CN102999902B - Optical navigation positioning navigation method based on CT registration result - Google Patents

Abstract

An optical navigation positioning system and navigation method based on CT registration results in the field of information processing technology, the system includes: preoperative CT image import module, image segmentation module, body surface initial registration module, preoperative CT image and surgical Two-dimensional ultrasound image module and intraoperative navigation module; the present invention uses a combination of virtual reality and intraoperative ultrasound to compensate for intraoperative positioning errors caused by factors such as breathing, thereby achieving accurate target points for coronary artery bypass grafting. Locate navigation. By manually segmenting and reconstructing the heart and coronary vascular tree in preoperative cardiac CT image data, and then building a virtual endoscopic and virtual endoscopic model with the help of optical navigator and intraoperative registration error correction based on CT and ultrasound. The augmented virtual reality environment integrated with the speculum can realize the precise positioning and navigation of the target point of coronary artery bypass grafting.

Description

Optical navigation positioning and navigation method based on CT registration results

technical field

The present invention relates to a system and method in the technical field of information processing, in particular to a navigation and positioning system and navigation method based on intraoperative ultrasound and preoperative CT registration results for assisting coronary artery bypass surgery.

Background technique

In recent years, the incidence of cardiovascular disease in my country has been increasing year by year, among which coronary heart disease is the most common cardiovascular disease. Among the treatment methods for coronary heart disease, coronary artery bypass grafting is currently the most important and mature treatment method in addition to drug therapy and interventional therapy. However, traditional coronary artery bypass surgery requires a median sternotomy, and if necessary, extracorporeal circulation and other operations are required to complete the operation. It includes disadvantages such as large incision, slow recovery, and many complications. As a new treatment method, navigation-assisted minimally invasive coronary artery bypass surgery only needs to make a few finger-thick incisions on the chest wall and use special surgical instruments to complete the operation, achieving beautiful incisions, small trauma, fast recovery, and less complications. Fewer symptoms and other requirements.

One of the main surgical difficulties in navigation-assisted minimally invasive coronary artery bypass grafting is how to quickly and accurately locate the target point. Improper or wrong positioning can directly affect the success or failure of the operation and the long-term efficacy. At present, the positioning of minimally invasive coronary artery bypass surgery mainly relies on preoperative images, and fails to use real-time information that reflects the real situation of the surgical area, and cannot solve the error problem caused by factors such as breathing, body surface marker displacement, and body position changes during the operation. As a result, the positioning effect is not ideal.

After searching the prior art, it was found that "3D-image guidance for minimally invasive robotic coronary artery bypass, Heart Surg Forum 2000-9732(3)" (three-dimensional image guidance robot-assisted minimally invasive coronary artery bypass surgery, Heart Surg Forum, 2000 TM.Peter first tried to use 3D images to study the preoperative planning of cardiac surgery. He obtained the surface model of the heart and skeleton by segmenting the CT images collected before the operation, and preliminarily established a minimally invasive Surgical navigation system for bypass surgery. At the same time, the system provides a virtual endoscope, using preoperative surgical registration technology, so that the relative positional relationship between the real endoscope and the model in the model test and the relationship between the virtual endoscope and the heart skeleton model modeled in the system The relative positional relationship is the same. But this system is only a prototype system, and there are many deficiencies, such as the real endoscopic 2D image is not effectively fused with the virtual 3D scene and so on.

"Flexible calibration of actuated stereoscopic endoscope for overlay in robot assisted surgery, MICCAI 2002 (1): LNCS 2488, T. Dohi and R. Kikinis (eds), 25-34." (Flexible calibration of stereoscopic endoscope in robot assisted surgery , 2002 Medical Imaging Computing and Computer-Aided Intervention Conference), Mourguess and Coste-Maniere tried to combine the endoscopic image of an animal's heart with the virtual endoscope of the three-dimensional model of the coronary tree of the heart with a certain transparency. Overlapping and fusion play a certain role in locating target points during surgery. However, the model of the coronary tree is static during the operation, and the fusion of the three-dimensional model of the coronary tree and the endoscopic image relies on identifying the marked points observed in the endoscope to make the two more accurate fusion. The robustness and accuracy are not very satisfactory.

Chinese Patent Document No. CN1650813 records the "robot surgical positioning method based on optical positioning for surgical navigation system". In this method, three marking positions are firstly designed on the robot base, and the three markings are selected by positioning the pointer with an optical tracker. When the robot probe and the optical tracker positioning pointer are docked, the same coordinate system in space is collected at the same time, and finally the coordinates of the optical tracker and the robot system are exchanged. This technology is mainly used in brain surgery. Due to the small anatomical deformation of brain surgery, the accuracy is easier to grasp; however, this technology cannot be applied to the field of cardiovascular system with large anatomical deformation.

Chinese Patent Document No. CN101703409A describes "a system and method for ultrasound-guided robot-assisted therapy". The system includes a surgical robot, a two-dimensional ultrasound instrument, a magnetic locator, and a workstation. Effective detection and automatic treatment of the region, thereby reducing the labor intensity of doctors and improving the accuracy of surgery. However, what is used in this method is a magnetic locator, and the magnetic locator is subject to various interferences such as the field, equipment, and other instruments when it is used, and errors are prone to occur. At the same time, 3D reconstructed ultrasound information will cause image distortion, and using the image information to guide surgery may cause deviations, and the amount of information generated is limited, so it cannot be used for some more delicate and complex operations, such as the precise operation of cardiac surgery.

It can be seen that although the robot-assisted treatment system has begun to be used in clinical treatment, its development in surgical target positioning is not perfect, and many clinical problems still need to be solved.

Contents of the invention

Aiming at the above-mentioned deficiencies existing in the prior art, the present invention proposes an optical navigation and positioning system and its navigation method based on CT registration results, and uses a combination of virtual reality and intraoperative ultrasound to compensate for surgical defects caused by factors such as breathing. Positioning error, so as to realize the precise positioning and navigation of the target point of coronary artery bypass grafting. By manually segmenting and reconstructing the heart and coronary vascular tree in the preoperative CT image data, and then building an internal model with the help of an optical navigator and intraoperative registration error correction based on the preoperative CT image and the two-dimensional ultrasound image The augmented virtual reality environment combined with endoscope and virtual endoscope can realize the precise positioning and navigation of the target point of coronary artery bypass grafting.

