Registration method, registration device, electronic device and computer-readable storage medium

Abstract

The application provides a registration method, a registration device, an electronic device and a computer readable storage medium, and relates to the field of surgical navigation. The method comprises the following steps: determining a first space main axis direction of an image space point cloud and a second space main axis direction of a surgical space point cloud, correcting the first space main axis direction to obtain a plurality of candidate directions, determining a first conversion matrix corresponding to each of the plurality of candidate directions according to the plurality of candidate directions and the second space main axis direction, performing registration on the image space point cloud and the surgical space point cloud according to the first conversion matrix corresponding to each of the plurality of candidate directions to obtain a second conversion matrix corresponding to each of the plurality of candidate directions, determining a coarse registration conversion matrix according to the first conversion matrix corresponding to each of the plurality of candidate directions and the second conversion matrix, and performing fine registration on the image space point cloud and the surgical space point cloud according to the coarse registration conversion matrix. The application realizes automatic registration of a surface and shortens the registration time.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of surgical navigation, and more particularly, to a registration method, a registration device, an electronic device and a computer readable storage medium in the field of surgical navigation. BACKGROUND

[0002] Space registration is the most critical technology of surgical navigation system, and space registration can be generally divided into point registration and markerless point registration. Point registration generally requires that marker points be specially pasted or placed on the surface of a navigation target (such as a patient, a human body model, etc.) before navigation, and then a surface scan is performed, which increases the radiation dose and the surgical cost of the navigation target. Compared with point registration, markerless point registration completes space registration by collecting point clouds of the surface of the navigation target, and therefore, markerless point registration is also called surface registration.

[0003] Surface registration is generally safer and more reliable, and does not cause additional trauma to the navigation target due to the placement of marker points, nor does it cause the interruption of surgical navigation due to the falling off of marker points. At present, most surface-based surgical navigation uses anatomical landmarks for coarse registration to provide a suitable initial pose for fine registration. However, manual selection of anatomical landmarks to obtain an initial pose increases the registration time and cannot achieve automatic surface registration. SUMMARY

[0004] The present application provides a registration method, a registration device, an electronic device and a computer readable storage medium, which can achieve automatic registration of the surface of a navigation target and shorten the registration time of the surface.

[0005] In a first aspect, a registration method is provided, which includes: determining a first space principal axis direction of an image space point cloud and a second space principal axis direction of a surgical space point cloud; correcting the first space principal axis direction to obtain a plurality of candidate directions; determining a first conversion matrix corresponding to each of the plurality of candidate directions according to the plurality of candidate directions and the second space principal axis direction; performing registration on the image space point cloud and the surgical space point cloud according to the first conversion matrix corresponding to each of the plurality of candidate directions to obtain a second conversion matrix corresponding to each of the plurality of candidate directions; determining a coarse registration conversion matrix according to the first conversion matrix corresponding to each of the plurality of candidate directions and the second conversion matrix; and performing fine registration on the image space point cloud and the surgical space point cloud according to the coarse registration conversion matrix.

[0006] With reference to the first aspect, in some possible implementation manners, the step of determining the first spatial principal axis direction of the image space point cloud and the second spatial principal axis direction of the surgery space point cloud comprises: downsampling the image space points to obtain a first simplified point cloud; downsampling the surgery space point cloud to obtain a second simplified point cloud; determining a first covariance matrix of the first simplified point cloud and a second covariance matrix of the second simplified point cloud; performing singular value decomposition on the first covariance matrix to obtain a first principal axis orthogonal matrix corresponding to the first simplified point cloud, and the first principal axis orthogonal matrix represents the first spatial principal axis direction; performing singular value decomposition on the second covariance matrix to obtain a second principal axis orthogonal matrix corresponding to the second simplified point cloud, and the second principal axis orthogonal matrix represents the second spatial principal axis direction.

[0007] With reference to the first aspect and the implementation manners, in some possible implementation manners, the step of correcting the first spatial principal axis direction to obtain a plurality of candidate directions comprises: determining a principal axis correction matrix in a manner of transforming the principal axis direction; and correcting the first spatial principal axis direction by using the principal axis correction matrix to obtain the plurality of candidate directions.

[0008] With reference to the first aspect and the implementation manners, in some possible implementation manners, the step of registering the image space point cloud and the surgery space point cloud according to the first conversion matrix corresponding to each of the plurality of candidate directions to obtain the second conversion matrix corresponding to each of the plurality of candidate directions comprises: transforming the first simplified point cloud by using the first conversion matrix corresponding to each of the plurality of candidate directions to obtain a plurality of groups of transformed point clouds; dividing each group of the transformed point clouds and the second simplified point cloud into a group to obtain a plurality of point cloud groups; and registering the transformed point cloud and the second simplified point cloud in each of the point cloud groups to obtain the second conversion matrix corresponding to each of the plurality of candidate directions.

[0009] With reference to the first aspect and the implementation manners, in some possible implementation manners, the step of determining the coarse registration conversion matrix according to the first conversion matrix and the second conversion matrix corresponding to each of the plurality of candidate directions comprises: determining a third conversion matrix corresponding to each of the plurality of candidate directions according to the first conversion matrix and the second conversion matrix corresponding to each of the plurality of candidate directions; determining a candidate direction that satisfies a principal axis determination condition from the plurality of candidate directions; and determining the third conversion matrix corresponding to the candidate direction that satisfies the principal axis determination condition as the coarse registration conversion matrix.

[0010] In some possible implementation manners, in combination with the first aspect and the foregoing implementation manners, the step of determining the candidate direction from the plurality of candidate directions that satisfies the principal axis determination condition comprises: obtaining a registration point pair corresponding to both the image space point cloud and the surgery space point cloud; determining a target distance corresponding to each of the plurality of candidate directions according to coordinates of two points in the registration point pair and a third conversion matrix corresponding to each of the plurality of candidate directions, wherein the target distance is an average Euclidean distance between the two points in the registration point pair; determining a minimum target distance from the target distances corresponding to the plurality of candidate directions; and determining the candidate direction corresponding to the minimum target distance from the plurality of candidate directions as the candidate direction that satisfies the principal axis determination condition.

[0011] In some possible implementation manners, in combination with the first aspect and the foregoing implementation manners, the step of obtaining the first outer surface point set of the navigation target in the surgery space; obtaining a second outer surface point set of a three-dimensional model of the navigation target in the image space; performing filtering processing on the first outer surface point set to obtain the surgery space point cloud; and performing down-sampling processing on the second outer surface point set to obtain the image space point cloud.

[0012] In a second aspect, a registration device is provided, which comprises:

[0013] A direction determination module is configured to determine a first spatial principal axis direction of the image space point cloud and a second spatial principal axis direction of the surgery space point cloud.

[0014] A direction correction module is configured to correct the first spatial principal axis direction to obtain a plurality of candidate directions.

[0015] A first calculation module is configured to determine a first conversion matrix corresponding to each of the plurality of candidate directions according to the plurality of candidate directions and the second spatial principal axis direction.

[0016] A second calculation module is configured to perform registration on the image space point cloud and the surgery space point cloud according to the first conversion matrix corresponding to each of the plurality of candidate directions to obtain a second conversion matrix corresponding to each of the plurality of candidate directions.

