Deviation correction method and system for optical locator
By using Kalman filtering and prediction models to correct the optical locator, the problem of inaccurate positioning in orthopedic surgery is solved, improving surgical precision and safety, and reducing surgical time and complications.
Patent Information
- Application Number
- CN202411040825.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-31
- Publication Date
- 2026-02-03
AI Technical Summary
In orthopedic surgery, vibration of the optical locator can lead to inaccurate surgical positioning, increasing the difficulty and risk of the surgery.
The Kalman filter model is combined with the position and attitude prediction model. Initial and measurement parameters are obtained through the data acquisition unit. The position prediction unit and attitude optimization unit are used for correction. The Kalman filter model is used to optimize the optimal position and attitude parameters. Finally, the optical positioner is corrected through the positioning correction unit.
It significantly improves the precision of surgical cutting and implant placement, reduces intraoperative tissue damage, lowers the risk of accidental nerve or blood vessel injury, and improves surgical stability and patient satisfaction.
Smart Images
Figure CN121445482A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of orthopedic surgery technology, specifically to a method and system for correcting the deviation of an optical locator. Background Technology
[0002] During orthopedic surgery, especially procedures such as bone sawing and grinding, bone vibration often occurs due to the interaction between surgical instruments and bone tissue. Simultaneously, the NDI optical positioning device used in surgery may also vibrate for various reasons. Combined with the patient's own physiological vibrations, these factors lead to instability and inaccuracy in surgical positioning, thus increasing the difficulty and risk of the surgery. Therefore, how to reduce positioning errors caused by vibration and improve the accuracy of surgical positioning has become an urgent problem to be solved in the field of orthopedic surgery.
[0003] Therefore, existing technologies need further development. Summary of the Invention
[0004] The purpose of this invention is to overcome the above-mentioned technical deficiencies and provide a method and system for correcting optical positioning devices to solve the technical problem of inaccurate surgical positioning caused by the shaking of optical positioning devices during surgery in related technologies.
[0005] To achieve the above technical objectives, the present invention adopts the following technical solution: A method for correcting the position of an optical locator is provided, comprising: acquiring initial position parameters, measured position parameters, initial attitude parameters, and measured attitude parameters of the optical locator; inputting the initial position parameters into a preset position prediction model to obtain the predicted position parameters for the next time step; inputting the predicted position parameters for the next time step and the measured position parameters for the current time step into a preset Kalman filter model to obtain the optimal position parameters for the next time step; preprocessing the initial attitude parameters and the measured attitude parameters, and then inputting them into the Kalman filter model to obtain the optimal attitude parameters; and correcting the positioning of the optical locator based on the optimal position parameters and the optimal attitude parameters to obtain the optimal pose of the optical locator.
[0006] Furthermore, the location prediction model is as follows:
[0007] e y n+1，n =F e y n，n ,
[0008]
[0009] Q = FQ a F T ,
[0010] P n+1，n =FP n，n F T+Q,
[0011] in, e y n+1，n This is the prediction of the registration point position at time n+1 from time n. e y n，n Let P be the real-time position of the preset registration point at time n, and F be the state transition matrix; n，n P is the square of the stability of the registration point location prediction at time n. n+1，n Let F be the square of the stability of the registration point location prediction at time n+1, F be the state transition matrix, and Q be the process noise covariance. Let V be the variance of the acceleration.
[0012] Furthermore, the state transition matrix F is specifically as follows:
[0013]
[0014] in, e x n+1，n , e y n+1，n , e z n+1，n It is a prediction of the future position and state at time n (time n+1). e x n+1，n ′, e y n+1，n ′, e z n+1，n ′ is the prediction of the future velocity state at time n (time n+1). e x n+1，n ′′, e y n+1，n ′′, e z n+1，n '' is the prediction of the future acceleration state at time n (time n+1). e x n，n , e y n，n , e z n，n Let n be the real-time position of the preset registration point at time n. e x n，n ′, e y n，n ′, e z n，n ' is the real-time velocity of the preset registration point at time n. e x n，n ", e y n，n ", e z n，n " is the real-time acceleration of the preset registration point at time n.
[0015] Furthermore, the Kalman filter model is as follows:
[0016] e x n，n = e x n，n-1 +K n (Z n -H e x n，n-1 ),
[0017]
[0018] K n =P n，n-1 H T HP n，n-1 H T +R n ) -1 ,
[0019] in, e x n+1，n This is the prediction of the registration point position at time n+1 from time n. e x n，n Z represents the real-time position of the preset registration point at time n. n Let H be the observation vector of the preset registration point of the optical locator, and let K be the observation matrix of the preset registration point of the optical locator. n For Kalman gain, R n This is the covariance matrix for measuring noise.
