CN114494602A - Collision detection method, system, computer device and storage medium - Google Patents

Abstract

The present application relates to the field of medical technology, and in particular, to a collision detection method, system, computer equipment and storage medium. A collision detection method includes: acquiring an object real-time position of a target object to be processed; registering a virtual object model of the target object to be processed and the object real-time position of the target object to be processed to determine the virtual object model The target position in the target coordinate system where the target object to be processed is located; obtain the real-time position of the target instrument; determine whether the target instrument will collide according to the target position and the real-time position of the instrument. The application effectively improves the safety and reliability of surgery, does not require additional sensors, reduces production, design and use costs, improves the operating space of the robot compared to the traditional non-contact method, and improves the efficiency of collaboration with doctors.

Description

Collision detection method, system, computer device and storage medium

technical field

The present application relates to the technical field of medical simulation control and graphics processing, and in particular, to a collision detection method, system, computer equipment and storage medium.

Background technique

Since the beginning of the 21st century, surgical robots have been changing traditional surgical methods, and this change is getting faster and faster today. Various types of robots have appeared in almost every clinical field, especially in laparoscopic-related surgical robots and orthopedic robots. With the wide application of surgical robots, their safety issues have become increasingly prominent. Among them, collision detection is a very important safety function, especially in the operating room where space is limited and the locations of personnel and equipment are complex.

The existing robot collision detection technologies include contact and non-contact. These technologies face the following problems in the application of surgical robots:

1. The contact technology requires the use of torque sensors in the manipulator joints, and the need for contact to detect collisions has certain safety risks.

2. Non-contact technology also requires the use of additional sensors to achieve collision detection functions, such as infrared sensors, ultrasonic sensors, etc. This technical solution requires the establishment of a virtual wall outside the working space of the robotic arm, which reduces the working efficiency of the robot, and Reduce the doctor's operating space.

These problems will affect the use experience of surgical robots, increase the cost of design, production and use, reduce the efficiency of collaboration between doctors and surgical robots, and even create risks that affect the safety of surgery.

SUMMARY OF THE INVENTION

Based on this, it is necessary to provide a collision detection method, apparatus, computer equipment and storage medium for the above technical problems.

In a first aspect, the present application provides a collision detection method, the collision detection method includes:

Get the object real-time position of the target object to be processed;

registering the virtual object model of the target object to be processed and the object real-time position of the target object to be processed to determine the target position of the virtual object model in the target coordinate system where the target object to be processed is located;

Obtain the real-time position of the device of the target device;

According to the target position and the real-time position of the instrument, it is determined whether the target instrument and the target object to be processed will collide.

In one embodiment, the virtual object model of the target object to be processed is registered with the object real-time position of the target object to be processed, so as to determine that the virtual object model is located where the target object to be processed is located The target position in the target coordinate system of , including:

acquiring multiple feature points pre-marked on the virtual object model;

The multiple feature points are matched with the real-time position of the target object to be processed to determine the target position of the virtual object model in the target coordinate system where the target object to be processed is located.

In one embodiment, the acquiring the real-time position of the target instrument includes:

acquiring the preset pose relationship between the reflective marker and the target device;

Based on the preset pose relationship, the real-time position of the device of the target device is obtained according to the real-time pose of the reflective marker.

In one embodiment, the virtual object model of the target object to be processed is registered with the object real-time position of the target object to be processed, so as to determine that the virtual object model is located where the target object to be processed is located Before the target position in the target coordinate system, also include:

Scanning to obtain the medical image of the target object to be processed;

Perform image reconstruction according to the medical image to obtain the virtual object model of the target object to be processed.

In one embodiment, the determining whether the target instrument will collide according to the target position and the real-time position of the instrument includes:

Using the K-dimensional space tree collision detection method and/or the AABB tree collision detection method, it is determined whether the target instrument will collide according to the target position and the real-time position of the instrument.

In one embodiment, the K-dimensional space tree collision detection method includes:

using the target position and the real-time position of the instrument as the queried point cloud and/or the target point cloud;

Create a K-dimensional space tree according to the queried point cloud;

Traverse the K-dimensional space tree, and determine whether the target instrument will collide according to the closest distance between the target point cloud and the queried point cloud.

In one embodiment, before taking the target position and the real-time position of the instrument as the queried point cloud and/or the target point cloud, the method includes:

The point cloud of the target position and the point cloud of the real-time position of the instrument are respectively down-sampled to obtain the sampling points of the target object to be processed and the target instrument.

In one embodiment, the down-sampling of the point cloud of the target position and the point cloud of the real-time position of the instrument respectively, to obtain the sampling points of the target object to be processed and the target instrument, includes:

Calculate the bounding box of the point cloud, and discretize the bounding box into several voxels;

The center point of each voxel or the point closest to the center point is used as the sampling point.

In one of the embodiments, the establishment of a K-dimensional space tree includes:

establish root node;

Calculate the variance value of the queried point cloud, and use the variance value as a segmentation feature;

Taking the median of the segmentation feature as the segmentation point;

Transfer the segmentation feature whose segmentation feature is smaller than the segmentation point to the left node of the root node, and transfer the segmentation feature larger than the segmentation point to the right node of the root node;

The above steps are recursively performed until all the segmentation features are established on the left node and the right node of the root node of the K-dimensional space tree.

In one embodiment, the traversal of the K-dimensional space tree, the traversal of the K-dimensional space tree, and the determination of the target instrument according to the closest distance between the target point cloud and the queried point cloud Whether collisions will occur, including:

Obtain the first target point cloud and the first queried point cloud;

Traverse each of the first queried point clouds, find the two closest points between the first queried point cloud and the first target point cloud, and use the distance between the two points as the first closest point between the point clouds distance, to determine whether a collision will occur at the first closest distance.

In one embodiment, the AABB tree collision detection method includes:

using the target position and the real-time position of the instrument as the queried point cloud and/or the target point cloud;

Build an axisymmetric bounding box tree according to the queried point cloud;

Traverse the axisymmetric bounding box tree, and determine whether the target instrument will collide according to the closest distance between the target point cloud and the queried point cloud.

In one of the embodiments, the establishing an axis-symmetric bounding box tree includes:

Take the triangle as the root node;

establishing a leaf node of the queried point cloud, and assigning an axisymmetric bounding box tree according to the associated object of the leaf node;

Find a predetermined existing node in the axisymmetric bounding box tree, and make the new leaf a sibling node of the predetermined existing node;

creating a branch node for the predetermined existing node and the new leaf, and assigning an axisymmetric bounding box of two nodes to the branch node;

attaching the new leaf to the two nodes;

removing the existing node from the axisymmetric bounding box tree and appending the existing node to the two nodes;

attaching the two nodes as child nodes of the parent node of the existing node;

The axisymmetric bounding box of the parent node is adjusted to ensure that the parent node contains the axisymmetric bounding box of all of the child nodes.

In one embodiment, traversing the axisymmetric bounding box tree, and judging whether the target instrument will collide according to the closest distance between the target point cloud and the queried point cloud, includes:

Obtain the second target point cloud and the second queried point cloud;

Build an axisymmetric bounding box tree of the second queried point cloud, traverse each of the second queried point clouds, and find the point of the second queried point cloud that is closest to the point of the second target point cloud The two points of , take the distance between the two points as the second closest distance between the point clouds.

In a second aspect, the present application further provides a collision detection system, the collision detection system includes: an image processor and a position collector, where the image processor is used to implement the steps of the above-mentioned method.

In a third aspect, the present application also provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

Get the object real-time position of the target object to be processed;

registering the virtual object model of the target object to be processed and the object real-time position of the target object to be processed to determine the target position of the virtual object model in the target coordinate system where the target object to be processed is located;

Obtain the real-time position of the device of the target device;

It is determined whether the target instrument will collide according to the target position and the real-time position of the instrument.

