CN102999902A - Optical navigation positioning system based on CT (computed tomography) registration results and navigation method thereby - Google Patents

Abstract

Disclosed are an optical navigation positioning system based on CT (computed tomography) registration results and a navigation method thereby. The system comprises a preoperative CT image guide input module, an image segmentation module, a body surface initial-registration module, a preoperative CT image and intraoperative two-dimensional ultrasound image module and an intraoperative navigation module. By combining virtual reality and intraoperative ultrasound, intraoperative positioning errors caused by factors such as breathing are compensated, and accordingly a target point for coronary artery bypass grafting is accurately positioned and navigated. Cardiac and coronary vessel tree in preoperative cardiac CT image data is manually segmented and reconstructed, an augmented virtual reality environment integrating endoscope and virtual endoscope is built by the aid of optical navigation apparatus and CT-ultrasound-based intraoperative registration error correction, and accordingly the target point of coronary artery bypass grafting is accurately positioned and navigated.

Description

Optical guidance positioning system and air navigation aid thereof based on the CT registration results

Technical field

What the present invention relates to is a kind of system and method for technical field of information processing, specifically a kind of navigation positioning system and air navigation aid thereof based on CT registration results before intraoperative ultrasound and the art for auxiliary coronary artery bypass surgery.

Background technology

The incidence of disease of China's cardiovascular disease rises year by year in recent years, and wherein coronary heart disease is modal angiocardiopathy.In the methods for the treatment of for coronary heart disease, bypass operation of coronary artery is present main, ripe therapeutic modality except drug therapy and interventional therapy.Yet traditional coronary artery bypass surgery needs mid-sternal incision, also need to just can finish operation by operations such as extracorporal circulatory systems in case of necessity.It comprises the shortcomings such as otch is large, it is slow to recover, complication is many.The auxiliary Wicresoft of navigation coronary bypass surgery as the novel therapeutic mode, only need by opening several finger thicknesses at the wall of the chest otch, utilize special operating theater instruments can finish operation, reached the requirements such as otch is attractive in appearance, wound is little, recovery is fast, few intercurrent disease.

How fast and accurately one of the main Operative Puzzle of auxiliary Wicresoft coronary artery bypass grafting of navigating is localizing objects point, and improper or wrong location can directly have influence on success or failure and the late result of operation.At present, the location of Wicresoft's coronary bypass surgery mainly relies on the front image of art, fail to utilize the real-time information of reflection operative region truth, can't solve the error problem that causes owing to factors such as breathing, skin-marker displacement and position chanPs in the art, thereby so that locating effect is undesirable.

Find through the retrieval to prior art, " 3D-image guidance for minimally invasive robotic coronary artery bypass; Heart Surg Forum 2000-9732 (3) " (3-D view guidance machine people assists method of off pump coronary bypass, forum of department of cardiac surgery, 2000) in TM.Peter attempt at first the research that the applying three-dimensional imaged image carries out the preplanning of cardiac operation art, he is by the CT Image Segmentation Using to gathering before the operation in patients, obtain the surface model of heart and skeleton, tentatively set up the operation guiding system of Wicresoft's bypass surgery.This system provides a virtual endoscope simultaneously, utilizes the technology of pre-operative surgical registration, so that the relative position relation of the skeleton of the heart model that modeling goes out in the relative position relation of model and virtual endoscope and the system in true endoscope and the model test is consistent.But this system is a prototype system, has many deficiencies, does not have and the virtual problems such as 3D scene effective integration such as true endoscope 2D image.

" 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. " (the flexible calibration of robotic assisted surgery neutral body endoscope, medical image in 2002 calculates with area of computer aided and gets involved conference) in Mourguess and Coste-Maniere attempting the virtual endoscope that the heart of an animal is set three-dimensional model at the image of Microendoscopic and the coronary artery of this heart is merged so that certain transparency is overlapping separately, adversary's Intraoperative position impact point has played certain effect.But the model of coronary artery tree is static in the art, and the fusion of the three-dimensional model of coronary artery tree and endoscope image is to rely on to identify the sign point of observing in the endoscope the two is merged comparatively accurately, and the robustness of this method and accuracy are very not satisfactory.

Chinese patent literature CN1650813, put down in writing " surgical navigation systems is based on the robotic surgery localization method of optical alignment ", the method is at first at three zone bits of robot base design, choose this three signs by the optical tracker positioning pointer, allow robot probe and optical tracker positioning pointer when docking gather simultaneously the space the same coordinate system, at last the coordinate interchange by optical tracker and robot system.This technology mainly use with cerebral surgery operation in because the deformation of cerebral surgery operation anatomy is little, easier assurance on the accuracy problem; But this technology can't be applied to the larger cardiovascular system field of anatomy deformation.

