Three-ring positioner and robot-assisted intervertebral foramen mirror positioning method

Abstract

The invention provides a three-ring positioner and a robot-assisted intervertebral foramen mirror positioning method, and relates to the field of auxiliary positioning. The method comprises the following steps: extracting 12 sparse feature points from an X-ray image shot based on a three-ring locator, and constructing an initial projection matrix; based on the sparse feature points, 24 dense feature points are extracted from the X-ray image, an LM algorithm is adopted to carry out iterative optimization on the initial projection matrix, and an optimized projection matrix model is constructed; selecting a needle insertion point on the X-ray image based on the optimized projection matrix model; calculating the required movement amount of the three-ring positioner moving to the needle feeding point; and controlling a mechanical arm of the robot to move to a target position according to the required movement amount. The method can improve the positioning precision and reduce the radiation quantity.

Description

Technical Field

[0001] The invention relates to the field of auxiliary positioning, and in particular to a three-ring positioner and a robot-assisted intervertebral foraminal endoscope positioning method. Background Art

[0002] Spinal diseases such as lumbar disc herniation are common and frequently occurring diseases in clinic. Transforaminal endoscopic lumbar disc surgery is a minimally invasive surgery for the treatment of lumbar disc herniation. It is currently the most minimally invasive, safest and most economical method for the treatment of spinal diseases such as disc herniation and lumbar spinal stenosis. According to the differences in body position and approach, transforaminal endoscopic lumbar disc surgery can be divided into coaxial endoscopic spine system (Yeung Endoscopic Spine System, YESS) technology, transforaminal endoscopic spine system (Transforaminal Endoscopic Spine System, TESSYS) technology and posterior interlaminar approach technology.

[0003] Due to the minimally invasive requirements of the perforaminal endoscope, the affected area cannot be directly incised for surgery. The usual practice is to first use a guide wire for puncture positioning, and then expand the channel to meet the intervention requirements of the perforaminal endoscope. However, the target area cannot be directly viewed during the puncture process. Traditional perforaminal endoscope puncture positioning requires doctors to use C-arm X-ray machines to repeatedly perform fluoroscopy and positioning punctures. The operation has low precision, long time, great harm to patients, high radiation, and high requirements on the doctor's experience and physical strength. Summary of the invention

[0004] The purpose of the present invention is to provide a three-ring positioner and a robot-assisted transforaminal endoscopic positioning method to solve the problem of low positioning accuracy.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A three-ring positioner, comprising: an end flange, an upper positioning plate, a lower positioning plate, a large circular ring, an indicator needle, a small circular ring and a middle circular ring;

[0007] The end flange is used to fix the upper positioning plate and the lower positioning plate;

[0008] The indicator needle is arranged on the upper positioning plate and the lower positioning plate; one end of the indicator needle is fixed to the small ring, and the other end of the indicator needle is fixed to the upper positioning plate;

[0009] The large circular ring is arranged on the upper surface of the upper positioning plate; the middle circular ring is arranged on the lower surface of the lower positioning plate;

[0010] The end of the robot's mechanical arm is connected to a three-ring locator, and the three-ring locator is placed under an X-ray machine to take an X-ray image.

[0011] A robot-assisted transforaminal endoscopic positioning method, the positioning method is applied to the above-mentioned three-ring positioner, and the positioning method comprises:

[0012] Extract 12 sparse feature points from the X-ray image taken by the three-ring locator and construct the initial projection matrix;

[0013] Based on the sparse feature points, 24 dense feature points are extracted from the X-ray image, and the initial projection matrix is ​​iteratively optimized using the LM algorithm to construct an optimized projection matrix model;

[0014] Based on the optimized projection matrix model, selecting a needle insertion point on the X-ray image;

[0015] Calculating the required movement amount of the three-ring locator to the needle insertion point;

[0016] According to the required movement amount, the robot's mechanical arm is controlled to move to the target position; the target position is the location of the needle insertion point.