In a fourth aspect, the present application also provides a computer-readable storage medium. The computer-readable storage medium has a computer program stored thereon, and when the computer program is executed by the processor, the following steps are implemented:

Get the object real-time position of the target object to be processed;

registering the virtual object model of the target object to be processed and the object real-time position of the target object to be processed to determine the target position of the virtual object model in the target coordinate system where the target object to be processed is located;

Obtain the real-time position of the device of the target device;

It is determined whether the target instrument will collide according to the target position and the real-time position of the instrument.

In a fifth aspect, the present application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the following steps:

Get the object real-time position of the target object to be processed;

registering the virtual object model of the target object to be processed and the object real-time position of the target object to be processed to determine the target position of the virtual object model in the target coordinate system where the target object to be processed is located;

Obtain the real-time position of the device of the target device;

It is determined whether the target instrument will collide according to the target position and the real-time position of the instrument.

The above collision detection method, device, computer equipment and storage medium, by acquiring the object real-time position of the target object to be processed; registering the virtual object model of the target object to be processed and the object real-time position of the target object to be processed, determine the target position of the virtual object model in the target coordinate system where the target object to be processed is located; obtain the real-time position of the target device; determine whether the target device will Collision. The intraoperative collision detection method for a surgical robot of the present application solves the technical problem in the prior art that additional sensors need to be added, which affects the operating space of the doctor and the robotic arm. It effectively improves the safety and reliability of surgery, does not require additional sensors, reduces the cost of production, design and use, improves the operating space of the robot compared to the traditional non-contact method, and improves the efficiency of collaboration with doctors.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application or in the traditional technology, the following briefly introduces the accompanying drawings that are used in the description of the embodiments or the traditional technology. Obviously, the drawings in the following description are only the For some embodiments of the application, for those of ordinary skill in the art, other drawings can also be obtained according to these drawings without any creative effort.

1 is an application environment diagram of a collision detection method in one embodiment;

2 is a schematic flowchart of a collision detection method in one embodiment;

3 is a schematic flowchart of steps of a collision detection method in one embodiment;

4 is a schematic flowchart of a real-time position registration process of a collision detection method in one embodiment;

5 is a schematic diagram of a registration application environment of the collision detection method in one embodiment;

6 is a schematic flowchart of a registration step of a collision detection method in one embodiment;

FIG. 7 is a schematic diagram of a process flow diagram of obtaining the position of a target device in a collision detection method in one embodiment;

8 is a schematic flowchart of a method for creating a virtual object model for collision detection in one embodiment;

9 is a schematic diagram of image segmentation of a collision detection method in one embodiment;

10 is a schematic diagram of a virtual object model of a collision detection method in one embodiment;

11 is a schematic flow chart of a K-dimensional space tree collision detection of a collision detection method in one embodiment;

12 is a schematic diagram of the K-dimensional space tree collision detection steps of the collision detection method in one embodiment;

13 is a schematic diagram of a downsampling flow chart of a collision detection method in one embodiment;

14 is a schematic diagram of downsampling of a collision detection method in one embodiment;

15 is a schematic diagram of a downsampling algorithm of a collision detection method in one embodiment;

16 is a schematic flowchart of a method for establishing a K-dimensional space tree for collision detection in one embodiment;

17 is a schematic diagram of a K-dimensional space tree of a collision detection method in one embodiment;

18 is a schematic diagram of nearest neighbor retrieval of a collision detection method in one embodiment;

19 is a schematic diagram of a distance control flow chart of a collision detection method in one embodiment;

20 is a schematic flowchart of a distance control step of a collision detection method in one embodiment;

Fig. 21 is the AABB tree collision detection flow schematic diagram of collision detection method in one embodiment;

22 is a schematic flowchart of the AABB tree collision detection steps of the collision detection method in one embodiment;

Figure 23 is a schematic flow chart of the establishment of the AABB tree of the collision detection method in one embodiment;

24 is a schematic diagram of the AABB tree distance control flow diagram of the collision detection method in one embodiment;

25 is a schematic diagram of spatial retrieval of a collision detection method in one embodiment;

26 is a schematic flowchart of the AABB tree distance control steps of the collision detection method in one embodiment;

Figure 27 is a diagram of the internal structure of a computer device in one embodiment.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present application more clearly understood, the present application will be described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, but not to limit the present application.

The collision detection method provided by the embodiment of the present application can be applied to the application environment shown in FIG. 1 . Wherein, the virtual object model is applied to a computer, which may include a main display, a keyboard, and a controller located in the navigation trolley. The following will take the collision detection system of the present invention for an orthopaedic surgery navigation system for knee joint replacement as an example for description, but in specific implementation, the osteotomy guide tool of the present invention is not limited to the osteotomy guidance of the orthopaedic navigation system for knee joint replacement The tool can also be used in other surgical navigation systems. In the orthopaedic surgery navigation system for knee replacement, a patient 118 is placed on an operating bed, an operating trolley 100 is provided on one side of the patient 118, and a robotic arm 102 for controlling the movement of the tool target 104 is placed on the operating trolley 100 As well as the base target 106, the tool target 104 is provided with an osteotomy guide tool 108, and the doctor can use an oscillating saw 110 or an electric drill to perform osteotomy and drilling operations through the osteotomy guide groove and guide hole of the osteotomy guide tool. On the other side of the patient, there are a main display 116 and a secondary display 114 arranged on the navigation trolley, and an NDI navigation device 112 is also arranged on the navigation trolley. Among them, NDI is an optical tracker, which can track the target in the tracking range in real time and obtain the pose information of the target in real time.

As shown in FIG. 2 and FIG. 3 , in one embodiment, a collision detection method is provided, and the method is applied to the collision detection system shown in FIG. 1 as an example to illustrate, including the following steps:

S202: Acquire the object real-time position of the target object to be processed.

Wherein, the target object is the affected area of the patient requiring surgery, for example, the patient's abdominal cavity, bone or breast and thyroid.

Specifically, the collision detection system can obtain the real-time position of the target object to be processed. For example, the orthopaedic surgery for a patient's knee joint replacement requires surgery on the affected part of the knee joint. The collision detection system can obtain the fracture area of the patient to be operated on. The fracture area can be preset with a certain range centered on the fracture site to form a fracture. area. The collision detection system acquires the real-time position of the target object, so as to subsequently perform surgery on the target object according to the real-time position of the target object.

S204: Register the virtual object model of the target object to be processed and the object real-time position of the target object to be processed to determine the target position of the virtual object model in the target coordinate system where the target object to be processed is located.

Among them, the registration is the registration of the actual body position of the patient and the image space position (ie, the "map" of the patient) during the entire surgical navigation process, and this process is called registration.

Specifically, the collision detection system establishes a virtual object model according to the target object to be processed, and then registers the virtual object model with the real-time position of the target object to be processed. The virtual object model is registered with the target object to be processed to determine the target position of the virtual object model in the target coordinate system where the target object to be processed is located, so that subsequent surgical operations can be based on the coordinates of the target object to be processed in the coordinate system. , and issue an action command to the target device.

S206: Acquire the real-time position of the target device.

Wherein, the target instrument is a surgical instrument used in a surgical operation, for example, the instrument may be a scalpel, a surgical saw, and the like.

Specifically, the collision detection system acquires the real-time position of the target instrument, that is, the position of the target instrument in the coordinate system of the virtual object model.

S208: Determine whether the target instrument and the target object to be processed will collide according to the target position and the real-time position of the instrument.