[0017] A third calculation module is configured to determine a coarse registration conversion matrix according to the first conversion matrix corresponding to each of the plurality of candidate directions and the second conversion matrix.

[0018] A point cloud registration module is configured to perform fine registration on the image space point cloud and the surgery space point cloud according to the coarse registration conversion matrix.

[0019] In some possible implementation manners, in combination with the second aspect, the direction determination module comprises:

[0020] a first down-sampling unit, configured to down-sample the image space points to obtain a first simplified point cloud;

[0021] a second down-sampling unit, configured to down-sample the surgery space point cloud to obtain a second simplified point cloud;

[0022] a covariance matrix calculation unit, configured to determine a first covariance matrix of the first simplified point cloud and a second covariance matrix of the second simplified point cloud;

[0023] a first decomposition unit, configured to perform singular value decomposition on the first covariance matrix to obtain a first principal axis orthogonal matrix corresponding to the first simplified point cloud, the first principal axis orthogonal matrix representing the first space principal axis direction;

[0024] a second decomposition unit, configured to perform singular value decomposition on the second covariance matrix to obtain a second principal axis orthogonal matrix corresponding to the second simplified point cloud, the second principal axis orthogonal matrix representing the second space principal axis direction.

[0025] With reference to the second aspect, in some possible implementation manners, the direction correction module includes:

[0026] a correction matrix construction unit, configured to determine a principal axis correction matrix by transforming the principal axis direction;

[0027] a principal axis direction correction unit, configured to correct the first space principal axis direction by using the principal axis correction matrix to obtain the multiple candidate directions.

[0028] With reference to the second aspect and the implementation manners, in some possible implementation manners, the second calculation module includes:

[0029] a point cloud conversion unit, configured to convert the first simplified point cloud by using a first conversion matrix corresponding to each of the multiple candidate directions to obtain multiple groups of converted point clouds;

[0030] a point cloud grouping unit, configured to group each of the converted point clouds and the second simplified point cloud to obtain multiple point cloud groups;

[0031] a first conversion matrix calculation unit, configured to register the converted point cloud and the second simplified point cloud in each of the point cloud groups to obtain a second conversion matrix corresponding to each of the multiple candidate directions.

[0032] With reference to the second aspect and the implementation manners, in some possible implementation manners, the third calculation module includes:

[0033] The second conversion matrix calculation unit is configured to determine a third conversion matrix corresponding to each of the plurality of candidate directions according to the first conversion matrix and the second conversion matrix corresponding to each of the plurality of candidate directions.

[0034] The direction selection unit is configured to determine a candidate direction satisfying the principal axis determination condition from the plurality of candidate directions.

[0035] The third conversion matrix calculation unit is configured to determine the third conversion matrix corresponding to the candidate direction satisfying the principal axis determination condition as the coarse registration conversion matrix.

[0036] With reference to the second aspect and the implementation manners of the second aspect, in some possible implementation manners, the direction selection unit includes:

[0037] The point pair acquisition sub-unit is configured to acquire a registration point pair corresponding to the image space point cloud and the surgical space point cloud.

[0038] The distance calculation sub-unit is configured to determine a target distance corresponding to each of the plurality of candidate directions according to the coordinates of the two points in the registration point pair and the third conversion matrix corresponding to each of the plurality of candidate directions, wherein the target distance is an average Euclidean distance between the two points in the registration point pair.

[0039] The distance selection sub-unit is configured to determine a minimum target distance in the target distances corresponding to the plurality of candidate directions.

[0040] The condition determination sub-unit is configured to determine a candidate direction corresponding to the minimum target distance in the plurality of candidate directions as the candidate direction satisfying the principal axis determination condition.

[0041] With reference to the second aspect and the implementation manners of the second aspect, in some possible implementation manners, the registration device further includes:

[0042] The first point set acquisition unit is configured to acquire a first outer surface point set of a navigation target in a surgical space.

[0043] The second point set acquisition unit is configured to acquire a second outer surface point set of a three-dimensional model of the navigation target in an image space.

[0044] The third downsampling unit is configured to perform filtering processing on the first outer surface point set to obtain the surgical space point cloud, and perform downsampling processing on the second outer surface point set to obtain the image space point cloud.

[0045] In a third aspect, an electronic device is provided, including a memory and a processor. The memory is configured to store executable program code, and the processor is configured to invoke and run the executable program code from the memory, so that the electronic device executes the method in the first aspect or any possible implementation manner of the first aspect.

[0046] In a fourth aspect, a computer program product is provided, including computer program code, which, when running on a computer, causes the computer to execute the method in the first aspect or any possible implementation manner of the first aspect.

[0047] In a fifth aspect, a computer-readable storage medium is provided, which stores computer program code, which, when running on a computer, causes the computer to execute the method in the first aspect or any possible implementation manner of the first aspect.

[0048] The registration method, registration device, electronic device and computer-readable storage medium provided by the embodiments of the present application have the following technical effects:

[0049] The embodiments of the present application determine the first space principal axis direction of the image space point cloud and the second space principal axis direction of the surgery space point cloud, correct the first space principal axis direction to obtain a plurality of candidate directions, determine the first conversion matrix corresponding to each of the plurality of candidate directions according to the plurality of candidate directions and the second space principal axis direction, perform registration on the image space point cloud and the surgery space point cloud according to the first conversion matrix corresponding to each of the plurality of candidate directions, obtain the second conversion matrix corresponding to each of the plurality of candidate directions, determine the coarse registration conversion matrix according to the first conversion matrix corresponding to each of the plurality of candidate directions and the second conversion matrix, and perform fine registration on the image space point cloud and the surgery space point cloud according to the coarse registration conversion matrix. Without surface labeling of the navigation target, the surface can be automatically registered. By performing coarse registration on the image space point cloud and the surgery space point cloud first and then performing fine registration, the registration time is shortened, the registration accuracy of fine registration is improved, the reliability of registration is enhanced, and the tedious manual registration process is avoided. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 A schematic flowchart of a registration method provided by the embodiments of the present application is shown;

[0051] Figure 2 A schematic diagram of a plurality of corrected first space principal axis directions is shown;

[0052] Figure 3-1 A surgery space point cloud used for coarse registration is shown;

[0053] Figure 3-2 An image space point cloud used in coarse registration is shown;

[0054] Figure 4-1 A surgical space point cloud used in fine registration is shown;

[0055] Figure 4-2 An image space point cloud used in fine registration is shown;

[0056] Figure 5 A flowchart of coarse registration is shown;

[0057] Figure 6 A flowchart of fine registration is shown;

[0058] Figure 7 is a structural schematic diagram of a registration device provided by an embodiment of the present application;

[0059] Figure 8 is a structural schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0060] The technical solutions in the present application will be described in detail below with reference to the drawings. In the description of the embodiments of the present application, unless otherwise specified, " / " represents the meaning of or, for example, A / B can represent A or B: "and / or" in the text only describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which means that there are three cases of A alone, A and B together, and B alone. In addition, in the description of the embodiments of the present application, "multiple" means two or more than two.

[0061] Hereinafter, the terms "first" and "second" are only used for description purposes, and cannot be understood as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with "first" and "second" can explicitly or implicitly include one or more of the features.