[0017] Optionally, 12 sparse feature points are extracted from the X-ray image taken by the three-ring locator to construct an initial projection matrix, specifically including:

[0018] Extracting the projections of the centerline of the indicator needle and the centerlines of the three circular rings in the X-ray image; wherein the projection of the centerline of the indicator needle in the X-ray image is a straight line; the projection of the centerline of the circular ring in the X-ray image is an ellipse; the three circular rings include a large circular ring, a middle circular ring and a small circular ring;

[0019] Based on the projection of the center line of the indicator needle and the center lines of the three circular rings, the intersection points of the straight lines passing through the center of each circular ring and parallel or perpendicular to the indicator needle and each of the circular rings are selected; the intersection points are used as sparse feature points;

[0020] An initial projection matrix is ​​constructed according to the sparse feature points using a direct linear transformation method.

[0021] Optionally, based on the sparse feature points, 24 dense feature points are extracted from the X-ray image, and the initial projection matrix is ​​iteratively optimized using the LM algorithm to construct an optimized projection matrix model, specifically including:

[0022] Dividing the dense feature points into a coordinate control group and an ellipse control group;

[0023] Determining the basic residuals of the coordinate control group and the basic residuals of the ellipse control group;

[0024] Obtaining an original step length of processing the initial projection matrix using a direct linear transformation method, and determining a transformed step length;

[0025] Determining a normalized residual according to the transformed step size;

[0026] According to the normalized residual, the current step size is optimized and iterated to construct an optimized projection matrix model.

[0027] Optionally, according to the normalized residual, the current step size is optimized and iterated to construct an optimized projection matrix model, specifically including:

[0028] Determining a current step size according to the normalized residual;

[0029] Determine whether the current step size is less than the step size threshold; if so, construct an optimized projection matrix model according to the normalized residual; if not, calculate the fitting accuracy of the optimization function within the current step size range; the optimization function is an expression of the optimized projection matrix model;

[0030] When the fitting accuracy is less than 0.5, the step size at the next moment is calculated according to the current step size; whether the fitting accuracy is greater than 0.75 is determined, if so, the damping factor is reduced, and the normalized residual is determined according to the reduced damping factor; if not, the damping factor is increased, and the normalized residual is determined according to the increased damping factor;

[0031] When the fitting accuracy is not less than 0.5, the damping factor is increased, and the normalized residual is determined according to the increased damping factor.

[0032] Optionally, the fitting accuracy pr is:

[0033]

[0034] Among them, ε(θ k +Δθ k+1 )The first optimization function; ε(θ k ) is the second optimization function; is ε(θ k ) derivative; Δθ k+1 is the current step length; J(θ k ) is the Jacobian matrix of the optimization function; r(θ k ) is the optimization function residual.

[0035] Among them, ε(θ k +Δθ k+1 ) is the first cost function optimization function; ε(θ k ) is the second cost function optimization function; is the cost function gradient; Δθ k+1 is the current step length; J(θ k ) is the Jacobian matrix of the kth iteration; r(θ k ) is the optimization function residual of the kth iteration.

[0036] Optionally, controlling the robot arm to move to the target position according to the required movement amount specifically includes:

[0037] Move the three-ring locator according to the required movement amount, and obtain the first position matrix of the three-ring locator in the original object coordinate system after the movement;

[0038] Obtain a second pose matrix when the three-ring locator is installed at the end of the robotic arm, and a third pose matrix of the end of the robotic arm in the base coordinate system before moving the three-ring locator;

[0039] Determine a pose matrix of the end of the robotic arm after the movement in a base coordinate system according to the first pose matrix, the second pose matrix and the third pose matrix;

[0040] The end of the robotic arm is controlled to move to the target position according to the posture matrix.

[0041] Optionally, the pose matrix T' tool-base for:

[0042] T' tool-base =T tool-base T l-tool T'(T l-tool ) -1 ;

[0043] Among them, T' is the first pose matrix; T l-tool is the second pose matrix; T tool-base is the third pose matrix.

