Robot-assisted screw placement precision evaluation method

Abstract

The invention discloses a robot-assisted screw placement precision evaluation method, which comprises the following steps: S1, acquiring a medical image of a patient through a C-arm machine, and implanting an implant through robot navigation in an operation; s2, multi-angle perspective images are obtained, and implants in the perspective images are respectively corresponding to the multi-angle perspective images; s3, extracting the implanted vertebral body part of each implant in each perspective image; s4, obtaining projection points of inner and outer points of the planned approach channel on the axis of each implant in each perspective image; s5, according to the projection points obtained in the step S4, based on epipolar constraint, obtaining the actual positions of the implant on the inner and outer points of the vertebral body; and S6, calculating the theoretical positions of the inner and outer points of the implant on the vertebral body according to the inner and outer points of the pre-operative planned approach channel, and calculating the implantation precision of the implant in combination with the S5. According to the method, rapid evaluation of the precision of the robot-assisted screw placement system can be realized, the non-visual observation of fluoroscopy in an operation is avoided, the operation is simple, and the radiation of a patient is greatly reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a robot-assisted nail placement accuracy evaluation method. Background Art

[0002] In conventional vertebroplasty surgery, the construction of the access channel is a key step. With the widespread application of orthopedic assisted screw placement robots, this step is increasingly completed with the assistance of robots. The accuracy of the access channel construction is crucial. The conventional method is to confirm the Kirschner wire through anteroposterior and lateral fluoroscopy or intraoperative three-dimensional CBCT after the Kirschner wire is inserted into the vertebral body. However, intraoperative fluoroscopy itself is not intuitive, and three-dimensional CBCT not only increases the complexity of the operation, but also increases the patient's radiation. Therefore, intraoperative rapid evaluation of the accuracy of the access channel construction of orthopedic assisted screw placement surgery robots is of great significance. Summary of the invention

[0003] Purpose of the invention: In view of the above-mentioned deficiencies, the present invention proposes a method for evaluating the accuracy of robot-assisted nail placement, which can realize the rapid evaluation of the accuracy of the robot-assisted nail placement system, avoid the non-intuitive observation of intraoperative fluoroscopy, simplify the operation, and greatly reduce the patient's radiation.

[0004] Technical solution:

[0005] The present invention provides a robot-assisted nail placement accuracy evaluation method, comprising:

[0006] S1. Obtain medical images of the patient through a C-arm machine, and place the implant using robot navigation during surgery;

[0007] S2, operating the C-arm machine to perform multi-angle perspective on the surgical area where the implant is placed to obtain corresponding perspective images, and extracting the implants in the perspective images corresponding to each angle;

[0008] S3, transforming the inner and outer points of the preoperatively planned approach channel on the medical image into perspective images at various angles according to the projection model of the C-arm machine, and extracting the implanted vertebral body parts of each implant in each perspective image obtained in S2;

[0009] S4, fitting straight lines to the implanted vertebral body parts of each implant in each perspective image obtained in S3, respectively, to obtain the axis of each implant in each perspective image, thereby obtaining the projection points of the inner and outer points of the planned access channel on the axis of each implant in each perspective image according to S3;

[0010] S5, according to the projection points obtained in S4, based on the epipolar constraint, obtaining the actual positions of the internal and external points of the implant in the medical image, and then obtaining the actual positions of the internal and external points of the implant on the vertebral body;

[0011] S6. Calculate the theoretical positions of the inner and outer points of the implant on the vertebral body according to the inner and outer points of the planned access channel, and calculate the implant placement accuracy in combination with S5.

[0012] Specifically, S3 includes:

[0013] S31, obtaining projection points of inner and outer points of the planned access channel in each perspective image according to the projection model of the C-arm machine and the posture of the C-arm at the corresponding angle;

[0014] S32, combining the implants in each perspective image obtained in S2 and the projection points in each perspective image obtained in S31 to extract the implanted vertebral body portion of each implant in each perspective image.