[0020] Furthermore, the initial attitude parameters include the initial attitude and the rate of change of the initial attitude; the measured attitude parameters include the measured attitude and the rate of change of the measured attitude.
[0021] Furthermore, the method for obtaining the optimal attitude parameters includes: inputting the initial attitude change rate into a preset first attitude prediction model to obtain the predicted parameters of the attitude change rate at the next moment, wherein the initial attitude change rate is calculated from the initial attitude at two moments; inputting the predicted parameters of the attitude change rate at the next moment and the measured attitude change rate into a Kalman filter model to obtain the optimal parameters of the attitude change rate; establishing a second attitude prediction model based on the optimal parameters of the attitude change rate; inputting the initial attitude and the measured attitude into the second attitude prediction model to obtain the attitude prediction parameters at the next moment; and inputting the attitude prediction parameters at the next moment into a Kalman filter model to obtain the optimal attitude parameters.
[0022] Furthermore, the preset first attitude prediction model includes:
[0023] e q n+1，n =q n，n ,
[0024] P′ n+1，n =F′P′ n，n F′ T +Q′,
[0025]
[0026] in, e q n+1，n q is the prediction of the attitude change rate at time n+1 from time n. n，n Let P' be the attitude change rate at time n, and F' be the state transition matrix; n，n P′ is the square of the stability of the attitude change rate prediction at time n. n+1，n Let be the square of the stability of the attitude change rate prediction at time n+1, Q' be the process noise covariance matrix, and R be the measurement noise covariance matrix.
[0027] Furthermore, the preset second pose prediction model includes:
[0028] e q′ n+1，n =F″ e q′ n，n ,
[0029] P″ n+1，n =F″P″ n，n F″ T +Q″,
[0030]
[0031] in, e q′ n+1，n Let q′ be the prediction of the attitude at time n+1 from time n. n，n Let F″ be the pose at time n, and F″ be the state transition matrix; P″ be the pose at time n. n，n P″ is the square of the stability of the attitude change rate prediction at time n. n+1，n Q″ represents the square of the stability of the attitude change rate prediction at time n+1, and Q″ represents the process noise covariance.
[0032] A correction system for an optical positioner includes: a data acquisition unit for acquiring initial position parameters, measured position parameters, initial attitude parameters, and measured attitude parameters of the optical positioner; a position prediction unit for inputting the initial position parameters into a preset position prediction model to obtain the predicted position parameters for the next moment; a position optimization unit for inputting the predicted position parameters for the next moment and the measured position parameters for the current moment into a preset Kalman filter model to obtain the optimal position parameters for the next moment; an attitude optimization unit for preprocessing the initial attitude parameters and the measured attitude parameters before inputting them into the Kalman filter model to obtain the optimal attitude parameters; and a positioning correction unit for correcting the positioning of the optical positioner based on the optimal position parameters and the optimal attitude parameters to obtain the optimal pose of the optical positioner.
[0033] A computer-readable storage medium storing computer-readable instructions, characterized in that, when executed by a processor, the computer-readable instructions implement the various steps of the correction method for the optical positioner described above.
[0034] Beneficial effects:
[0035] 1. The optical positioning device correction method of the present invention can reduce alignment errors caused by jitter, significantly improve the accuracy of surgical cutting and implant placement, thereby improving surgical precision and the quality of surgical results, thereby reducing intraoperative tissue damage, helping patients recover faster after surgery, reducing complications, and thus improving patient satisfaction and quality of life.
[0036] 2. The correction method of the optical positioner of the present invention can reduce the risk of accidental damage to nerves or blood vessels during surgery, improve patient safety, and reduce surgical risks.