[0044] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects: the present invention provides a three-ring locator and a robot-assisted percutaneous endoscopic positioning method, which only needs to take an X-ray image, and by extracting sparse feature points and dense feature points in the X-ray image, it is possible to construct a projection matrix for moving the three-ring locator to the target position, thereby completing the rapid positioning of the percutaneous endoscopic channel, avoiding repeated X-ray radiography of the human body, and reducing the amount of radiation; in addition, the projection matrix is ​​iteratively optimized using the LM algorithm to further improve the positioning accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0046] Figure 1An exploded view of the three-ring positioner provided by the present invention;

[0047] Figure 2 A three-dimensional diagram of the three-ring positioner provided by the present invention;

[0048] Figure 3 A flow chart of the robot-assisted transforaminal endoscopic positioning method provided by the present invention;

[0049] Figure 4 This is a schematic diagram of the distribution of sparse 3D feature points provided by the present invention;

[0050] Figure 5 A schematic diagram of dense 3D feature points provided by the present invention;

[0051] Figure 6 This is a flow chart of the LM algorithm provided by the present invention. DETAILED DESCRIPTION

[0052] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] The purpose of the present invention is to provide a three-ring positioner and a robot-assisted transforaminal endoscopic positioning method, which can improve positioning accuracy and reduce radiation dose.

[0054] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0055] like Figure 1-Figure 2 As shown, the present invention provides a three-ring locator, comprising: an end flange 1, an upper positioning plate 2, a lower positioning plate 3, a large circular ring 4, an indicator needle 5, a small circular ring 6 and a middle circular ring 7; the end flange 1 is used to fix the upper positioning plate 2 and the lower positioning plate 3; the indicator needle 5 is arranged on the upper positioning plate 2 and the lower positioning plate 3; one end of the indicator needle 5 fixes the small circular ring 6, and the other end of the indicator needle 5 fixes the upper positioning plate 2; the large circular ring 4 is arranged on the upper surface of the upper positioning plate 2; the middle circular ring 7 is arranged on the lower surface of the lower positioning plate 3; the end of the robot's mechanical arm is connected to the three-ring locator, and the three-ring locator is placed under an X-ray machine to take X-ray images.

[0056] In practical applications, a three-ring locator is used to design a DLT-LM algorithm based on X-ray images. The three-ring locator is connected with a robotic arm and placed under an X-ray machine. After shooting, the X-ray image is used for spatial positioning.

[0057] The imaging principle of X-ray images can be approximately described by the pinhole camera model. The X-ray pinhole camera model transforms the feature points in the object coordinate system into the X-ray image coordinate system through 3D-2D transformation. Its 3D-2D transformation equation is:

[0058] p i =Aq i (1)

[0059]

[0060] Among them, p i (u i ,v i ), i = 1, 2, ..., n represents the 2D homogeneous coordinates of the feature points in the X-ray image coordinate system, q i (x i ,y i ,z i ), i = 1, 2, ..., n are the 3D homogeneous coordinates of the feature points in the object coordinates, and A is the 3D-2D transformation matrix. The X-ray image projection model calculation can be described as: given n groups of corresponding 2D feature points p i (i=1,2,…,n) and 3D feature point q i (i=1,2,…,n), solve the 3D-2D transformation matrix A, A has a total of 11 parameters to be solved.

[0061] like Figure 3 As shown, based on the above three-ring positioner, the present invention also provides a robot-assisted transforaminal endoscopic positioning method, comprising:

[0062] Step 301: Extract 12 sparse feature points from the X-ray image taken by the three-ring locator and construct an initial projection matrix.

[0063] In practical applications, the direct linear transformation method is used to calculate the X-ray image projection model. Assuming that n groups p are known, i and q i , where the 3D-2D transformation equation of a set of feature points is expanded to obtain:

[0064]

[0065] Writing the 3D-2D transformation equation of n sets of feature points into a matrix yields:

[0066]

[0067] Among them, A1~A 11 is an unknown quantity, so we can simplify Formula 4 and use matrix operation rules to get:

[0068] MX=B (5)

[0069] X=(M T M) -1 M T B (6)

[0070] in,

[0071] If the equation has a solution, then 2n>11 is required, so at least 6 groups of p are needed. i and q i , the direct linear transformation method can be used to solve the 3D-2D transformation matrix A. The feature points corresponding to this method are called sparse feature points, such as Figure 4 shown.