[0062] The surgical navigation system can be divided into two stages of preoperative and intraoperative. The image space is determined by three-dimensional reconstruction of the navigation target (such as patient, human body model, etc.) medical image in preoperative, and the position information of the reflective marker ball on the surgical tool is detected in real time by the CT device to determine the surgical space in intraoperative. The key step of the surgical navigation system is the registration of the surgical space to the image space, which directly affects the overall accuracy of navigation, and the purpose is to obtain the conversion relationship between the image space and the surgical space. The accuracy of space registration determines the positioning accuracy of the surgical navigation system to a great extent.

[0063] Spatial registration is the most critical technology of surgical navigation system, and spatial registration can be generally divided into point registration and markerless point registration. Point registration generally needs to specially paste or place marker points on the surface of the navigation target before surgical navigation, and then perform surface scanning, which increases the radiation dose and surgical cost of the navigation target. Compared with point registration, markerless point registration completes spatial registration by collecting point cloud of the surface of the navigation target, and therefore, markerless point registration is also called surface registration.

[0064] Surface registration is generally safer and more reliable, and does not cause additional trauma to the navigation target due to placement of marker points, and does not cause interruption of surgical navigation due to falling of marker points. At present, most of the surface-based surgical navigation uses anatomical landmarks for coarse registration to provide a suitable initial pose for fine registration. However, manual selection of anatomical landmarks to obtain an initial pose increases the registration time and cannot realize automatic surface registration.

[0065] Based on the problems in the above related technologies, the embodiment of the present application provides a registration method, a registration device, an electronic device and a computer readable storage medium.

[0066] The following is an embodiment of a registration method provided by the embodiment of the present application.

[0067] Figure 1 A schematic flowchart of a registration method provided by the embodiment of the present application is shown. As shown in the example, Figure 1 The registration method provided by the embodiment of the present application is applied to a surgical navigation system, and the space conversion involved in the process of implementing registration by the registration method provided by the embodiment of the present application can be conversion from a surgical space to an image space or conversion from an image space to a surgical space. The embodiment of the present application takes the conversion from an image space to a surgical space as an example, and the registration method includes the following solutions:

[0068] S110: Determine a first space principal axis direction of the image space point cloud and a second space principal axis direction of the surgical space point cloud.

[0069] The process of implementing registration by the registration method provided by the embodiment of the present application is as follows: first, coarse registration of point cloud data is performed, and then fine registration of point cloud data is performed. The coarse registration of point cloud data includes two steps, the first step is to perform point cloud data registration by using a principal component analysis registration algorithm (PCA), and the second step is to perform point cloud data registration by using a deep learning registration algorithm (FMR).

[0070] In an exemplary embodiment, the surgical space point cloud refers to a surface point set of a navigation target in a surgical space, a CT device (Computed Tomography) tracks a space in which a surgical tool collects the surface point set of the navigation target in real time, and the space is referred to as the surgical space. The surgical space can be represented by a surgical space coordinate system S n . The surgical tool has a reflective marker ball thereon. After the surgical tool contacts the surface of the navigation target and moves in the surgical space, the CT device collects the position of the reflective marker ball in real time, thereby obtaining real-time position information of the surgical tool, i.e., obtaining the surgical space point cloud of the navigation target. The image space point cloud refers to a surface point set of a three-dimensional model of the navigation target in an image space, and the space in which the surgical navigation system performs three-dimensional reconstruction on the navigation target is referred to as the image space. The image space can be represented by an image space coordinate system S ct . The navigation target can be a patient, a human model, or other objects.

[0071] After the surgical space point cloud and the image space point cloud of the navigation target are obtained, the surgical space point cloud is represented as Q, and the image space point cloud is represented as P.

[0072] The PCA algorithm is used to register P and Q. Based on the PCA registration method, the main axis direction of the point cloud data is mainly used to initially register P and Q. In the process of initially registering P and Q, the first space main axis direction of the image space point cloud and the second space main axis direction of the surgical space point cloud are calculated. The first space main axis direction is the direction of the main axis of the image space point cloud, and the second space main axis direction is the direction of the main axis of the surgical space point cloud. The main axis of the image space point cloud and the main axis of the surgical space point cloud each include a first main axis, a second main axis, and a third main axis. The first main axis corresponds to the X axis, the second main axis corresponds to the Y axis, and the third main axis corresponds to the Z axis. The main axes of P in S ct include a first main axis x p , a second main axis y p , and a third main axis z p . The main axes of Q in S n include a first main axis x q , a second main axis y q , and a third main axis z q . Wherein z p =x p ×y p , z q =x q ×z q .

[0073] In one possible implementation, the above S110 includes the following solutions:

[0074] The spatial points of the above image are downsampled to obtain the first simplified point cloud;

[0075] The surgical space point cloud was downsampled to obtain a second simplified point cloud;

[0076] Determine the first covariance matrix of the first simplified point cloud and the second covariance matrix of the second simplified point cloud;

[0077] Singular value decomposition is performed on the first covariance matrix to obtain the first principal axis orthogonal matrix corresponding to the first simplified point cloud. The first principal axis orthogonal matrix represents the direction of the first spatial principal axis.

[0078] Singular value decomposition is performed on the second covariance matrix to obtain the second principal axis orthogonal matrix corresponding to the second simplified point cloud. The second principal axis orthogonal matrix represents the direction of the second spatial principal axis.

[0079] In an exemplary embodiment, P is downsampled to obtain a first simplified point cloud, denoted as P1; Q is downsampled to obtain a second simplified point cloud, denoted as Q1. Figure 3-1 As shown, Figure 3-1 The surgical space point cloud used for coarse registration is shown. Figure 3-1 The point cloud in the image is Q1; for example... Figure 3-2 As shown, Figure 3-2 The image space point cloud used for coarse registration is shown. Figure 3-2 The point cloud in the image is P1.

[0080] After obtaining P1 and Q1, P1 and Q1 are used as inputs to the PCA algorithm. The PCA algorithm is used to solve for principal components in P1 and Q1, thereby achieving initial registration of P1 and Q1 and obtaining the first and second spatial principal axis directions. By downsampling P and Q, the number of point clouds used for coarse registration is reduced, shortening both the time for coarse registration and the time for subsequent fine registration.

[0081] The initial registration of P1 and Q1 using the PCA algorithm includes:

[0082] Calculate the first covariance matrix of P1 and the second covariance matrix of Q1. Based on the covariance matrices, calculate the principal eigencomponents, which represent the principal axis directions of the point cloud. The process for calculating the first covariance matrix of P1 and the second covariance matrix of Q1 is as follows:

[0083] Calculate the first center coordinates of P1 and the second center coordinates of Q1. Then, determine the first covariance matrix based on the first center coordinates and the second covariance matrix based on the second center coordinates.

[0084] M1 represents the number of point clouds of Q1, N1 represents the number of point clouds of P1, N1 < N, and M1 < M. The first center coordinates are represented as The second center coordinates are represented as The center coordinates are calculated by formula set (1) as follows:

[0085]

[0086] The covariance matrix is calculated by formula set (2) as follows:

[0087]

[0088] The first covariance matrix is represented as The second covariance matrix is represented as

[0089] After calculating the first covariance matrix of P1 and the second covariance matrix of Q1, singular value decomposition (SVD) is performed on the first covariance matrix to obtain the first principal axis orthogonal matrix corresponding to P1, and singular value decomposition is performed on the second covariance matrix to obtain the second principal axis orthogonal matrix corresponding to Q1.