Specifically, the collision detection system obtains the target position of the target object and the position of the target device in the coordinate system of the virtual object model. After knowing the positions of different targets in the coordinate system, it can determine whether the target device will collide. The basis for the collision can be Depending on the position of the two targets in the coordinate system, there are events such as infinite approach and contact.

In this embodiment, the collision detection system obtains the object real-time position of the target object to be processed; and then performs registration according to the virtual object model of the target object to be processed and the real-time position of the target object to be processed to determine the location of the target object to be processed. The target position in the coordinate system of the virtual object model; in addition, the position of the target device in the coordinate system is obtained, and it is judged whether the two positions in the coordinate system collide. By judging whether the target instrument will collide according to the target position of the target object and the real-time position of the instrument, the technical problem that additional sensors need to be added in the prior art and affects the operation space of the doctor and the robot arm is solved. It effectively improves the safety and reliability of surgery, does not require additional sensors, reduces production, design and use costs, improves the operating space of the robot compared to traditional non-contact methods, and improves the efficiency of collaboration with doctors.

As shown in Figure 4, Figure 5 and Figure 6, in one embodiment, step S204 registers the virtual object model of the target object to be processed and the object real-time position of the target object to be processed to determine whether the virtual object model is Process the target position in the target coordinate system where the target object is located, including:

S302: Acquire multiple feature points pre-marked on the virtual object model.

Specifically, the virtual object model is pre-marked with multiple feature points for registration, and the doctor operates the surgical instrument on the patient's bone part to register the multiple feature points with the bone part, so that the virtual object model is registered with the bone part , complete the surgical instrument registration, so as to obtain the real-time position of the surgical instrument at the end of the robotic arm through the virtual object model. For example, place the operating trolley and the navigation trolley at a suitable position beside the operating table, and install the femoral target, tibial target, base target, osteotomy guide tool, tool target and other equipment; the doctor scans the bone CT scan model of the patient's leg. Import the computer to form a 3D model for preoperative planning, plan the coordinates of the osteotomy plane, select the appropriate type of prosthesis, and plan the installation position of the prosthesis; the doctor uses the target pen to point the patient's femur and tibia feature points, and the NDI navigation device uses the base target to target As a benchmark, record the position of the patient's bone feature point and send the bone feature point position to the computer.

S304: Match the multiple feature points with the real-time position of the target object to be processed to determine the target position of the virtual object model in the target coordinate system where the target object to be processed is located.

Specifically, the target object to be processed is the surgical site of the patient, such as the femur and the tibia. The computer obtains the actual orientation of the femur and tibia of the patient's surgical position through the calculation method of feature point matching, and corresponds the actual orientation to the CT image orientation of the femur and tibia. The femoral target and the tibial target can track the actual position of the bone in real time. During the operation, as long as the relative position of the target and the bone is fixed, the movement of the bone will not affect the surgical effect.

In this embodiment, the collision detection system obtains multiple feature points pre-marked on the virtual object model, and then matches the multiple feature points with the real-time position of the target object to be processed, and then completes the registration. In the model, the actual position of the target object to be processed can be tracked in real time, which is convenient for subsequent surgical operations.

As shown in FIG. 7 , in one embodiment, step S206 acquires the real-time position of the target instrument, including:

S402: Acquire a preset pose relationship between the reflective marker and the target device.

Specifically, the collision detection system obtains the pose relationship between the reflective marker and the target instrument. For example, the collision detection system first registers the virtual object model of the patient's bone part with the bone part of the patient's surgical posture, and then, according to the reflective marker The controller obtains the current position and attitude of the target instrument (such as a robotic arm).

S404: Based on the preset pose relationship, obtain the real-time position of the target device according to the real-time pose of the reflective marker.

Specifically, the collision detection system obtains the real-time pose of the surgical instrument according to the reflective marker and the preset pose relationship between the surgical instrument and the reflective marker determined by registration, and then obtains the real-time pose of the surgical instrument according to the pose of the surgical instrument and human tissue. The distance between is used for collision detection. For example, the collision detection system controls the movement of the robot arm according to the pre-planned movement path of the robot arm, and obtains the current position and posture of the robot arm through the reflective marker; calculates the difference between the current position and posture of the robot arm and the bone position of the patient on the operating bed. If the distance exceeds the threshold, the robotic arm will continue to move, and continue to execute the pre-planned robotic arm movement path to control the robotic arm movement, otherwise stop the robotic arm.

In this embodiment, the collision detection system obtains the pose relationship between the reflective marker and the target device, and then obtains the real-time position of the target device according to the real-time pose of the reflective marker. In the virtual object model, it is convenient to judge whether the target device will collide.

As shown in FIG. 8 and FIG. 9, in one embodiment, in step S2O4, the virtual object model of the target object to be processed is registered with the object real-time position of the target object to be processed, so as to determine the position of the virtual object model in the target object to be processed. Before the target position in the target coordinate system where the object is located, it also includes:

S502: Scanning to obtain a medical image of the target object to be processed.

Specifically, the preoperative collision detection system obtains the medical image of the target object to be processed by scanning, wherein the target object to be processed may be the target tissue of the patient, including but not limited to joints, skin, and the like. For example, the collision detection system segments the image of the knee joint to obtain the segmentation contour of the patient's target tissue. Further, the segmentation algorithm may use a traditional image processing algorithm or a deep learning algorithm.

S504: Perform image reconstruction according to the medical image to obtain a virtual object model of the target object to be processed.

Specifically, the collision detection system reconstructs a virtual object model of the patient's knee joint through the acquired knee joint segmentation contour, as shown in FIG. 10 , where the virtual object model is a three-dimensional model of the knee joint.

In this embodiment, the collision detection system obtains the medical image of the target object to be processed by scanning, and then forms a virtual object model from the medical image, which facilitates the registration of the virtual object model with the intraoperative patient target tissue.

In one embodiment, step S208 determines whether the target instrument will collide according to the target position and the real-time position of the instrument, including: adopting the K-dimensional space tree collision detection method and/or adopting the AABB tree collision detection method, according to the target position and the real-time instrument position. The position determines whether the target instrument will collide.

Among them, the K-dimensional space tree is KD-tree, the full name of k-dimensional tree, which is a modification of the binary search tree, which can store the instance points in the k-dimensional space for fast retrieval. . AABB tree is AABB-tree, full name axis aligned bounding box tree, axis-symmetric bounding box, is a modification of binary search tree, each node contains geometric primitives, AABB tree is constructed by calculating the entire input graph is initialized with an AABB of the set of primitives, then all primitives are sorted along the longest axis of the box, and the primitives are divided into two sets of equal size, this process is applied recursively until the AABB contains a single primitive.

Specifically, the collision detection system judges whether the target instrument will collide according to the target position and the real-time position of the instrument, and uses the K-dimensional space tree collision detection method and/or the AABB tree collision detection method to determine whether the target instrument will collide.

In this embodiment, the collision detection system uses the K-dimensional space tree collision detection method and/or the AABB tree collision detection method to determine whether the target instrument will collide, and can track the intraoperative pose of the target instrument and patient tissue in real time ; Real-time detection of distance, eliminating the risk of collision.

As shown in FIG. 11 and FIG. 12 , in one embodiment, the K-dimensional space tree collision detection method includes:

S602: Use the target position and the real-time position of the instrument as the queried point cloud and/or the target point cloud.

Specifically, the collision detection system determines whether the target instrument will collide according to the target position and the real-time position of the instrument, and uses the target position and the real-time position of the instrument as the queried point cloud and the target point cloud, or uses the target position and the real-time position of the instrument as the queried point cloud and the target point cloud. The real-time position of the instrument is used as the target point cloud and the queried point cloud.

S604: Create a K-dimensional space tree according to the queried point cloud.