The present invention is achieved through the following technical solutions:

The invention relates to an optical navigation and positioning system based on CT registration results, including: a preoperative CT image import module, an image segmentation module, a body surface initial registration module, a preoperative CT image and an intraoperative two-dimensional ultrasonic image registration module and the intraoperative navigation module, wherein: the preoperative CT image import module receives the DICOM format image files obtained by the preoperative imaging examination, generates a preoperative image package and outputs it to the image segmentation module and the body surface initial registration module respectively, and the preoperative CT image The image and intraoperative 2D ultrasound image registration module are respectively connected to the image segmentation module and body surface initial registration module and receive the 3D dynamic coronary tree with surgical target points and the initial body surface registration results. The intraoperative two-dimensional ultrasound image registration module outputs the preoperative CT transformation matrix to the intraoperative navigation module, and outputs accurate surgical navigation information through the intraoperative navigation module.

The preoperative image package includes: several preoperative CT images in DIOM format for one or several cardiac cycles in the target area and three-dimensional images reconstructed from DICOM data, wherein: the preoperative CT image data of one cardiac cycle is composed of The preoperative images of several phases are composed.

The three-dimensional dynamic coronary tree with operation target points includes: a dynamic heart model and a blood vessel tree model generated by an image segmentation module.

The preoperative image import module includes: a DICOM image read-in unit and a preoperative image three-dimensional rendering unit, wherein: the DICOM image read-in unit imports a series of DICOM format image files of one cardiac cycle of the heart obtained by the preoperative imaging examination into The DICOM data is analyzed and transmitted to the preoperative image 3D rendering unit, which reconstructs a 3D stereoscopic image based on the DICOM data and merges it into a preoperative image package, which is output to the image segmentation module and body surface initial registration module respectively.

The image segmentation module includes: a heart segmentation unit and a vascular tree segmentation unit, wherein: the heart segmentation unit manually outlines the heart outline frame by frame based on clinical experience on the cardiac CT images in a cardiac cycle in the preoperative image package, and obtains Based on the clinical experience, the vascular tree segmentation unit manually outlines the vascular tree frame by frame based on the vascular tree CT images within one cardiac cycle in the preoperative image package and manually marks the surgery according to clinical needs. Target points, so as to reconstruct the vascular tree model according to the obtained vascular tree CT image segmentation results and output it to the preoperative CT image and intraoperative 2D ultrasound image together with the dynamic heart model as a three-dimensional dynamic coronary tree with surgical target points Registration module.

The body surface initial registration module includes: a registration marker selection unit and a conversion matrix calculation unit, wherein: the registration marker selection unit selects the registration markers set on the body surface of the surgical object, and obtains its preoperative The coordinates of the CT image data in the image coordinate system and the space coordinates of the real surgical object, the transformation matrix calculation unit uses the rigid body registration algorithm to realize the registration between the two coordinates according to the coordinates of the image coordinate system and the space coordinates, that is, to realize the registration between the image space and the space coordinates. The registration of the surgical object space, and use these two sets of coordinates to calculate the initial transformation matrix as the initial registration result of the body surface, and output it to the preoperative CT image and intraoperative two-dimensional ultrasound registration module.

The preoperative CT image and intraoperative two-dimensional ultrasonic image registration module includes: an intraoperative two-dimensional ultrasonic image read-in unit, a two-dimensional ultrasonic image and preoperative CT image registration unit, and an ECG (electrocardiogram, electrocardiogram) read-in unit. unit, wherein: the two-dimensional ultrasonic image and preoperative CT image registration unit receives the two-dimensional ultrasonic input image acquired by the intraoperative two-dimensional ultrasonic image read-in unit and the current heart beating phase acquired by the electrocardiogram read-in unit, through the input electrocardiogram signal The corresponding several phases in the three-dimensional dynamic coronary tree with surgical target points are matched with the current beating phase of the heart, and the conversion matrix between the two is calculated and output to the intraoperative navigation module, where each conversion matrix corresponds to the cardiac cycle A phased heart in , and the 3D coronary tree with surgical target points.

The intraoperative navigation module includes: a transformation matrix selection unit and a navigation display unit, wherein: the transformation matrix selection unit selects a corresponding transformation matrix according to the current heart beating phase, and outputs the obtained transformation matrix to the navigation display The navigation and display unit calculates the distance and relative positional relationship between the current virtual instrument and the actual target point and displays it on video, so as to guide the operation to be completed accurately.

The present invention relates to the navigation method of above-mentioned system, comprises the following steps:

In the first step, after the heart segmentation unit in the image segmentation module obtains a series of CT image data (DICOM format) of one cardiac cycle in the target area through the preoperative CT image import module, the heart CT images in the cardiac cycle are gradually analyzed based on clinical experience. The frame of the heart is manually outlined, the segmentation result is obtained, and the dynamic heart model is reconstructed according to the segmentation result.

In the second step, the vascular tree segmentation unit in the image segmentation module manually outlines the outline of the vascular tree frame by frame based on clinical experience on the CT images of the vascular tree within one cardiac cycle in the preoperative image package, and manually marks the surgical target points according to clinical needs. A segmentation result is obtained, and a vascular tree model is reconstructed according to the segmentation result, and a three-dimensional virtual scene including a dynamic vascular tree model is constructed, and the three-dimensional virtual scene constitutes a virtual endoscopic image.

The third step, the body surface initial registration module uses the body surface registration to obtain the conversion matrix of the preoperative CT image coordinate system and the real-time two-dimensional ultrasound image coordinate system of the surgical object, that is, the coordinate points in two different spatial coordinate systems, through The features are mapped to each other and realize the one-to-one correspondence, so as to achieve the correspondence between the final preoperative CT image and the real-time two-dimensional ultrasound image of the surgical object; specifically, by selecting several registration marker points in the preoperative CT image, find the corresponding image in the real space. Register the points corresponding to the marked points in the center and use the optical navigator to obtain the coordinates of these points in the real space. Using these two sets of different coordinate systems, but the position coordinates of the point sets that correspond one-to-one, obtain the conversion between the two spaces Matrix T.