[0037] 3. The correction method of the optical positioner of the present invention can make the surgical process more stable and reduce the operation time, because the doctor can complete the surgical steps faster and reduce the time for adjustment and correction. Attached Figure Description
[0038] Figure 1 This is a flowchart of the correction method for the optical positioner used in Embodiment 1 of the present invention;
[0039] Figure 2 This is a schematic diagram of the position filtering method in the correction method of the optical positioner used in Embodiment 2 of the present invention;
[0040] Figure 3 This is a flowchart of the position filtering method in the optical positioner correction method used in Embodiment 2 of the present invention;
[0041] Figure 4This is a schematic diagram of the attitude filtering method in the optical positioner correction method used in Embodiment 3 of the present invention;
[0042] Figure 5 This is a schematic diagram of the optimal position in the correction method of the optical positioner used in Embodiment 1 of the present invention;
[0043] Figure 6 This is a schematic diagram of the optimal pose in the correction method of the optical positioner used in the embodiments of the present invention;
[0044] Figure 7 This is a schematic diagram of the correction effect of the correction system of the optical positioner used in the embodiment of the present invention;
[0045] Figure 8 The flowchart of obtaining the optimal attitude parameters in the correction method of the optical positioner used in Embodiment 1 of the invention. Detailed Implementation
[0046] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.
[0047] Example 1:
[0048] This invention provides a method for correcting the deviation of an optical positioner. Please refer to [link / reference]. Figure 1 ,include:
[0049] S100 acquires the initial position parameters, measured position parameters, initial attitude parameters, and measured attitude parameters of the optical positioner;
[0050] In practice, the obtained position is the pose of the optical positioner, which includes the coordinates (x, y, z) of the optical positioner in three directions in the three-dimensional coordinate system and the quaternion pose (q0, q1, q2, q3), a total of 7 parameters. Since it is difficult to calculate the position and pose together in actual use, the correction method in this embodiment corrects the pose of the optical positioner from the two aspects of position and pose filtering.
[0051] S200 inputs the initial position parameters into a preset position prediction model to obtain the predicted position parameters for the next moment;
[0052] Specifically, the location prediction model is as follows:
[0053] e y n+1，n =Fe y n，n ,
[0054]
[0055] Q = FQ a F T ,
[0056] P n+1，n =FP n，n F T +Q,
[0057] in, e x n+1，n This is the prediction of the registration point position at time n+1 from time n. e x n，n Let P be the real-time position of the preset registration point at time n, and F be the state transition matrix; n，n P is the square of the stability of the predicted registration point position at time n. n+1，n Let F be the square of the stability of the predicted registration point position at time n+1, F be the state transition matrix, and Q be the process noise covariance. Let V be the variance of the acceleration. Thus, by using a mathematical model to deduce the values of unknown data based on the known linear relationship of the data, we can obtain a more accurate prediction of the next moment.
[0058] In practical application, the following problem equation, namely the state transition matrix F, needs to be defined first:
[0059]
[0060] e x n+1，n ′= e x n，n ′+ e x n，n ''Δt,
[0061] e x n+1，n ′′= e x n，n ′′,
[0062]
[0063] e y n+1，n ′= e y n，n ′+ e y n，n ''Δt,
[0064] e y n+1，n ′′=e y n，n ′,
[0065]
[0066] e z n+1，n ′= e z n，n ′+ e z n，n ''Δt,
[0067] e z n+1，n ′′= e z n，n ′′,
[0068] in, e x n+1，n , e y n+1，n , e z n+1，n It is a prediction of the future position and state at time n (time n+1). e x n+1，n ′, e y n+1，n ′, e z n+1，n ′ is the prediction of the future velocity state at time n (time n+1). e x n+1，n ′′, e y n+1，n ′′, e z n+1，n '' is the prediction of the future acceleration state at time n (time n+1). e x n，n , e y n，n , e z n，n Let n be the real-time position of the preset registration point at time n. e x n，n ′, e y n，n ′, e z n，n ' is the real-time velocity of the preset registration point at time n. e x n，n ", e y n，n ", e z n，n ″ represents the real-time acceleration of the preset registration point at time n, and Δt represents the time interval between time n and time n-1.
[0069] Then, an extrapolation equation is constructed. Based on the known linear relationship of the data (i.e., the problem equation), the value of the unknown data (i.e., the position prediction at the next moment) is calculated through a mathematical model. Specifically, the extrapolation equation is as follows:
[0070] e x n+1，n =F e x n，n ,
[0071] in, e x n+1，n This is the prediction of the registration point position at time n+1 from time n. e x n，n Let be the real-time position of the preset registration point at time n, and F be the state transition matrix, i.e., the problem equation.