Chinese patent literature CN101703409A, put down in writing " a kind of system and method for ultrasound-guided robot-assisted treatment ", this system comprises operating robot, two-dimensional ultrasound instrument, magnetic orientator and workstation, by they are combined, effective detection and robotization treatment to affected area have been realized, thereby reduce doctor's labour intensity, and improve the degree of accuracy that operation is implemented.But what use in the method is magnetic orientator, and magnetic orientator is subject to the various interference such as place, apparatus, other instruments in use, produces easily error.The ultrasound information of three-dimensional reconstruction can produce the anamorphose problem simultaneously, may cause a deviation with its image information guided operation, and it is limited to produce quantity of information, can't be used for some comparatively operations of elaborate, such as the accurate operation of heart operation.

As seen, although the robot assisted therapy system has begun to be applied in the clinical treatment, its development on the surgical target location is not perfect, and a lot of clinical difficult problems still have to be solved.

Summary of the invention

The present invention is directed to the prior art above shortcomings, a kind of optical guidance positioning system and air navigation aid thereof based on the CT registration results proposed, the mode that adopts virtual reality and intraoperative ultrasound to combine, the Intraoperative position error that compensation causes owing to factors such as breathings, thus realization is to the accurate location navigation of the impact point of off pump coronary artery bypass.By the heart in the CT view data before the art and coronary artery vascular tree are manually cut apart reconstruction, then the optical guidance instrument with based on art before build an enhancing reality environment that endoscope and virtual endoscope are merged mutually under registration error is proofreaied and correct in the art of CT image and two-dimensional ultrasonic image the help, thereby can realize the accurate location navigation to the impact point of off pump coronary artery bypass.

The present invention is achieved by the following technical solutions:

The present invention relates to a kind of optical guidance positioning system based on the CT registration results, comprise: the CT image imports module before the art, the image segmentation module, body surface initial registration module, navigation module in two-dimensional ultrasonic image registration module and the art in the front CT image of art and the art, wherein: the CT image imports module and receives the DICOM format-pattern file that the front imaging examination of art obtains before the art, the image bag also exports respectively image segmentation module and body surface initial registration module to before generating art, before the art in CT image and the art two-dimensional ultrasonic image registration module link to each other with body surface initial registration module with the image segmentation module respectively and receive with the Three-Dimensional Dynamic coronary artery of surgical target point and set and body surface initial registration result, the front CT transition matrix of two-dimensional ultrasonic image registration module output art navigation module to the art in CT image and the art is exported accurate surgical navigational information by navigation module in the art before the art.

The image bag comprises before the described art: CT image and the 3 D stereoscopic image that obtains according to the DICOM data reconstruction before the art of some width of cloth DIOM forms of one or several cardiac cycles the target area in, wherein: the front CT view data of cardiac cycle art by the art of several phase places before image form.

Described Three-Dimensional Dynamic coronary artery tree with surgical target point comprises: by dynamic heart model and the vascular tree model of image segmentation module generation.

Image importing module comprises before the described art: the DICOM image reads in unit and the front image three-dimensional drawing unit of art, wherein: the DICOM image read in the unit with art before a series of DICOM format-pattern files of one cardiac cycle of heart of obtaining of imaging examination import parse the DICOM data and transfer to art before the image three-dimensional drawing unit, image bag and export respectively image segmentation module and body surface initial registration module to before the image three-dimensional drawing unit becomes 3 D stereoscopic image and merges into art according to the DICOM data reconstruction before the art.

Described image segmentation module comprises: cardiac segmentation unit and vascular tree cutting unit, wherein: the cardiac segmentation unit to art before in the image bag cardiac CT image in the cardiac cycle manually sketch the contours of frame by frame cardiac silhouette based on clinical experience, and rebuild according to the cardiac CT image segmentation result that obtains and to obtain the dynamic heart model; The vascular tree cutting unit to art before in the image bag vascular tree CT image in the cardiac cycle manually sketch the contours of frame by frame the vascular tree profile and manually mark the surgical target point according to clinical demand based on clinical experience, obtain the vascular tree model and export before the art two-dimensional ultrasonic image registration module in the CT image and art to as setting with the Three-Dimensional Dynamic coronary artery of surgical target point in the lump with the dynamic heart model thereby rebuild according to the vascular tree CT image segmentation result that obtains.

Described body surface initial registration module comprises: register mark point selection unit and transform matrix calculations unit, wherein: the register mark point on the surgical object body surface is selected to be arranged in register mark point selection unit, and obtain the front CT view data of its art in the volume coordinate of image coordinate system coordinate and true surgical object, the transform matrix calculations unit is according to image coordinate system coordinate and volume coordinate, utilize the rigid registration algorithm to realize registration between these two coordinates, namely realize the registration in image space and surgical object space, and utilize these two groups of coordinate Calculation to go out the initial conversion matrix as body surface initial registration result, export two-dimensional ultrasound registration module in the front CT image of art and the art to.