The fourth step is to match several phases in the three-dimensional dynamic coronary tree with the surgical target point with the current heart beating phase by inputting the electrocardiogram signal, that is, the above-mentioned transformation matrix T is used as the initial transformation matrix, and Ti (i= 1, 2, ... N) is the transformation matrix corrected for each phase of the CT image in the preoperative image package on the basis of the transformation matrix T, that is, for one of the phases of the CT image i, Ti×T can further Correct the registration error between the phase CT image and the real-time two-dimensional ultrasound image of the surgical object, where N is the phase number of the preoperative CT image in the preoperative CT image data of one cardiac cycle, so as to obtain a set of corrections for T The transformation matrix; the specific steps include:

4.1) Acquisition of a series of real-time images of the surgical object to the heart through the ultrasonic probe, wherein each of the collected real-time images of the surgical object corresponds to a corresponding preoperative CT image, and each phase of the preoperative CT image corresponds to A series of two-dimensional ultrasound images were obtained, and for phase i preoperative CT image data, the surface contour of the inner wall of the heart was extracted;

4.2) Extract the contour of the inner wall of the heart for each two-dimensional ultrasound image corresponding to phase i, and the inner wall extracted from this series of two-dimensional ultrasound images forms a set of points, and extracts the feature points from the point set set, the surface contour of the inner wall of the heart extracted from the preoperative CT image is another set of points;

4.3) Register the point set on the two-dimensional ultrasound image to the point set on the preoperative CT image through the iterative closest point algorithm (ICP algorithm), and obtain the transformation matrix Ti between the two point sets for subsequent fine matching The starting matrix for quasi-processing.

In the fifth step, the real endoscopic image is fused with the virtual endoscopic image through external optical positioning.

The in vitro optical positioning refers to the use of NDI optical positioning instrument for infrared reflection positioning, so as to realize the three-dimensional positioning of the research object.

The fusion refers to: by fusing the two scenes with different transparency, an enhanced virtual reality environment is formed, so that the virtual scene containing the beating heart model seen by the virtual endoscope is the same as that seen by the real endoscope. At the same time, from the set of transformation matrices obtained in step 4.3, select the corresponding transformation matrix Ti according to the current heartbeat phase, so as to establish the mapping transformation relationship between the image space and the real space, and finally use the mapping transformation relationship for Real-time display of the distance and relative positional relationship between the instrument and the target point, so as to guide the accurate implementation of the operation.

technical effect

The advantages of the present invention include: 1. It is the first to introduce intraoperative ultrasound in robot-assisted coronary artery bypass surgery navigation, and correct errors caused by factors such as breathing through the registration of intraoperative two-dimensional ultrasound images and preoperative CT images; 2. It solves the uncertainty problem of relying on personal experience to locate the target point of the bridge in clinical practice.

Description of drawings

Fig. 1 is a schematic diagram of the module structure of the present invention.

Fig. 2 is a plurality of three-dimensional schematic diagrams of the heart (including the corresponding positions of the coronary tree and the electrocardiogram) of the embodiment.

Fig. 3 is a schematic diagram of an optical navigator and a registration module.

Fig. 4 is a schematic diagram for introducing the flow of the embodiment.

Fig. 5 is the fusion schematic diagram of embodiment;

In the figure: a is the real endoscopic image; b is the virtual endoscopic image; c is the overlapping image of the two.

detailed description

The embodiments of the present invention are described in detail below. This embodiment is implemented on the premise of the technical solution of the present invention, and detailed implementation methods and specific operating procedures are provided, but the protection scope of the present invention is not limited to the following implementation example.

Example 1

As shown in Figure 1, this embodiment includes: including: preoperative CT image import module, image segmentation module, body surface initial registration module, preoperative CT image and intraoperative two-dimensional ultrasound image registration module and intraoperative navigation module , wherein: the preoperative CT image import module receives the DICOM format image files obtained by the preoperative imaging examination, generates a preoperative image package and outputs it to the image segmentation module and the body surface initial registration module respectively, and the preoperative CT image and the intraoperative two The three-dimensional ultrasound image registration module is connected with the image segmentation module and the body surface initial registration module respectively, and receives the three-dimensional dynamic coronary tree with the surgical target point and the initial registration result of the body surface. The preoperative CT image and the intraoperative two-dimensional ultrasound The image registration module outputs the preoperative CT transformation matrix to the intraoperative navigation module, and the precise surgical navigation information is output through the intraoperative navigation module.

The preoperative image package includes: several preoperative CT images in DIOM format of one or several cardiac cycles in the target area and three-dimensional images reconstructed according to DICOM data.

The three-dimensional dynamic coronary tree with operation target points includes: a dynamic heart model and a blood vessel tree model generated by an image segmentation module.

The preoperative CT image importing module includes: a DICOM image reading unit and a preoperative image three-dimensional rendering unit, wherein: the DICOM image reading unit is a series of DICOM format image files of one cardiac cycle of the heart obtained by the preoperative imaging examination Import and analyze the DICOM data and transmit it to the preoperative image 3D rendering unit. The preoperative image 3D rendering unit reconstructs a 3D stereoscopic image based on the DICOM data and merges it into a preoperative image package and outputs it to the image segmentation module and body surface initial registration module respectively. .

The image segmentation module includes: a heart segmentation unit and a vascular tree segmentation unit, wherein: the heart segmentation unit manually outlines the heart outline frame by frame based on clinical experience on the cardiac CT images in a cardiac cycle in the preoperative image package, and obtains Based on the clinical experience, the vascular tree segmentation unit manually outlines the vascular tree frame by frame based on the vascular tree CT images within one cardiac cycle in the preoperative image package and manually marks the surgery according to clinical needs. Target points, so as to reconstruct the vascular tree model according to the obtained vascular tree CT image segmentation results and output it to the preoperative CT image and Intraoperative 2D ultrasound image registration module.

The body surface initial registration module includes: a registration marker selection unit and a transformation matrix calculation unit, wherein: the registration marker selection unit selects the registration markers set on the body surface of the surgical object, and obtains its CT The image data is in the coordinates of the image coordinate system and the space coordinates, and the conversion matrix calculation unit uses the rigid body registration algorithm to realize the registration between the two coordinates according to the coordinates of the image coordinate system and the space coordinates, that is, to realize the registration between the image space and the surgical object space. The two sets of coordinates are used to calculate the initial transformation matrix as the initial registration result of the body surface, and output to the preoperative CT image and intraoperative two-dimensional ultrasound image registration module.