[0072] It should be noted that the state transition matrix F is specifically as follows:
[0073]
[0074] in, e x n+1，n , e y n+1，n , e z n+1，n It is a prediction of the future position and state at time n (time n+1). e x n+1，n ′, e y n+1，n ′, e z n+1，n ′ is the prediction of the future velocity state at time n (time n+1). e x n+1，n ′′, e y n+1，n ′′, e z n+1，n '' is the prediction of the future acceleration state at time n (time n+1). e x n，n , e y n，n , e z n，n Let n be the real-time position of the preset registration point at time n. e x n，n ′, e y n，n ′, e z n，n ' is the real-time velocity of the preset registration point at time n. e x n，n ", e y n，n ", ez n，n " is the real-time acceleration of the preset registration point at time n.
[0075] Thus, by obtaining the position coordinates of two adjacent moments based on existing measurements, the velocity (i.e., ...) can be obtained. e x n，n ′, e y n，n ′, e z n，n ′) and acceleration (i.e. e x n，n ", e y n，n ", e z n，n By combining the parameters ("") with position, velocity, and acceleration to predict the state at the next moment, the position at the next moment can be more accurately located (i.e., ...). e x n+1，n , e y n+1，n , e z n+1，n ).
[0076] In practice, since there are no significant speed changes during the instrument handling process in orthopedic surgery, the above-mentioned acceleration model can meet the requirements for predictive filtering during surgery.
[0077] After predicting the position at the next moment, it is necessary to establish a covariance extrapolation equation to account for the uncertainty of the prediction. Specifically, the covariance extrapolation equation is as follows:
[0078] P n+1，n =FP n，n F T +Q,
[0079] Among them, P n，n P is the square of the stability of the predicted registration point position at time n. n+1，n F is the square of the stability of the predicted registration point position at time n+1, F is the state transition matrix, and Q is the process noise covariance.
[0080] Specifically, Q=FQ a F T The variance of acceleration is expressed using the state transition matrix. Projecting onto our dynamic model yields Q a .
[0081] Specifically, Kalman filtering is based on linear algebra and the hidden Markov model. Its basic dynamic system can be represented by a Markov chain, which is based on a linear operator disturbed by Gaussian noise (i.e., normally distributed noise).
[0082] Since the vibrations generated by the bones during surgery generally follow a normal distribution, we can use an improved Kalman filter to denoise the vibrations generated during surgery, which can provide a more accurate prediction of the location at the next moment.
[0083] S300 inputs the predicted position parameters for the next time step and the measured position parameters for the current time step into a preset Kalman filter model to obtain the optimal position parameters for the next time step.
[0084] It should be noted that after obtaining the predicted position at the next time moment and the square of the stability of the predicted position of the registration point at time n+1 through the above extrapolation equation, the predicted value and the observed value are input into the improved Kalman filter model to obtain the optimal position parameters for the next time moment.
[0085] In practice, the Kalman filter model is specifically as follows:
[0086] e x n，n = e x n，n-1 +K n (Z n -H e x n，n-1 ),
[0087]
[0088] K n =P n，n-1 H T HP n，n-1 H T +R n ) -1 ,
[0089] in, e x n+1，n This is the prediction of the registration point position at time n+1 from time n. e x n，n Z represents the real-time position of the preset registration point at time n. n Let H be the observation vector of the preset registration point of the optical locator, and let K be the observation matrix of the preset registration point of the optical locator. n For Kalman gain, R n This is the covariance matrix for measuring noise.
[0090] In this way, after obtaining the position coordinates through position measurement, the Kalman filter model can be selected based on the characteristics of the position coordinates to reduce the error in predicting the measurement values.
[0091] Specifically, the measurement only provides the X, Y, and Z coordinates of the optical locator; therefore, the observation matrix H of the preset registration points of the optical locator is set as follows:
[0092]
[0093] Since the measurement covariance is relatively balanced in actual use, the measurement covariance R is set as follows:
[0094]
[0095] Example 2:
[0096] See Figure 2 The position filtering method in this embodiment is as follows:
[0097] Set the initial value of the input. e x 0，0 and P 0，0 For any value, input the initial value into the extrapolation equation to obtain the time update (i.e., the predicted position value for the next time step and the square of the stability of the position prediction).
[0098] It should be noted that since the position information obtained by the optical locator is only three-dimensional coordinates, that is, the x, y, z values in the three-dimensional coordinate system, the method for obtaining the velocity and acceleration mentioned in the extrapolation equation is to process them by buffering two frames. For example, we obtain the position coordinates s2 at time t2, s1 at time t1, and s0 at time t0, where time is the interval between each two frames, such as t1-t0. The velocity is expressed as: v0=(s1-s0) / time, v1=(s2-s1) / time, and the acceleration is expressed as: a=(v1-v0) / time.