[0090] The first principal axis orthogonal matrix and the second principal axis orthogonal matrix are calculated by formula set (3) as follows:

[0091]

[0092] wherein, the first principal axis orthogonal matrix is represented as the second principal axis orthogonal matrix is represented as D represents a diagonal matrix of m*n, the diagonal elements are singular values of C, V represents an orthogonal matrix of n*n, m=3, and n=3.

[0093] After obtaining the first principal axis orthogonal matrix corresponding to P1 and the second principal axis orthogonal matrix corresponding to Q1, the first space principal axis direction is represented by the first principal axis orthogonal matrix, and the second space principal axis direction is represented by the second principal axis orthogonal matrix, that is, the first space principal axis direction and the second space principal axis direction are calculated.

[0094] S120: correcting the first space principal axis direction to obtain a plurality of candidate directions.

[0095] During the initial registration of P1 and Q1 using the PCA algorithm, there may be a problem of principal axis reversal, meaning that the direction of the principal axis of P may also be reversed. After obtaining the first spatial principal axis direction of P and the second spatial principal axis direction of Q, this embodiment of the application sequentially transforms the first principal axis x of P. p Second principal axis y p 2 in total 2 There are four principal axis transformation relationships, and the first spatial principal axis direction of P is corrected based on these four relationships, resulting in four corrected first spatial principal axis directions. These four corrected first spatial principal axis directions are the aforementioned multiple candidate directions, and there are four candidate directions. For example... Figure 2 As shown, Figure 2 A schematic diagram of various corrected first spatial principal axis directions is shown. The four candidate directions are shown as (a), (b), (c) and (d), respectively. For example, the first candidate direction corresponds to (a), the second candidate direction corresponds to (b), the third candidate direction corresponds to (c), and the fourth candidate direction corresponds to (d).

[0096] In one possible implementation, S120 above includes the following scheme:

[0097] The principal axis correction matrix is ​​determined by changing the principal axis direction.

[0098] The aforementioned principal axis correction matrix is ​​used to correct the first spatial principal axis direction, resulting in the aforementioned multiple candidate directions.

[0099] In an exemplary embodiment, since the PCA algorithm may have a principal axis reversal problem after transforming P1 and Q1, it may not be able to adjust the point cloud to a suitable pose. Therefore, a principal axis correction matrix can be constructed to transform the first spatial principal axis direction of P to solve the principal axis reversal problem. Specifically, the first spatial principal axis direction of P transformed by the principal axis correction matrix is ​​the positive direction of the principal axis of P.

[0100] The process of constructing the principal axis correction matrix includes: determining the principal axis correction matrix by transforming the principal axis direction, that is, sequentially transforming the first principal axis x of P. p Second principal axis y p There are a total of 4 principal axis transformation relationships, among which the third principal axis z p =x p ×y p Thus, the principal axis correction matrix is ​​obtained.

[0101] The principal axis correction matrix is ​​represented by the set K. That is, K has 4 possible cases, including k1, k2, k3 and k4.

[0102] After the first spatial principal axis direction and the second spatial principal axis direction are calculated, the first spatial principal axis direction is calibrated by k1, k2, k3 and k4 respectively to obtain four kinds of candidate directions. Among them,

[0103] The first kind of candidate direction is represented as

[0104] The second kind of candidate direction is represented as

[0105] The third kind of candidate direction is represented as

[0106] The fourth kind of candidate direction is represented as

[0107] S130: According to the above-mentioned multiple candidate directions and the above-mentioned second spatial principal axis direction, the first conversion matrix corresponding to each of the above-mentioned multiple candidate directions is determined.

[0108] After obtaining the four kinds of candidate directions, the first conversion matrix corresponding to each of the four kinds of candidate directions is calculated according to the four kinds of candidate directions and the second spatial principal axis direction, which is represented as V11, V12, V13 and V14 respectively. For example, the first candidate direction corresponds to V11, the second candidate direction corresponds to V12, the third candidate direction corresponds to V13, and the fourth candidate direction corresponds to V14.

[0109] The first conversion matrix includes a rotation matrix R1 and a translation vector T1. After obtaining the four kinds of candidate directions, V11 corresponding to the first candidate direction, V12 corresponding to the second candidate direction, V13 corresponding to the third candidate direction, and V14 corresponding to the fourth candidate direction are calculated by formula set (4). Formula set (4) is as follows:

[0110]

[0111] V11 includes a rotation matrix R 11 and a translation vector T 11 , V12 includes a rotation matrix R 12 and a translation vector T 12 , V13 includes a rotation matrix R 13 and a translation vector T 13 , and V14 includes a rotation matrix R 14 and a translation vector T 14 . For example, the The calculation methods of V12, V13 and V14 are the same as that of V11, which will not be described here.

[0112] S140: registering the image space point cloud and the surgery space point cloud according to the first conversion matrix corresponding to each of the plurality of candidate directions, to obtain a second conversion matrix corresponding to each of the plurality of candidate directions.

[0113] According to the first conversion matrix corresponding to each of the four candidate directions, P and Q are registered, specifically, according to the first conversion matrix corresponding to each of the four candidate directions, P1 and Q1 are registered based on a deep learning registration algorithm (Feature-metric Registration, FMR) to obtain a second conversion matrix corresponding to each of the four candidate directions. Wherein the second conversion matrix corresponding to the first candidate direction is represented as V21, the second conversion matrix corresponding to the second candidate direction is represented as V22, the second conversion matrix corresponding to the third candidate direction is represented as V23, and the second conversion matrix corresponding to the fourth candidate direction is represented as V24.

[0114] In one possible implementation, the determination of S140 includes the following solutions:

[0115] The first simplified point cloud is converted by using the first conversion matrix corresponding to each of the plurality of candidate directions respectively, to obtain a plurality of groups of converted point clouds;

[0116] Each group of converted point clouds and the second simplified point cloud are divided into a group, to obtain a plurality of point cloud groups;

[0117] The converted point cloud and the second simplified point cloud in each point cloud group are registered respectively, to obtain a second conversion matrix corresponding to each of the plurality of candidate directions.

[0118] In an exemplary embodiment, P1 is converted by using the first conversion matrix corresponding to each of the four candidate directions respectively, to obtain four groups of converted point clouds, and the converted point cloud is represented as P2. Then, each group of P2 and Q1 is divided into a group, to obtain four point cloud groups, that is, the first group of P2 and Q1 is divided into a group, to obtain a point cloud group 1, the second group of P2 and Q1 is divided into a group, to obtain a point cloud group 2, the third group of P2 and Q1 is divided into a group, to obtain a point cloud group 3, and the fourth group of P2 and Q1 is divided into a group, to obtain a point cloud group 4.