[0072] 3D sparse feature point extraction is performed in the object coordinate system, and the intersection of the rings with a straight line passing through the center of each ring and parallel or perpendicular to the indicator needle 5 is selected.

[0073] The main steps for extracting 2D sparse feature points in X-ray images are:

[0074] 1. Extract the indicator needle 5 and three circular rings in the X-ray image. The projection of the center line of the indicator needle 5 in the 2D image is still a straight line, and the projection of the center line of the circular ring in the 2D image becomes an ellipse, so the ellipse is used to fit the circular ring.

[0075] 2. Take the intersection of the straight line passing through the center of each 2D ring and parallel or perpendicular to the indicator pin 5 and the 2D ring. Each 2D ring can get 4 feature points, and a total of 12 sparse 2D feature points p for the three rings i (i=1,2,…,12). Then, the 3D-2D transformation matrix A, i.e. the initial projection matrix, can be obtained by using the direct linear transformation method.

[0076] Step 302: Based on the sparse feature points, 24 dense feature points are extracted from the X-ray image, and the initial projection matrix is ​​iteratively optimized using the LM algorithm to construct an optimized projection matrix model.

[0077] The variable to be optimized in the present invention is the 3D-2D conversion matrix A, so the optimization variable θ=[A1 A2 … A 11 ] T , the objective function is f i (θ) = e i (i=1,2,…,n), then the residual is r i (θ)=[f i (θ)-e i ](i=1,2,…,n), where e iis the value of the objective function at the current variable, the optimization function can be written as:

[0078]

[0079] r(θ)=[r1(θ) r2(θ) … r n (θ)] T (8)

[0080] Among them, ε(θ) is the cost function optimization function, r(θ) is the residual function matrix optimization function residual, r i (θ) is the residual of the i-th feature point.

[0081] Let θ k is the kth iteration result, i.e., the kth optimization variable, H(θ k ) is ε(θ k ), the iterative formula of θ is:

[0082]

[0083] Among them, θ k+1 is the k+1th iteration result, r p (θ) is the residual error of the p-th feature point, θ q is the qth element of θ, is the gradient of the cost function, and J(θ) is the Jacobian matrix of the optimization function.

[0084] Simplifying the calculation of the Hessian matrix, the iterative formula of θ is approximately

[0085]

[0086] Among them, r(θ k ) is the optimization function residual of the kth iteration.

[0087] The damping factor μ (μ>0) is introduced in the solution process, and the iterative formula (11) of θ is adjusted to:

[0088] θ k+1 ≈θ k -[J(θ) T J(θ)+μI] -1 J(θ) T r(θ) (12)

[0089] Where I is the identity matrix and μ ensures that [J(θ) T J(θ)+μI] is positive definite, and the iteration is carried out in the direction of the descent of the optimization function.

[0090] When designing the optimization function of the 3D-2D conversion matrix of X-ray images, the objective function is established using the ellipse equation and some feature points extracted from the 2D image. In order to make the optimization function more stable, when extracting feature points, the extraction density is increased on the basis of sparse feature points, and dense feature points are established, such as Figure 5 shown.

[0091] Based on the extraction of sparse 3D feature points, feature points corresponding to each 45° angle are added to form a total of 24 dense 3D feature points.

[0092] Since the 3D feature points q1 and q2 correspond exactly to the 2D feature point coordinates p1 and p2 on the image, the 24 dense 3D feature points are divided into two groups, q1 and q2 are the coordinate corresponding groups, and the rest are the ellipse corresponding groups. The 3D feature points in the ellipse corresponding groups are all located on the ellipse curve of the 2D image. The 3D feature points and 2D feature points in the coordinate corresponding groups are still represented by q1 and q2. i (i=1,2) and p i (i=1,2) indicates that all pi are image coordinates corresponding to qi. The 3D feature points of the small circle 6, the middle circle 7 and the large circle 4 in the ellipse corresponding group are represented by and express.