The registration marker points refer to: when acquiring CT image data before operation, paste 6-8 metal marker points on the body surface of the surgical object, and the metal marker points are evenly distributed on the thoracic body surface of the surgical object, Metal markers are highlighted in the preoperative CT image data.

The space coordinates mentioned above are specifically realized by setting an NDI spectrum optical navigator in the registration marker selection unit.

The preoperative CT image and intraoperative two-dimensional ultrasonic image registration module includes: an intraoperative two-dimensional ultrasonic image read-in unit, a two-dimensional ultrasonic image and preoperative CT image registration unit, and an ECG (electrocardiogram, electrocardiogram) read-in unit. unit, wherein: the two-dimensional ultrasound image and preoperative CT image registration unit receives the two-dimensional ultrasound input image collected by the intraoperative ultrasound image read-in unit and the current heart beating phase collected by the electrocardiogram read-in unit, through the input electrocardiogram signal will have The corresponding several phases in the three-dimensional dynamic coronary tree of the surgical target point are matched with the current beating phase of the heart, and the preoperative CT transformation matrix between the two is calculated and output to the intraoperative navigation module. Each group of preoperative CT The conversion matrix corresponds to a current beating phase of the heart and a corresponding phase in the three-dimensional dynamic coronary tree with the surgical target point.

The intraoperative navigation module includes: a transformation matrix selection unit and a navigation display unit, wherein: the transformation matrix selection unit selects a corresponding group of transformation matrices in the preoperative CT transformation matrix according to the current cardiac beating phase, and transforms the obtained group of The matrix is output to the navigation display unit, and the distance and relative position relationship between the current virtual instrument and the actual target point are calculated by the navigation display unit and displayed on video, so as to guide the operation to be completed accurately.

Navigation method, including the following steps:

In the first step, after the heart segmentation unit in the image segmentation module obtains a series of CT image data (DICOM format) of one cardiac cycle in the target area through the preoperative CT image import module, the heart CT images in the cardiac cycle are gradually analyzed based on clinical experience. The frame of the heart is manually outlined, the segmentation result is obtained, and the dynamic heart model is reconstructed according to the segmentation result.

In the second step, the vascular tree segmentation unit in the image segmentation module manually outlines the outline of the vascular tree frame by frame based on clinical experience on the CT images of the vascular tree within one cardiac cycle in the preoperative image package, and manually marks the surgical target points according to clinical needs. A segmentation result is obtained, and a vascular tree model is reconstructed according to the segmentation result to construct a three-dimensional virtual scene including a dynamic vascular tree model, and the three-dimensional virtual scene constitutes a virtual endoscopic image.

The third step, the body surface initial registration module uses the body surface registration to obtain the transformation matrix between the coordinates of the preoperative CT image data in the image coordinate system and the spatial coordinates of the real surgical object, that is, the coordinate points in two different spatial coordinate systems, Through feature mutual mapping and one-to-one correspondence, the correspondence between the final preoperative CT image and the real-time two-dimensional ultrasound image of the surgical object is achieved; specifically, by selecting a number of registration markers in the preoperative CT image, find the matching points in the real space. Register the points corresponding to the marked points in the image and use the optical navigator to obtain the coordinates of these points in the real space. Using these two sets of different coordinate systems, but the position coordinates of the point sets that correspond one-to-one, the distance between the two spaces is obtained. transformation matrix.

The body surface registration adopts a rigid body registration algorithm, that is, in two-dimensional space, the transformation formula of point (x ₁ , y ₁ ) to point (x ₂ , y ₂ ) through rigid body transformation is:

Among them, θ is the rotation angle, and (t _x , _ty ) ^T is the translation amount.

The transformation matrix T: [X ₂ ]=T[X ₁ ], wherein: X ₁ and X ₂ correspond to the coordinates of a certain point in the two-dimensional ultrasound image space of the surgical object and the preoperative CT image data space, respectively.

Step 4: Since the preoperative CT image data of one cardiac cycle consists of several phases of preoperative images, the corresponding several phases in the three-dimensional dynamic coronary tree with the surgical target point are compared with the current heart beat by inputting the electrocardiogram signal. Phase matching, the specific steps include:

4.1) Acquisition of a series of real-time images of the surgical object to the heart through the ultrasonic probe, wherein each of the collected real-time images of the surgical object corresponds to a corresponding preoperative CT image, and each phase of the preoperative CT image corresponds to A series of two-dimensional ultrasound images, for the preoperative CT image data of phase i, the surface contour of the inner wall of the heart is extracted, that is, for a preoperative CT image i of a phase, Ti×T can further correct the phase CT The registration error between the image and the real-time two-dimensional ultrasound image of the surgical object, where N is the phase number of the preoperative CT image in the preoperative CT image data of one cardiac cycle, and the specific process is as follows:

A calibrated ultrasound probe is used to obtain intraoperative two-dimensional ultrasound images, and the two-dimensional ultrasound image coordinate system is converted to the preoperative CT image data coordinate system by T, so as to be fused with the preoperative CT image. Let Ti (i=1, 2,...N) be the transformation matrix after correcting the preoperative CT image of each phase in the preoperative image package on the basis of the transformation matrix T, that is, for one phase, and one The preoperative CT image i corresponding to Ti in the group transformation matrix, Ti×T can further correct the registration error between the phase CT image and the real-time two-dimensional ultrasound image of the surgical object. The specific process is as follows: a series of two-dimensional ultrasound images are collected on the heart through the calibrated ultrasound probe. Due to the electrocardiogram signal, each collected two-dimensional ultrasound image corresponds to a corresponding preoperative CT image. After completion, each phase of the preoperative CT image corresponds to a series of 2D ultrasound images. For the preoperative CT image data of phase i, the surface contour of the inner wall of the heart is extracted. Then each two-dimensional ultrasound image corresponding to the phase extracts the contour of the inner wall of the heart. The inner wall extracted from this series of 2D ultrasound images constitutes one set of points, and the surface contour of the inner wall of the heart extracted from the preoperative CT images is another set of points. Through the iterative closest point (ICP) algorithm, a new transformation matrix Ti is obtained by registering the point set on the two-dimensional ultrasound image to the point set on the preoperative CT image, and Ti×T is based on T, and the accuracy is improved. further improvement;