[0119] P2 and Q1 in the point cloud group 1 are registered by using the FMR algorithm, to obtain the second conversion matrix V21 corresponding to the first candidate direction;

[0120] P2 and Q1 in the point cloud group 2 are registered by using the FMR algorithm, to obtain the second conversion matrix V22 corresponding to the second candidate direction;

[0121] P2 and Q1 in the point cloud group 3 are registered by using the FMR algorithm, to obtain the second conversion matrix V23 corresponding to the third candidate direction;

[0122] The P2 and the Q1 in the point cloud set 4 are registered by using the FMR algorithm to obtain a second conversion matrix V24 corresponding to the fourth candidate direction.

[0123] The process of coarsely registering the P2 and the Q1 by using the FMR algorithm is represented by formula (5):

[0124]

[0125] Wherein, is a feature metric error between the P2 and the Q1, F(P2) is a feature function corresponding to the P2, K is a feature dimension, and F(.) is a global feature of the point cloud, The second conversion matrix obtained by the FMR algorithm is fused by the R2 and the T2, that is, the second conversion matrix includes the rotation matrix R2 and the translation vector T2.

[0126] S150: Determine a coarse registration conversion matrix according to the first conversion matrix and the second conversion matrix corresponding to each of the plurality of candidate directions.

[0127] After obtaining the first conversion matrix and the second conversion matrix corresponding to each of the four candidate directions, a suitable candidate direction is selected as a target principal axis direction of the P from the four candidate directions, and then a coarse registration conversion matrix is calculated through the first conversion matrix and the second conversion matrix corresponding to the target principal axis direction. The coarse registration conversion matrix is also called a coarse registration application conversion matrix. The coarse registration conversion matrix is represented as Vs, assuming that the first conversion matrix and the second conversion matrix corresponding to the target principal axis direction are V11 and V21 respectively, and Vs = V11*V21.

[0128] In one possible implementation manner, the S150 includes the following scheme:

[0129] Determine a third conversion matrix corresponding to each of the plurality of candidate directions according to the first conversion matrix and the second conversion matrix corresponding to each of the plurality of candidate directions.

[0130] Determine a candidate direction satisfying a principal axis determination condition from the plurality of candidate directions.

[0131] Determine the third conversion matrix corresponding to the candidate direction satisfying the principal axis determination condition as the coarse registration conversion matrix.

[0132] Since Q and P are obtained by different devices, it is difficult to ensure that the obtained point clouds are completely consistent in distribution range, resulting in that the principal axis directions of the two space point clouds are not completely parallel. Therefore, after the point clouds are registered by using the FMR algorithm, the principal axis corresponding to P is selected from the four principal axes of P by using the pre-set principal axis determination condition, so as to obtain the target principal axis direction.

[0133] After the third conversion matrix corresponding to each of the four candidate directions is obtained, the third conversion matrix corresponding to each of the four candidate directions is calculated according to the first conversion matrix and the second conversion matrix corresponding to each of the four candidate directions. For example, the third conversion matrix corresponding to the first candidate direction is represented as V31, V31 = V11*V21; the third conversion matrix corresponding to the second candidate direction is represented as V32, V32 = V12*V22; the third conversion matrix corresponding to the third candidate direction is represented as V33, V33 = V13*V23; and the third conversion matrix corresponding to the fourth candidate direction is represented as V34, V34 = V14*V24.

[0134] The principal axis determination condition is used as the selection condition of the target principal axis direction, the candidate direction satisfying the principal axis determination condition is obtained from the four candidate directions, the candidate direction satisfying the principal axis determination condition is determined as the target principal axis direction, and then the third conversion matrix corresponding to the first conversion matrix and the second conversion matrix of the target principal axis direction is determined as the coarse registration conversion matrix. For example, the target principal axis direction is the second candidate direction, and the third conversion matrix corresponding to the second candidate direction is V32, that is, Vs = V32.

[0135] In a possible implementation, the above determining the candidate direction satisfying the principal axis determination condition from the above multiple candidate directions includes the following scheme:

[0136] The registration point pair corresponding to the image space point cloud and the surgical space point cloud is obtained.

[0137] According to the coordinates of the two points in the registration point pair and the third conversion matrix corresponding to each of the multiple candidate directions, a target distance corresponding to each of the multiple candidate directions is determined, wherein the target distance is the average Euclidean distance between the two points in the registration point pair.

[0138] The minimum target distance in the target distances corresponding to the multiple candidate directions is determined.

[0139] The candidate direction corresponding to the minimum target distance in the multiple candidate directions is determined as the candidate direction satisfying the principal axis determination condition.

[0140] Each registration point pair includes two mutually paired points in two point clouds, and the registration point pair is specifically obtained from P2 and Q1 in each point cloud group after registration is completed by the FMR algorithm, so as to obtain four groups of registration point pairs, that is, each group of registration point pairs corresponds to one of the four candidate directions. After obtaining the four groups of registration point pairs each corresponding to one of the four candidate directions, the coordinates of the two points in each registration point pair are obtained, and then based on the calculation formula of the average Euclidean distance, the target distance corresponding to each of the four candidate directions is calculated according to the coordinates of the two points in each registration point pair and the third conversion matrix corresponding to each of the four candidate directions. The target distance is the average Euclidean distance between the two points in the registration point pair. The calculation formula of the average Euclidean distance is shown in formula (6):

[0141]

[0142] wherein, is In S n the corresponding points, Q1 is obtained by tracking the surgical tool by the CT device, the acquisition range is random, the acquisition environment is not accurate enough, the shape of the navigation target formed by Q1 cannot be ensured to be similar enough to the shape of the three-dimensional model formed by P1, and S RE may not be able to correctly identify the case that the main shafts are aligned in the same direction. Further registration of the point cloud based on the FMR algorithm further shortens the spatial position difference and reduces the registration time.

[0143] In order to screen out the main shaft corresponding to P from the four main shafts of Q, which is correctly paired with the main shaft of Q, and obtain the target main shaft direction, the minimum average Euclidean distance between the two points in the registration point pair after registration is used as the main shaft determination condition, and the main shaft corresponding to P from the four main shafts of Q, which is correctly paired with the main shaft of Q, is screened out according to the minimum average Euclidean distance, and the target main shaft direction is obtained.

[0144] After the target distance corresponding to each of the four candidate directions is calculated, four target distances are obtained, the minimum target distance is determined from the four target distances, and then the main shaft pair corresponding to the minimum target distance in the four target distances is determined as the main shaft pair that satisfies the main shaft determination condition, so as to obtain the target main shaft direction.

[0145] For example, based on formula (6) and P2 and Q1 in point cloud group 1, the target distance corresponding to the first candidate direction calculated is S1 RE ; based on formula (6) and P2 and Q1 in point cloud group 2, the target distance corresponding to the second candidate direction calculated is S2 RE ; based on formula (6) and P2 and Q1 in point cloud group 3, the target distance corresponding to the third candidate direction calculated is S3 RE; based on formula (6) and P2 and Q1 in point cloud group 4, the target distance corresponding to the fourth candidate direction calculated is S4 RE .

[0146] S1 RE , S2 RE , S3 RE and S4 RE are calculated RE , and S4 RE is the smallest among the four target distances, then the fourth candidate direction is the candidate direction satisfying the main axis determination condition, that is, the fourth candidate direction is the target main axis direction,

[0147] S160: according to the coarse registration conversion matrix, fine registration is performed on the image space point cloud and the surgical space point cloud.