The two-dimensional ultrasonic image registration module comprises in the front CT image of described art and the art: two-dimensional ultrasonic image reads in the unit in the art, CT image registration unit and ECG (electrocardiogram before two-dimensional ultrasonic image and the art, cardiogram) reads in the unit, wherein: two-dimensional ultrasonic image reads in the two-dimensional ultrasound input picture of unit collection and the current heartbeat phase place that cardiogram reads in the unit collection in the front CT image registration unit reception of two-dimensional ultrasonic image and the art art, to mate with corresponding several phase places and current heartbeat phase place in the Three-Dimensional Dynamic coronary artery tree of surgical target point by the input ECG signal, calculate transition matrix between the two and export navigation module in the art to, wherein the heart of each transition matrix phase place in corresponding cardiac cycle and set with the three-dimensional coronary artery of surgical target point.

Navigation module comprises in the described art: transition matrix selected cell and navigation display unit, wherein: the transition matrix selected cell is according to a current heartbeat Selecting phasing corresponding transition matrix with it, and export this transition matrix of gained to the navigation display unit, calculate distance and the relative position relation between current virtual apparatus and the actual target point and carry out the video demonstration by the navigation display unit, thereby guided surgery is accurately finished.

The present invention relates to the air navigation aid of said system, may further comprise the steps:

After cardiac segmentation unit in the first step, the image segmentation module imports module a series of CT view data (DICOM form) of acquisition one cardiac cycle of target area by the front CT image of art, cardiac CT image in this cardiac cycle is manually sketched the contours of cardiac silhouette frame by frame based on clinical experience, obtain segmentation result, and reconstruction obtains the dynamic heart model according to segmentation result.

Vascular tree cutting unit in second step, the image segmentation module to art before in the image bag vascular tree CT image in the cardiac cycle manually sketch the contours of frame by frame the vascular tree profile and manually mark the surgical target point according to clinical demand based on clinical experience, obtain segmentation result, thereby rebuild acquisition vascular tree model according to segmentation result, construct a three-dimensional virtual scene that comprises the dynamic vascular tree-model, this three-dimensional virtual scene has consisted of a virtual endoscopic images.

The 3rd step, body surface initial registration module utilize the body surface registration to obtain the transition matrix of the front CT image coordinate system of art and surgical object Real-time Two-dimensional ultrasonoscopy coordinate system, i.e. coordinate points under two different spaces coordinate systems, by mutually mapping and realize one-to-one relationship of feature, reach the corresponding of CT image and surgical object Real-time Two-dimensional ultrasonoscopy before the final art; Specifically by choosing several register mark points in the CT image in the preoperative, find the point corresponding with register mark point in the image and utilize the optical guidance instrument to obtain these at the coordinate of real space in real space, utilize this two groups of different coordinates, but the position coordinates of point set is tried to achieve two transition matrix T between the space one to one.

The 4th step, to be complementary with corresponding several phase places and current heartbeat phase place in the Three-Dimensional Dynamic coronary artery tree of surgical target point by the input ECG signal, namely with above-mentioned transition matrix T as the initial conversion matrix, if Ti(i=1,2, N) be on the basis of transition matrix T for art before transition matrix behind the CT correct image of each phase place in the image bag, namely for the CT image i of one of them phase place, Ti * T can further proofread and correct the registration error between this phase place CT image and the surgical object Real-time Two-dimensional ultrasonoscopy, wherein N is the phase place number of the front CT image of art in the front CT view data of cardiac cycle art, thereby obtains one group of transition matrix that T is proofreaied and correct; Concrete steps comprise:

4.1) by ultrasonic probe heart is gathered the surgical object realtime graphic of a series of surgical objects, wherein the surgical object realtime graphic that collects of each width of cloth all before corresponding some corresponding arts before the art of CT image, each phase place the CT image corresponding a series of two-dimensional ultrasonic image, for CT view data before the art of phase place i, extract the surface profile of wall of the heart;

4.2) corresponding each width of cloth two-dimensional ultrasonic image of phase place i extracted the profile of wall of the heart, the inwall that extracts from this a series of two-dimensional ultrasonic image has formed one group of point set, extract the unique point point set from this point is concentrated, the wall of the heart surface profile that extracts the CT image before art is another group point set;

4.3) by iterative closest point algorithms (ICP algorithm) point set on the two-dimensional ultrasonic image is registrated to the point set on the CT image before the art, obtain the transition matrix Ti between these two point sets, be used for the initial matrix of follow-up smart registration process.

The 5th goes on foot, by external optical alignment true endoscopic images and virtual endoscopic images is merged.

Described external optical alignment refers to adopt the NDI optical orientator to carry out the infrared reflection location, thereby realizes the three-dimensional localization to research object.

Described fusion refers to: by two width of cloth scenes are merged mutually with different transparencies, just formed one and strengthened reality environment, so that the real scene that the virtual scene that comprises the heartbeat model that virtual endoscope is seen and true endoscope are seen is consistent; The transition matrix Ti corresponding according to current heartbeat Selecting phasing from one group of transition matrix that step 4.3 obtains simultaneously, thereby set up the Mapping and Converting relation between image space and the real space, this Mapping and Converting relation is used for demonstrating in real time distance and the relative position relation between apparatus and the impact point the most at last, thereby guided surgery is accurately implemented.