[0093] The coordinate corresponding group directly uses the coordinate value as the design basis of the objective function, which can be

[0094]

[0095] Among them, f u,i (θ) and f v,i (θ) is the objective function, x i ,y i and z i is the coordinate of the 3D feature point, e u,i 、e v,i is the residual value.

[0096] The basic residual of the coordinate correspondence group is

[0097]

[0098] Among them, r' u,i (θ) and r' v,i (θ) is the residual.

[0099] The ellipse correspondence group uses the characteristic that the 3D feature points fall on the ellipse curve after projection to design the objective function, which can be:

[0100] Among them, f E,i (θ) is the objective function, d E , eE , g E ,h E , j E are the coefficients of the equation, is the 2D coordinate, e E,i is the function value of the equation, S, M, L represent the codes for small, medium and large rings respectively.

[0101]

[0102] in, are the homogeneous coordinates of the corresponding group of projection points of the ellipse, is the 3D homogeneous coordinate, are homogeneous coordinate coefficients, is the 2D coordinate, are 3D coordinates and A' is the coefficient.

[0103] The basic residual r' of the ellipse corresponding group E,i (θ) is

[0104] r E,i (θ) = f E,i (θ)-e E,i ,(E=S,M,L) (17)

[0105] Since the dimensions of the basic residuals of the coordinate correspondence group and the ellipse correspondence group are inconsistent, it is not reliable to directly use the basic residuals to design the optimization function. Therefore, it is necessary to normalize the basic residuals of the coordinate correspondence group and the ellipse correspondence group.

[0106] Assume that the initial value of θ obtained by direct linear transformation is θ 0 , according to θ 0 The basic residual obtained is r' u,i (θ 0 ), r' c,i (θ 0 ) and r' E,i (θ 0 ). In the subsequent residual calculations, they are divided by the average value of each group of basic residuals, that is, at the kth iteration, θ k The corresponding residuals are

[0107]

[0108] Among them, r c and r f is the intermediate quantity, r' S,i (θ k ), r' M,i (θ k ) and r' L,i (θ k) are the residuals corresponding to small, medium and large circles respectively.

[0109] At this time, the X-ray image 3D-2D conversion matrix optimization function can be designed as

[0110]

[0111] Where r(θ) is the vertical arrangement matrix of the coordinate corresponding group residuals and the ellipse corresponding group residuals, with a size of 26×1.

[0112] The LM algorithm process is as follows Figure 6 As shown, first calculate θ 0 The basic residual of , and then get the normalized residual, and calculate the current step size Δθ k+1 When the current step length is less than the threshold h, the algorithm ends. Otherwise, the optimization function fitting accuracy within the current step length range is calculated. If the accuracy is less than 0.5, the damping factor is increased and the step length is recalculated. If it is greater than 0.5, the result is adjusted according to the step length. If the accuracy is higher, the damping factor is reduced to improve the iteration efficiency.

[0113] When calculating the iteration step, the optimization function ε(θ) is expanded using the Taylor formula and simplified by ignoring the higher-order terms above the quadratic term. In addition, due to the high computational complexity of the Hessian matrix, the Hessian matrix is ​​expressed as J(θ) T J(θ) is used to simplify the calculation. Therefore, it is necessary to consider the optimization function fitting accuracy pr after simplified calculation, and the formula is

[0114]

[0115] The numerator is the step length Δθ k+1 After that, the actual change of the optimization function, and the denominator is the gradient of the simplified calculation of the optimization function Multiply the current step size Δθ k+1 ,ε(θ k +Δθ k+1 ) is the first cost function optimization function. When pr is less than 0.5, it means that the simplified optimization function ε(θ k ) after a step length Δθ k+1 The change is greater than ε(θ k ), so the fitting result is poor, and it is necessary to increase the damping factor and reduce the step size; when pr is greater than 0.5, it is considered that the fitting effect is good, but when pr is greater than 0.75, it means that the simplified optimization function ε(θ k ) after a step length Δθ k+1 The change in the value of is approximately equal to or less than the actual change, so it can be considered that the current step length Δθ k+1 Too conservative. It is necessary to appropriately reduce the damping factor, increase the step size, and improve the iteration efficiency.