[0148] As shown in Figure 4-1 , Figure 4-1 the surgical space point cloud used for fine registration is shown, Figure 4-1 the point cloud in which is Q; as shown in Figure 4-2 , Figure 4-2 the image space point cloud used for fine registration is shown, Figure 4-2 the point cloud in which is P. After obtaining the coarse registration conversion matrix, P is transformed by using the coarse registration conversion matrix to obtain the conversion point cloud corresponding to P, which is denoted as P3.

[0149] After obtaining P3, P3 and Q are fine registered by using the ICP (Iterative Closest Point, point cloud registration algorithm) algorithm. The ICP algorithm constructs a rigid transformation matrix by searching for the point pair relationship between P3 and Q, and accurately aligns P3 and Q by iteratively minimizing the average difference between the corresponding points in the two spaces. The registration principle of the ICP algorithm is expressed by the following formula (7):

[0150]

[0151] wherein V4 represents the globally minimized conversion matrix of P3 and Q, q i ∈Q.

[0152] In one possible implementation, before S110, the registration method further includes the following scheme:

[0153] obtain a first outer surface point set of the navigation target in the surgical space;

[0154] obtain a second outer surface point set of the three-dimensional model of the navigation target in the image space;

[0155] The first outer surface point set is filtered to obtain the surgical space point cloud;

[0156] The second outer surface point set is down-sampled to obtain the image space point cloud.

[0157] The first outer surface point set is collected by a CT device tracking movement of a surgical tool on a navigation target surface, and the second outer surface point set is a point set of a three-dimensional model surface of the navigation target. The first outer surface point set is denoted as Q0, and the second outer surface point set is denoted as P0. The Q0 is filtered to obtain Q, and the P0 is down-sampled to obtain P, thereby reducing the number of point clouds for coarse registration and shortening the coarse registration time. The filtering process refers to denoising and removing outliers from the Q0.

[0158] According to the technical scheme, the first space principal axis direction of the image space point cloud and the second space principal axis direction of the surgical space point cloud are determined, the first space principal axis direction is corrected to obtain a plurality of candidate directions, the first conversion matrix corresponding to each of the plurality of candidate directions is determined according to the plurality of candidate directions and the second space principal axis direction, the image space point cloud and the surgical space point cloud are registered according to the first conversion matrix corresponding to each of the plurality of candidate directions, the second conversion matrix corresponding to each of the plurality of candidate directions is obtained, the coarse registration conversion matrix is determined according to the first conversion matrix corresponding to each of the plurality of candidate directions and the second conversion matrix, and the image space point cloud and the surgical space point cloud are fine-registered according to the coarse registration conversion matrix. According to the technical scheme, the surface of the navigation target does not need to be marked, and the automatic registration of the surface can be realized. The image space point cloud and the surgical space point cloud are coarsely registered first, and then fine-registered, which shortens the registration time, improves the registration accuracy of fine registration, enhances the reliability of registration, and avoids the cumbersome manual registration process.

[0159] In the technical scheme, first, the PCA algorithm is used to register the image space point cloud and the surgery space point, to obtain a first space main shaft direction of the image space point cloud and a second space main shaft direction of the surgery space point cloud, then the first space main shaft direction is corrected to obtain a plurality of to-be-selected directions, then according to the plurality of to-be-selected directions and the second space main shaft direction, a plurality of first conversion matrices corresponding to the plurality of to-be-selected directions are calculated, and then based on the FMR algorithm and according to the plurality of first conversion matrices, the surgery space point cloud and the image space point cloud are registered, so that the PCA algorithm and the FMR algorithm are utilized to realize fast registration, and the problem of long registration time of the ICP algorithm is improved. In addition, the main shaft correction matrix is constructed in the registration of the image space point cloud and the surgery space point by the PCA algorithm, which belongs to the improvement of the PCA algorithm, the main shaft correction matrix solves the problem of reverse main shaft, provides a reliable initial value for the ICP algorithm, and can complete the coarse registration without selecting the anatomical marker point by artificial selection, realizes the surface automatic registration without the marker point, does not cause trauma to the navigation target during the surface registration, and has the advantages of safety and reliability.

[0160] The following is another embodiment of a registration method provided by the embodiment.

[0161] As shown in Figure 5 , a flowchart of coarse registration is shown, as shown in Figure 5 , a flowchart of fine registration is shown. In combination with Figure 6 and Figure 6 , the registration method provided by the embodiment first performs coarse registration of point cloud data, and then performs fine registration of point cloud data. The coarse registration of point cloud data includes two steps, the first step is to register the point cloud data by using the PCA algorithm, and the second step is to register the point cloud data by using the FMR algorithm. The specific registration process is as follows: Figure 5 Figure 6 Obtain the original CT sequence of the three-dimensional model of the navigation target in the image space, extract the point cloud from the original CT sequence, obtain the second outer surface point set P0, downsample P0 to obtain the image space point cloud P, and downsample P to obtain the first simplified point cloud P1.

[0162] Obtain the first outer surface point set Q0 of the navigation target in the surgery space, filter Q0 to obtain the surgery space point cloud Q, and downsample Q to obtain the second simplified point cloud Q1. P1 and Q1 are to-be-registered space point clouds in

[0163] Obtain the first outer surface point set Q0 of the navigation target in the surgery space, filter Q0 to obtain the surgery space point cloud Q, and downsample Q to obtain the second simplified point cloud Q1. P1 and Q1 are to-be-registered space point clouds in Figure 5 .

[0164] ​After obtaining P1 and Q1, the PCA algorithm is used to perform initial registration on Q1 and P1. During the registration of P1 and Q1 by the PCA algorithm, the first spatial principal axis direction of P1 and the second spatial principal axis direction of Q1 are calculated, and then the first spatial principal axis direction is corrected by the constructed principal axis correction matrix to obtain four corrected first spatial principal axis directions, each of which is referred to as a candidate direction, that is, there are four candidate directions, and then a plurality of candidate directions each corresponding to a first conversion matrix is calculated according to the four candidate directions and the second spatial principal axis direction. After the initial registration of Q1 and P1 by the PCA algorithm is completed, the first conversion matrix corresponding to each of the four candidate directions is obtained.

[0165] After obtaining the first conversion matrix corresponding to each of the four candidate directions, the first conversion matrix corresponding to each of the four candidate directions is used to convert P1 respectively to obtain four groups of converted point clouds corresponding to P1, and the converted point cloud is represented as P2.

[0166] Then, the FMR algorithm is used to register P2 and Q1 in each point cloud group respectively to obtain a second conversion matrix corresponding to each candidate direction, thereby obtaining a second conversion matrix corresponding to each of the four candidate directions.

[0167] Next, the third conversion matrix corresponding to each of the four candidate directions is calculated through the first conversion matrix and the second conversion matrix corresponding to each of the four candidate directions, and the third conversion matrix is represented as V3. According to the above formula (5), the candidate direction satisfying the principal axis determination condition is selected from the four candidate directions through principal axis correction to obtain a target principal axis direction, and then a coarse registration conversion matrix is calculated according to the first conversion matrix and the second conversion matrix corresponding to the target principal axis direction.

