, and avoiding puncture failure or damage to surrounding tissues due to error accumulation.

[0076] This application further proposes that in step S50, the generation mechanism for generating the needle tip dynamic correction speed command includes: the needle tip dynamic correction speed command is synthesized by a feedforward tracking component, an error elimination component, and an active impedance dissipation component; wherein: the feedforward tracking component is directly mapped to the analytical feedforward speed, which is used to actively match respiratory dynamics; the error elimination component is obtained by multiplying the spatial tracking error vector with the proportional position stiffness gain of the system; the active impedance dissipation component (i.e., the speed deviation damping term) is obtained by multiplying the difference vector between the three-dimensional physical motion speed of the needle tip and the analytical feedforward speed with the active relative damping gain of the system, and this component is introduced into the synthesis operation in the form of negative feedback to provide directional damping and smooth out mechanical chattering during the puncture process.

[0077] The generation of the needle tip dynamic correction speed command is calculated according to the following formula:

[0078] In the above formula, The three-dimensional velocity command vector for dynamic correction of the needle tip at time t is the final motion command output to the actuator. The proportional position gain scalar, expressed in units of negative one second; This represents the active relative damping gain constant. Simultaneously, the system can generate the desired angular velocity vector to maintain the needle tip attitude based on the deviation between the real-time needle tip attitude and the target puncture attitude. . It can be generated by the proportional control of attitude error or the attitude closed-loop controller, and combined with the three-dimensional velocity command vector for tip dynamic correction. Together they are used to drive the actuator, thereby achieving coordinated correction of position and attitude in Cartesian space.

[0079] Furthermore, to ensure the stable operation of the active impedance control term, the proportional position gain scalar... With active relative damping gain constant Tuning is performed based on the critical damping configuration principle: firstly, the damping is gradually increased during offline debugging. Record the critical gain at the point where the system exhibits constant-amplitude oscillations while tracking a step signal. The oscillation period was then used to set the initial value based on the critical proportionality method. And on this basis, by setting (in The system is configured to be in a slightly overdamped state by setting a virtual mass at the tip, thereby completely suppressing mechanical chattering. The virtual mass at the tip... The virtual mass can be set to 0.5 to 2 times the sum of the physical mass of the puncture needle and the equivalent inertia of the actuator end effector in the needle tip direction. Specifically, if the physical mass of the puncture needle is known, and the equivalent inertia of the robot end effector in the needle tip axis can be obtained through mechanical design parameters or simple calibration, then the two can be directly added together and multiplied by an appropriate safety factor to obtain the virtual mass. If precise inertia data is lacking, a simplified rigid body dynamics model of the actuator and puncture needle can be established in a simulation environment first. The virtual mass can be adjusted through step response experiments so that the system exhibits a critically damped or slightly overdamped state in the position step response, thereby determining a suitable virtual mass value. This method does not require precise dynamic modeling and can be implemented by those skilled in the art based on conventional system identification methods. The three-dimensional velocity command vector for the dynamic correction of the needle tip is generated. The specific mechanism for mapping and driving the motion of the actuators includes: the system acquiring the current joint angle vectors of the actuators (such as multi-joint puncture robots) in real time. The Jacobian matrix under the current configuration is calculated using the robot's forward kinematics. Subsequently, the Jacobian matrix is... Find the damped least squares pseudo-inverse To avoid kinematic singularities; finally, through mapping equations Parse the joint angular velocity commands required by the underlying motor of the actuator. And send it to the servo driver, thereby completing the precise dynamic correction motion of the needle tip in Cartesian space (wherein) (The desired angular velocity vector to maintain the predetermined puncture posture of the needle tip).

[0080] This represents the analytical feedforward target velocity vector, which provides the predicted motion trend of the target point under undisturbed conditions, serving as the basic component of the command. The proportional position gain scalar, expressed in negative first power in seconds, is used to adjust the spatial tracking error vector. The response intensity is adjusted to eliminate positional deviation between the needle tip and the target.

[0081] In the formula This term constitutes the active impedance control term. This term introduces the active relative damping gain constant. And combined with the three-dimensional physical motion speed of the needle tip With analytical feedforward speed The deviation between the actual and expected feedforward speeds is used to adjust the command. This control term effectively suppresses unwanted oscillations or overshoots and enhances system stability when there is a difference between the actual needle tip speed and the expected feedforward speed. For example, when the actual needle tip speed is faster than the feedforward speed, this term generates a reverse deceleration command; conversely, it generates an acceleration command, thus making the actual needle tip speed more closely follow the predicted target speed.