[0015] More specifically, in S32, extracting the implanted vertebral body portion of each implant in the perspective image is specifically:

[0016] According to the projection points of the inner and outer points of the planned approach channel in each perspective image, the corresponding reference unit vector is constructed, and the pixel points of the implant area in each perspective image are traversed to determine whether its projection on the reference unit vector is between the two projection points mentioned above. If so, the pixel point is retained, otherwise it is discarded. After the traversal is completed, the inserted vertebral body part of each implant in each perspective image is extracted.

[0017] Specifically, in S5, the actual positions of the internal and external points of the implant in the medical image are obtained based on the epipolar constraint, as follows:

[0018] S51, according to the projection points of the inner and outer points of the planned access channel obtained in S4 on the axis of each implant in each perspective image, calculate the corresponding straight line in the perspective image of other viewing angles, and use the intersection point of the straight line and the axis of the corresponding implant in the perspective image corresponding to it obtained in S4 as the point corresponding to the projection point in the perspective image of the viewing angle;

[0019] S52. According to the corresponding points in the perspective images of each viewing angle obtained in S51, several straight lines passing through the optical center and the corresponding points are obtained, and the intersection points of these straight lines are fitted to obtain the actual positions of the corresponding points in the medical image, that is, the actual positions of the inner and outer points of the implant in the medical image.

[0020] Specifically, in said S5, after obtaining the actual positions of the inner and outer points of the implant in the medical image, the positions are transformed to the reference of the optical array on the C-arm or the optical array on the patient, that is, the actual positions of the inner and outer points of the implant on the vertebral body are obtained;

[0021] In S6, according to the planned inner and outer points of the approach channel, the points are transformed to the reference of the optical array on the C-arm or the optical array on the patient, so as to obtain the theoretical positions of the inner and outer points of the implant on the vertebral body.

[0022] Specifically, in S6, the implant placement accuracy includes implant angle error and implant distance error, wherein the implant angle error is the angle between the line connecting the theoretical positions of the inner and outer points of the implant on the vertebral body and the line connecting the actual positions of the inner and outer points of the implant on the vertebral body, and the implant distance error is the average value of the distances from the theoretical positions of the inner and outer points of the implant on the vertebral body to the lines connecting the actual positions of the inner and outer points of the implant on the vertebral body.

[0023] Specifically, in S2, the fluoroscopic image is segmented by a threshold value and then connected domain detection is performed to obtain the implant in the fluoroscopic image.

[0024] More specifically, the connected domain is defined as a region where image pixels are connected, and the specific feature of the connection between pixels is that four domains or eight neighborhoods are connected.

[0025] More specifically, after the connected domain is detected, the number of pixels in each connected domain is calculated, and the connected domains with the number of pixels less than a set threshold are deleted, so as to obtain the implant in the fluoroscopic image.

[0026] Beneficial effects: The present invention realizes rapid evaluation of the intraoperative accuracy of the robot nail making system through perspective images such as frontal and lateral positions, which has the following significant benefits:

[0027] 1) Avoid repeated fluoroscopic confirmation through multiple C-arms, reducing iatrogenic radiation during surgery;

[0028] 2) The present invention can achieve sub-millimeter quantitative evaluation, improve the accuracy and safety of surgery, and greatly improve the safety and success rate of surgical procedures with high precision requirements (such as cervical spine surgery, etc.). BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the technical solutions in the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are merely embodiments of the present invention, and ordinary technicians in this field can obtain other drawings based on these drawings without creative work.