[0168] After obtaining the coarse registration conversion matrix, P is converted by using the coarse registration conversion matrix to obtain a converted matrix corresponding to P, and the converted matrix corresponding to P is represented as P3. Finally, the ICP algorithm is used to perform fine registration on P3 and Q.

[0169] By using the PCA algorithm and the FMR algorithm to perform coarse registration on the point cloud data, the utilization of the fast registration characteristics of the PCA algorithm and the FMR algorithm is realized, and the problem of long registration time of the ICP algorithm can be improved. In addition, the problem of reverse principal axis is solved through principal axis correction, thereby providing a reliable initial value for the ICP algorithm, and the coarse registration can be completed without selecting anatomical marker points by artificial selection, realizing surface automatic registration without marker points, and the navigation target will not be damaged during surface registration, which has the advantages of safety and reliability.

[0170] The following is an embodiment of the device of the present disclosure, which can be used to perform the method embodiments of the present disclosure. For details not disclosed in the device embodiments of the present disclosure, please refer to the method embodiments of the present disclosure.

[0171] Figure 7 A structural schematic diagram of a registration device provided by an embodiment of the present application is shown. As shown in the example, Figure 7 The registration device 700 includes:

[0172] A direction determination module 710 is configured to determine a first spatial principal axis direction of an image space point cloud and a second spatial principal axis direction of a surgical space point cloud.

[0173] A direction correction module 720 is configured to correct the first spatial principal axis direction to obtain a plurality of candidate directions.

[0174] A first calculation module 730 is configured to determine a first conversion matrix corresponding to each of the plurality of candidate directions according to the plurality of candidate directions and the second spatial principal axis direction.

[0175] A second calculation module 740 is configured to perform registration on the image space point cloud and the surgical space point cloud according to the first conversion matrix corresponding to each of the plurality of candidate directions to obtain a second conversion matrix corresponding to each of the plurality of candidate directions.

[0176] A third calculation module 750 is configured to determine a coarse registration conversion matrix according to the first conversion matrix corresponding to each of the plurality of candidate directions and the second conversion matrix.

[0177] A point cloud registration module 760 is configured to perform fine registration on the image space point cloud and the surgical space point cloud according to the coarse registration conversion matrix.

[0178] In a possible implementation manner, the direction determination module 710 includes:

[0179] A first down-sampling unit is configured to down-sample the image space points to obtain a first simplified point cloud.

[0180] A second down-sampling unit is configured to down-sample the surgical space point cloud to obtain a second simplified point cloud.

[0181] A covariance matrix calculation unit is configured to determine a first covariance matrix of the first simplified point cloud and a second covariance matrix of the second simplified point cloud.

[0182] A first decomposition unit is configured to perform singular value decomposition on the first covariance matrix to obtain a first principal axis orthogonal matrix corresponding to the first simplified point cloud, and the first principal axis orthogonal matrix represents the first spatial principal axis direction.

[0183] A second decomposition unit is configured to perform singular value decomposition on the second covariance matrix to obtain a second principal axis orthogonal matrix corresponding to the second simplified point cloud, and the second principal axis orthogonal matrix represents a second spatial principal axis direction.

[0184] In a possible implementation, the direction correction module 720 includes:

[0185] A correction matrix construction unit is configured to determine a principal axis correction matrix by transforming the principal axis direction.

[0186] A principal axis direction correction unit is configured to correct the first spatial principal axis direction by using the principal axis correction matrix to obtain the plurality of candidate directions.

[0187] In a possible implementation, the second calculation module 740 includes:

[0188] A point cloud conversion unit is configured to convert the first simplified point cloud by using a first conversion matrix corresponding to each of the plurality of candidate directions to obtain a plurality of groups of converted point clouds.

[0189] A point cloud grouping unit is configured to group each of the converted point clouds and the second simplified point cloud to obtain a plurality of point cloud groups.

[0190] A first conversion matrix calculation unit is configured to register the converted point cloud and the second simplified point cloud in each of the point cloud groups to obtain a second conversion matrix corresponding to each of the plurality of candidate directions.

[0191] In a possible implementation, the third calculation module 750 includes:

[0192] A second conversion matrix calculation unit is configured to determine a third conversion matrix corresponding to each of the plurality of candidate directions according to the first conversion matrix and the second conversion matrix corresponding to each of the plurality of candidate directions.

[0193] A direction selection unit is configured to select a candidate direction that satisfies a principal axis determination condition from the plurality of candidate directions.

[0194] A third conversion matrix calculation unit is configured to determine the third conversion matrix corresponding to the candidate direction that satisfies the principal axis determination condition as the coarse registration conversion matrix.

[0195] In a possible implementation, the direction selection unit includes:

[0196] A point pair acquisition subunit is configured to acquire a registration point pair corresponding to the image space point cloud and the surgical space point cloud.

[0197] The distance calculation subunit is configured to determine a target distance corresponding to each of the plurality of candidate directions according to the coordinates of the two points in the registration point pair and the third conversion matrix corresponding to each of the plurality of candidate directions, wherein the target distance is an average Euclidean distance between the two points in the registration point pair.

[0198] The distance selection subunit is configured to determine a minimum target distance in the target distances corresponding to the plurality of candidate directions.

[0199] The condition determination subunit is configured to determine the candidate direction corresponding to the minimum target distance as the candidate direction satisfying the main shaft determination condition.

[0200] In a possible implementation, the registration device 700 further includes:

[0201] The first point set acquisition unit is configured to acquire a first external surface point set of a navigation target in a surgical space.

[0202] The second point set acquisition unit is configured to acquire a second external surface point set of a three-dimensional model of the navigation target in an image space.

[0203] The third downsampling unit is configured to perform filtering processing on the first external surface point set to obtain the surgical space point cloud, and perform downsampling processing on the second external surface point set to obtain the image space point cloud.

[0204] It should be noted that the registration device provided in the embodiments described above is used to execute the registration method, and the division of the functional modules is only used as an example, and in actual applications, the functions can be distributed to different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the registration device and the registration method provided in the embodiments described above belong to the same concept, and therefore, for details not disclosed in the device embodiments, please refer to the registration method embodiments described above, which will not be described here.

[0205] The serial numbers of the embodiments of the present disclosure are only for description, and do not represent the advantages or disadvantages of the embodiments.

[0206] Figure 8 A structural schematic diagram of an electronic device provided by an embodiment of the present application is shown.

[0207] As shown in the example, Figure 8 The electronic device 800 includes a memory 801 and a processor 802, wherein the memory 801 stores executable program code 8011, and the processor 802 is configured to call and execute the executable program code 8011 to execute a registration method.

[0208] The embodiment can divide the function modules of the electronic device according to the method examples described above, for example, can correspond to each function module, or can integrate two or more functions in one processing module, and the integrated module can be implemented in the form of hardware. It should be noted that the division of the modules in the embodiment is illustrative, and is only a logical function division. In actual implementation, another division manner can be used.

[0209] In the case of dividing each function module corresponding to each function, the electronic device can include a direction determination module, a direction correction module, a first calculation module, a second calculation module, a third calculation module, a point cloud registration module, and the like. It should be noted that all related contents of each step involved in the method embodiments described above can be cited to the function description of the corresponding function module, and will not be described here.

