Detection method and detection device of surgical navigation system and surgical navigation system

Abstract

The embodiment of the invention provides a detection method and a detection device of a surgical navigation system and the surgical navigation system. The detection method comprises the following steps: registering a real-time face point cloud and a preoperative reference face point cloud based on a four-point one-induced point cloud registration algorithm, and determining a target registration transformation matrix; according to the target registration transformation matrix, the real-time face point cloud of the target object and the real-time pose of the target head frame, the real-time relative position of the face of the target object and the target head frame is determined; according to the initial relative position and the real-time relative position, the navigation deviation of the surgical navigation system to the target object is detected. According to the method, global registration is carried out by adopting a four-point one-dimensional point cloud registration (4PCS) algorithm, and through geometric consistency constraint based on a coplanar four-point set, the problems that an existing face verification mode depends on manual probe collection, operation is complex, efficiency is low, and relative position deviation of a head frame and the face is difficult to find in time are comprehensively solved; therefore, the accuracy and robustness of the surgical navigation system are ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of surgical navigation systems, and particularly relates to a detection method and device of a surgical navigation system and the surgical navigation system. BACKGROUND

[0002] Neurosurgery requires high navigation accuracy, and usually requires fixing the patient's head through a head holder and installing optical markers (Markers) on the head holder for positioning to establish a surgical navigation coordinate system. The stability of the head holder Marker directly affects the overall accuracy of the navigation system. If the patient's head moves slightly during fixation or the head holder position relative to the face is offset, it may cause the navigation coordinate to be inconsistent with the actual patient anatomy, resulting in serious surgical risks.

[0003] In clinical practice, the common verification method mainly relies on attaching additional marker points to the patient's face and collecting facial point clouds point by point through a navigation probe, and then comparing them with preoperative image models to determine whether there is relative displacement between the head holder and the face. Attaching additional marker points to the patient's face not only prolongs the preoperative preparation time, but also easily causes errors due to improper operation, and may cause discomfort to the patient. Collecting facial point clouds point by point through a navigation probe relies on manual operation by doctors, which is not only inefficient but also has poor repeatability. The verification accuracy is limited by manual operation and marker stability, and it is difficult to find small relative shifts between the head holder and the patient's face in time, and it cannot fully meet the requirements of real-time and sub-millimeter accuracy in neurosurgery. SUMMARY

[0004] The present application provides a detection method and device of a surgical navigation system and the surgical navigation system, which adopts a four-point consistency point cloud registration (4PCS) algorithm to globally register real-time facial point clouds collected by a structured light camera with preoperative reference facial point clouds. Through geometric consistency constraints based on coplanar four-point sets, the problem of existing facial verification methods relying on manual probe collection, complex operation, low efficiency, and difficulty in finding the relative position offset between the head holder and the face in time is comprehensively solved, thereby ensuring the accuracy and robustness of the surgical navigation system.

[0005] The detection method of the surgical navigation system provided by the embodiment of the present application comprises: before the target object is navigated by the surgical navigation system, recording a preoperative reference facial point cloud of the target object, a preoperative reference pose of a target head frame, and an initial relative position of a face of the target object and the target head frame, the target head frame being used for fixing a head of the target object; in a case where the target object is navigated by the surgical navigation system, recording a real-time facial point cloud of the target object and a real-time pose of the target head frame; based on a four-point consistent point cloud registration algorithm, registering the real-time facial point cloud and the preoperative reference facial point cloud, and determining a target registration transformation matrix; according to the target registration transformation matrix, the real-time facial point cloud of the target object and the real-time pose of the target head frame, determining a real-time relative position of the face of the target object and the target head frame; and detecting navigation deviation of the target object by the surgical navigation system according to the initial relative position and the real-time relative position.

[0006] In some embodiments, the surgical navigation system comprises a binocular structured light camera, and the recording, before the target object is navigated by the surgical navigation system, of the preoperative reference facial point cloud of the target object, the preoperative reference pose of the target head frame and the initial relative position comprises: before the target object is navigated by the surgical navigation system, scanning a face of the target object by the binocular structured light camera to obtain a three-dimensional model of the face in an initial camera coordinate system, so as to determine the preoperative reference facial point cloud of the target object; before the target object is navigated by the surgical navigation system, identifying, by the binocular structured light camera, a mark point on the target head frame to obtain a pose of the mark point in the initial camera coordinate system, so as to determine the preoperative reference pose of the target head frame; and determining the initial relative position according to the preoperative reference facial point cloud of the target object and the preoperative reference pose of the target head frame.

[0007] In some embodiments, the surgical navigation system comprises a binocular structured light camera, and the recording, in a case where the target object is navigated by the surgical navigation system, of the real-time facial point cloud of the target object and the real-time pose of the target head frame comprises: in a case where the target object is navigated by the surgical navigation system, scanning a face of the target object by the binocular structured light camera to obtain a three-dimensional model of the face in an experimental camera coordinate system, so as to determine the real-time facial point cloud of the target object; in a case where the target object is navigated by the surgical navigation system, identifying, by the binocular structured light camera, a mark point on the target head frame to obtain a pose of the mark point in the experimental camera coordinate system, so as to determine the real-time pose of the target head frame.