Technique effect

Advantage of the present invention comprises: 1. take the lead in introducing intraoperative ultrasound in robot assisted coronary bypass surgery navigation, proofread and correct the error that the factor such as breathing causes by the registration of CT image before two-dimensional ultrasonic image in the art and the art; 2. solved and relied on clinically the personal experience to locate the uncertain problem of bridging impact point all the time.

Description of drawings

Fig. 1 is modular structure synoptic diagram of the present invention.

Fig. 2 is several three-dimensional cardiac synoptic diagram of embodiment (containing coronary artery tree and cardiogram correspondence position).

Fig. 3 is optical guidance instrument and registration module synoptic diagram.

Fig. 4 is that the embodiment flow process is introduced synoptic diagram.

Fig. 5 is that embodiment merges synoptic diagram;

Among the figure: a is real endoscopic images; B is virtual endoscopic images; C is both coincidence pattern pictures.

Embodiment

The below elaborates to embodiments of the invention, and present embodiment is implemented under take technical solution of the present invention as prerequisite, provided detailed embodiment and concrete operating process, but protection scope of the present invention is not limited to following embodiment.

Embodiment 1

As shown in Figure 1, present embodiment comprises: comprising: the CT image imports module before the art, the image segmentation module, body surface initial registration module, navigation module in two-dimensional ultrasonic image registration module and the art in the front CT image of art and the art, wherein: the CT image imports module and receives the DICOM format-pattern file that the front imaging examination of art obtains before the art, the image bag also exports respectively image segmentation module and body surface initial registration module to before generating art, before the art in CT image and the art two-dimensional ultrasonic image registration module link to each other with body surface initial registration module with the image segmentation module respectively and receive with the Three-Dimensional Dynamic coronary artery of surgical target point and set and body surface initial registration result, the front CT transition matrix of two-dimensional ultrasonic image registration module output art navigation module to the art in CT image and the art is exported accurate surgical navigational information by navigation module in the art before the art.

The image bag comprises before the described art: the front CT image of the art of some width of cloth DIOM forms of one or several cardiac cycles and the 3 D stereoscopic image that obtains according to the DICOM data reconstruction the target area in.

Described Three-Dimensional Dynamic coronary artery tree with surgical target point comprises: by dynamic heart model and the vascular tree model of image segmentation module generation.

CT image importing module comprises before the described art: the DICOM image reads in unit and the front image three-dimensional drawing unit of art, wherein: the DICOM image read in the unit with art before a series of DICOM format-pattern files of one cardiac cycle of heart of obtaining of imaging examination import parse the DICOM data and transfer to art before the image three-dimensional drawing unit, image bag and export respectively image segmentation module and body surface initial registration module to before the image three-dimensional drawing unit becomes 3 D stereoscopic image and merges into art according to the DICOM data reconstruction before the art.

Described image segmentation module comprises: cardiac segmentation unit and vascular tree cutting unit, wherein: the cardiac segmentation unit to art before in the image bag cardiac CT image in the cardiac cycle manually sketch the contours of frame by frame cardiac silhouette based on clinical experience, and rebuild according to the cardiac CT image segmentation result that obtains and to obtain the dynamic heart model; The vascular tree cutting unit to art before in the image bag vascular tree CT image in the cardiac cycle manually sketch the contours of frame by frame the vascular tree profile and manually mark the surgical target point according to clinical demand based on clinical experience, obtain the vascular tree model and set as the Three-Dimensional Dynamic coronary artery with surgical target point as shown in Figure 2 in the lump with the dynamic heart model to export before the art two-dimensional ultrasonic image registration module in the CT image and art to thereby rebuild according to the vascular tree CT image segmentation result that obtains.

Described body surface initial registration module comprises: register mark point selection unit and transform matrix calculations unit, wherein: the register mark point on the surgical object body surface is selected to be arranged in register mark point selection unit, and obtain its in the CT view data in image coordinate system coordinate and volume coordinate, the transform matrix calculations unit is according to image coordinate system coordinate and volume coordinate, utilize the rigid registration algorithm to realize registration between these two coordinates, namely realize the registration in image space and surgical object space, and utilize these two groups of coordinate Calculation to go out the initial conversion matrix as body surface initial registration result, export two-dimensional ultrasonic image registration module in the front CT image of art and the art to.

Described register mark point refers to: when obtaining in the preoperative the CT view data, body surface at surgical object sticks 6-8 metal marker point, the metal marker point is evenly distributed on the thorax body surface of surgical object, metal marker point can be in the preoperative highlighted demonstration in the CT view data.

Described volume coordinate specifically is achieved by NDI spectra optical guidance instrument is set in register mark point selection unit.

The two-dimensional ultrasonic image registration module comprises in the front CT image of described art and the art: two-dimensional ultrasonic image reads in the unit in the art, CT image registration unit and ECG (electrocardiogram before two-dimensional ultrasonic image and the art, cardiogram) reads in the unit, wherein: CT image registration unit receives the intraoperative ultrasound image and reads in the two-dimensional ultrasound input picture of unit collection and the current heartbeat phase place that cardiogram reads in the unit collection before two-dimensional ultrasonic image and the art, to mate with corresponding several phase places and current heartbeat phase place in the Three-Dimensional Dynamic coronary artery tree of surgical target point by the input ECG signal, calculate the front CT transition matrix of art between the two and export navigation module in the art to, corresponding phase during wherein each is organized the corresponding a kind of current heartbeat phase place of the front CT transition matrix of art and sets with the Three-Dimensional Dynamic coronary artery of surgical target point.