[0030] Figure 1 It is a flow chart of the robot-assisted nail placement accuracy evaluation method of the present invention;

[0031] Figure 2 This is an example diagram of a robot navigation and positioning system;

[0032] Figure 3 This is an example diagram of multi-angle perspective projection;

[0033] Figure 4This is an example of a perspective image after the robot has completed nail placement. Figure 4 (a) is the frontal view after the robot nail placement is completed; Figure 4 (b) is the lateral view after the robot nail placement is completed;

[0034] Figure 5 Example image extracted for an implant, where Figure 5 (a) for Figure 4 An example of an image obtained by threshold segmentation. Figure 5 (b) is an example diagram for judging connectivity between pixels. Figure 5 (c) is an example of the results of implant connectivity detection. Figure 5 (d) is an example of the results of implant extraction;

[0035] Figure 6 An example diagram of segmenting an implant placed in a vertebral body, where: Figure 6 (a) is an example of transforming the internal and external points of the planned channel in the image to the image. Figure 6 (b) is an example of the result of segmenting the implant inserted into the vertebral body;

[0036] Figure 7 An example diagram of implant placement accuracy evaluation. DETAILED DESCRIPTION

[0037] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present application is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.

[0038] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present invention should have the common meanings understood by persons having ordinary skills in the field to which the present invention belongs.

[0039] The robot-assisted nail placement accuracy evaluation method of the present invention is as follows Figure 1 As shown, including:

[0040] S1. Obtain medical images of the patient through a C-arm machine, and place the implant using robot navigation during surgery;

[0041] like Figure 2 As shown, the robot navigation and positioning system mainly links the patient's medical image, the patient during surgery, and the C-arm machine, so that the access channel information in the patient's medical image can be converted to the reference of the patient during surgery. The patient during surgery is linked to the execution robot arm system in real time through the optical array, so as to realize the conversion of the access channel information planned in the patient's medical image to the execution component system to complete the nail placement operation.

[0042] In the present invention, the implanted object is generally a Kirschner wire, but it may also be other implants that need to be implanted into the human body.

[0043] S2, operating the C-arm machine to perform multi-angle perspective on the surgical area where the implant is placed to obtain corresponding perspective images, and extracting the implants in the perspective images corresponding to each angle;

[0044] After the implant is placed by robot navigation during surgery, the C-arm machine is used to perform multi-angle perspective of the surgical area where the implant is placed, such as Figure 4 As shown in FIG, during perspective, the optical array on the C-arm and the optical array on the patient need to be seen at the same time, that is, the perspective process of multiple angles essentially constitutes a two-dimensional perspective model, such as Figure 3 As shown, the two-dimensional perspective of the C-arm itself is a process of sampling a three-dimensional image into a two-dimensional image, which can be expressed by a projection model. The calibration of the C-arm has been completed before it leaves the factory, that is, the transformation relationship between the projection model and the optical array on the C-arm is known;

[0045] In the present invention, the implants in the perspective images corresponding to each angle are extracted, specifically, the perspective images are segmented by threshold and then connected domain detection is performed to obtain the implants in the perspective images.

[0046] In the present invention, the connected areas in the binary image after threshold segmentation of the perspective image also include some impurity noise areas that do not belong to the implant, such as Figure 5 As shown in B in (a), therefore, it is necessary to perform connected domain detection. After connected domain detection, several connected domains can be obtained in the binary image, such as Figure 5 (c) as shown.

[0047] In the present invention, a connected domain is defined as an area where image pixels are connected. Specifically, the connected features between pixels are four domains or eight neighborhoods. This embodiment uses four neighborhoods, such as Figure 5 For pixel p1 in (b), its neighborhood is region 1, region 2, region 3 and region 4, then p1 and p2 are adjacent, p1 and p3 are not adjacent, and p1 and p4 are not adjacent.

[0048] The area detected by the connected domain is φ 1 -φ 5 , and because the image area of ​​the implant region is relatively large, the number of pixels in each connected domain is calculated, and the connected domains with the number of pixels less than the set threshold are deleted to obtain the region corresponding to the implant, such as Figure 5 As shown in (d), the implant is obtained in the fluoroscopic image.

[0049] In the present invention, the threshold value may be set to 50.