[0008] In some embodiments, the four-point consistency point cloud registration algorithm is configured to register the point cloud set in the real-time facial point cloud and the point cloud set in the preoperative reference facial point cloud to determine a target registration transformation matrix, including: screening a reference approximate coplanar four-point set from the preoperative reference facial point cloud; preliminarily screening a target approximate coplanar four-point set from the real-time facial point cloud according to the reference approximate coplanar four-point set; calculating a candidate rigid transformation matrix of each target approximate coplanar four-point set and the reference approximate coplanar four-point set; evaluating each candidate rigid transformation matrix based on a target evaluation function to obtain a score value of each candidate rigid transformation matrix; and determining the target registration transformation matrix from a plurality of candidate rigid transformation matrices according to the score value of each candidate rigid transformation matrix.

[0009] In some embodiments, the screening of the reference approximate coplanar four-point set from the preoperative reference facial point cloud includes: determining an overlapping region in the preoperative reference facial point cloud according to a preset overlap integral; randomly selecting a first feature point, a second feature point, and a third feature point in the overlapping region; determining an initial plane according to the first feature point, the second feature point, and the third feature point; screening a fourth feature point in the overlapping region, the fourth feature point having a non-planar error value lower than a first preset error value with respect to the initial plane; and determining the reference approximate coplanar four-point set according to the first feature point, the second feature point, the third feature point, and the fourth feature point.

[0010] In some embodiments, the preliminary screening of the target approximate coplanar four-point set from the real-time facial point cloud according to the reference approximate coplanar four-point set includes: preliminarily screening a target approximate coplanar four-point set from the real-time facial point cloud according to a point pair distance of the target approximate coplanar four-point set, the point pair distance of the target approximate coplanar four-point set having an error value lower than a second preset error value with respect to the point pair distance of the reference approximate coplanar four-point set; and preliminarily screening a target approximate coplanar four-point set from the real-time facial point cloud according to an affine invariant ratio of the target approximate coplanar four-point set, the affine invariant ratio of the target approximate coplanar four-point set having an error value lower than a third preset error value with respect to the affine invariant ratio of the reference approximate coplanar four-point set.

[0011] In some embodiments, the evaluating each of the candidate rigid transformation matrix based on the target evaluation function to obtain a score value of each of the candidate rigid transformation matrix comprises: transforming the reference approximately coplanar four-point set by the candidate rigid transformation matrix to obtain a candidate registration point set corresponding to the candidate rigid transformation matrix; and evaluating the consistency of each of the candidate registration point set and the reference approximately coplanar four-point set based on a geometric consistency evaluation function to obtain the score value of each of the candidate rigid transformation matrix.

[0012] In some embodiments, the determining the real-time relative position of the face of the target object and the target head frame according to the target registration transformation matrix, the real-time face point cloud of the target object and the real-time pose of the target head frame comprises: determining updated feature coordinates of the target head frame according to the real-time pose of the target head frame; determining updated feature coordinates of the face of the target object according to the target registration transformation matrix and the real-time face point cloud of the target object; and determining the real-time relative position of the face of the target object and the target head frame according to the updated feature coordinates of the target head frame and the updated feature coordinates of the face of the target object.

[0013] The detection device provided by the embodiment of the present application comprises a memory and a processor, the memory stores a computer program, and the processor is configured to execute the above detection method through the computer program.

[0014] The surgical navigation system provided by the embodiment of the present application comprises the above detection device.

[0015] The detection method of the surgical navigation system provided by the embodiment of the present application comprises: recording a preoperative reference face point cloud of a target object, a preoperative reference pose of a target head frame and an initial relative position of the face of the target object and the target head frame before the target object is navigated by the surgical navigation system, the target head frame being used for fixing the head of the target object; recording a real-time face point cloud of the target object and a real-time pose of the target head frame in the case that the target object is navigated by the surgical navigation system; registering the real-time face point cloud and the preoperative reference face point cloud based on a four-point consistent point cloud registration algorithm to determine a target registration transformation matrix; determining the real-time relative position of the face of the target object and the target head frame according to the target registration transformation matrix, the real-time face point cloud of the target object and the real-time pose of the target head frame; and detecting navigation deviation of the target object by the surgical navigation system according to the initial relative position and the real-time relative position.

[0016] The embodiment of the present application adopts a four-point consistency point cloud registration (4PCS) algorithm to globally register real-time facial point clouds collected by a structured light camera and preoperative reference facial point clouds. Through geometric consistency constraints based on coplanar four-point sets, the problems of existing facial verification methods, such as dependence on manual probe collection, complex operation, low efficiency, and difficulty in timely discovering the relative position offset between a head frame and a face, are comprehensively solved, thereby ensuring the accuracy and robustness of a surgical navigation system. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0018] Figure 1 The working step schematic diagram of the detection method of the surgical navigation system provided by the embodiment of the present application.

[0019] Figure 2 The structural schematic diagram of the surgical navigation system provided by the embodiment of the present application.