Specifically, the collision detection system takes the target position or the real-time position of the instrument as the queried point cloud, and creates a K-dimensional space tree according to the queried point cloud, so as to determine the sample points of the target object to be processed and the target instrument through the K-dimensional space tree. The location relationship of the sampling points.

S604: Traverse the K-dimensional space tree, and determine whether the target equipment will collide according to the closest distance between the target point cloud and the queried point cloud.

Specifically, the K-dimensional space tree is traversed, and the closest distance between the sampling point of the target device and the sampling point of the target object to be processed is queried according to binary search, and it is judged whether it is smaller than a preset threshold, so as to perform collision detection and control the movement of the target device// stop.

In this embodiment, by creating a K-dimensional space tree, the collision detection system imports the sampling points of the target object to be processed and the target equipment into the K-dimensional space tree to obtain the closest distance between the target object to be processed and the target equipment, and judges Whether the distance is less than a threshold value controls further movement of the target instrument.

In one embodiment, before taking the target position and the real-time position of the instrument as the queried point cloud and/or the target point cloud, the method includes:

The point cloud of the target position and the point cloud of the real-time position of the instrument are respectively down-sampled to obtain the sampling points of the target object to be processed and the target instrument.

Among them, the point cloud is a collection of points on the surface of the object. Downsampling is to voxelize the three-dimensional space, and then sample a point in each voxel, usually the center point or the point closest to the center can be used as the sampling point.

Specifically, the collision detection system downsamples the point cloud data of the target instrument and the point cloud data of the bones of the patient pose respectively. Among them, the point cloud downsampling adopts the point cloud voxel downsampling algorithm for calculation and sampling. Since the collision only needs to consider the surface of the object, the point cloud only has points on the surface of the object, and only these points are downsampled to obtain the target object to be processed and The sampling point of the target device.

In this embodiment, the collision detection system can reduce the number of points while ensuring the shape characteristics of the point cloud and improve the calculation speed of collision detection by down-sampling the point cloud of the target object to be processed and the target equipment.

As shown in FIGS. 13 , 14 and 15 , in one embodiment, the point cloud of the target position and the point cloud of the real-time position of the instrument are down-sampled to obtain the target object to be processed and the point cloud of the instrument. Sampling points for target devices, including:

S702: Calculate the bounding box of the point cloud, and discretize the bounding box into several voxels.

Specifically, the collision detection system calculates the bounding box of the point cloud, and then discretizes the bounding box into small voxels. Among them, the length, width and height of the bounding box of the voxel can be set as required, and can also be obtained by setting the number of grid points in the three directions of the bounding box.

S704: Use the center point of each voxel or the point closest to the center point as the sampling point.

Specifically, the collision detection system creates voxels, calculates the bounding box of the point cloud, and then discretizes the bounding box into small voxels. Each small voxel contains several points, and the collision detection system calculates the bounding box of the voxel. A point closest to the center point is taken out as a sampling point, and the remaining points in the small voxels are discarded.

In this embodiment, the collision detection system obtains the center point of the voxel or the point closest to the center as the sampling point, and discards the points in the small voxel, which reduces the amount of calculation when performing collision detection in real time.

As shown in Figure 16, in one embodiment, establishing a K-dimensional space tree includes:

S802: Establish a root node.

Specifically, when the collision detection system establishes a K-dimensional space tree, the root node of the K-dimensional space tree must be established first.

S804: Calculate the variance value of the queried point cloud, and use the variance value as a segmentation feature.

Specifically, after establishing the root node of the K-dimensional space tree, the collision detection system calculates the variance value of the queried point cloud, and the variance calculation formula is:

Among them, x is the arithmetic mean, which is equivalent to doing the variance of all the values of the x-axis of a point set sequence, and doing the variance of all the values of the y-axis, and then comparing the variance to determine the feature, and n is the root node. The obtained variance value is used as the segmentation feature.

S806: Use the median of the segmentation features as the segmentation point.

Specifically, after the collision detection system establishes the root node of the K-dimensional space tree, the variance value is calculated, the obtained variance value is used as the segmentation feature, and the median of the segmentation feature is used as the segmentation point.

S808: Transfer the segmentation feature whose segmentation feature is smaller than the segmentation point to the left node of the root node, and the segmentation feature whose segmentation feature is larger than the segmentation point is transmitted to the right node of the root node.

Specifically, after the collision detection system establishes the root node of the K-dimensional space tree, the median of the segmentation features is used as the segmentation point, and the point set sequence data set smaller than the segmentation point is passed to the left node of the root node, and the transmission is greater than the segmentation point. Gives the right node of the root node.

S810: Recursively execute the above steps until all segmentation features are established on the left node and the right node of the root node of the K-dimensional space tree.

As shown in FIG. 17 , specifically, the collision detection system recursively executes steps S704 to S708 until all data in the point set sequence are established on the nodes of the K-dimensional space tree. For example, six points A, B, C, D, E and F form a two-dimensional data set. The feature with the largest variance of these six points is the x-coordinate value, and the point A whose x-coordinate value is the median is the root node. , Divide point B into the left subtree, and divide C, D, E and F into the right subtree. The median node of the x-axis value of C, D, E and F is C(70,10), so C is used as A’s right node. The maximum variance feature of C, D, E and F is the y-axis feature, so D, E and F form the right subtree of C, and the maximum variance of D, E and F is the x-axis value, so point D is the median value The nodes are the right nodes of D, E, and F in the C subtree, while E is the left node of D, and F is the right node of D.

Specifically, the process of searching for nearest neighbors in a K-dimensional space tree is as follows: starting from the root node, the algorithm recursively moves down the tree in the same way as inserting a search point, i.e. to the left or to the right depending on whether the point is smaller than or greater than the current node in the split dimension. Once the algorithm reaches a leaf node, it checks the node and if the distance is better, saves the node as "current best". The algorithm expands the recursion of the tree, performing the following steps at each node: if the current node is closer than the current best node, it becomes the current best node. The algorithm checks if there might be any point on the other side of the split plane that is closer to the search point than the current best point.

As shown in Figure 18, it should be noted that the K-dimensional space tree is accomplished by intersecting the segmentation hyperplane with a hypersphere surrounding the search point, the radius of the hypersphere being equal to the current closest distance. Since the hyperplanes are all axis-aligned, this is implemented as a simple comparison to see if the distance between the split coordinates of the search point and the current node is less than the distance from the search point to the current best point. If the hypersphere crosses the plane, there may be a closer point on the other side of the plane, so the algorithm must move down another branch of the tree from the current node to find a closer point, following the same recursive process as the entire search. If the hypersphere does not intersect the splitting plane, the algorithm continues up the tree and eliminates the entire branch on the other side of the node. When the algorithm completes this process for the root node, the search is complete.

In this embodiment, the collision detection system establishes a root node; takes the feature with the largest variance as the segmentation feature; takes the median of the segmentation features as the segmentation point; The segmentation feature is passed to the left node of the root node, and the segmentation feature greater than the median is passed to the right node of the root node; recursively execute until all eigenvalues in the segmentation feature are established to all the K-dimensional space tree. Up to the root node described above, the establishment of the K-dimensional space tree is completed. It is convenient to determine the nearest node of the K-dimensional space tree to the patient's affected area, which is convenient for subsequent collision detection.

As shown in FIG. 19 and FIG. 20 , in one embodiment, traversing the K-dimensional space tree, and determining whether the target equipment will collide according to the closest distance between the target point cloud and the queried point cloud, including:

S902: Obtain the first target point cloud and the first queried point cloud.

The target point cloud is the point cloud of the affected part of the patient, for example, the point cloud of the affected part of the patient's knee joint; the queried point cloud is the point cloud of the end of the target instrument, such as the point cloud of the end of the surgical robotic arm.

Specifically, the collision detection system acquires the target point cloud and the queried point cloud, and forms the target point cloud group and the queried point cloud group.