[0099] Therefore, we have a complete measurement input starting from time t2. By substituting the result obtained from the extrapolation equation and the measurement input into the filtering algorithm, we can obtain the measurement update value (i.e., the optimal position parameter).
[0100] Specifically, the filtering algorithm is as follows:
[0101] e x n，n = e x n，n-1 +K n (Z n -H e x n，n-1 ),
[0102]
[0103] K n =P n，n-1 H T HP n，n-1 H T +R n ) -1 ,
[0104] Where R is the covariance matrix of the measurement noise, which can be obtained from the measurement equipment manufacturer, calculated from data, derived theoretically, or obtained through the calibration process.
[0105] See Figure 3 In practice, the specific method for filtering (correcting) position parameters is as follows:
[0106] Step 0: Initialization;
[0107] Specifically, initialization is performed only once, providing the initial state. e x 0，0 ) and initial state variance (P 0，0 Two parameters are initialized before prediction.
[0108] Step 1: Measurement;
[0109] The measurement process also provides two parameters: the measured value (Z). n ) and measurement variance (R n );
[0110] Step 2: State update, output the optimal position parameters and the uncertainty (i.e., variance) of the optimal position parameters.
[0111] The input for state update is: measured value (Z) n Measurement variance (R) n ), state prediction ( e x n，n-1 ) and state prediction variance (P n，n-1 ).
[0112] Based on the above inputs, the state update process calculates the Kalman gain and generates the output: the position prediction for the next time step. e x n，n ) and location prediction variance (P n，n ).
[0113] Step 3: State Prediction
[0114] The prediction process uses a system dynamic model (i.e., extrapolation equations) to extrapolate the current position prediction and its variance to the next time step, outputting... e x n+1，n Pn+1，n .
[0115] Step 4: Measure and obtain the measured value (Z). n ) and measurement variance (R n );
[0116] Step 5: Status Update
[0117] Input: Output of step 3, measured value (Z) n ) and measurement variance (R n It should be noted that the output of step 3... e x n+1，n P n+1，n At the current point in time, it is represented as e x n，n-1 P n，n-1 ;
[0118] Output: State estimate ( e x n，n ), State estimation variance (P n，n ).
[0119] Step 6: State prediction;
[0120] The output of step 5, along with the input state extrapolation and covariance extrapolation equations, yields the following results: e x n+1，n P n+1，n :
[0121] Step 7: Measurement;
[0122] Step 8: Status update.
[0123] It should be noted that in the first iteration, the initialization parameters are used as the current system's position prediction values and variance (i.e., step 0, initialization), while in the second iteration, the position predictions generated in step 2 are used. e x n，n ) and location prediction variance (P n，n The output of the current time step position prediction is used as the input for the next time step position prediction, and so on, continuously updating the position prediction and its variance for the next time step. See [link to documentation]. Figure 5 This allows us to obtain a more accurate prediction of the position at the next moment.
[0124] S400 preprocesses the initial attitude parameters and the measured attitude parameters, then inputs them into the Kalman filter model to obtain the optimal attitude parameters;
[0125] In practice, the initial attitude parameters include the initial attitude and the initial attitude change rate; the measured attitude parameters include the measured attitude and the measured attitude change rate.
[0126] It should be noted that since we can only obtain quaternions and not angular velocity vectors, and since the attitude of the optical positioning device changes uniformly during surgery, we obtain the following equation as the problem equation for the attitude parameters:
[0127]
[0128] q tv =q t-1 / q t-2 ,
[0129] Where, q tv It also satisfies the unit quaternion requirement. During the operation, there will be no significant changes in speed. Only the changes at two moments are needed. Therefore, the acceleration model described above can meet the requirements for predictive filtering during surgery.
[0130] It should be noted that since we directly obtain the actual quaternion, regarding the q we mentioned at the beginning... t-1 / q t-2 The method for obtaining the position parameters is similar to that for obtaining the position parameters mentioned above. The method of caching the first few frames is adopted. Since the position filtering caches the first two frames, in order to maintain consistency with the position filtering, the attitude filtering also uses the method of caching the first two frames.
[0131] Therefore, our observation equation is directly:
[0132] Z t =q t .
[0133] In the optical positioner correction method of this embodiment, the method for obtaining the optimal attitude parameters includes:
[0134] S410 inputs the initial attitude change rate into a preset first attitude prediction model to obtain the prediction parameters of the attitude change rate at the next moment, wherein the initial attitude change rate is calculated from the initial attitude at two moments.