[0020] Figure 3 The coordinate system definition schematic diagram of the preoperative reference facial point cloud and the real-time facial point cloud provided by the embodiment of the present application.

[0021] Figure 4 The schematic diagram of the affine transformation invariance defining the approximate coplanar four points provided by the embodiment of the present application.

[0022] Figure 5 The schematic diagram of the affine transformation invariance defining the approximate coplanar four points provided by the embodiment of the present application. DETAILED DESCRIPTION

[0023] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings of the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0024] Neurosurgery requires high navigation accuracy, usually requires the patient's head to be fixed through the headrest, and the optical marker (Marker) for positioning is installed on the headrest to establish the surgical navigation coordinate system. The stability of the headrest Marker directly affects the overall accuracy of the navigation system. If the patient's head moves slightly during the fixing process, or the headrest position relative to the face is offset, it may cause the navigation coordinates to be inconsistent with the actual patient anatomy, resulting in serious surgical risks.

[0025] In clinical practice, the common verification method mainly relies on attaching additional marker points to the patient's face and collecting the face point cloud point by point through the navigation probe, and then comparing it with the preoperative image model to determine whether there is relative displacement between the headrest and the face. Attaching additional marker points to the patient's face not only prolongs the preoperative preparation time, but also easily causes errors due to improper operation, and may also cause discomfort to the patient. Collecting face point cloud point by point through the navigation probe relies on the manual operation of the doctor, which is not only inefficient, but also has poor repeatability. The verification accuracy is limited by manual operation and marker stability, and it is difficult to find the small relative offset between the headrest and the patient's face in time, which cannot fully meet the requirements of neurosurgery for real-time and sub-millimeter accuracy.

[0026] To solve the above technical problems, the embodiments of the present application provide a detection method, a detection device and a surgical navigation system of a surgical navigation system, which realizes the detection of the surgical navigation system based on the automatic and non-contact face verification method of the structured light camera, and improves the reliability and clinical applicability of the verification.

[0027] The embodiments of the present application use the four-point congruent point cloud registration (4PCS) algorithm to globally register the real-time face point cloud collected by the structured light camera and the preoperative reference face point cloud. Through the geometric congruence constraint based on the coplanar four-point set, the problems of the existing face verification method, such as relying on manual probe collection, complex operation, low efficiency and difficulty in finding the relative position offset between the headrest and the face in time, are comprehensively solved, thereby ensuring the accuracy and robustness of the surgical navigation system.

[0028] First, some of the nouns or terms appearing in the process of describing the embodiments of the present application are explained as follows: 4PCS (4-Points Congruent Sets) algorithm is a point cloud rough registration method based on affine invariance, which realizes efficient alignment by finding the geometric congruence of the coplanar four-point base set in the source point cloud and the target point cloud. This algorithm significantly reduces the matching calculation amount through the topological constraint of the coplanar four-point set, and is suitable for rough registration of large-scale point clouds with high overlap rate.

[0029] The embodiment of the present application can be applied to the application scenarios in which the surgical navigation system needs to make rigid contact with the target, such as the fine surgical operations of the orthopedic joint replacement and dental implant of the surgical navigation system, and the fine operations such as assembly and processing of the industrial surgical navigation system.

[0030] The detection method of the surgical navigation system provided by the exemplary embodiment of the present application will be described below in combination with the above-described application scenarios with reference to the accompanying drawings. It should be noted that the above-described application scenarios are only shown for the purpose of facilitating the understanding of the principles of the present application, and the embodiments of the present application are not limited in this respect.

[0031] Figure 1 The working steps of the detection method of the surgical navigation system provided by the embodiment of the present application are shown in the schematic diagram. Figure 1 As shown in the schematic diagram, the detection method of the surgical navigation system provided by the embodiment of the present application comprises: Step 01: Before the target object is navigated by the surgical navigation system, record the preoperative reference facial point cloud of the target object, the preoperative reference pose of the target head frame, and the initial relative position of the face of the target object and the target head frame, and the target head frame is used to fix the head of the target object; Step 02: In the case where the target object is navigated by the surgical navigation system, record the real-time facial point cloud of the target object and the real-time pose of the target head frame; Step 03: Based on the four-point consistent point cloud registration algorithm, register the real-time facial point cloud and the preoperative reference facial point cloud to determine the target registration transformation matrix; Step 04: According to the target registration transformation matrix, the real-time facial point cloud of the target object and the real-time pose of the target head frame, determine the real-time relative position of the face of the target object and the target head frame; Step 05: According to the initial relative position and the real-time relative position, detect the navigation deviation of the target object by the surgical navigation system.

[0032] Specifically, before the target object is navigated by the surgical navigation system, record the preoperative reference facial point cloud of the target object as the registration reference, record the spatial position and attitude of the head frame in the preoperative coordinate system as the reference pose of the target head frame, and record the initial relative position of the face of the target object and the target head frame as the baseline for subsequent deviation detection.