The calibrated ultrasound probe means that in order to integrate the real-time two-dimensional ultrasound image into the navigation system, it is necessary to obtain a conversion matrix from the coordinate system of the two-dimensional ultrasound image to the coordinate system of the navigator. Let TM _{td ← ui} be the transformation matrix from the two-dimensional ultrasound image coordinate system to the coordinate system of the tracking device (optical radiating sphere or electromagnetic sensor) fixed on the ultrasound probe, and TM _{ui ← td} be the tracking device from the ultrasound probe The conversion matrix from the coordinate system to the world coordinate system (navigator coordinate system), the coordinates of a point in the two-dimensional ultrasound image can be converted to the coordinates in the world coordinate system by the following formula.

$\left[\begin{array}{c}{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}\\ {\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}\\ {\mathrm{<span\; class="notranslate">\; z</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}\\ \mathrm{<span\; class="notranslate">\; 1</span>}\end{array}\right]<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; tm</span>}}_{\mathrm{<span\; class="notranslate">\; ui</span>}<span\; class="notranslate">\; \&LeftArrow;</span>\mathrm{<span\; class="notranslate">\; td</span>}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; tm</span>}}_{\mathrm{<span\; class="notranslate">\; td</span>}<span\; class="notranslate">\; \&LeftArrow;</span>\mathrm{<span\; class="notranslate">\; ui</span>}}\left[\begin{array}{c}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; u</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}\\ {\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; u</span>}}_{\mathrm{<span\; class="notranslate">\; v</span>}}\\ \mathrm{<span\; class="notranslate">\; 0</span>}\\ \mathrm{<span\; class="notranslate">\; 1</span>}\end{array}\right]<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

Among them, (u _k , u _v ) is the coordinate of this point in the two-dimensional ultrasound image coordinate system, (s _x , s _y ) is the proportional coefficient of the x-axis and y-axis, (x _w , y _w , z _w ) are its coordinates in the world coordinate system. Ultrasound probe calibration is required to obtain the transformation matrix TM _{td ← ui} from the two-dimensional ultrasound image coordinate system to the coordinate system of the tracking device fixed on the ultrasound probe;

After the above steps are completed, the point set on the two-dimensional ultrasound image is registered to the point set on the preoperative CT image through the iterative closest point algorithm (ICP algorithm), and the transformation matrix Ti between the two point sets is obtained for subsequent The starting matrix for fine registration processing, the specific steps include: assuming two point sets P and Q to be registered, p _i and q _i are points in two point sets respectively, i=1...n, the key to the registration problem is to find the optimal solution such that R and T at minimum hour;

After the initial ICP registration is completed, a transformation matrix Ti×T is obtained for each phase of the preoperative CT image i. After that, in the navigation phase, fine registration will be carried out continuously in real time, and the initial transformation matrix for each fine registration is Ti×T. For a preoperative CT image of a certain phase, the goal is to find an optimal transformation matrix that maximizes the similarity measure, denoted as T′i, so that it can finally satisfy the error caused by factors such as breathing, so as to meet the target The positioning of blood vessels requires that the similarity measure used in this process is the normalized mutual information.

The normalized mutual information is defined as:

$\mathrm{<span\; class="notranslate">\; NMI</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; )</span>$

Among them, M is the image gray point set sampled on the preoperative CT in the area where the two-dimensional ultrasound image is located, R is the real-time two-dimensional ultrasound image gray point set, H(M) is the Shannon entropy of M, i _M represents the gray value of M image pixels, Represents the probability of the pixel gray value i _M in the M image; H(R) is the Shannon entropy of R, i _R represents the gray value of the R image pixel, Represents the probability of the pixel gray value i _R in the R image; H(M,R) is the joint entropy of M and R, $h(m,R)=-\underset{i_m}{\mathrm{\&Sigma;}}\underset{i_R}{\mathrm{\&Sigma;}}{p}_{(i_m,i_R)}\log \left[p_{(i_m,i_R)}\right],$ Represents the probability that the gray value of the pixel in the M image is i _M , and the gray value of the pixel in the R image is i _R . For the M image and R image to be registered, let I _M (X _M ) and I _R (X _R ) be the gray functions of M and R respectively, and X _M and X _R represent the coordinates in the M and R image spaces respectively , then X _R =T′i×X _M , that is, I _R (X _R )=I _R (T′i×X _M ).

The whole registration process is to find T'i, so that NMI (M, R) is the largest. Since Ti×T is obtained through registration with intraoperative two-dimensional ultrasound images, it already has a small error and is a better initial optimization position, so it can quickly converge to the optimal solution, thus obtaining a final exact transformation matrix.

The fifth step is to fuse the images of the real endoscope and the virtual endoscope through external optical positioning.

As shown in Figure 5, the fusion refers to: by fusing the two scenes with different transparency, an augmented virtual reality environment is formed, so that the virtual scene containing the beating heart model seen by the virtual endoscope is consistent with the The real scene seen by the real endoscope is consistent; at the same time, from the set of transformation matrices obtained in step 4.3, the corresponding transformation matrix Ti is selected according to the current heartbeat phase, so as to establish the mapping transformation relationship between the image space and the real space, and finally the The mapping conversion relationship is used to display the distance and relative positional relationship between the instrument and the target point in real time, so as to guide the accurate implementation of the operation.

As shown in FIG. 3 , the in vitro optical positioning refers to the use of NDI optical positioning instrument for infrared reflection positioning, so as to realize the three-dimensional positioning of the research object. The NDI optical navigator includes: a light source 8 and a receiver 9 , and a registration tool 10 . Among them, the farthest effective distance between the light source and the receiver is 3000mm, and the maximum area can reach 1470×1856mm ² . The registration tool is composed of a reflective ball 11 and a long registration needle 12 at the front end. Before the CT scan, the patient evenly attaches 6-8 metal markers, so that the metal markers are highlighted in the preoperative CT image data. During the operation, the patient lies on the operating table in the same position and connects to the life detection equipment. After general anesthesia, the patient begins to use the coordinates of the metal marker points in the image coordinate system obtained in the image, and at the same time use the NDI optical navigator to obtain The coordinates of the metal marker points in the real research object space, the registration of this coordinate system is realized by using the rigid body registration algorithm, so as to realize the registration of the image space and the operation object space.