[0050] S3, transforming the inner and outer points of the preoperatively planned approach channel into perspective images of various angles according to the projection model of the C-arm machine, and extracting the implanted vertebral body parts of each implant in each perspective image obtained in S2;

[0051] In the present invention, S2 has extracted the implant. This embodiment takes the Kirschner wire as an example. Generally, the Kirschner wire is inserted into the vertebral body about 2-3 cm. In order to facilitate the actual fluoroscopy, the external Kirschner wire will be bent and clamped on the sterile cloth, which will cause the Kirschner wire to bend to a certain extent. However, the tip of the Kirschner wire, which is about 2 cm in length, will not bend because it is inserted into the vertebral body. Therefore, the present invention needs to separate the Kirschner wire inserted into the vertebral body.

[0052] Specifically include:

[0053] S31, according to the projection model of the C-arm machine and the posture of the C-arm at the corresponding angle, the projection points of the inner and outer points of the access channel planned on the medical image in each perspective image can be obtained;

[0054] In the present invention, an operator such as a doctor can plan the inner and outer points of the access channel in the medical image, which can be done before and during the operation.

[0055] Taking the first perspective projection as an example, refer to Figure 6 (a) as follows:

[0056]

[0057] in, are the projection points of the inner and outer points of the planned approach channel in the perspective image obtained by the first perspective projection; F 2 The mapping relationship between points in the medical image and points on the perspective image can be obtained through the projection model, specifically according to the transformation relationship of the aforementioned projection model relative to the optical array on the C-arm and the position and posture of the optical array on the C-arm obtained in real time.

[0058] Specifically, according to C img With C carm The relative transformation relationship between Transform the inner and outer points of the planned approach channel to the optical array reference on the C-arm, that is:

[0059]

[0060] in, They are respectively the inner and outer points of the approach channel transformed to the reference of the optical array on the C-arm;

[0061] Afterwards, according to the transformation relationship between the aforementioned projection model and the optical array on the C-arm and the position and posture of the optical array on the C-arm acquired in real time, the mapping relationship between the point in the medical image and the point on the perspective image can be obtained. 2 .

[0062] S32, combining the implants in each perspective image obtained in S2 and the projection points in each perspective image obtained in S31 to extract the implanted vertebral body portion of each implant in each perspective image;

[0063] Since the implant is placed according to the planned internal and external points, the internal and external points in the medical image are mapped to the position on the perspective image, which appears near the implant area. Then, the implant area in each perspective image obtained by S2 and the projection point in each perspective image obtained by S31 are extracted to obtain the implant area in each perspective image, as follows:

[0064] According to the projection points of the inner and outer points of the planned approach channel in each perspective image, the corresponding reference unit vector is constructed, and the pixel points of the implant area in each perspective image are traversed to determine whether its projection on the reference unit vector is between the two projection points mentioned above. If so, the pixel point is retained, otherwise it is removed. After the traversal is completed, the implanted vertebral body part of each implant in each perspective image can be extracted, such as Figure 6 (b) as shown.

[0065] Furthermore, in this embodiment, the reference unit vector constructed is:

[0066]

[0067] For each pixel in the region Construct the vector:

[0068]

[0069] It can be seen that the pixel length of the projection points of the inner and outer points of the planned approach channel in each perspective image is Then, it is determined whether the projection of a certain pixel point in the implant area in the perspective image on the reference unit vector is located between the two projection points, as follows:

[0070] Calculating vectors exist The length of the projection on If d i ≥0 and d i ≤d pixel , then keep the pixel, otherwise discard it.

[0071] S4, fitting straight lines to the implanted vertebral body parts of each implant in each perspective image obtained in S3, respectively, to obtain the axis of each implant in each perspective image, thereby obtaining the projection points of the inner and outer points of the planned access channel on the axis of each implant in each perspective image according to the transformation calculation of S3;

[0072] In this embodiment, the perspective image after the robot navigation implant is as follows Figure 4 As shown in (a) and 4(b), by combining S1 and S3, if the corresponding implant axis is extracted on the fluoroscopic image, the implant axis can be mapped to the reference of the optical array on the C-arm or the optical array on the patient, so that the accuracy deviation between the current implant axis and the planned access channel can be calculated.