[0033] In the case that the surgical navigation system navigates the target object, the real-time face point cloud of the target object and the real-time pose of the target head frame are tracked and recorded, a four-point consistency point cloud registration algorithm (4PCS) is adopted to register the real-time face point cloud and the preoperative reference face point cloud, a target registration transformation matrix is determined so as to obtain a registered face position, the real-time relative position of the face of the target object and the target head frame is accurately calculated, and the initial relative position and the real-time relative position are compared to detect the navigation deviation of the surgical navigation system to the target object.

[0034] The embodiment of the present application detects the navigation deviation caused by the displacement of the head frame in the surgical process by preoperative recording and intraoperative real-time registration, adopts the 4PCS algorithm to globally register the real-time face point cloud collected by the structured light camera and the preoperative reference face point cloud, and through the geometric consistency constraint based on the coplanar four-point set, the 4PCS can realize fast and stable global alignment without initial pose information, avoiding the convergence failure problem of the ICP algorithm when the initial error is large. At the same time, the 4PCS algorithm has low computational complexity and can meet the dual requirements of real-time and accuracy of the surgical navigation system, significantly improving the efficiency and reliability of the relative position verification of the face and the Marker.

[0035] Figure 2 The surgical navigation system provided by the embodiment of the present application is shown in Figure 2 As shown, the surgical navigation system includes an infrared binocular structured light camera, a head frame, a registration calculation module and an error calculation module. The infrared binocular structured light camera is used to collect real-time three-dimensional point cloud data of the patient's face and identify the Marker on the head frame, and provide high-precision pose information. The structured light function of the camera itself can quickly construct a face structure model, providing data support for subsequent registration and comparison. The head frame is installed on the patient's head frame and is used as a head fixation and coordinate reference. Through registration with the face point cloud identified by the structured light camera, it is determined whether there is a relative position deviation between the head frame and the face.

[0036] The registration calculation module compares the recorded face structure and the face structure during verification based on the point cloud registration algorithm, calculates the spatial transformation matrix, and calculates the relative position of the face during verification and the Marker. The error calculation module calculates the current verification Marker origin position by applying the spatial transformation matrix to the origin position of the Marker, and calculates the error between the recorded Marker point and the origin position of the verification Marker point.

[0037] The surgical navigation system provided by the embodiment of the present application combines the three-dimensional reconstruction function of the structured light camera and is used for non-contact verification of the relative position relationship between the head frame and the patient's face during the surgical process.

[0038] Figure 3This is a schematic diagram illustrating the coordinate system definition of the preoperative reference facial point cloud and the real-time facial point cloud provided in embodiments of the present invention. Figure 3 As shown, the coordinate system is specifically defined as follows: C Head : Facial coordinate system recorded before the experiment.

[0039] C Marker : Marker coordinate system attached to the face recorded before the experiment.

[0040] C Head’ The experiment verified the facial coordinate system in real time.

[0041] C Marker’ : The marker coordinate system is attached to the face for real-time verification during the experiment.

[0042] C Camera Camera coordinate system.

[0043] In this invention, the binocular infrared structured light camera has undergone an internal calibration process to ensure that the depth information acquired by the camera has a unified dimension and spatial consistency with the spatial geometric coordinates. After calibration, the facial point cloud obtained by scanning and the reference marker array are both located in the same spatial coordinate system, and the coordinate relationship can be defined in the camera coordinate system C. Camera It is expressed directly in Chinese.

[0044] The following reference Figure 3 The defined coordinate system will be used to further explain steps 01 to 05 above.

[0045] In some embodiments, step 01 above includes: Step 011: Before the surgical navigation system navigates the target object, the face of the target object is scanned by a binocular structured light camera to obtain a three-dimensional facial model in the initial camera coordinate system, so as to determine the preoperative reference facial point cloud of the target object. Step 012: Before the surgical navigation system navigates the target object, the marker points on the target head frame are identified by a binocular structured light camera to obtain the pose of the marker points in the initial camera coordinate system, so as to determine the preoperative reference pose of the target head frame; Step 013: Determine the initial relative position based on the preoperative reference facial point cloud of the target subject and the preoperative reference pose of the target head frame.

[0046] Specifically, before surgical navigation, a binocular structured light camera is used to identify the markers on the head frame, obtaining the markers' positions on the C-axis. Camera The rigid body pose is stored as a 4×4 homogeneous transformation matrix, denoted as: .

[0047] Then, a three-dimensional surface model of the patient's face is generated using active projection gratings and binocular parallax reconstruction, and recorded on a C-cell. Camera From this, the relative position of the face and the marker can be obtained.

[0048] In some embodiments, step 02 above includes: Step 021: With the surgical navigation system navigating the target object, the face of the target object is scanned by a binocular structured light camera to obtain a three-dimensional facial model in the experimental camera coordinate system, so as to determine the real-time facial point cloud of the target object. Step 022: With the surgical navigation system navigating the target object, the marker points on the target head frame are identified by a binocular structured light camera to obtain the pose of the marker points in the experimental camera coordinate system, so as to determine the real-time pose of the target head frame.