Through the real-time position information of the endoscope and the manipulator transmitted by the optical locator, the software can display the relative position of the manipulator and the model on the computer screen in real time. In order to realize the fusion of real endoscope and virtual endoscope, it is necessary to obtain the transformation matrix from the world coordinate system to the endoscope coordinate system and the matrix from the endoscope coordinate system to the endoscope projection coordinate system, that is, it is necessary to carry out Calibration of endoscopes. After the endoscope calibration is completed, through the input information such as the position and orientation of the endoscope, the state of the virtual endoscope and the real endoscope in the realization are exactly the same. Therefore, the virtual scene containing the beating heart model seen by the virtual endoscope is consistent with the real scene seen by the real endoscope. By combining the two scenes with different transparency, an augmented virtual reality environment is formed, which is effective To guide the accurate implementation of surgery.

The above-mentioned endoscope calibration: the acquisition of the endoscope image is the result obtained by projecting a 3D scene on a 2D projection plane.

$\mathrm{<span\; class="notranslate">\; \&lambda;</span>}\left[\begin{array}{c}\mathrm{<span\; class="notranslate">\; u</span>}\\ \mathrm{<span\; class="notranslate">\; v</span>}\\ \mathrm{<span\; class="notranslate">\; 1</span>}\end{array}\right]<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; int</span>}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; ext</span>}}<span\; class="notranslate">\; \&times;</span>\left[\begin{array}{c}\mathrm{<span\; class="notranslate">\; x</span>}\\ \mathrm{<span\; class="notranslate">\; the\; y</span>}\\ \mathrm{<span\; class="notranslate">\; z</span>}\\ \mathrm{<span\; class="notranslate">\; 1</span>}\end{array}\right]<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 6</span>}<span\; class="notranslate">\; )</span>$

Among them, X=[xyz 1] ^T is the homogeneous coordinate system of a point in the 3D scene, x, y, and z represent the coordinates of the point on the x-axis, y-axis, and z-axis respectively, and [uv 1] ^T represents the The coordinates of the point in the 2D projection plane, u and v represent the coordinates of the point on the x-axis and y-axis respectively, and λ is the scaling factor of the secondary coordinate system of the 2D projection plane. The above formula includes a conversion matrix M _ext for converting from the world coordinate system to the endoscope coordinate system and a conversion matrix M _int for converting from the endoscope coordinate system to the endoscope projection coordinate system. M _ext is a 4×4 transformation matrix, which can be expressed as:

${\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; ext</span>}}<span\; class="notranslate">\; =</span>\left[\begin{array}{cccc}{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}& {\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}& {\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}& {\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}\\ {\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; 4</span>}}& {\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; 5</span>}}& {\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; 6</span>}}& {\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}\\ {\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; 7</span>}}& {\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; 8</span>}}& {\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; 9</span>}}& {\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; z</span>}}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 1</span>}\end{array}\right]<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; )</span>$

Among them, r ₁ ... r ₉ are rotation factors, t _x , _ty , t _z are translation vectors.

M _int can be expressed as:

${\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; int</span>}}<span\; class="notranslate">\; =</span>\left[\begin{array}{cccc}\mathrm{<span\; class="notranslate">\; sf</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& {\mathrm{<span\; class="notranslate">\; u</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}& \mathrm{<span\; class="notranslate">\; 0</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; f</span>}& {\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}& \mathrm{<span\; class="notranslate">\; 0</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 1</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}\end{array}\right]<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; )</span>$

Where, f is the distance from the focal point of the lens to the center of the mirror, s is the aspect ratio of the field of view of the lens, and (u ₀ , v ₀ ) is the coordinate of the center of the mirror in the 2D projection coordinate system. The calibration of the endoscope is to determine the two matrices M _int and M _ext .

Using this positioning technology can realize the positioning of the target coronary artery relatively quickly and accurately, with an average accuracy of about 3mm, which greatly reduces the time for finding coronary artery disease in robot-assisted coronary artery bypass grafting, and makes the operation safer and more efficient.

In order to avoid errors caused by factors such as respiration and heartbeat during the real-time two-dimensional ultrasound registration of the preoperative CT image and the surgical object, intraoperative ultrasound is introduced into robotic coronary artery bypass surgery navigation, which is the highlight of the present invention.

The beneficial effects it can obtain: 1. Realize the optical navigation positioning method based on the registration results of intraoperative ultrasound and preoperative CT; 2. Solve the problem of relying on personal experience to locate the bridge target point in clinical practice, making surgical operations Safer and more efficient.

Claims (1)

1. an optical guidance positioning navigation method based on CT registration result, it is characterised in that comprise the following steps:

Cardiac segmentation unit in the first step, image segmentation module imports module by preoperative CT image and obtains one, the target area heart After the preoperative CT view data of a series of DICOM format in dynamic cycle, CT image preoperative to the heart in this cardiac cycle based on Clinical experience the most manually sketches the contours of cardiac silhouette, it is thus achieved that segmentation result, and obtains dynamic heart model according to segmentation result reconstruction；

Vessel tree segmentation unit in second step, image segmentation module is preoperative to the vascular tree in a cardiac cycle in preoperative image bag CT image is the most manually sketched the contours of vascular tree profile and manually marks surgical target point according to clinical demand based on clinical experience, it is thus achieved that Segmentation result, and rebuild acquisition vessel tree model according to segmentation result, construct a three-dimensional comprising dynamic vascular tree-model Scene, this three-dimensional virtual scene constitutes a virtual endoscopic images；