[0082] Through the above technical solution, this application can effectively compensate for the dynamic velocity difference between the actual movement of the needle tip and the predicted movement of the target point during puncture. This active impedance control term, based on traditional proportional control according to position error, adds a dynamic response to velocity deviation. This allows the generated dynamic correction velocity command for the needle tip to not only correct position deviation but also actively suppress oscillations and instabilities caused by velocity mismatch. This significantly improves the accuracy and robustness of target point tracking during puncture in complex respiratory dynamic environments, ensuring the smoothness and safety of the puncture process, thereby avoiding overshoot or lag caused by velocity deviation and improving the overall puncture success rate.

[0083] The following example will provide a more detailed explanation of the above technical solution: In a robot-assisted liver biopsy, user A needs to precisely puncture a target lesion within the patient's liver. Due to the patient's respiratory movements, the liver lesion undergoes continuous periodic displacement, and the insertion of the puncture needle also causes additional tissue deformation. These factors all pose challenges to puncture accuracy. This method aims to dynamically correct the puncture target point to achieve high-precision puncture.

[0084] First, during the puncture, the system continuously acquires real-time three-dimensional pose data of the puncture needle and dynamic needle insertion point coordinates on the skin surface driven by respiratory movements. For example, the three-dimensional position coordinate vector of the puncture needle tail sensor 100 is monitored in real time through an optical tracking system. and three-dimensional rotation matrix Simultaneously, dynamic needle insertion points on the skin surface are obtained through image registration or surface marker tracking. Based on this data, the system analyzes and obtains the real-time three-dimensional position coordinate vector of the needle tip. ,in This is the physical nominal length of the puncture needle. The real-time orientation vector of the needle tip is a unit vector. This was also calculated. Next, the system calculated the actual effective puncture depth. To eliminate measurement noise, the system performs a one-dimensional low-pass filter on the time derivative of the effective puncture depth. Processing yields the instantaneous effective needle advance scalar speed. Simultaneously, a three-dimensional low-pass filter is applied to the time derivative of the real-time three-dimensional position coordinate vector of the needle tip. Processing yields the three-dimensional real physical motion velocity vector of the needle tip. This step solves the problem of accurately obtaining the motion state of the puncture needle, providing a reliable input for subsequent dynamic calibration.

[0085] Next, to predict the dynamic location of the target site without puncture interference, the system acquires surface respiratory substitution signals. For example, it acquires the patient's chest and abdominal respiratory displacement, velocity, and acceleration scalars via breathing belts or external sensors. , and These signals are first processed by a state estimator to filter out differential noise and improve signal purity. Unlike existing techniques that rely solely on linear synchronization mapping, this method combines the amplitude mapping relationship between soft tissue interiors and exteriors with the viscoelastic phase hysteresis characteristics. Specifically, the system utilizes the three-dimensional coordinate vector of the target's resting state. And the in vivo and in vitro amplitude coupling scalar coefficients in the main respiratory direction and the secondary hysteresis direction. and The unit vector of the primary respiratory displacement direction and the secondary motion direction and and the tissue phase lag time constant Predicting the three-dimensional coordinate vector of a dynamic target without interference. Simultaneously, analyze the feedforward target velocity vector. It is also calculated. The minus sign here represents the physical hysteresis effect. By considering viscoelastic phase hysteresis and multi-directional motion, this method overcomes the systematic errors and feedback hysteresis problems existing in target prediction methods.

[0086] After the puncture needle enters the tissue, the interaction between the needle and the tissue causes additional deformation at the target site. To accurately quantify this deformation, the system uses the target site calibration coordinates from the previous control cycle. The real-time three-dimensional position coordinate vector of the needle tip at the current moment Establish spatial relative position vectors Unlike existing methods that treat the puncture needle as a tracker of undisturbed motion, this method decouples the soft tissue deformation model into radial pushing volume displacement and axial frictional drag causally activated by the instantaneous velocity of needle insertion. Specifically, the projection distance scalar of the relative position vector along the needle insertion axis... and the radial perpendicular vector deviating from the needle body axis It is calculated. Then the initial offset vector of tissue deformation before clipping is calculated. Calculated, where It is the physical solid radius of the puncture needle. It is the dimensionless frictional coupling coefficient of soft tissue. This is the motion state activation sensitivity coefficient. Finally, it is determined by the vector amplitude saturation limiting function `sat` and the maximum permissible tissue deformation physical limiting threshold. The final tissue deformation offset vector at time t is obtained. This sophisticated deformation model takes into account the nonlinear decoupled deformation of needle-tissue contact mechanics, significantly improving the accuracy of target location estimation.