[0049] Specifically, when the surgical navigation system navigates the target object, the 3D reconstruction function of the structured light camera is used to determine the real-time facial point cloud of the target object and record it in C. Camera Down.

[0050] The marker ' on the head unit is identified using a binocular structured light camera, and the position of the marker ' at C is obtained. Camera The rigid body pose is stored as a 4×4 homogeneous transformation matrix, denoted as:

[0051] In some embodiments, step 03 above includes: Step 031: Select a set of four approximately coplanar reference points from the preoperative reference facial point cloud; Step 032: Based on the reference set of approximately coplanar four points, initially filter the target set of approximately coplanar four points from the real-time facial point cloud; Step 033: Calculate the candidate rigid body transformation matrix for each target approximate coplanar four-point set and the reference approximate coplanar four-point set; Step 034: Based on the objective evaluation function, evaluate each candidate rigid body transformation matrix to obtain a score value for each candidate rigid body transformation matrix; Step 035: Determine the target registration transformation matrix from multiple candidate rigid body transformation matrices based on the score value of each candidate rigid body transformation matrix.

[0052] Specifically, when the surgical navigation system navigates the target object, the 4PCS registration method is used to solve for the transformation matrix from the real-time verification facial coordinate system to the facial coordinate system recorded before the experiment, which can be used as the target registration transformation matrix. Specifically:

[0053] in: Rotation matrix: Represents the rotation relationship between coordinate systems. Translation vector: Represents the translation relationship between coordinate systems.

[0054] In the 4PCS registration method, the affine transformation of approximately coplanar four points has the following property: given three collinear points {a, b, c}, the ratio ||a| ... b‖ / ‖a c is invariant. Huttenlocher uses this invariant to extract all 2D affine invariant sets for four points on a plane, where these invariants are equivalent under affine transformations. A similar method is used in 3D space. Given a coplanar 4-point basis, we find its affine equivalent (congruent) 4-point set in another point cloud dataset. All affine invariant 4-point sets are supersets of congruent 4-point sets in 3D space. Subsequently, we verify whether such 4-point sets are (approximately) consistent with the selected basis set. First, the method for extracting 2D affine invariant 4-point sets is briefly introduced, and then the extraction process for 3D affine invariant 4-point sets is described in detail.

[0055] Let X ≡ {a, b, c, d} be a set of coplanar points, not all of which are collinear. Define two independent ratios for the three collinear points. Let ab and cd be two lines intersecting at point e. These two ratios r1 and r2 are:

[0056]

[0057] Figure 4 A schematic diagram illustrating the definition of approximately coplanar four points for affine transformation invariance. (See diagram below.) Figure 4 As shown, the two ratios r1 and r2 are invariant to affine transformations and can uniquely define four points.

[0058] Figure 5 This is a schematic diagram illustrating the extraction of approximately coplanar four points using affine transformation invariance. (See diagram below.) Figure 5 As shown, given a set Q consisting of n points, and two affine invariant ratios... and This can efficiently extract all four point sets defined by these two invariants, where k is the number of four point sets, as shown below: For each pair of points , ∈Q, calculate the two intermediate points:

[0059]

[0060] For any 4 points obtained from the reference point cloud , At any two points in the point cloud to be registered , From the above formula, we can obtain 2 (Depend on (Calculated), 2 Point (by) (Calculated). For two line segments (4 points), if any one of the points in the first line segment... With any one of the second line segments If the two line segments overlap or are within a very small distance, then the four endpoints of these two line segments can be considered to be the corresponding points of the four points in the reference point cloud.

[0061] Given a 4-point basis B ∈ R3 of (approximately) coplanar points chosen from a point set P and another point set Q, extract the set of all 4 points from Q that are approximately equal to B. First, given B, compute the ratio of its two affine invariants in this plane, as before.

[0062] Then, from point Q, all points related to B are extracted through affine transformation. Although this method can obtain the coplanar point set of the required 4 points, there will be a certain number of spurious matches.

[0063] To remove non-congruent bases, examine their original positions in the point cloud and verify that the corresponding set is consistent with the base set B within a certain distance threshold. Then, using each basis in B and Q, compute the optimal alignment rigid transformation by least squares.

[0064] The above procedure requires a large amount of memory, which is unsuitable for large point clouds. Therefore, the embodiments of this application make the following improvement: rigid transformation preserves the Euclidean distance between points. Given a cardinality B ≡ {a, b, c, d}, first calculate the distance d1 = ||a||a. b‖ and d2=‖c d‖. Now we only consider the pairs of points in Q with a distance of d1 or d2, and the maximum error is δ.

[0065] Given a preoperative reference facial point cloud P and a real-time facial point cloud Q, a set of four approximately coplanar points is extracted from the preoperative reference facial point cloud P as a reference set of approximately coplanar four points B. Based on the reference set of approximately coplanar four points B, a set of four points approximately equal to the reference set of approximately coplanar four points B is extracted from the real-time facial point cloud Q. U ≡ all subsets {U1, U2, ..., Us} of the four points in Q, which may be consistent with B. For each Ui, the correspondence information between B and Ui is used to solve for the candidate rigid body transformation matrix Ti between B and Ui.