3rd step, body surface initial registration module utilize body surface registration to obtain preoperative CT view data at image coordinate system coordinate with true The transition matrix of the space coordinates of real surgical object, the coordinate points under i.e. two different spaces coordinate systems, mapped mutually also by feature Realize one-to-one relationship, reach final preoperative CT image corresponding with surgical object Real-time Two-dimensional ultrasonoscopy, i.e. especially by CT image is chosen several register mark points in the preoperative, finds the point corresponding with register mark point in image also in real space Utilize optical guidance instrument to obtain these coordinates at real space, utilize this two groups of different coordinates, but point set one to one Position coordinates, try to achieve the transition matrix between two spaces；

4th step, by input ECG signal by several phase places corresponding in the Three-Dimensional Dynamic arteria coronaria tree with surgical target point and Current cardiac phase place of beating matches, i.e. using transition matrix T as initial conversion matrix, if Ti is on the basis of transition matrix T On for the transition matrix obtained after the preoperative CT correct image of each phase place, i.e. the preoperative CT for a phase place schemes As i, Ti × T can further correct the registration error between this phase place CT image and surgical object Real-time Two-dimensional ultrasonoscopy, tool Body step includes:

4.1) by the surgical object real time imaging of ultrasonic probe surgical object a series of to cardiac acquisition, each of which width collects Surgical object real time imaging some corresponding preoperative CT image all corresponding, the preoperative CT image corresponding of each phase place The two-dimensional ultrasonic image of series, for the preoperative CT view data of phase place i, extracts the surface profile of wall of the heart, i.e. Obtain the two-dimensional ultrasonic image in art by a ultrasonic probe demarcated, by T, two-dimensional ultrasonic image coordinate system is transformed into Preoperative CT image data coordinate system, thus merge with preoperative CT image；If Ti, i=1,2 ... N, wherein N is one The phase place number of pre-operative image in the preoperative CT view data of cardiac cycle, on the basis of transition matrix T for preoperative image Transition matrix after the preoperative CT correct image of each phase place in bag, i.e. for a phase place and one group conversion square Preoperative CT image i, Ti × T corresponding for Ti in Zhen can further correct this phase place CT image and surgical object real-time two Registration error between dimension ultrasonoscopy；

The described ultrasonic probe demarcated refers to: in order to enable that real-time two-dimensional ultrasonic image is integrated into navigation system, need to try to achieve The transition matrix of navigator coordinate system it is tied to from two-dimensional ultrasonic image coordinate；If TM_td←uiIt is to be tied to admittedly from two-dimensional ultrasonic image coordinate The transition matrix of the coordinate system of the tracing equipment being scheduled on ultrasonic probe, TM_ui←tdIt it is the tracing equipment coordinate system from ultrasonic probe To the transition matrix of world coordinate system, the coordinate of a point in two-dimensional ultrasonic image can be transformed into the world by equation below and sit Coordinate under mark system；

$\left[\begin{array}{c}{x}_w\\ y_w\\ z_w\\ 1\end{array}\right]={\mathrm{TM}}_{ui\&LeftArrow;td}\&times;{\mathrm{TM}}_{td\&LeftArrow;ui}\left[\begin{array}{c}{s}_x\&times;u_k\\ s_y\&times;u_v\\ 0\\ 1\end{array}\right]$

Wherein, (u_k,u_v) it is this coordinate in two-dimensional ultrasonic image coordinate system, (s_x,s_y) it is the ratio of x-axis and y-axis Coefficient, (x_w,y_w,z_w) it is its coordinate in world coordinate system；Ultrasonic probe is demarcated and is sought to try to achieve from two-dimensional ultrasonic image coordinate It is tied to the transition matrix TM of the coordinate system of the tracing equipment being fixed on ultrasonic probe_td←ui；

4.2) each width two-dimensional ultrasonic image corresponding to phase place i is extracted the profile of wall of the heart, from this series of two-dimensional ultrasound The inwall extracted in image constitutes one group of point set, extracts characteristic point point set from this some concentration, extracts from preoperative CT image Wall of the heart surface profile is that another organizes point set；

4.3) point set point set on two-dimensional ultrasonic image being registrated on preoperative CT image by iterative closest point algorithm, obtain this two Transition matrix Ti between individual point set, for the initial matrix of follow-up smart registration process, this iterative closest point algorithm is by two-dimensional ultrasound Point set on image is registrated to the point set on preoperative CT image, obtains the transition matrix Ti between the two point set, for follow-up The initial matrix of essence registration process, concrete steps include: assume two point set P and Q subject to registration,p_i And q_iBeing the point of two somes concentrations respectively, i=1 ... n, the key of registration problems is just to solve for optimal solution and makesR and T time minimum；

After completing registration at the beginning of ICP, the preoperative CT image i of each phase place has each obtained a transition matrix Ti × T； Afterwards in navigation stage, it will real-time is carried out continuously essence registration, and the initial conversion matrix of essence registration is exactly Ti × T each time；Right In the preoperative CT image of some phase place, aim at and find an optimum transition matrix making similarity measure maximum, be designated as T ' i, Can finally meet the error caused due to factors such as breathings, thus meet the positioning requirements to target blood, in the process The similarity measure used is normalized mutual information；

Normalized mutual information is:Wherein: M is that two-dimensional ultrasonic image region is in art The gradation of image point set of front CT up-sampling gained, R is real-time two-dimensional ultrasonic image gray scale point set, and H (M) is the Shannon entropy of M,i_MRepresent the gray value of M image slices vegetarian refreshments,Representing pixel gray value in M image is i_M Probability；H (R) is the Shannon entropy of R,i_RRepresent the gray value of R image slices vegetarian refreshments,Generation In table R image, pixel gray value is i_RProbability；H (M, R) is the combination entropy of M and R, Representing pixel gray value in M image is i_M, pixel in R image Point gray value is i_RProbability；For M image subject to registration and R image, if I_M(X_M)、I_R(X_R) be respectively M, R ash Degree function, X_M、X_RRepresent the coordinate in M, R image space, then X respectively_R=T ' i × X_M, i.e. I_R(X_R)=I_R(T′i×X_M)； Whole registration process namely finds T ' i so that NMI (M, R) is maximum；Due to Ti × T be by with two-dimensional ultrasound figure in art The registration of picture obtains, and has been provided with less error, is an optimizing position the most initial, therefore can quickly converge to Excellent solution, thus obtain a final accurate transition matrix；