[0135] S420 inputs the predicted parameters of the attitude change rate at the next moment and the measured attitude change rate into the Kalman filter model to obtain the optimal parameters of the attitude change rate;
[0136] S430 inputs the initial attitude and the measured attitude into a preset second attitude prediction model to obtain the attitude prediction parameters for the next moment;
[0137] S440 inputs the attitude prediction parameters for the next time step into the Kalman filter model, uses the attitude change rate as the observation matrix in the Kalman filter model, and obtains the optimal attitude parameters.
[0138] It should be noted that attitude filtering is divided into two steps, q tv Filtering and q t Filtering, where q tv The filtering uses the first attitude prediction model for q tv Predict the state at the next time step and set q tv The predicted state output value at the next time step is input into the Kalman filter model to obtain q. tv The filtered output result is used as q. t Since the observation matrix in the second attitude prediction model can only obtain quaternions and not angular velocity vectors through measurement, the above method only requires measuring the attitude quaternions corresponding to two time moments to obtain the attitude prediction for the next time moment, making the measurement and calculation process simpler.
[0139] In the optical positioner correction method of this embodiment, the preset first attitude prediction model includes:
[0140] e q n+1，n =q n，n ,
[0141] P′ n+1，n =F′P′ n，n F′ T +Q′,
[0142]
[0143] in, e q n+1，n q is the prediction of the attitude change rate at time n+1 from time n. n，n Let P' be the attitude change rate at time n, and F' be the state transition matrix; n，n P′ is the square of the stability of the predicted attitude change rate at time n. n+1，n Let Q' be the square of the stability of the attitude change rate prediction at time n+1, Q' be the process noise covariance matrix, and R be the measurement noise covariance matrix.
[0144] In this way, since the orientation of the optical locator changes at an almost uniform speed during surgery, the orientation change rate at the next moment can be directly derived from the orientation change rate at the current moment. The deterministic prediction of the orientation change rate at the next moment can be made by using the process noise covariance and measurement noise covariance. This can be mutually verified with the prediction of the next moment from the current moment, making the prediction of the next moment more accurate.
[0145] Preferably, the initial input
[0146] In the optical positioner correction method of this embodiment, the preset second attitude prediction model includes:
[0147] e q′ n+1，n =F″ e q′ n，n ,
[0148] P″ n+1，n =F″P″ n，n F″ T +Q″,
[0149]
[0150] in, e q′ n+1，n Let q′ be the prediction of the attitude at time n+1 from time n. n，n Let F″ be the pose at time n, and F″ be the state transition matrix; P″ be the pose at time n. n，n P″ is the square of the stability of the predicted attitude change rate at time n. n+1，n F″ is the square of the stability of the attitude change rate prediction at time n+1, and Q″ is the process noise covariance. Specifically, F″ = e x n，n Initial input In this way, by selecting a matrix suitable for attitude changes as the state transition mechanism, the second attitude prediction model can make the prediction of the attitude at the next moment more accurate.
[0151] Example 3:
[0152] See Figure 4 The attitude filtering method in this embodiment is as follows:
[0153] Step 0:
[0154] q tv The filtering is achieved by extrapolating the initial state and the initial state variance using the equation. e q n+1，n =q n，n , P′ n+1，n =F′P′ n， n F′ T +Q′ to q tv Make predictions;
[0155] q t The filtering is achieved by extrapolating the initial state and the initial state variance using the equation. e q′ n+1，n =F″ e q′ n，n , P″ n+1，n =F″P″ n，n F″T Use +Q″ for prediction;
[0156] Step 1:
[0157] Measurement:
[0158] q tv Input measurement value q t-1 / q t-2 and measurement variance R tv ;
[0159] q t Input measurement value q t-1 and measurement variance R t ;
[0160] Step 2:
[0161] State update: Output the optimal pose and the uncertainty (i.e., variance) of the optimal pose.
[0162] Specifically, input the q obtained in step 0. tv The predicted value, q obtained in step 0 t Predicted value, measured value q t-1 / q t-2 and measurement variance R tv Through K n =P n，n-1 H T HP n，n-1 H T +R n ) -1 Calculate the Kalman gain; via K n =P n，n-1 H T HP n，n-1 H T +R n ) -1 Calculate the optimal pose; through Calculate the uncertainty of the optimal pose.
[0163] Specifically, q tv The predicted values are input into the Kalman filter model to obtain... e x tv P tv and put e x tv P tv Substituting it into the Kalman filter model, as the observation matrix H, for q t Make a prediction and obtain e x t P t .