[0073] S3 has transformed the inner and outer points of the planned access channel into perspective images at different angles. The projection points of the inner and outer points of the planned access channel in each perspective image are and You can pass Fitting to obtain the vertical point corresponding to the axis of the implant and That is, the projection points of the inner and outer points of the planned access channel on the axis of each implant in each perspective image.

[0074] S5. Based on the projection points of the inner and outer points of the planned access channel obtained in S4 on the axis of each implant in each perspective image, the actual positions of the inner and outer points of the implant in the medical image are obtained based on the epipolar constraint, and the actual positions of the inner and outer points of the implant on the vertebral body are obtained by transforming the projection points to the same reference as the theoretical positions of the inner and outer points of the implant on the vertebral body;

[0075] In the present invention, the epipolar constraint is that a point on a perspective image of one viewing angle corresponds to a straight line on perspective images of other viewing angles, which includes:

[0076] S51, according to the projection points of the inner and outer points of the planned access channel obtained in S4 on the axis of each implant in each perspective image, calculate the corresponding straight line in the perspective image of other viewing angles, and use the intersection point of the straight line and the axis of the corresponding implant in the perspective image obtained in S4 as the point corresponding to the projection point in the perspective image of the viewing angle;

[0077] In this embodiment, Figure 3 As shown, there are three viewing angles in total, then the point on viewing angle 1 Corresponding to the straight line on view 2 straight line and The intersection point is Corresponding point on perspective 2 image The same can be said

[0078] S52, obtaining a number of straight lines passing through the optical center and the corresponding point according to the corresponding points in the perspective images of each viewing angle obtained in S51, and fitting the intersection points of the straight lines to obtain the actual position of the corresponding point in the medical image, that is, the actual position of the inner and outer points of the implant in the medical image;

[0079] According to the point obtained by S4 And the point obtained by S51 and The actual position of the corresponding point of the implant on the vertebral body can be obtained as follows:

[0080] Reference Figure 3 , is the position of the optical array on the C-arm when the C-arm performs the ith perspective. According to the transformation relationship between the aforementioned projection model and the optical array on the C-arm, the following mapping relationship can be established:

[0081]

[0082] in, is a point on the perspective image obtained by the i-th perspective, which is composed of point p in space 1 In the i-th perspective projection, we get For the center and point of light The straight line F 1 Represents a point on a perspective image Straight line with space The mapping relationship;

[0083] Then, multiple straight lines are obtained through the points on the perspective image at multiple angles. By fitting the intersection points of the multiple straight lines, the actual position of the point in space can be obtained, as follows:

[0084]

[0085] Among them, p 1 carm is the actual position of the point in space, and E represents the intersection point of all straight line fittings.

[0086] S6, calculating the theoretical positions of the inner and outer points of the implant on the vertebral body according to the inner and outer points of the planned access channel, and calculating the implant placement accuracy, i.e., the robot-assisted screw placement accuracy, in combination with the actual positions of the inner and outer points of the implant on the vertebral body obtained in S5;

[0087] In the present invention, the theoretical positions of the inner and outer points of the planned access channel are obtained by transforming the inner and outer points of the planned access channel to the reference of the optical array on the C-arm or the optical array on the patient.