[0116] Step 303: Based on the optimized projection matrix model, select a needle insertion point on the X-ray image.

[0117] Step 304: Calculate the required movement amount of the three-ring locator to the needle insertion point.

[0118] Step 305: According to the required movement amount, the robot arm is controlled to move to the target position, wherein the target position is the position of the needle insertion point.

[0119] It can be seen that the robot-assisted transforaminal endoscopic positioning method provided by the present invention is an X-ray image projection model calculation method based on DLT-LM, which can realize posterior approach transforaminal endoscopic positioning in combination with the three-ring positioner and the robotic arm. After the X-ray image shooting and projection model calculation are completed using the three-ring positioner, the needle insertion point can be selected on the X-ray image, and the movement amount required for the three-ring positioner to move to the needle insertion point can be calculated, thereby controlling the robotic arm to move to the specified position.

[0120] After the X-ray image calibration is completed, select the needle insertion point (u t ,v t ), in three-dimensional space, this point corresponds to a ray that emanates from the optical center and intersects the small ring plane z in turn. S = 0, middle ring plane z M =-33, where the z coordinates of the plane where the small and medium rings are located. The intersection points of the ray where the selected needle point is located and the plane where the small and medium rings 7 are located are q S and q M . Assume that the X-axis direction of the original object coordinate system is The Y-axis direction is The Z-axis direction is The origin coordinate is O. The direction of the Z axis after movement is In order to minimize the change in posture before and after the movement, the direction of the Y axis after the movement is taken as The direction of the X axis after moving is The coordinates of the origin after movement are O'=q S , thus we can get the first pose matrix T' of the three-ring locator in the original object coordinate system after it moves. Let the TCP matrix of the three-ring locator installed at the end of the robot arm be the second pose matrix T l-tool , the third pose matrix of the end of the manipulator in the base coordinate system before moving the three-ring locator is T tool-base , which can be read through the robot control box, then the pose matrix of the end of the robot in the base coordinate system after movement is

[0121] T' tool-base =T tool-base T l-tool T'(Tl-tool ) -1 (twenty one)

[0122] By controlling the end of the robotic arm to move to the specified position, the needle insertion point on the X-ray image can be positioned.

[0123] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0124] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A three-ring positioner, characterized in that: include: End flange, upper positioning plate, lower positioning plate, large ring, indicator needle, small ring and middle ring; The end flange is used to fix the upper positioning plate and the lower positioning plate; The indicator needle is arranged on the upper positioning plate and the lower positioning plate; one end of the indicator needle is fixed to the small ring, and the other end of the indicator needle is fixed to the upper positioning plate; The large circular ring is arranged on the upper surface of the upper positioning plate; the middle circular ring is arranged on the lower surface of the lower positioning plate; The end of the robot's mechanical arm is connected to a three-ring locator, and the three-ring locator is placed under an X-ray machine to take an X-ray image.

2. A robot-assisted transforaminal endoscopic positioning method, characterized in that: The positioning method is applied to the three-ring positioner according to claim 1, and the positioning method comprises: Extract 12 sparse feature points from the X-ray image taken by the three-ring locator and construct the initial projection matrix; Based on the sparse feature points, 24 dense feature points are extracted from the X-ray image, and the initial projection matrix is ​​iteratively optimized using the LM algorithm to construct an optimized projection matrix model; Based on the optimized projection matrix model, selecting a needle insertion point on the X-ray image; Calculating the required movement amount of the three-ring locator to the needle insertion point; According to the required movement amount, the robot's mechanical arm is controlled to move to the target position; the target position is the location of the needle insertion point.