S904: Traverse each first queried point cloud, find the two closest points between the point of the first queried point cloud and the point of the first target point cloud, use the distance between the two points as the first closest distance between the point clouds, and determine the Whether a collision will occur at the first closest distance.

Specifically, the collision detection system traverses each point of the queried point cloud group, and finds the closest point in the target point cloud group and the queried point cloud group. Through the closest point between the target point cloud group and the queried point cloud group, the minimum distance of the closest point is obtained, and the closest distance between the target point cloud group and the queried point cloud group is determined. For example, in orthopaedic surgery, after registration, the point cloud at the end of the surgical robotic arm is obtained as the queried point cloud, and the point cloud of the patient's knee joint is the target point cloud. Query the point cloud to establish a corresponding K-dimensional space tree, and then use the established K-dimensional space tree to traverse and query the closest distance between each point in the K-dimensional space tree and the target point cloud, and update the closest distance between the two point clouds.

In this embodiment, the collision detection system acquires the first target point cloud and the first queried point cloud, establishes a K-dimensional space tree of the queried point cloud, and traverses each first queried point cloud to find the first queried point cloud. The point of the point cloud is the closest two points to the point of the first target point cloud, and the distance between the two points is taken as the first closest distance between the point clouds. By determining the first closest distance, and when the first closest distance changes, it can be updated in real time, so as to know the possible collision situation in time, and improve the safety and reliability of the operation.

As shown in Figure 21 and Figure 22, in one embodiment, the AABB tree collision detection method includes:

S1002: Use the target position and the real-time position of the instrument as a queried point cloud and/or a target point cloud.

Specifically, the collision detection system determines whether the target instrument will collide according to the target position and the real-time position of the instrument, and uses the target position and the real-time position of the instrument as the queried point cloud and the target point cloud, or uses the target position and the real-time position of the instrument as the queried point cloud and the target point cloud. The real-time position of the instrument is used as the target point cloud and the queried point cloud.

S1004: Build an axisymmetric bounding box tree according to the queried point cloud.

Specifically, the collision detection system establishes an axisymmetric bounding box tree according to the queried point cloud, and imports the sampling points of the target object to be processed and the target instrument into the axisymmetric bounding box tree, so as to determine the sampling points of the target object to be processed through the axisymmetric bounding box tree The positional relationship to the sampling point of the target instrument.

S1006: Traverse the axisymmetric bounding box tree, and determine whether the target instrument will collide according to the closest distance between the target point cloud and the queried point cloud.

Specifically, the collision detection system traverses the axisymmetric bounding box tree, determines the closest distance between the sampling point of the target object to be processed and the sampling point of the target instrument, and judges whether the closest distance is less than a preset threshold, so as to control the movement/stop of the target instrument.

In this embodiment, the collision detection system creates an axisymmetric bounding box tree, and imports the sampling points of the target object to be processed and the target device into the axisymmetric bounding box tree to obtain the closest distance between the target object to be processed and the target device, Real-time tracking of the intraoperative pose of the robotic arm and patient tissue, real-time detection of distance, and elimination of collision risks.

As shown in Figure 23, in one embodiment, an axisymmetric bounding box tree is established, including:

S1102: Use the triangle as the root node.

Specifically, the root node contains geometric primitives, for example, the root node contains triangle primitives.

S1104: Establish leaf nodes of the queried point cloud, and assign an axis-symmetric bounding box tree according to the associated objects of the leaf nodes.

Specifically, the collision detection system establishes the leaf node of the queried point cloud, and assigns an AABB to the leaf node according to the associated object of the leaf node. Import the surgical instrument point cloud data so that the point cloud data becomes the associated object of the leaf node.

S1106: Find a predetermined existing node in the axisymmetric bounding box tree, and make the new leaf a sibling node of the predetermined existing node.

Specifically, the collision detection system finds an existing node in the axisymmetric bounding box tree, the existing node is a leaf node or branch except the root node, and compares the existing node and the new leaf node as sibling nodes to determine whether there is intersection.

S1108: Create branch nodes for predetermined existing nodes and new leaves, and assign axisymmetric bounding boxes of two nodes to the branch nodes.

Specifically, the collision detection system finds an existing node in the axisymmetric bounding box tree. The existing node is a leaf node or branch except the root node, and the existing node and the new leaf node are regarded as sibling nodes. After comparing each sibling node , if there is no intersection, judge the size of the bounding box area after the combination of sibling nodes, the smaller two bounding boxes create new branch nodes respectively, and assign an AABB containing two nodes to other nodes, which essentially combines two AABBs to locate new branch nodes and new leaves.

S1110: Attach a new leaf to both nodes.

Specifically, the collision detection system finds the existing node in the axisymmetric bounding box tree, attaches the new leaf added later to the new branch node, takes the existing node and the new leaf node as sibling nodes, and the smaller two bounding boxes A new branch node is created separately, and an AABB containing two nodes is assigned to the other node, which essentially combines the two AABBs for the new branch node and the new leaf to locate, and the new leaf is attached to the branch node.

S1112: Remove the existing node from the axisymmetric bounding box tree and append the existing node to the two nodes.

Specifically, the collision detection system removes the compared existing nodes in the axisymmetric bounding box tree and appends the removed existing nodes to the new branch nodes.

S1114: Attach two nodes as child nodes of the parent node of the existing node.

Specifically, the collision detection system removes the compared existing nodes in the axisymmetric bounding box tree, appends the removed existing nodes to a new branch node, and then appends the new branch node as the previous node on the existing node Child node of parent node.

S1116: Adjust the axisymmetric bounding box of the parent node to ensure that the parent node contains the axisymmetric bounding boxes of all child nodes.

Specifically, the collision detection system removes the compared existing nodes in the axisymmetric bounding box tree, appends the removed existing nodes to a new branch node, and then appends the new branch node as the previous node on the existing node Child node of parent node. Back on the axisymmetric bounding box tree, adjust the axisymmetric bounding boxes of all parent nodes to ensure that the parent node contains the axisymmetric bounding boxes of all child nodes. For example, an axisymmetric bounding box tree is built by the following process:

The first triangle is added to the tree as the root node, and the second and third triangles are added to the tree structure. The second triangle and the third triangle need to be compared with the root node to determine whether the area of the root node has an intersection with the first triangle. If there is no intersection, compare the size of the bounding box containing the first triangle and the second triangle, the first triangle and the third triangle, the area of the first triangle and the second triangle is smaller, establish the parent node B of B1 and B2, delete the root node , take B1 and B2 as the left and right child nodes of B.

In the same way, compare B and C, there is no intersection, and establish root nodes A, B and C as left and right child nodes respectively.

Further, after the AABB tree is established, the closest distance between point clouds is queried through the AABB tree data structure retrieval. Sub-geometry objects in the AABB tree are triangular patches in the model. For example, query the distance from each point to the conical model in the graph, traverse from the root node, retrieve the distance between the target point cloud and the root node, then traverse left and right and branch nodes, then traverse left and right leaf nodes, and so on. Among them, the minimum value is the closest distance.

Among them, the AABB tree contains nine triangular patches of the tetrahedron model, and the distance between each point in the target spherical point cloud and these nine triangular patches is retrieved by the above method, and the minimum value is the closest between the two. distance.

In this embodiment, after the collision detection system establishes the AABB tree, it can dynamically construct the spatial position of the parent node through the update of the primitive pose, thereby reducing the time for rebuilding the tree structure during the movement of the robotic arm, which can improve the collision rate. detection efficiency.

As shown in Fig. 24, Fig. 25 and Fig. 26, in one embodiment, the axisymmetric bounding box tree is traversed, and the target is determined according to the closest distance between the target point cloud and the queried point cloud Whether the instrument will collide, including:

S1202: Acquire a second target point cloud and a second queried point cloud.