[0066] To verify the effectiveness of each candidate rigid body transformation matrix Ti, each candidate rigid body transformation matrix is ​​evaluated based on the target evaluation function, and a score value is obtained for each candidate rigid body transformation matrix. The higher the score, the higher the registration accuracy. Finally, the one with the highest score Ti is selected as the target registration transformation matrix.

[0067] In some implementations, step 031 above includes: determining the overlapping region in the preoperative reference facial point cloud based on a preset overlap integral; In the overlapping region, the first feature point, the second feature point, and the third feature point are randomly selected; The initial plane is determined based on the first feature point, the second feature point, and the third feature point; In the overlapping region, a fourth feature point is selected, and the non-planarity error value between the fourth feature point and the initial plane is lower than the first preset error value. Based on the first feature point, the second feature point, the third feature point, and the fourth feature point, determine the set of four reference approximately coplanar points.

[0068] Specifically, select a basis B consisting of 4 coplanar points. P. Because it's unlikely that there are four coplanar points. Randomly select three points, and choose the remaining points, so that the four points together form a (approximately) coplanar point. A wider base is created by selecting points that are relatively far apart, resulting in more stable coplanar points. Allowing some non-planarity, let the four previously selected points... Let the first three points define a plane. The distance from the fourth point to the plane is the non-planarity error, and the non-planarity error value needs to be lower than the first preset error value.

[0069] The mathematical expression for nonplanarity error is:

[0070] in,

[0071] Let be the plane normal vector defined by the first three points.

[0072] However, if the selected points are too far apart, they may not all fall within the overlapping region (for partial matching). This maximum distance is estimated using an overlap score f. If no estimate of f is provided, the algorithm will run in a way that reduces guesswork for f=1, 0.5, 0.25, ... until the desired error tolerance is reached. With f known, first select a set of three points that may fall within the overlapping region, then select a fourth point as before.

[0073] In some embodiments, step 032 above includes: Based on the point-to-point distance of the target approximately coplanar four-point set, the target approximately coplanar four-point set is initially screened from the real-time facial point cloud. The error value between the point-to-point distance of the target approximately coplanar four-point set and the point-to-point distance of the reference approximately coplanar four-point set is lower than the second preset error value. Based on the affine invariant ratio of the target's approximately coplanar four-point set, the target's approximately coplanar four-point set is initially screened from the real-time facial point cloud. The error value between the affine invariant ratio of the target's approximately coplanar four-point set and the affine invariant ratio of the reference approximately coplanar four-point set is lower than the third preset error value.

[0074] Specifically, the target approximately coplanar set of four points B={a,b,c,d} and its affine invariant ratio r1=∥a e∥∥a b∥、r2=∥c e∥∥c d∥. Search in the real-time facial point cloud Q for all point pair combinations that satisfy the following conditions: Distance constraint: The distance between point pairs must be approximately equal to the distance between point pairs in the base set (tolerance is the second preset error value):

[0075] Calculate intermediate points: For the pair of points (q1, q2) in the target point cloud Q that satisfy the distance constraint, calculate their affine invariant intermediate points:

[0076] There exists another pair of points (q3, q4) whose intermediate points e1' or e2' satisfy ||ei ej'∥≤ third preset error value. At this point, it can be considered that... With any one of the second line segments If the two line segments overlap or are within a very small distance, then the four endpoints corresponding to these two line segments can be considered to be the corresponding points of the four points in the target approximately coplanar four-point set B, which can be used as the initial screening target approximately coplanar four-point set.

[0077] In some embodiments, step 033 above includes: The reference set of four approximately coplanar points is transformed using a candidate rigid body transformation matrix to obtain the set of candidate registration points corresponding to the candidate rigid body transformation matrix. The evaluation function based on geometric consistency assesses the consistency between each candidate registration point set and the reference approximately coplanar four-point set, and obtains the score value of each candidate rigid body transformation matrix.

[0078] To verify the effectiveness of each candidate rigid transformation Ti, this invention uses a scoring function based on geometric consistency. The benchmark point cloud... After Ti transformation, the result is obtained ,statistics Center and target point cloud The distance between any point in the middle is less than a preset threshold. The proportion of points:

[0079] A higher score indicates higher registration accuracy. The highest score, Ti, is ultimately selected as the target registration transformation matrix.

[0080] In some implementations, step 04 above includes: Based on the real-time pose of the target headframe, determine the updated feature coordinates of the target headframe; Based on the target registration transformation matrix and the real-time facial point cloud of the target object, determine the updated feature coordinates of the target object's face; Based on the updated feature coordinates of the target headgear and the updated feature coordinates of the target object's face, the real-time relative position between the target object's face and the target headgear is determined.