3. The robot-assisted transforaminal endoscopic positioning method according to claim 2, characterized in that: Based on the X-ray image taken by the three-ring locator, 12 sparse feature points are extracted to construct the initial projection matrix, which includes: Extracting the projections of the centerline of the indicator needle and the centerlines of the three circular rings in the X-ray image; wherein the projection of the centerline of the indicator needle in the X-ray image is a straight line; the projection of the centerline of the circular ring in the X-ray image is an ellipse; the three circular rings include a large circular ring, a middle circular ring and a small circular ring; Based on the projection of the center line of the indicator needle and the center lines of the three circular rings, the intersection points of the straight lines passing through the center of each circular ring and parallel or perpendicular to the indicator needle and each of the circular rings are selected; the intersection points are used as sparse feature points; An initial projection matrix is ​​constructed according to the sparse feature points using a direct linear transformation method.

4. The robot-assisted transforaminal endoscopic positioning method according to claim 2, characterized in that: Based on the sparse feature points, 24 dense feature points are extracted from the X-ray image, and the initial projection matrix is ​​iteratively optimized using the LM algorithm to construct an optimized projection matrix model, specifically including: Dividing the dense feature points into a coordinate control group and an ellipse control group; Determining the basic residuals of the coordinate control group and the basic residuals of the ellipse control group; Obtaining an original step length of processing the initial projection matrix using a direct linear transformation method, and determining a transformed step length; Determining a normalized residual according to the transformed step size; According to the normalized residual, the current step size is optimized and iterated to construct an optimized projection matrix model.

5. The robot-assisted transforaminal endoscopic positioning method according to claim 4, characterized in that: According to the normalized residual, the current step size is optimized and iterated to construct an optimized projection matrix model, which specifically includes: Determining a current step size according to the normalized residual; Determine whether the current step size is less than the step size threshold; if so, construct an optimized projection matrix model according to the normalized residual; if not, calculate the fitting accuracy of the optimization function within the current step size range; the optimization function is an expression of the optimized projection matrix model; When the fitting accuracy is less than 0.5, the step size at the next moment is calculated according to the current step size; whether the fitting accuracy is greater than 0.75 is determined, if so, the damping factor is reduced, and the normalized residual is determined according to the reduced damping factor; if not, the damping factor is increased, and the normalized residual is determined according to the increased damping factor; When the fitting accuracy is not less than 0.5, the damping factor is increased, and the normalized residual is determined according to the increased damping factor.

6. The robot-assisted transforaminal endoscopic positioning method according to claim 5, characterized in that: The fitting accuracy pr is: Among them, ε(θ k +Δθ k+1 ) is the first cost function optimization function; ε(θ k ) is the second cost function optimization function; is the cost function gradient; Δθ k+1 is the current step length; J(θ k ) is the Jacobian matrix of the kth iteration; r(θ k ) is the optimization function residual of the kth iteration.

7. The robot-assisted transforaminal endoscopic positioning method according to claim 2, characterized in that: According to the required movement amount, the robot's mechanical arm is controlled to move to the target position, specifically including: Move the three-ring locator according to the required movement amount, and obtain the first position matrix of the three-ring locator in the original object coordinate system after the movement; Obtain a second pose matrix when the three-ring locator is installed at the end of the robotic arm, and a third pose matrix of the end of the robotic arm in the base coordinate system before moving the three-ring locator; Determine a pose matrix of the end of the robotic arm after the movement in a base coordinate system according to the first pose matrix, the second pose matrix and the third pose matrix; The end of the robotic arm is controlled to move to the target position according to the posture matrix.

8. The robot-assisted transforaminal endoscopic positioning method according to claim 7, characterized in that: The pose matrix T to ' ol-base for: T′ tool-base =T tool-base T l-tool T'(T l-tool ) -1 ; Among them, T' is the first pose matrix; T l-tool is the second pose matrix; T tool-base is the third pose matrix.

Images