Navigation module comprises in the described art: transition matrix selected cell and navigation display unit, wherein: the transition matrix selected cell according to current heartbeat Selecting phasing art before corresponding one group of transition matrix in the CT transition matrix, and export one group of transition matrix of gained to the navigation display unit, calculate distance and the relative position relation between current virtual apparatus and the actual target point and carry out the video demonstration by the navigation display unit, thereby guided surgery is accurately finished.

Air navigation aid may further comprise the steps:

After cardiac segmentation unit in the first step, the image segmentation module imports module a series of CT view data (DICOM form) of acquisition one cardiac cycle of target area by the front CT image of art, cardiac CT image in this cardiac cycle is manually sketched the contours of cardiac silhouette frame by frame based on clinical experience, obtain segmentation result, and reconstruction obtains the dynamic heart model according to segmentation result.

Vascular tree cutting unit in second step, the image segmentation module to art before in the image bag vascular tree CT image in the cardiac cycle manually sketch the contours of frame by frame the vascular tree profile and manually mark the surgical target point according to clinical demand based on clinical experience, obtain segmentation result, and according to segmentation result reconstruction acquisition vascular tree model, construct a three-dimensional virtual scene that comprises the dynamic vascular tree-model, this three-dimensional virtual scene has consisted of a virtual endoscopic images.

The CT view data was at the transition matrix of image coordinate system coordinate with the volume coordinate of true surgical object before the 3rd step, body surface initial registration module utilized the body surface registration to obtain art, i.e. coordinate points under two different spaces coordinate systems, by mutually mapping and realize one-to-one relationship of feature, reach the corresponding of CT image and surgical object Real-time Two-dimensional ultrasonoscopy before the final art; Specifically by choosing several register mark points in the CT image in the preoperative, find the point corresponding with register mark point in the image and utilize the optical guidance instrument to obtain these at the coordinate of real space in real space, utilize this two groups of different coordinates, but the position coordinates of point set is tried to achieve two transition matrixes between the space one to one.

Described body surface registration adopts the rigid registration algorithm, namely in two-dimensional space, and point (x ₁, y ₁) arrive point (x through rigid body translation ₂, y ₂) transformation for mula be:

Wherein, θ is the anglec of rotation, (t _x, t _y) ^TBe translational movement.

Described transition matrix T:[X ₂]=T[X ₁], wherein: X ₁And X ₂Certain point coordinate of CT image data space before the corresponding surgical object two-dimensional ultrasonic image space of difference and the art.

The 4th step, since before a cardiac cycle art CT view data by the art of several phase places before image form, therefore will be complementary with corresponding several phase places and current heartbeat phase place in the Three-Dimensional Dynamic coronary artery tree of surgical target point by the input ECG signal, concrete steps comprise:

4.1) by ultrasonic probe heart is gathered the surgical object realtime graphic of a series of surgical objects, the surgical object realtime graphic that collects of each width of cloth CT image before corresponding some corresponding arts all wherein, before the art of each phase place the CT image corresponding a series of two-dimensional ultrasonic image, for CT view data before the art of phase place i, extract the surface profile of wall of the heart, namely for CT image i before the art of a phase place, Ti * T can further proofread and correct the registration error between this phase place CT image and the surgical object Real-time Two-dimensional ultrasonoscopy, wherein N is the phase place number of the front CT image of art in the front CT view data of cardiac cycle art, and idiographic flow is:

Demarcate good ultrasonic probe by one and obtain two-dimensional ultrasonic image in the art, by T the two-dimensional ultrasonic image coordinate system is transformed into CT image data coordinate system before the art, thus with art before the CT image merge.If Ti(i=1,2, N) be on the basis of transition matrix T for art before the transition matrix behind the CT correct image before the art of each phase place in the image bag, namely for CT image i before a phase place, corresponding with the Ti in the one group of transition matrix art, Ti * T can further proofread and correct the registration error between this phase place CT image and the surgical object Real-time Two-dimensional ultrasonoscopy.Idiographic flow is: by the ultrasonic probe of demarcating heart is gathered a series of two-dimensional ultrasonic image, because ECG signal, the two-dimensional ultrasonic image that each width of cloth gathers is the front CT image of corresponding some corresponding arts all.After finishing, before the art of each phase place the CT image corresponding a series of two-dimensional ultrasonic image.For CT view data before the art of phase place i, extract the surface profile of wall of the heart.Then corresponding each width of cloth two-dimensional ultrasonic image of this phase place extracts the profile of wall of the heart.The inwall that extracts from this a series of two-dimensional ultrasonic image has formed one group of point set, and the wall of the heart surface profile that extracts the CT image before art is another group point set.By iterative closest point (ICP) algorithm the point set on the two-dimensional ultrasonic image is registrated to before the art and has obtained a new transition matrix Ti on the point set on the CT image, and Ti * T is on the basis of T, precision has had further raising;