5th step, by external optical alignment, true endoscope is merged with virtual endoscopic images, i.e. by by two width scenes Blend with different transparencys, just define one and strengthen reality environment so that what virtual endoscope was seen comprises heart jumping The real scene that the virtual scene of movable model is seen with true endoscope is consistent；Simultaneously from one group of transition matrix that step 4.3 obtains Beat transition matrix Ti corresponding to Selecting phasing according to current cardiac, thus set up the Mapping and Converting between image space and real space Relation, this Mapping and Converting relation is for demonstrating the distance between apparatus and impact point and relative position relation in real time the most at last, thus Operation is guided accurately to implement；

Described external optical alignment refers to use NDI optical orientator to carry out infrared reflection location, thus realizes object of study Three-dimensional localization；

Described fusion refers to: by two width scenes being blended with different transparencys, just defines one and strengthens virtual reality ring Border so that the virtual scene that what virtual endoscope was seen comprise heartbeat model is consistent with the real scene that true endoscope is seen； Beat transition matrix Ti corresponding to Selecting phasing according to current cardiac from one group of transition matrix that step 4.3 obtains simultaneously, thus build Vertical Mapping and Converting relation between image space and real space, this Mapping and Converting relation demonstrates apparatus and mesh in real time the most at last Distance between punctuate and relative position relation, thus guide operation accurately to implement；

Described method is realized by following system, and this system includes: preoperative CT image imports module, image segmentation module, body surface Initial registration module, preoperative CT image and navigation module in two-dimensional ultrasonic image registration module in art and art, wherein: preoperative CT Image imports module and receives the DICOM format image file that preoperative imaging examination obtains, and generates preoperative image bag and exports respectively To image segmentation module and body surface initial registration module, preoperative CT image and two-dimensional ultrasonic image registration module in art respectively with image Segmentation module is connected with body surface initial registration module and receives the Three-Dimensional Dynamic arteria coronaria tree with surgical target point and body surface is initially joined Quasi-result, preoperative CT image, passes through to navigation module in art with the preoperative CT transition matrix of two-dimensional ultrasound registration module output in art In art, navigation module exports accurate surgical navigational information；

Described preoperative image bag includes: in target area, some width DIOM forms of one or several cardiac cycles is preoperative CT image and the 3 D stereoscopic image obtained according to DICOM data reconstruction, wherein: the preoperative CT image of a cardiac cycle Data are made up of the pre-operative image of several phase places；

The described Three-Dimensional Dynamic arteria coronaria tree with surgical target point includes: the dynamic heart model generated by image segmentation module and blood Pipe tree-model；

Described preoperative CT image imports module and includes: DICOM image reads in unit and preoperative image three-dimensional drawing unit, wherein: DICOM image reads in a series of DICOM format images of one cardiac cycle of heart that preoperative imaging examination is obtained by unit File imports and parses DICOM data and transmit extremely preoperative image three-dimensional drawing unit, preoperative image three-dimensional drawing unit according to DICOM data reconstruction becomes 3 D stereoscopic image and merges into preoperative image bag and export to image segmentation module and body surface initial respectively Registration module；

Described image segmentation module includes: cardiac segmentation unit and vessel tree segmentation unit, wherein: cardiac segmentation unit is to preoperative In image bag, the heart preoperative CT image in a cardiac cycle carries out manual segmentation frame by frame based on clinical experience, and according to acquisition Heart preoperative CT image segmentation result is rebuild and is obtained dynamic heart model；Vessel tree segmentation unit is aroused in interest in preoperative image bag one Vascular tree preoperative CT image in cycle carries out manual segmentation frame by frame based on clinical experience, and according to the preoperative CT of vascular tree obtained Image segmentation result is rebuild and is obtained vessel tree model and be preced with as the Three-Dimensional Dynamic with surgical target point in the lump with dynamic heart model The output of arteries and veins tree is to preoperative CT image and two-dimensional ultrasonic image registration module in art；

Described body surface initial registration module includes: register mark point selection unit and transform matrix calculations unit, wherein: registration mark The note register mark point that is arranged on surgical object body surface of point selection Unit selection, and obtain its in the preoperative CT view data at image Coordinate system coordinate and space coordinates, transform matrix calculations unit, according to image coordinate system coordinate and space coordinates, utilizes rigid body to join Quasi-algorithm realizes the registration between the two coordinate, i.e. realizes the registration of image space and surgical object space, and utilizes these two groups seats Mark calculates initial conversion matrix as body surface initial registration result, output to preoperative CT image and two-dimensional ultrasonic image registration in art Module；

Described preoperative CT image includes with two-dimensional ultrasonic image registration module in art: in art two-dimensional ultrasonic image read in unit, two Dimension ultrasonoscopy reads in unit with preoperative CT image registration unit and electrocardiogram, wherein: two-dimensional ultrasonic image and preoperative CT image Registration unit receives intraoperative ultrasound image and reads in the two-dimensional ultrasound input picture of unit collection and the current of electrocardiogram reading unit collection Heartbeat phase place, by input ECG signal by several phase places corresponding in the Three-Dimensional Dynamic arteria coronaria tree with surgical target point Phase place of beating with current cardiac is mated, and calculates preoperative CT image transition matrix between the two and exports to art navigation mould Block, the corresponding a kind of current cardiac of each of which group preoperative CT transition matrix phase place of beating is preced with the Three-Dimensional Dynamic with surgical target point Corresponding phase in arteries and veins tree；

In described art, navigation module includes: transition matrix selects unit and navigation display unit, wherein: transition matrix selects unit Beat one group of transition matrix corresponding in Selecting phasing preoperative CT transition matrix according to current cardiac, and by the one of gained group of conversion square Battle array output to the display unit that navigates, by navigation display unit calculate the distance between current virtual apparatus and actual target point and Relative position relation also carries out video and shows, thus guides operation to be precisely accomplished；

Described body surface registration uses rigid registration algorithm, i.e. in two-dimensional space, point (x₁,y₁) through rigid body translation to point (x₂,y₂) transformation for mula be:Wherein: θ is the anglec of rotation, (t_x,t_y)^TFor Translational movement；

Described transition matrix T:[X₂]=T [X₁], wherein: X₁And X₂The most corresponding surgical object two-dimensional ultrasonic image space Certain point coordinates with preoperative CT image data space.