Specifically, the collision detection system acquires the second target point cloud and the second queried point cloud, and forms a second target point cloud group and a second queried point cloud group.

S1204: Establish an axis-symmetric bounding box tree of the second queried point cloud, traverse each second queried point cloud, find the two points closest to the point of the second queried point cloud and the point of the second target point cloud, The point distance is used as the second closest distance between point clouds.

Specifically, the collision detection system acquires the second queried point cloud, builds an AABB tree according to the acquired second queried point cloud group, traverses each point of the second queried point cloud group on the AABB tree, and finds the second target The closest point in the point cloud group to the second queried point cloud group. Through the closest point in the second target point cloud group and the second queried point cloud group, the minimum distance of the closest point is obtained, that is, the second closest distance between the second point clouds.

In this embodiment, after establishing the AABB tree, the collision detection system eliminates the collision risk according to the second closest distance between the second target point cloud and the second queried point cloud, and can track the collision between the robotic arm and the patient tissue in real time. Intraoperative posture.

It should be understood that, although the steps in the flowcharts involved in the above embodiments are sequentially displayed according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, the execution of these steps is not strictly limited to the order, and these steps may be performed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but may be performed at different times The execution order of these steps or phases is not necessarily sequential, but may be performed alternately or alternately with other steps or at least a part of the steps or phases in the other steps.

Based on the same inventive concept, an embodiment of the present application also provides a collision detection system for implementing the above-mentioned collision detection method. The implementation scheme for solving the problem provided by the collision detection system is similar to the implementation scheme described in the above method, so the specific limitations in one or more collision detection system embodiments provided below can refer to the above for collision detection method. limitations, which are not repeated here.

In one of the embodiments, a collision detection system is provided. The collision detection system includes an image processor and a position collector, and the image processor is used to implement the steps of the above method.

Specifically, the position collector acquires the real-time position of the surgical instrument by using the reflective marker and the positional relationship between the surgical instrument and the reflective marker determined by registration. The image processor is used to control the position collector to implement the steps of the above method.

Based on the same inventive concept, an embodiment of the present application also provides a surgical system for implementing the above-mentioned collision detection method. The implementation scheme for solving the problem provided by the surgical system is similar to the implementation scheme described in the above method, so the specific limitations in one or more surgical system embodiments provided below can refer to the above limitations on the collision detection method, It is not repeated here.

In one of the embodiments, a surgical system is provided, comprising: a robotic arm and a navigation device, the navigation device sends the coordinates of the preoperative planning to the robotic arm, and the robotic arm moves to a predetermined position through tool target positioning.

Specifically, the operating trolley and the navigation trolley are placed at suitable positions beside the hospital bed, and the femoral target, tibial target, base target, sterile bag, osteotomy guide tool, tool target, etc. are installed. The doctor imports the CT scan model of the patient's bone into the computer for preoperative planning, for example: planning the coordinates of the osteotomy plane, and selecting the appropriate type of prosthesis and planning the installation orientation of the prosthesis. The computer includes the main monitor 2, the keyboard and the navigation trolley. controller inside. The doctor uses the target pen to point the feature points of the patient's femur and tibia. The NDI navigation device uses the base target as a reference to record the position of the patient's bone feature points, and sends the position of the bone feature points to the computer, and then the computer obtains the femur and tibia through the feature matching algorithm. The actual orientation of the tibia. And corresponding to the CT image orientation of the femur and tibia, then the navigation system associates the actual orientation of the femur and tibia with the corresponding targets installed on the femur and tibia, so that the femoral target and the tibial target can track the actual position of the bone in real time. The navigation device sends the coordinates of the preoperatively planned osteotomy plane to the robotic arm. The robotic arm locates the osteotomy plane through the tool target and moves to a predetermined position. The robotic arm enters the holding state, and the doctor can use the oscillating saw or electric drill to guide the tool through the osteotomy. The osteotomy guide groove and guide hole are used for osteotomy and drilling operations. After the osteotomy and drilling are completed, the doctor can install the prosthesis and perform other surgical procedures.

The various modules in the collision detection system and the surgical system described above may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

In one of the embodiments, a computer device is provided, the computer device may be a server, and the internal structure diagram thereof may be as shown in FIG. 27 . The computer device includes a processor, memory, a network interface, and a database connected by a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium, an internal memory. The nonvolatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the execution of the operating system and computer programs in the non-volatile storage medium. The database of the computer device is used to store periodic task assignment data, such as configuration files, theoretical operating parameters and theoretical deviation value ranges, task attribute information, and the like. The network interface of the computer device is used to communicate with an external terminal through a network connection. The computer program, when executed by a processor, implements a collision detection method.

Those skilled in the art can understand that the structure shown in FIG. 27 is only a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer equipment to which the solution of the present application is applied. Include more or fewer components than shown in the figures, or combine certain components, or have a different arrangement of components.

In one embodiment, a computer device is provided, including a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

Get the object real-time position of the target object to be processed;

The virtual object model of the target object to be processed is registered with the object real-time position of the target object to be processed to determine the target position of the virtual object model in the target coordinate system where the target object to be processed is located;

Obtain the real-time position of the device of the target device;

Determine whether the target instrument and the target object to be processed will collide according to the target position and the real-time position of the instrument.

In one embodiment, when the processor executes the computer program, the virtual object model of the target object to be processed is registered with the object real-time position of the target object to be processed, so as to determine the target coordinate system where the virtual object model is located in the target object to be processed target locations in , including:

Obtain multiple feature points pre-marked on the virtual object model;

The multiple feature points are matched with the real-time position of the target object to be processed to determine the target position of the virtual object model in the target coordinate system where the target object to be processed is located.

In one embodiment, when the processor executes the computer program, the real-time position of the instrument of the target instrument is obtained, including:

Obtain the preset pose relationship between the reflective marker and the target device;

Based on the preset pose relationship, the real-time position of the target device is obtained according to the real-time pose of the reflective marker.

In one embodiment, when the processor executes the computer program, the virtual object model of the target object to be processed is registered with the object real-time position of the target object to be processed, so as to determine the target coordinate system where the virtual object model is located in the target object to be processed Before the target location in, also include:

Scanning to obtain the medical image of the target object to be processed;

The virtual object model of the target object to be processed is obtained by image reconstruction according to the medical image.

In one embodiment, when the processor executes the computer program, determining whether the target instrument will collide according to the target position and the real-time position of the instrument, including:

Using the K-dimensional space tree collision detection method and/or the AABB tree collision detection method, it is determined whether the target instrument will collide according to the target position and the real-time position of the instrument.

In one embodiment, a K-dimensional space tree collision detection method is implemented when the processor executes the computer program, including:

Use the target position and the real-time position of the instrument as the queried point cloud and/or the target point cloud;

Create a K-dimensional space tree based on the queried point cloud;

Traverse the K-dimensional space tree, and judge whether the target instrument will collide according to the closest distance between the target point cloud and the queried point cloud.

In one embodiment, before the processor executes the computer program, the target position and the real-time position of the instrument are used as the queried point cloud and/or the target point cloud, including:

The point cloud of the target position and the point cloud of the real-time position of the instrument are down-sampled respectively to obtain the sampling points of the target object to be processed and the target instrument.

In one embodiment, when the processor executes the computer program, down-sampling is performed on the point cloud of the target position and the point cloud of the real-time position of the instrument respectively, so as to obtain the sampling points of the target object to be processed and the target instrument, including:

Calculate the bounding box of the point cloud and discretize the bounding box into several voxels;

Take the center point of each voxel or the point closest to the center point as the sampling point.