[0088] Continue to refer to Figure 2 , C img is the image coordinate system corresponding to the medical image, C pat and C carm are the positions of the optical array on the patient and the C-arm of the C-arm machine, respectively, where C pat and C carm The relative transformation relationship between It can be obtained by real-time identification and tracking of the optical array, C img With C carm The relative transformation relationship between Generally, the C-arm machine is calibrated and remains unchanged when it leaves the factory, that is, the image obtained by the C-arm machine and the C-arm are a rigid body. and If the planned entry channel is an inner or outer point, the inner or outer points of the planned entry channel can be and The coordinate system is converted to the coordinate system corresponding to the optical array on the patient, that is, the inner and outer points of the planned access channel are transformed to the intraoperative image, as follows:

[0089]

[0090] in, and This is the theoretical position of the medial and lateral points of the implant on the vertebral body.

[0091] In the present invention, obtaining the theoretical positions of the inner and outer points of the implant on the vertebral body can be performed in any step.

[0092] In the present invention, the actual positions of the internal and external points of the implant in the medical image obtained in S52 are transformed to the same reference as the theoretical positions of the internal and external points of the implant on the vertebral body, so as to obtain the actual positions of the internal and external points of the implant on the vertebral body.

[0093] In the present invention, reference is made to Figure 7 The calculated implant placement accuracy includes the implant angle error and the implant distance error. The implant angle error is the line connecting the theoretical positions of the inner and outer points of the implant on the vertebral body, i.e., the axis A. bat The line connecting the actual position of the inner and outer points of the implant on the vertebral body is its axis B. bat The angle between bat ,B bat ), the implant distance error is the line connecting the theoretical position of the inner and outer points of the implant on the vertebral body to the actual position of the inner and outer points of the implant on the vertebral body, i.e., its axis B bat The distance d 1 d 2 The average value (d 1 +d 2) / 2.

[0094] In the present invention, the Kirschner wire is a tool commonly used in clinical nail making surgery. The axis calculation and positioning algorithm of the Kirschner wire image in the present invention has important reference significance for the recognition and positioning of the Kirschner wire axis in medical images.

[0095] The present invention uses a C-arm machine to perform multi-angle perspective of the surgical area where the implant is placed during surgery to obtain corresponding perspective images, and can identify the implant on the multi-angle perspective images, and then associate the implant image information on each perspective image based on the polar line constraint, and finally accurately obtain the actual position of the inner point of the implant on the vertebral body, thereby realizing a rapid evaluation of the accuracy of the robot-assisted nail placement system, avoiding the non-intuitive observation of the intraoperative perspective itself, simplifying the operation, and greatly reducing the patient's radiation exposure.

[0096] The present invention can also realize efficient rapid evaluation and recording of intraoperative accuracy of robot-assisted nail making surgery based on the above-mentioned accuracy evaluation method. Through long-term and large-scale quantitative accuracy evaluation records, it provides a large amount of real intraoperative accuracy data for subsequent research on the problem of drift of robot surgery accuracy over time, which is of great significance for the subsequent optimization and upgrading of the robot navigation system.

[0097] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present invention (including the claims) is limited to these examples. Under the concept of the present invention, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present invention as described above, which are not provided in detail for the sake of simplicity.

[0098] The embodiments of the present invention are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present invention should be included in the protection scope of the present invention.

Claims

1. A robot-assisted nail placement accuracy evaluation method, characterized in that: include: S1. Obtain medical images of the patient through a C-arm machine, and place the implant using robot navigation during surgery; S2, operating the C-arm machine to perform multi-angle perspective on the surgical area where the implant is placed to obtain corresponding perspective images, and extracting the implants in the perspective images corresponding to each angle; S3, transforming the inner and outer points of the approach channel planned on the medical image into perspective images of various angles according to the projection model of the C-arm machine, and extracting the implanted vertebral body parts of each implant in each perspective image obtained in S2; S4, fitting straight lines to the implanted vertebral body parts of each implant in each perspective image obtained in S3, respectively, to obtain the axis of each implant in each perspective image, thereby obtaining the projection points of the inner and outer points of the planned access channel on the axis of each implant in each perspective image according to S3; S5, according to the projection points obtained in S4, based on the epipolar constraint, obtaining the actual positions of the internal and external points of the implant in the medical image, and then obtaining the actual positions of the internal and external points of the implant on the vertebral body; S6. Calculate the theoretical positions of the inner and outer points of the implant on the vertebral body according to the inner and outer points of the planned access channel, and calculate the implant placement accuracy in combination with S5.