[0081] Specifically, the registration method described above will be used to obtain... The matrix is ​​applied to the coordinates of Marker' in the camera coordinate system. Marker' is calibrated via a binocular structured light camera, thus allowing us to obtain C. Camera The transformation matrix below , That is, Marker' in C Camera The coordinates of the origin P below PointOnCamera We can obtain the transformation matrix through the camera. Therefore, in C Head The coordinates of the Marker can be obtained using the formula. The distance error between Marker' and the point whose origin coordinates are in the same coordinate system is calculated using Euclidean geometric distance, as shown in the following formula:

[0082] In step 05 above, the distance error between the Marker' and the origin coordinates of the Marker in the same coordinate system is calculated based on the Euclidean geometric distance. This can determine the navigation deviation of the surgical navigation system for the target object. The larger the distance error, the more serious the navigation deviation of the surgical navigation system for the target object.

[0083] This application embodiment utilizes the structured light function built into the camera to quickly acquire three-dimensional facial information during surgery and compare it with preoperative reference information, thereby detecting whether there is a relative positional shift between the head frame and the patient's face, avoiding navigation inaccuracy caused by a loose head frame.

[0084] After registration, this invention uses a 3D structured light camera to scan the face structure and record the relative position of the face and the marker, which serves as a reference for the navigation system. During the verification phase, the face structure is scanned again using structured light, and registration is performed with the previous reference. The resulting transformation matrix is ​​applied to the origin center of the marker. By comparing the differences between the two coordinate scans, the overall positioning accuracy of the system is directly quantified, providing intuitive and reliable verification results.

[0085] Compared to traditional methods that rely on probe sampling or solely on headframe stability assessment, this invention combines a dual verification mechanism of rapid structured light detection and point-to-point error comparison. This mechanism can detect headframe shifts in real time and quickly assess the accuracy of the navigation system, significantly improving the convenience and efficiency of clinical operations.

[0086] This invention can serve as a general enhancement module for navigation surgical systems. It requires no additional expensive equipment and can complete the verification process using only existing structured light cameras and probes, making it easy to promote and apply in various neurosurgical scenarios.

[0087] In summary, this invention combines structured light verification with facial sampling comparison to form a verification method that provides dual protection of "headset stability + navigation accuracy". This not only overcomes the shortcomings of traditional facial verification methods, such as low efficiency, reliance on manual labor, and difficulty in timely detection of problems, but also significantly improves the safety, accuracy, and clinical applicability of the navigation system.

[0088] The detection device provided in this embodiment of the invention includes a memory and a processor. The memory stores a computer program, and the processor is configured to execute the detection method of the above-described embodiments through the computer program. The embodiments of the detection device include the detection method of the surgical navigation system described above, which will not be repeated here.

[0089] For example, the detection method of the surgical navigation system provided in this embodiment of the invention can be executed by a computer device, which can be a terminal or a server. The terminal can be a smartphone, tablet, laptop, desktop computer, smart TV, smart speaker, wearable smart device, personal computer (PC), smart vehicle terminal, etc. The server can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery network (CDN), and big data and artificial intelligence platforms. However, it is not limited to these. The terminal and server can be directly or indirectly connected via wired or wireless communication, which is not limited in this embodiment of the invention.

[0090] The surgical navigation system provided in this embodiment of the invention includes the detection device provided in this embodiment of the invention. The implementation methods of the surgical navigation system include the detection method and detection device of the surgical navigation system described above, and will not be repeated here.

[0091] For example, the surgical navigation system provided in the embodiments of the present invention may include a surgical navigation system for performing delicate surgical operations such as orthopedic joint replacement and dental implantation, a rehabilitation surgical navigation system that interacts with the human body, and an industrial surgical navigation system for performing delicate operations such as assembly and processing.

[0092] Surgical navigation systems can automatically adjust filtering and control parameters based on the rate of change of external force and the stage of surgery, without relying on manual adjustments. This improves surgical safety, reduces the risk of accidental cutting and scratching, and enhances the reliability and precision of surgical navigation systems in complex bone tissue manipulations. Surgical navigation systems are highly versatile, applicable not only to orthopedic / dental surgeries but also extend to other high-precision contact tasks.

[0093] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0094] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0095] In the embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0096] In this embodiment of the invention, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0097] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0098] In addition, the functional units in the embodiments of the present invention can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a unit.

[0099] If the functionality is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer or a server) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0100] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A detection method for a surgical navigation system, characterized in that, The detection method includes: Before the surgical navigation system navigates the target object, the preoperative reference facial point cloud of the target object, the preoperative reference pose of the target head frame, and the initial relative position of the target object's face and the target head frame are recorded. The target head frame is used to fix the head of the target object. When the surgical navigation system navigates the target object, the real-time facial point cloud of the target object and the real-time pose of the target head frame are recorded. Based on the four-point consistency point cloud registration algorithm, the real-time facial point cloud and the preoperative reference facial point cloud are registered to determine the target registration transformation matrix. Based on the target registration transformation matrix, the real-time facial point cloud of the target object, and the real-time pose of the target headgear, the real-time relative position of the target object's face and the target headgear is determined. Based on the initial relative position and the real-time relative position, the navigation offset of the surgical navigation system to the target object is detected.