In one embodiment, when the processor executes the computer program, the establishment of a K-dimensional space tree includes:

establish root node;

Calculate the variance value of the queried point cloud, and use the variance value as the segmentation feature;

Take the median of the segmentation feature as the segmentation point;

Pass the segmentation feature whose segmentation feature is smaller than the segmentation point to the left node of the root node, and the segmentation feature larger than the segmentation point to the right node of the root node;

The above steps are performed recursively until all segmentation features are established on the left and right nodes of the root node of the K-dimensional space tree.

In one embodiment, when the processor executes the computer program, it implements traversal of the K-dimensional space tree, and determines whether the target instrument will collide according to the closest distance between the target point cloud and the queried point cloud, including:

Obtain the first target point cloud and the first queried point cloud;

Traverse each first queried point cloud, find the two closest points between the first queried point cloud and the first target point cloud, take the distance between the two points as the first closest distance between the point clouds, and determine the first Whether a collision will occur at the closest distance.

In one embodiment, the processor implements the AABB tree collision detection method when executing the computer program, including:

Use the target position and the real-time position of the instrument as the queried point cloud and/or the target point cloud;

Build an axisymmetric bounding box tree based on the queried point cloud;

Traverse the axisymmetric bounding box tree, and judge whether the target instrument will collide according to the closest distance between the target point cloud and the queried point cloud.

In one embodiment, when the processor executes the computer program, the establishment of an axis-symmetric bounding box tree includes:

Take the triangle as the root node;

Establish leaf nodes of the queried point cloud, and assign axisymmetric bounding box trees according to the associated objects of the leaf nodes;

Find a predetermined existing node in the axisymmetric bounding box tree, and make the new leaf a sibling node of the predetermined existing node;

Create branch nodes for predetermined existing nodes and new leaves, and assign axisymmetric bounding boxes of two nodes to the branch nodes;

Append the new leaf to both nodes;

Remove the existing node from the axisymmetric bounding box tree and append the existing node to two nodes;

Append two nodes as children of the parent node of the existing node;

Adjusts the axisymmetric bounding box of the parent node to ensure that the parent node contains the axisymmetric bounding boxes of all child nodes.

In one embodiment, when the processor executes the computer program, the axisymmetric bounding box tree is traversed, and whether the target instrument will collide is determined according to the closest distance between the target point cloud and the queried point cloud, include:

Obtain the second target point cloud and the second queried point cloud;

Establish an axis-symmetric bounding box tree of the second queried point cloud, traverse each second queried point cloud, find the two closest points between the second queried point cloud and the second target point cloud, and calculate the distance between the two points. as the second closest distance between point clouds.

In one embodiment, when the computer program is executed by the processor, the virtual object model of the target object to be processed is registered with the real-time position of the target object to be processed, so as to determine the target coordinates of the virtual object model where the target object to be processed is located target locations in the system, including:

Get the object real-time position of the target object to be processed;

The virtual object model of the target object to be processed is registered with the object real-time position of the target object to be processed to determine the target position of the virtual object model in the target coordinate system where the target object to be processed is located;

Obtain the real-time position of the device of the target device;

According to the target position and the real-time position of the instrument, it is determined whether the target instrument will collide with the target object to be processed.

In one embodiment, when the computer program is executed by the processor, the virtual object model of the target object to be processed is registered with the real-time position of the target object to be processed, so as to determine the target coordinates of the virtual object model where the target object to be processed is located target locations in the system, including:

Obtain multiple feature points pre-marked on the virtual object model;

The multiple feature points are matched with the real-time position of the target object to be processed to determine the target position of the virtual object model in the target coordinate system where the target object to be processed is located.

In one embodiment, the computer program, when executed by the processor, achieves obtaining the real-time instrument position of the target instrument, including:

Obtain the preset pose relationship between the reflective marker and the target device;

Based on the preset pose relationship, the real-time position of the target device is obtained according to the real-time pose of the reflective marker.

In one embodiment, when the computer program is executed by the processor, the virtual object model of the target object to be processed is registered with the real-time position of the target object to be processed, so as to determine the target coordinates of the virtual object model where the target object to be processed is located Before the target location in the system, also include:

Scanning to obtain the medical image of the target object to be processed;

The virtual object model of the target object to be processed is obtained by image reconstruction according to the medical image.

In one embodiment, when the computer program is executed by the processor, it can determine whether the target instrument will collide according to the target position and the real-time position of the instrument, including:

Using the K-dimensional space tree collision detection method and/or the AABB tree collision detection method, it is determined whether the target instrument will collide according to the target position and the real-time position of the instrument.

In one embodiment, when the computer program is executed by the processor, a K-dimensional space tree collision detection method is implemented, including:

Use the target position and the real-time position of the instrument as the queried point cloud and/or the target point cloud;

Create a K-dimensional space tree based on the queried point cloud;

Traverse the K-dimensional space tree, and judge whether the target instrument will collide according to the closest distance between the target point cloud and the queried point cloud.

In one embodiment, the computer program, when executed by the processor, includes:

The point cloud of the target position and the point cloud of the real-time position of the instrument are down-sampled respectively to obtain the sampling points of the target object to be processed and the target instrument.

In one embodiment, when the computer program is executed by the processor, down-sampling is performed on the point cloud of the target position and the point cloud of the real-time position of the instrument respectively, so as to obtain the sampling points of the target object to be processed and the target instrument, including:

Calculate the bounding box of the point cloud and discretize the bounding box into several voxels;

Take the center point of each voxel or the point closest to the center point as the sampling point.

In one embodiment, the computer program, when executed by the processor, implements the establishment of a K-dimensional space tree, including:

establish root node;

Calculate the variance value of the queried point cloud, and use the variance value as the segmentation feature;

Take the median of the segmentation feature as the segmentation point;

Pass the segmentation feature whose segmentation feature is smaller than the segmentation point to the left node of the root node, and the segmentation feature larger than the segmentation point to the right node of the root node;

The above steps are performed recursively until all segmentation features are established on the left and right nodes of the root node of the K-dimensional space tree.

In one embodiment, when the computer program is executed by the processor, the K-dimensional space tree is traversed, and whether the target instrument will collide is determined according to the closest distance between the target point cloud and the queried point cloud, including:

Obtain the first target point cloud and the first queried point cloud;

Traverse each first queried point cloud, find the two closest points between the first queried point cloud and the first target point cloud, take the distance between the two points as the first closest distance between the point clouds, and determine the first Whether a collision will occur at the closest distance.

In one embodiment, when the computer program is executed by the processor, the AABB tree collision detection method includes:

Use the target position and the real-time position of the instrument as the queried point cloud and/or the target point cloud;

Build an axisymmetric bounding box tree based on the queried point cloud;

Traverse the axisymmetric bounding box tree, and judge whether the target instrument will collide according to the closest distance between the target point cloud and the queried point cloud.

In one embodiment, the computer program, when executed by the processor, implements the establishment of an axisymmetric bounding box tree, including:

Take the triangle as the root node;

Establish leaf nodes of the queried point cloud, and assign axisymmetric bounding box trees according to the associated objects of the leaf nodes;

Find a predetermined existing node in the axisymmetric bounding box tree, and make the new leaf a sibling node of the predetermined existing node;

Create branch nodes for predetermined existing nodes and new leaves, and assign axisymmetric bounding boxes of two nodes to the branch nodes;

Append the new leaf to both nodes;

Remove the existing node from the axisymmetric bounding box tree and append the existing node to two nodes;

Append two nodes as children of the parent node of the existing node;

Adjusts the axisymmetric bounding box of the parent node to ensure that the parent node contains the axisymmetric bounding boxes of all child nodes.