The ultrasonic probe that described demarcation is good refers to: for real-time two-dimensional ultrasonic image being integrated into navigational system, need to try to achieve the transition matrix that is tied to the navigating instrument coordinate system from the two-dimensional ultrasonic image coordinate.If TM _{Td ← ui}The transition matrix that is tied to the coordinate system of the tracing equipment that is fixed on the ultrasonic probe (optics give out light ball or electromagnetic sensor) from the two-dimensional ultrasonic image coordinate, TM _{Ui ← td}Be the transition matrix that tracing equipment coordinate from the ultrasonic probe is tied to world coordinate system (navigating instrument coordinate system), the coordinate of a point in the two-dimensional ultrasonic image can be transformed into by following formula the coordinate under the world coordinate system.

$\left[\begin{array}{c}{x}_w\\ y_w\\ z_w\\ 1\end{array}\right]={\mathrm{TM}}_{\mathrm{ui}\←\mathrm{td}}\×{\mathrm{TM}}_{\mathrm{td}\←\mathrm{ui}}\left[\begin{array}{c}{s}_x\×u_k\\ s_y\×u_v\\ 0\\ 1\end{array}\right]---\left(3\right)$

Wherein, (u _k, u _v) be this coordinate in the two-dimensional ultrasonic image coordinate system, (s _x, s _y) be the scale-up factor of x axle and y axle, (x _w, y _w, z _w) be its coordinate in world coordinate system.Ultrasonic probe is demarcated and will be tried to achieve exactly the transition matrix TM that is tied to the coordinate system that is fixed on the tracing equipment on the ultrasonic probe from the two-dimensional ultrasonic image coordinate _{Td ← ui}

Finish after the above-mentioned steps by iterative closest point algorithms (ICP algorithm) point set on the two-dimensional ultrasonic image is registrated to the point set on the CT image before the art, obtain the transition matrix Ti between these two point sets, the initial matrix that is used for follow-up smart registration process, concrete steps comprise: suppose two point set P and Q subject to registration

p _iAnd q _iRespectively the point that two points are concentrated, i=1 ... n, the key of registration problems be exactly find the solution optimum solution so that

Hour R and T;

After having finished the first registration of ICP, CT image i has obtained a transition matrix Ti * T separately before the art of each phase place.Afterwards in navigation stage, will be real-time carry out continuously smart registration, the initial conversion matrix of smart registration is exactly Ti * T each time.For CT image before the art of some phase places, target is to seek a transition matrix that makes the optimum of similarity measure maximum, be designated as T ' i, can finally satisfy the error that causes owing to factors such as breathings, thereby satisfy the positioning requirements to target blood, the similarity measure that adopts in this process is normalized mutual information.

Normalized mutual information is defined as:

$\mathrm{NMI}(M,R)=\frac{H\left(M\right)+H\left(R\right)}{H(M,R)}---\left(5\right)$

Wherein, M is the in the preoperative gradation of image point set of CT up-sampling gained of two-dimensional ultrasonic image region, and 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-scale value of M image slices vegetarian refreshments,

Represent that the pixel gray-scale value is i in the M image _MProbability; H(R) be the Shannon entropy of R,

i _RRepresent the gray-scale value of R image slices vegetarian refreshments,

Represent that the pixel gray-scale value is i in the R image _RProbability; H(M, R) be the combination entropy of M and R, $H(M,R)=-\underset{i_M}{\mathrm{\Σ}}\underset{i_R}{\mathrm{\Σ}}{p}_{(i_M,i_R)}\log \left[p_{(i_M,i_R)}\right],$

Represent that the pixel gray-scale value is i in the M image _M, the pixel gray-scale value is i in the R image _RProbability.For M image subject to registration and R image, establish I _M(X _M), I _R(X _R) be respectively the gray scale function of M, R, X _M, X _RRepresent respectively the coordinate in M, the R image space, then X _R=T ' i * X _M, i.e. I _R(X _R)=I _R(T ' i * X _M).

Whole registration process is namely sought T ' i, so that NMI(M, R) maximum.Since Ti * T be by with art in the registration of two-dimensional ultrasonic image obtain, had less error, be a preferably initial optimizing position, therefore can converge to fast optimum solution, thereby obtain a final accurate transition matrix.

The 5th goes on foot, by external optical alignment true endoscope and virtual endoscopic images is merged.

As shown in Figure 5, described fusion refers to: by two width of cloth scenes are merged mutually with different transparencies, just formed one and strengthened reality environment, so that the real scene that the virtual scene that comprises the heartbeat model that virtual endoscope is seen and true endoscope are seen is consistent; The transition matrix Ti corresponding according to current heartbeat Selecting phasing from one group of transition matrix that step 4.3 obtains simultaneously, thereby set up the Mapping and Converting relation between image space and the real space, this Mapping and Converting relation is used for demonstrating in real time distance and the relative position relation between apparatus and the impact point the most at last, thereby guided surgery is accurately implemented.