2. The robot-assisted nail placement accuracy evaluation method according to claim 1, characterized in that: The S3 includes: S31, obtaining projection points of inner and outer points of the planned access channel in each perspective image according to the projection model of the C-arm machine and the posture of the C-arm at the corresponding angle; S32, combining the implants in each perspective image obtained in S2 and the projection points in each perspective image obtained in S31 to extract the implanted vertebral body portion of each implant in each perspective image.

3. The robot-assisted nail placement accuracy evaluation method according to claim 2, characterized in that: In S32, extracting the implanted vertebral body portion of each implant in the perspective image is specifically: According to the projection points of the inner and outer points of the planned approach channel in each perspective image, the corresponding reference unit vector is constructed, and the pixel points of the implant area in each perspective image are traversed to determine whether its projection on the reference unit vector is between the two projection points mentioned above. If so, the pixel point is retained, otherwise it is discarded. After the traversal is completed, the inserted vertebral body part of each implant in each perspective image is extracted.

4. The robot-assisted nail placement accuracy evaluation method according to claim 1, characterized in that: In S5, the actual positions of the internal and external points of the implant in the medical image are obtained based on the epipolar constraint, as follows: S51, according to the projection points of the inner and outer points of the planned access channel obtained in S4 on the axis of each implant in each perspective image, calculate the corresponding straight line in the perspective image of other viewing angles, and use the intersection point of the straight line and the axis of the corresponding implant in the perspective image corresponding to it obtained in S4 as the point corresponding to the projection point in the perspective image of the viewing angle; S52. According to the corresponding points in the perspective images of each viewing angle obtained in S51, several straight lines passing through the optical center and the corresponding points are obtained, and the intersection points of these straight lines are fitted to obtain the actual positions of the corresponding points in the medical image, that is, the actual positions of the inner and outer points of the implant in the medical image.

5. The robot-assisted nail placement accuracy evaluation method according to claim 1, characterized in that: In said S5, after obtaining the actual positions of the inner and outer points of the implant in the medical image, the positions are transformed to the reference of the optical array on the C-arm or the optical array on the patient, that is, the actual positions of the inner and outer points of the implant on the vertebral body are obtained; In S6, according to the planned inner and outer points of the approach channel, the points are transformed to the reference of the optical array on the C-arm or the optical array on the patient, so as to obtain the theoretical positions of the inner and outer points of the implant on the vertebral body.

6. The robot-assisted nail placement accuracy evaluation method according to claim 1, characterized in that: In S6, the implant placement accuracy includes the implant angle error and the implant distance error, wherein the implant angle error is the angle between the line connecting the theoretical positions of the inner and outer points of the implant on the vertebral body and the line connecting the actual positions of the inner and outer points of the implant on the vertebral body, and the implant distance error is the average value of the distances from the lines connecting the theoretical positions of the inner and outer points of the implant on the vertebral body to the actual positions of the inner and outer points of the implant on the vertebral body.

7. The robot-assisted nail placement accuracy evaluation method according to claim 1, characterized in that: In S2, the fluoroscopic image is segmented by a threshold value and then connected domain detection is performed to obtain the implant in the fluoroscopic image.

8. The robot-assisted nail placement accuracy evaluation method according to claim 7, characterized in that: The connected domain is defined as an area where image pixels are connected, and the specific feature of the connection between pixels is that four areas or eight neighborhoods are connected.

9. The robot-assisted nail placement accuracy evaluation method according to claim 7, characterized in that: After the connected domain is detected, the number of pixels in each connected domain is calculated, and the connected domains with the number of pixels less than a set threshold are deleted, so as to obtain the implant in the fluoroscopic image.

Images