In one embodiment, the computer program traverses the axisymmetric bounding box tree when executed by the processor, and determines whether the target instrument will collide according to the closest distance between the target point cloud and the queried point cloud ,include:

Obtain the second target point cloud and the second queried point cloud;

Establish an axis-symmetric bounding box tree of the second queried point cloud, traverse each second queried point cloud, find the two closest points between the second queried point cloud and the second target point cloud, and calculate the distance between the two points. as the second closest distance between point clouds.

It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all Information and data authorized by the user or fully authorized by the parties.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage In the medium, when the computer program is executed, it may include the processes of the above-mentioned method embodiments. Wherein, any reference to a memory, a database or other media used in the various embodiments provided in this application may include at least one of a non-volatile memory and a volatile memory. Non-volatile memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetic variable memory (Magnetoresistive Random Memory) Access Memory, MRAM), Ferroelectric Random Access Memory (FRAM), Phase Change Memory (Phase Change Memory, PCM), graphene memory, etc. Volatile memory may include random access memory (Random Access Memory, RAM) or external cache memory, and the like. By way of illustration and not limitation, the RAM may be in various forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM). The database involved in the various embodiments provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a blockchain-based distributed database, etc., but is not limited thereto. The processors involved in the various embodiments provided in this application may be general-purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, data processing logic devices based on quantum computing, etc., and are not limited to this.

The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, all possible combinations of the technical features in the above embodiments are not described. However, as long as there is no contradiction in the combination of these technical features, all It is considered to be the range described in this specification.

The above-mentioned embodiments only represent several embodiments of the present application, and the descriptions thereof are relatively specific and detailed, but should not be construed as a limitation on the scope of the patent of the present application. It should be pointed out that for those skilled in the art, without departing from the concept of the present application, several modifications and improvements can be made, which all belong to the protection scope of the present application. Therefore, the scope of protection of the present application should be determined by the appended claims.

Claims (17)

1. a collision detection method, is characterized in that, described collision detection method comprises: Get the object real-time position of the target object to be processed; registering the virtual object model of the target object to be processed and the object real-time position of the target object to be processed to determine the target position of the virtual object model in the target coordinate system where the target object to be processed is located; Obtain the real-time position of the device of the target device; According to the target position and the real-time position of the instrument, it is determined whether the target instrument and the target object to be processed will collide. 2 . The collision detection method according to claim 1 , wherein the virtual object model of the target object to be processed is registered with the object real-time position of the target object to be processed to determine the virtual object model. 3 . The target position of the object model in the target coordinate system where the target object to be processed is located, including: acquiring multiple feature points pre-marked on the virtual object model; The multiple feature points are matched with the real-time position of the target object to be processed to determine the target position of the virtual object model in the target coordinate system where the target object to be processed is located. 3. The collision detection method according to claim 1, wherein the acquiring the real-time position of the device of the target device comprises: acquiring the preset pose relationship between the reflective marker and the target device; Based on the preset pose relationship, the real-time position of the device of the target device is obtained according to the real-time pose of the reflective marker. 4 . The collision detection method according to claim 1 , wherein the virtual object model of the target object to be processed is registered with the object real-time position of the target object to be processed to determine the virtual object model of the target object to be processed. 5 . Before the target position in the target coordinate system where the target object to be processed is located, the object model further includes: Scanning to obtain the medical image of the target object to be processed; Perform image reconstruction according to the medical image to obtain the virtual object model of the target object to be processed. 5 . The collision detection method according to claim 1 , wherein determining whether the target device will collide according to the target position and the real-time position of the device comprises: 6 . Using the K-dimensional space tree collision detection method and/or the AABB tree collision detection method, it is determined whether the target instrument will collide according to the target position and the real-time position of the instrument. 6. The collision detection method according to claim 5, wherein the K-dimensional space tree collision detection method comprises: using the target position and the real-time position of the instrument as the queried point cloud and/or the target point cloud; Create a K-dimensional space tree according to the queried point cloud; Traverse the K-dimensional space tree, and determine whether the target instrument will collide according to the closest distance between the target point cloud and the queried point cloud. 7. The collision detection method according to claim 6, characterized in that before taking the target position and the real-time position of the instrument as the queried point cloud and/or the target point cloud, the method comprises: The point cloud of the target position and the point cloud of the real-time position of the instrument are respectively down-sampled to obtain the sampling points of the target object to be processed and the target instrument. 8 . The collision detection method according to claim 7 , wherein the point cloud of the target position and the point cloud of the real-time position of the instrument are respectively down-sampled to obtain the target object to be processed and the object to be processed. 9 . Describe the sampling points for the target device, including: Calculate the bounding box of the point cloud, and discretize the bounding box into several voxels; The center point of each voxel or the point closest to the center point is used as the sampling point. 9. The collision detection method according to claim 6, wherein the establishing a K-dimensional space tree comprises: establish root node; Calculate the variance value of the queried point cloud, and use the variance value as a segmentation feature; Taking the median of the segmentation feature as the segmentation point; Transfer the segmentation feature whose segmentation feature is smaller than the segmentation point to the left node of the root node, and transfer the segmentation feature larger than the segmentation point to the right node of the root node; The above steps are recursively performed until all the segmentation features are established on the left node and the right node of the root node of the K-dimensional space tree. 10 . The collision detection method according to claim 6 , wherein the traversal of the K-dimensional space tree is performed, and the target device is judged according to the closest distance between the target point cloud and the queried point cloud. 11 . Whether collisions will occur, including: Obtain the first target point cloud and the first queried point cloud; Traverse each of the first queried point clouds, find the two closest points between the first queried point cloud and the first target point cloud, and use the distance between the two points as the first closest point between the point clouds distance, to determine whether a collision will occur at the first closest distance. 11. The collision detection method according to claim 5, wherein the AABB tree collision detection method comprises: using the target position and the real-time position of the instrument as the queried point cloud and/or the target point cloud; Build an axisymmetric bounding box tree according to the queried point cloud; Traverse the axisymmetric bounding box tree, and determine whether the target instrument will collide according to the closest distance between the target point cloud and the queried point cloud. 12. The collision detection method according to claim 11, wherein the establishing an axis-symmetric bounding box tree comprises: Take the triangle as the root node; establishing a leaf node of the queried point cloud, and assigning an axisymmetric bounding box tree according to the associated object of the leaf node; Find a predetermined existing node in the axisymmetric bounding box tree, and make the new leaf a sibling node of the predetermined existing node; creating a branch node for the predetermined existing node and the new leaf, and assigning an axisymmetric bounding box of two nodes to the branch node; attaching the new leaf to the two nodes; removing the existing node from the axisymmetric bounding box tree and appending the existing node to the two nodes; attaching the two nodes as child nodes of the parent node of the existing node; The axisymmetric bounding box of the parent node is adjusted to ensure that the parent node contains the axisymmetric bounding box of all of the child nodes. 13 . The collision detection method according to claim 11 , wherein the axisymmetric bounding box tree is traversed, and whether the target device is determined according to the closest distance between the target point cloud and the queried point cloud. 14 . Collision will occur, including: Obtain the second target point cloud and the second queried point cloud; Build an axisymmetric bounding box tree of the second queried point cloud, traverse each of the second queried point clouds, and find the point of the second queried point cloud that is closest to the point of the second target point cloud The two points of , take the distance between the two points as the second closest distance between the point clouds. 14. A collision detection system, characterized in that the collision detection system comprises: an image processor and a position collector, the image processor being used to implement the steps of any one of 1 to 13 of the method. 15. A computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein the processor implements the method according to any one of claims 1 to 13 when the processor executes the computer program. step. 16. A computer-readable storage medium on which a computer program is stored, characterized in that, when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 13 are implemented. 17. A computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements the steps of the method of any one of claims 1 to 13.

Images