As shown in Figure 3, described external optical alignment refers to adopt the NDI optical orientator to carry out the infrared reflection location, thereby realizes the three-dimensional localization to research object.This NDI optical guidance instrument comprises: light source 8 and receiver 9, registration tools 10.Wherein light source and receiver farthest coverage be 3000mm, maximum area can reach 1470 * 1856mm ²Registration tools is comprised of witch ball 11 parts and long registration syringe needle 12 parts of front end.The patient evenly attaches 6-8 metal marker point before carrying out CT scan, metal marker is put in the preoperative highlighted demonstration in the CT view data like this.The patient plows on operating table with same position when performing the operation, the life of delivering a child checkout equipment, after applying general anaesthetic, begin to utilize and in image, obtain the coordinate of metal marker point in image coordinate system, utilize simultaneously metal marker point that NDI optical guidance instrument obtains at the coordinate in true research object space, utilize the rigid registration algorithm to realize the registration of this coordinate system, thereby realize the registration in image space and surgical object space.

The endoscope that imports in real time by optical orientator and the positional information of mechanical arm, software can show the relative position of mechanical arm and model in real time in computer screen.In order to realize the fusion of true endoscope and virtual endoscope, need to try to achieve from world coordinates and be tied to the transition matrix of endoscope coordinate system and be tied to the matrix of endoscope projected coordinate system from the endoscope coordinate, namely need to carry out the demarcation of endoscope.After finishing endoscope and demarcating, the endoscope position by input towards etc. information, virtual endoscope is in full accord with the state of true endoscope in the realization.Thereby so that the real scene that the virtual scene that comprises the heartbeat model that virtual endoscope is seen and true endoscope are seen is consistent, by two width of cloth scenes are merged mutually with different transparencies, just formed one and strengthened reality environment, effectively guided surgery is accurately implemented.

Above-mentioned endoscope is demarcated: obtaining of endoscopic images is that a 3D scene is in the result of a 2D projection plane gained.

$\mathrm{\λ}\left[\begin{array}{c}u\\ v\\ 1\end{array}\right]=M_{\mathrm{int}}\×M_{\mathrm{ext}}\×\left[\begin{array}{c}x\\ y\\ z\\ 1\end{array}\right]---\left(6\right)$

Wherein, X=[x y z 1] ^TBe the homogeneous coordinate system of a point in the 3D scene, x, y, z represent respectively the some coordinate on x axle, y axle, z axle, [u v 1] ^TRepresented the coordinate of this point in the 2D projection plane, u, v represent respectively the some coordinate on x axle, y axle, and λ is the secondly zoom factor of coordinate system of 2D projection plane.Comprised a transition matrix M who is transformed into the endoscope coordinate system from world coordinate system in the following formula _ExtAnd transition matrix M from endoscope coordinate system conversion endoscope projected coordinate system _IntM _ExtBe 4 * 4 transition matrixes, can be expressed as:

$M_{\mathrm{ext}}=\left[\begin{array}{cccc}{r}_1& r_2& r_3& t_x\\ r_4& r_5& r_6& t_y\\ r_7& r_8& r_9& t_z\\ 0& 0& 0& 1\end{array}\right]---\left(7\right)$

Wherein, r ₁... r ₉Be twiddle factor, t _x, t _y, t _zBe translation vector.

M _IntCan be expressed as:

$M_{\mathrm{int}}=\left[\begin{array}{cccc}\mathrm{sf}& 0& u_0& 0\\ 0& f& v_0& 0\\ 0& 0& 1& 0\end{array}\right]---\left(8\right)$

Wherein, f is the distance that the camera lens focus arrives the minute surface center, and s is the ratio of width to height in the camera lens visual field, (u ₀, v ₀) be the coordinate of minute surface center under the 2D projected coordinate system.The demarcation of endoscope namely is to determine M _IntAnd M _ExtThese two matrixes.

Use this location technology can realize comparatively quickly and accurately that accuracy of the mean reaches about 3mm to target location coronarius, greatly reduced the time of seeking the pathology coronary artery in the robot assisted coronary bypass, make operation safer, efficient.

Before the art in CT image and the ultrasonic registration process of surgical object Real-time Two-dimensional because the error that the factors such as breathings, heartbeat cause is introduced the robot coronary artery bypass surgery with intraoperative ultrasound and is navigated, this is salient point of the present invention.

Its obtainable beneficial effect: 1. realized the optical guidance localization method based on CT registration results before intraoperative ultrasound and the art; 2. solved and rely on clinically the personal experience to locate bridging impact point problem all the time, so that operation technique is safer, efficient.