[0164] Step 3:
[0165] State prediction: by e x n，n P n，n roll out e x n+1，n P n+1，n ;
[0166] Step 4:
[0167] q tv and q t The measurements obtained are: measured value (Z) and measurement variance (R).
[0168] Step 5:
[0169] State update: Calculate the Kalman gain, optimal pose, and uncertainty of the optimal pose;
[0170] Repeat steps 3-5 above to update the optimal attitude parameters for the next time step in real time.
[0171] The S500 corrects the positioning of the optical positioner based on the optimal position parameters and optimal attitude parameters, thereby obtaining the optimal pose of the optical positioner.
[0172] See Figure 6 Since we correct (i.e. filter) the position and attitude parameters of the optical locator separately, we obtain the optimal position and attitude parameters. By combining the two, we can obtain the optimal estimated pose of the optical locator, complete the correction of the optical locator, and enable the optical locator and skeleton to be accurately registered.
[0173] This embodiment provides a correction system for an optical locator, comprising: a data acquisition unit for acquiring initial position parameters, measured position parameters, initial attitude parameters, and measured attitude parameters of the optical locator; a position prediction unit for inputting the initial position parameters into a preset position prediction model to obtain predicted position parameters for the next time step; a position optimization unit for inputting the predicted position parameters for the next time step and the measured position parameters for the current time step into a preset Kalman filter model to obtain optimal position parameters for the next time step; an attitude optimization unit for preprocessing the initial attitude parameters and the measured attitude parameters before inputting them into the Kalman filter model to obtain optimal attitude parameters; and a positioning correction unit for correcting the positioning of the optical locator based on the optimal position parameters and optimal attitude parameters to obtain the optimal pose of the optical locator.
[0174] It should be noted that orthopedic surgical robots need to plan their procedures based on images before surgery to save intraoperative time and facilitate positioning. This preoperative planning includes the following steps:
[0175] Import the patient's images, which are generally CT images that provide excellent bone visualization, but other modalities, such as MRI, can also be used.
[0176] Based on the imported patient images, obtain the skeletal model of the patient's affected area.
[0177] Install NDI optical locator devices around the patient's affected area.
[0178] Obtain bony landmarks from the patient.
[0179] Preset registration points
[0180] By aligning the patient's bony landmarks with preset registration points, the bones under the optical locator device are registered onto the bone model.
[0181] It is understandable that the NDI optical locator and the skeleton can be considered as a completely rigid whole, so the pose of the skeleton can be determined by locating the NDI optical locator.
[0182] Finally, see Figure 7 The method of correction of an optical locator based on imported images, which is provided in this embodiment, is used to filter out jitter, reduce errors caused by bone jitter, and improve the accuracy of registration.
[0183] This embodiment provides a computer-readable storage medium storing computer-readable instructions, characterized in that, when executed by a processor, the computer-readable instructions implement the various steps of the optical positioner correction method described above.
[0184] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0185] Optionally, specific examples in this embodiment can refer to the examples described in the above embodiments, and will not be repeated here.
[0186] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.
[0187] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.
[0188] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.
Claims
1. A method for correcting the deviation of an optical positioner, characterized in that, include: Obtain the initial position parameters, measured position parameters, initial attitude parameters, and measured attitude parameters of the optical positioner; The initial position parameters are input into a preset position prediction model to obtain the predicted position parameters for the next moment. The predicted position parameters at the next time step and the measured position parameters at the current time step are input into a preset Kalman filter model to obtain the optimal position parameters at the next time step. After preprocessing the initial attitude parameters and the measured attitude parameters, they are input into the Kalman filter model to obtain the optimal attitude parameters; Based on the optimal position parameters and optimal attitude parameters, the positioning of the optical positioner is corrected to obtain the optimal pose of the optical positioner.
2. The correction method for the optical positioner according to claim 1, characterized in that, The location prediction model is specifically as follows: Q=FQ a F T , P n+1,n =FP n,n F T +Q, in, e x n+1，n This is the prediction of the registration point position at time n+1 from time n. e x n,n Let P be the real-time position of the preset registration point at time n, and F be the state transition matrix; n,n P is the square of the stability of the predicted registration point position at time n. n+1,n Let F be the square of the stability of the predicted registration point position at time n+1, F be the state transition matrix, and Q be the process noise covariance. Let V be the variance of the acceleration.