[0210] The electronic device provided by the embodiment is used to execute the registration method described above, and therefore can achieve the same effect as the implementation method described above.

[0211] In the case of using an integrated unit, the electronic device can include a processing module and a storage module. The processing module can be used to control and manage the actions of the electronic device. The storage module can be used to support the electronic device to execute program codes and data.

[0212] The processing module can be a processor or a controller, which can implement or execute various exemplary logical blocks, modules and circuits described in combination with the disclosure of the present application. The processor can also be a combination of computing functions, such as one or more microprocessor combinations, digital signal processing (DSP) and microprocessor combinations, and the like. The storage module can be a memory.

[0213] The embodiment also provides a computer readable storage medium, which stores computer program codes. When the computer program codes run on a computer, the computer executes the related method steps described above to implement one of the registration methods in the embodiments described above.

[0214] The embodiment also provides a computer program product, which makes the computer execute the related steps described above to implement one of the registration methods in the embodiments described above when the computer program product runs on the computer.

[0215] In addition, the electronic device provided by the embodiment of the present application can be a chip, an assembly or a module. The electronic device can include a connected processor and a memory. The memory is used to store instructions. When the electronic device runs, the processor can call and execute the instructions to make the chip execute one of the registration methods in the embodiments described above.

[0216] Among them, the electronic device, computer readable storage medium, computer program product or chip provided by the embodiment are used for executing the corresponding method provided above, so the beneficial effects achieved thereby can refer to the beneficial effects in the corresponding method provided above, which will not be repeated here.

[0217] Through the description of the above embodiments, those skilled in the art can understand that, for the convenience and brevity, only the above division of functional modules is taken as an example, and in actual application, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above.

[0218] In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented in other ways. For example, the apparatus embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another device, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or other forms.

[0219] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed in the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method of registration, characterized by, The registration method comprises: determining a first spatial principal axis direction of an image space point cloud and a second spatial principal axis direction of a surgery space point cloud; correcting the first spatial principal axis direction to obtain a plurality of candidate directions; determining a first conversion matrix corresponding to each of the plurality of candidate directions according to the plurality of candidate directions and the second spatial principal axis direction; registering the image space point cloud and the surgery space point cloud according to the first conversion matrix corresponding to each of the plurality of candidate directions to obtain a second conversion matrix corresponding to each of the plurality of candidate directions; determining a third conversion matrix corresponding to each of the plurality of candidate directions according to the first conversion matrix and the second conversion matrix corresponding to each of the plurality of candidate directions; obtaining a registration point pair corresponding to the image space point cloud and the surgery space point cloud; determining a target distance corresponding to each of the plurality of candidate directions according to the coordinates of the two points in the registration point pair and the third conversion matrix corresponding to each of the plurality of candidate directions, wherein the target distance is an average Euclidean distance between the two points in the registration point pair; determining a minimum target distance among the target distances corresponding to each of the plurality of candidate directions; determining a candidate direction corresponding to the minimum target distance as a candidate direction satisfying a principal axis determination condition among the plurality of candidate directions; determining a third conversion matrix corresponding to the candidate direction satisfying the principal axis determination condition as a coarse registration conversion matrix; performing fine registration on the image space point cloud and the surgery space point cloud according to the coarse registration conversion matrix.

2. The registration method of claim 1, wherein, The step of determining a first spatial principal axis direction of an image space point cloud and a second spatial principal axis direction of a surgery space point cloud comprises: down-sampling the image space points to obtain a first simplified point cloud; down-sampling the surgery space point cloud to obtain a second simplified point cloud; determining a first covariance matrix of the first simplified point cloud and a second covariance matrix of the second simplified point cloud; performing singular value decomposition on the first covariance matrix to obtain a first principal axis orthogonal matrix corresponding to the first simplified point cloud, wherein the first principal axis orthogonal matrix represents the first spatial principal axis direction; performing singular value decomposition on the second covariance matrix to obtain a second principal axis orthogonal matrix corresponding to the second simplified point cloud, wherein the second principal axis orthogonal matrix represents the second spatial principal axis direction.

3. The method of registration of claim 2, wherein, The step of correcting the first spatial principal axis direction to obtain a plurality of candidate directions comprises: determining a principal axis correction matrix by transforming the principal axis direction; correcting the first spatial principal axis direction by using the principal axis correction matrix to obtain the plurality of candidate directions.

4. The registration method of claim 2, wherein, The step of registering the image space point cloud and the surgery space point cloud according to the first conversion matrix corresponding to each of the plurality of candidate directions to obtain a second conversion matrix corresponding to each of the plurality of candidate directions comprises: transforming the first simplified point cloud by using the first conversion matrix corresponding to each of the plurality of candidate directions to obtain a plurality of groups of transformed point clouds; dividing each group of the transformed point clouds and the second simplified point cloud into a group to obtain a plurality of point cloud groups; Respectively register the converted point cloud and the second simplified point cloud in each of the point cloud groups to obtain a second conversion matrix corresponding to each of the multiple candidate directions.

5. The registration method of any one of claims 1 or 2, wherein, Before the step of determining the target principal axis direction according to the image space point cloud and the surgery space point cloud, the registration method further comprises: obtaining a first outer surface point set of a navigation target in a surgery space; obtaining a second outer surface point set of a three-dimensional model of the navigation target in an image space; performing filtering processing on the first outer surface point set to obtain the surgery space point cloud; performing down-sampling processing on the second outer surface point set to obtain the image space point cloud.

6. A registration device characterized by The registration device comprises: a direction determination module configured to determine a first space principal axis direction of an image space point cloud and a second space principal axis direction of a surgery space point cloud; a direction correction module configured to correct the first space principal axis direction to obtain multiple candidate directions; a first calculation module configured to determine a first conversion matrix corresponding to each of the multiple candidate directions according to the multiple candidate directions and the second space principal axis direction; a second calculation module configured to register the image space point cloud and the surgery space point cloud according to the first conversion matrix corresponding to each of the multiple candidate directions to obtain a second conversion matrix corresponding to each of the multiple candidate directions; a third calculation module configured to determine a third conversion matrix corresponding to each of the multiple candidate directions according to the first conversion matrix corresponding to each of the multiple candidate directions and the second conversion matrix, obtain a registration point pair corresponding to both the image space point cloud and the surgery space point cloud, determine a target distance corresponding to each of the multiple candidate directions according to coordinates of two points in the registration point pair and the third conversion matrix corresponding to each of the multiple candidate directions, wherein the target distance is an average Euclidean distance between the two points in the registration point pair, determine a minimum target distance in the target distances corresponding to each of the multiple candidate directions, determine a candidate direction corresponding to the minimum target distance as a candidate direction satisfying a principal axis determination condition, and determine a third conversion matrix corresponding to the candidate direction satisfying the principal axis determination condition as a coarse registration conversion matrix; a point cloud registration module configured to perform fine registration on the image space point cloud and the surgery space point cloud according to the coarse registration conversion matrix.

7. An electronic device, comprising: The electronic device comprises: a memory configured to store executable program codes; a processor configured to call and run the executable program codes from the memory, so that the electronic device performs the method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, when the computer program is executed, the method according to any one of claims 1 to 5 is realized.

Images