[0087] Subsequently, the system will generate the dynamic target point three-dimensional coordinate vector without puncture interference. With the final tissue deformation offset vector By superimposing the coordinates, the corrected coordinates of the target point within the current period can be obtained. This calibration coordinate system incorporates deformation caused by respiratory motion and needle-tissue interaction, providing the most accurate estimate of the target location. This is then combined with the current real-time three-dimensional position coordinate vector of the needle tip. Extract the spatial tracking error vector This error vector directly reflects the deviation between the needle tip and the actual target point.

[0088] Finally, to drive the actuator to move and correct the needle tip position, the system generates a dynamic correction speed command for the needle tip. This command is based not only on the analytical feedforward target velocity vector. and spatial tracking error vector It also introduces a three-dimensional real physical velocity vector of the needle tip. With analytical feedforward target velocity vector The active impedance control term is formed by the deviation. Specifically, the three-dimensional velocity command vector for dynamic correction of the needle tip. Calculated, where It is a proportional position gain scalar. It is an active relative damping gain constant. This active impedance control term can effectively suppress mechanical chattering and improve the stability and safety of the system. Through the above series of steps, this method achieves precise dynamic correction of the puncture target under respiratory dynamics, significantly improving the accuracy and safety of interventional surgery, and overcoming the problems of inaccurate target prediction, feedback lag, and mechanical chattering in existing technologies.

[0089] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for dynamic correction of puncture target points under respiratory dynamics, characterized in that, Includes the following steps: S10: Obtain the real-time three-dimensional pose data of the puncture needle and the dynamic needle insertion point coordinates on the skin surface driven by respiratory motion, and analyze the three-dimensional physical motion speed of the needle tip, the effective puncture depth, and the needle insertion instantaneous speed after low-pass filtering. S20 acquires the surface respiratory substitution signal, filters out differential noise based on the state estimator, and combines the amplitude mapping relationship between soft tissue and viscoelastic phase hysteresis characteristics to predict the dynamic target coordinates and analytical feedforward velocity without puncture interference. S30, based on the target point real coordinate feedback of the previous control cycle, establish a spatial relative position vector, decouple the soft tissue deformation model into radial pushing volume dispersal and axial friction drag activated by the instantaneous speed of the needle insertion, and calculate the tissue deformation offset vector after physical limiting processing. S40, superimpose the dynamic target coordinates under no puncture interference with the tissue deformation offset vector to obtain the corrected coordinates of the target in the current cycle, and extract the spatial tracking error vector by combining the current three-dimensional coordinates of the needle tip. S50, based on analytical feedforward speed and spatial tracking error vector, introduces an active impedance control term consisting of the deviation between the three-dimensional physical motion speed of the needle tip and the analytical feedforward speed, to generate a dynamic correction speed command for the needle tip to drive the actuator movement.

2. The method for dynamic correction of puncture target under respiratory dynamics according to claim 1, characterized in that, In step S10, the analytical mechanism for obtaining the effective puncture depth includes: Based on rigid body kinematics, and combining the pose of the needle tail sensor with the effective nominal length of the puncture needle, the real-time three-dimensional position coordinate vector of the needle tip and the real-time attitude direction unit vector of the needle tip are mapped and obtained. Based on the relative deviation between the real-time three-dimensional position coordinate vector of the needle tip and the dynamic needle insertion point coordinates on the skin surface, and by projecting it onto the unit vector of the real-time posture direction of the needle tip, the non-negative effective puncture depth is extracted.

3. The method for dynamic correction of puncture target under respiratory dynamics according to claim 2, characterized in that, In step S10, the analytical mechanism for obtaining the three-dimensional physical motion velocity of the needle tip and the instantaneous needle insertion velocity after low-pass filtering includes: The effective puncture depth and the real-time three-dimensional position coordinate vector of the needle tip are extracted based on the rate of change in the time domain. A state observation or low-pass filtering mechanism is introduced to eliminate signal boundary jumps and differential high-frequency noise, thereby outputting the instantaneous speed of needle insertion and the three-dimensional physical motion speed of the needle tip.