3. The correction method for the optical positioner according to claim 2, characterized in that, The state transition matrix F is specifically: Among them, ex n+1,n , e y n+1,n , e z n+1,n It is a prediction of the future position and state at time n (time n+1). e x n+1,n ′, e y n+1,n ′, e z n+1,n ′ is the prediction of the future velocity state at time n (time n+1). e x n+1,n ", e y n+1,n ", e z n+1,n "It is a prediction of the future acceleration state at time n (time n+1)." e x n,n , e y n,n , e z n,n Let n be the real-time position of the preset registration point at time n. e x n,n ′, e y n,n ′, e z n,n ' is the real-time velocity of the preset registration point at time n. e x n,n ", e y n,n ", e z n,n " is the real-time acceleration of the preset registration point at time n.
4. The correction method for the optical positioner according to claim 3, characterized in that, The Kalman filter model is specifically as follows: e x n,n = e x n,n-1 +K n (Z n -H e x n,n-1 ), K n =P n,n-1 H T (HP n,n-1 H T +R n ) -1 , in, e x n+1,n This is the prediction of the registration point position at time n+1 from time n. e x n,n Z represents the real-time position of the preset registration point at time n. n Let H be the observation vector of the preset registration point of the optical locator, and let K be the observation matrix of the preset registration point of the optical locator. n For Kalman gain, R n This is the covariance matrix for measuring noise.
5. The correction method for the optical positioner according to claim 1, characterized in that, The initial attitude parameters include the initial attitude and the initial attitude change rate; the measured attitude parameters include the measured attitude and the measured attitude change rate.
6. The correction method for the optical positioner according to claim 5, characterized in that, The method for obtaining the optimal attitude parameters includes: The initial attitude change rate is input into a preset first attitude prediction model to obtain the prediction parameters of the attitude change rate at the next moment. The initial attitude change rate is calculated from the initial attitude at two moments. The predicted parameters of the attitude change rate at the next moment and the measured attitude change rate are input into the Kalman filter model to obtain the optimal parameters of the attitude change rate. The initial attitude and the measured attitude are input into a preset second attitude prediction model to obtain the attitude prediction parameters for the next moment. The attitude prediction parameters for the next time step are input into the Kalman filter model, and the attitude change rate is used as the observation matrix in the Kalman filter model to obtain the optimal attitude parameters.
7. The correction method for the optical positioner according to claim 6, characterized in that, The preset first attitude prediction model includes: e q n+1,n =q n,n , P′ n+1,n =F′P′ n,n F′ T +Q′, in, e q n+1,n q is the prediction of the attitude change rate at time n+1 from time n. n,n Let P' be the attitude change rate at time n, and F' be the state transition matrix; n,n P′ is the square of the stability of the predicted attitude change rate at time n. n+1,n Let Q' be the square of the stability of the attitude change rate prediction at time n+1, Q' be the process noise covariance matrix, and R be the measurement noise covariance matrix.
8. The correction method for the optical positioner according to claim 1, characterized in that, The preset second attitude prediction model includes: e q′ n+1,n =F″ e q′ n,n , P″ n+1,n =F″P″ n,n F″ T +Q″, in, e q′ n+1,n Let q′ be the prediction of the attitude at time n+1 from time n. n,n Let F″ be the pose at time n, and F″ be the state transition matrix; P″ be the pose at time n. n,n P″ is the square of the stability of the predicted attitude change rate at time n. n+1,n Q″ is the square of the stability of the attitude change rate prediction at time n+1, and Q″ is the process noise covariance.
9. A correction system for an optical positioner, characterized in that, include: The data acquisition unit is used to acquire the initial position parameters, measured position parameters, initial attitude parameters, and measured attitude parameters of the optical positioner. A position prediction unit is used to input the initial position parameters into a preset position prediction model to obtain the predicted position parameters for the next moment. A position optimization unit is used to input the predicted position parameters at the next time moment and the measured position parameters at the current time moment into a preset Kalman filter model to obtain the optimal position parameters at the next time moment. An attitude optimization unit is used to preprocess the initial attitude parameters and the measured attitude parameters, and then input them into the Kalman filter model to obtain the optimal attitude parameters. A positioning correction unit is used to correct the positioning of the optical positioner based on the optimal position parameters and the optimal attitude parameters, so as to obtain the optimal pose of the optical positioner.
10. A computer-readable storage medium storing computer-readable instructions thereon, characterized in that, When the computer-readable instructions are executed by a processor, they implement the steps of the correction method for the optical positioner as described in any one of claims 1-8.