2. The detection method for the surgical navigation system as described in claim 1, characterized in that, The surgical navigation system includes a binocular structured light camera. Before navigating the target object, the system records the preoperative reference facial point cloud, the preoperative reference pose of the target head frame, and the initial relative position of the target object, including: Before the surgical navigation system navigates the target object, the face of the target object is scanned by the binocular structured light camera to obtain a three-dimensional facial model in the initial camera coordinate system, so as to determine the preoperative reference facial point cloud of the target object; Before the surgical navigation system navigates the target object, the binocular structured light camera identifies the marker points on the target head frame to obtain the pose of the marker points in the initial camera coordinate system, so as to determine the preoperative reference pose of the target head frame; The initial relative position is determined based on the preoperative reference facial point cloud of the target object and the preoperative reference pose of the target head frame.

3. The detection method for the surgical navigation system as described in claim 1, characterized in that, The surgical navigation system includes a binocular structured light camera. When the surgical navigation system navigates the target object, recording the real-time facial point cloud of the target object and the real-time pose of the target headgear includes: When the surgical navigation system navigates the target object, the face of the target object is scanned by the binocular structured light camera to obtain a three-dimensional facial model in the experimental camera coordinate system, so as to determine the real-time facial point cloud of the target object; When the surgical navigation system navigates the target object, the binocular structured light camera identifies the marker points on the target head frame to obtain the pose of the marker points in the experimental camera coordinate system, thereby determining the real-time pose of the target head frame.

4. The detection method for the surgical navigation system as described in claim 1, characterized in that, The four-point consistency point cloud registration algorithm registers the point cloud set in the real-time facial point cloud with the point cloud set in the preoperative reference facial point cloud to determine the target registration transformation matrix, including: From the preoperative reference facial point cloud, a set of approximately coplanar four points is selected; Based on the reference set of approximately coplanar four points, a preliminary selection of the target set of approximately coplanar four points is made from the real-time facial point cloud. Calculate the candidate rigid body transformation matrix between each of the target approximately coplanar four-point sets and the reference approximately coplanar four-point sets; Based on the objective evaluation function, each candidate rigid body transformation matrix is ​​evaluated to obtain a score value for each candidate rigid body transformation matrix; The target registration transformation matrix is ​​determined from the plurality of candidate rigid body transformation matrices based on the score value of each candidate rigid body transformation matrix.

5. The detection method for the surgical navigation system as described in claim 4, characterized in that, The step of selecting a set of approximately coplanar four points from the preoperative reference facial point cloud includes: The overlapping region in the preoperative reference facial point cloud is determined based on the preset overlap integral. In the overlapping region, a first feature point, a second feature point, and a third feature point are randomly selected; The initial plane is determined based on the first feature point, the second feature point, and the third feature point; In the overlapping region, a fourth feature point is selected, wherein the non-planarity error value between the fourth feature point and the initial plane is lower than a first preset error value; The reference approximately coplanar four-point set is determined based on the first feature point, the second feature point, the third feature point, and the fourth feature point.

6. The detection method for the surgical navigation system as described in claim 4, characterized in that, The preliminary screening of target approximately coplanar four-point sets from the real-time facial point cloud based on the reference approximately coplanar four-point set includes: Based on the point-to-point distance of the target approximately coplanar four-point set, a target approximately coplanar four-point set is initially screened from the real-time facial point cloud. The error value between the point-to-point distance of the target approximately coplanar four-point set and the point-to-point distance of the reference approximately coplanar four-point set is lower than a second preset error value. Based on the affine invariant ratio of the target approximately coplanar four-point set, a preliminary selection of the target approximately coplanar four-point set is made from the real-time facial point cloud. The error value between the affine invariant ratio of the target approximately coplanar four-point set and the affine invariant ratio of the reference approximately coplanar four-point set is lower than a third preset error value.

7. The detection method for the surgical navigation system as described in claim 4, characterized in that, The evaluation of each candidate rigid body transformation matrix based on the target evaluation function to obtain a score value for each candidate rigid body transformation matrix includes: The reference approximately coplanar four-point set is transformed by the candidate rigid body transformation matrix to obtain the candidate registration point set corresponding to the candidate rigid body transformation matrix. The evaluation function based on geometric consistency evaluates the consistency between each candidate registration point set and the reference approximately coplanar four-point set, and obtains a score value for each candidate rigid body transformation matrix.

8. The detection method for the surgical navigation system as described in claim 1, characterized in that, Determining the real-time relative position of the target object's face and the target headframe based on the target registration transformation matrix, the real-time facial point cloud of the target object, and the real-time pose of the target headframe includes: Based on the real-time pose of the target headframe, determine the updated feature coordinates of the target headframe; Based on the target registration transformation matrix and the real-time facial point cloud of the target object, the updated feature coordinates of the target object's face are determined; Based on the updated feature coordinates of the target headgear and the updated feature coordinates of the target object's face, the real-time relative position between the target object's face and the target headgear is determined.

9. A detection device, characterized in that, The detection device includes a memory and a processor, the memory storing a computer program, and the processor being configured to execute the detection method according to any one of claims 1 to 8 through the computer program.

10. A surgical navigation system, characterized in that, The surgical navigation system includes the detection device as described in claim 9.

Images