4. The method for dynamic correction of puncture target under respiratory dynamics according to claim 3, characterized in that, In step S20, the mechanism for predicting the dynamic target coordinates without puncture interference includes: Based on the three-dimensional coordinate vector of the target at rest, the internal and external displacement mapping components in the main respiratory direction are superimposed, and the phase lag displacement components in the secondary lag direction are subtracted to synthesize the dynamic target coordinates without puncture interference. Among them, the phase lag displacement component is generated by the instantaneous velocity of the surface respiratory substitute signal being constrained by the tissue lag time constant and the in vivo and in vitro amplitude coupling coefficient.

5. The method for dynamic correction of puncture target under respiratory dynamics according to claim 4, characterized in that, In step S20, the mechanism for predicting the analytical feedforward speed includes: The mechanism for synthesizing dynamic target coordinates is analyzed in the time domain. The instantaneous velocity of the surface respiratory substitute signal is used to generate the feedforward velocity component in the main respiratory direction, and the instantaneous acceleration of the surface respiratory substitute signal is used to generate the feedforward velocity component in the secondary lag direction. The two are superimposed to generate the analytical feedforward velocity.

6. The method for dynamic correction of puncture target under respiratory dynamics according to claim 5, characterized in that, In step S30, the basic decoupling mechanism for calculating the tissue deformation offset vector includes: Using the correction coordinates with a single sampling period delay as the target point's true coordinate feedback for the previous control period, the difference between the correction coordinates and the current real-time three-dimensional position coordinate vector of the needle tip is used to construct a spatial relative position vector containing time delay memory, thereby blocking the algebraic loop within the same sampling period. The spatial relative position vector is orthogonally decomposed relative to the unit vector of the real-time attitude direction of the needle tip to obtain the axial projection scalar and the radial perpendicular vector. The initial tissue deformation offset vector is constructed as the vector sum of the radial pushing volume displacement term and the axial friction drag term, wherein: the radial pushing volume displacement term acts in a direction parallel to the radial perpendicular vector, and the axial friction drag term acts in a direction parallel to the unit vector of the real-time attitude direction of the needle tip.

7. The method for dynamic correction of puncture target under respiratory dynamics according to claim 6, characterized in that, The constraint rules and limiting mechanisms for the radial pushing volume displacement term and the axial friction drag term include: The magnitude of the radial pushing volume displacement term decreases nonlinearly with the increase of the magnitude of the radial perpendicular vector, and the physical geometric scale of the puncture needle is introduced as a regularization constraint in the denominator of the magnitude decay calculation to eliminate the computational singularity when the radial distance approaches zero. The amplitude of the axial friction drag term is constrained by the soft tissue friction coupling coefficient and the effective puncture depth and decreases with the relative spatial distance. In addition, the axial friction drag term adopts a smooth state activation mechanism with the needle insertion instantaneous speed as the independent variable to modulate the direction and amplitude, so as to ensure that the drag effect conforms to the causal mechanical direction of contact friction. The generated initial tissue deformation offset vector is subjected to amplitude saturation limiting operation based on the maximum yield limit of the tissue to cut off the discrete recursive positive feedback oscillation caused by time delay memory, and the final tissue deformation offset vector is output.

8. The method for dynamic correction of puncture target under respiratory dynamics according to claim 7, characterized in that, In step S40, the corrected coordinates of the target point within the current period are obtained, and the spatial tracking error vector is extracted by combining the current three-dimensional coordinates of the needle tip. The extraction mechanism includes: The dynamic target coordinates under no puncture interference are superimposed with the tissue deformation offset vector by spatial vector addition. The superposition result is used as the corrected coordinates of the target in the current sampling period and as the historical state input for the next control period for iteration. The spatial tracking error vector that needs to be eliminated by subtracting the correction coordinates from the current real-time three-dimensional position coordinate vector of the needle tip is obtained.

9. The method for dynamic correction of puncture target under respiratory dynamics according to claim 8, characterized in that, In step S50, the mechanism for generating the needle tip dynamic correction speed command includes: The tip dynamic correction speed command is synthesized from a feedforward tracking component, an error cancellation component, and an active impedance dissipation component; where: The feedforward tracking component is directly mapped to the analytical feedforward velocity, which is used to actively match respiratory dynamics; The error cancellation component is obtained by multiplying the spatial tracking error vector by the proportional position stiffness gain of the system; The active impedance dissipation component is obtained by multiplying the difference vector between the three-dimensional physical motion velocity of the needle tip and the analytical feedforward velocity with the active relative damping gain of the system. This component is introduced into the synthesis operation in the form of negative feedback to provide directional damping and smooth out mechanical chattering during the puncture process.

Images