Claims (10)

1. optical guidance positioning system based on the CT registration results, it is characterized in that, comprise: the CT image imports module before the art, the image segmentation module, body surface initial registration module, navigation module in two-dimensional ultrasonic image registration module and the art in the front CT image of art and the art, wherein: the CT image imports module and receives the DICOM format-pattern file that the front imaging examination of art obtains before the art, the image bag also exports respectively image segmentation module and body surface initial registration module to before generating art, before the art in CT image and the art two-dimensional ultrasonic image registration module link to each other with body surface initial registration module with the image segmentation module respectively and receive with the Three-Dimensional Dynamic coronary artery of surgical target point and set and body surface initial registration result, the front CT transition matrix of two-dimensional ultrasound registration module output art navigation module to the art in CT image and the art is exported accurate surgical navigational information by navigation module in the art before the art;

The image bag comprises before the described art: CT image and the 3 D stereoscopic image that obtains according to the DICOM data reconstruction before the art of some width of cloth DIOM forms of one or several cardiac cycles the target area in, wherein: the front CT view data of the art of a cardiac cycle by the art of several phase places before image form;

Described Three-Dimensional Dynamic coronary artery tree with surgical target point comprises: by dynamic heart model and the vascular tree model of image segmentation module generation.

2. system according to claim 1, it is characterized in that, CT image importing module comprises before the described art: the DICOM image reads in unit and the front image three-dimensional drawing unit of art, wherein: the DICOM image read in the unit with art before a series of DICOM format-pattern files of one cardiac cycle of heart of obtaining of imaging examination import parse the DICOM data and transfer to art before the image three-dimensional drawing unit, image bag and export respectively image segmentation module and body surface initial registration module to before the image three-dimensional drawing unit becomes 3 D stereoscopic image and merges into art according to the DICOM data reconstruction before the art.

3. system according to claim 1, it is characterized in that, described image segmentation module comprises: cardiac segmentation unit and vascular tree cutting unit, wherein: the cardiac segmentation unit to art before in the image bag before the heart art in the cardiac cycle CT image manually cut apart frame by frame based on clinical experience, and rebuild according to CT image segmentation result before the heart art that obtains and to obtain the dynamic heart model; The vascular tree cutting unit to art before in the image bag before the vascular tree art in the cardiac cycle CT image manually cut apart frame by frame based on clinical experience, and rebuild according to CT image segmentation result before the vascular tree art that obtains and to obtain the vascular tree model and to export before the art two-dimensional ultrasonic image registration module in the CT image and art to as setting with the Three-Dimensional Dynamic coronary artery of surgical target point in the lump with the dynamic heart model.

4. system according to claim 1, it is characterized in that, described body surface initial registration module comprises: register mark point selection unit and transform matrix calculations unit, wherein: the register mark point on the surgical object body surface is selected to be arranged in register mark point selection unit, and obtain its in the preoperative the CT view data in image coordinate system coordinate and volume coordinate, the transform matrix calculations unit is according to image coordinate system coordinate and volume coordinate, utilize the rigid registration algorithm to realize registration between these two coordinates, namely realize the registration in image space and surgical object space, and utilize these two groups of coordinate Calculation to go out the initial conversion matrix as body surface initial registration result, export two-dimensional ultrasonic image registration module in the front CT image of art and the art to.

5. system according to claim 1, it is characterized in that, the two-dimensional ultrasonic image registration module comprises in the front CT image of described art and the art: two-dimensional ultrasonic image reads in the unit in the art, CT image registration unit and cardiogram read in the unit before two-dimensional ultrasonic image and the art, wherein: CT image registration unit receives the intraoperative ultrasound image and reads in the two-dimensional ultrasound input picture of unit collection and the current heartbeat phase place that cardiogram reads in the unit collection before two-dimensional ultrasonic image and the art, to mate with corresponding several phase places and current heartbeat phase place in the Three-Dimensional Dynamic coronary artery tree of surgical target point by the input ECG signal, calculate the front CT image transitions matrix of art between the two and export navigation module in the art to, corresponding phase during wherein each is organized the corresponding a kind of current heartbeat phase place of the front CT transition matrix of art and sets with the Three-Dimensional Dynamic coronary artery of surgical target point.

6. system according to claim 1, it is characterized in that, navigation module comprises in the described art: transition matrix selected cell and navigation display unit, wherein: the transition matrix selected cell according to current heartbeat Selecting phasing art before corresponding one group of transition matrix in the CT transition matrix, and export one group of transition matrix of gained to the navigation display unit, calculate distance and the relative position relation between current virtual apparatus and the actual target point and carry out the video demonstration by the navigation display unit, thereby guided surgery is accurately finished.

7. the air navigation aid according to the described system of above-mentioned arbitrary claim is characterized in that, may further comprise the steps:

After the front CT view data of art of cardiac segmentation unit in the first step, the image segmentation module by a series of DICOM forms of front CT image importing module acquisition one cardiac cycle of target area of art, CT image before the heart art in this cardiac cycle is manually sketched the contours of cardiac silhouette frame by frame based on clinical experience, obtain segmentation result, and reconstruction obtains the dynamic heart model according to segmentation result;

Vascular tree cutting unit in second step, the image segmentation module to art before in the image bag before the vascular tree art in the cardiac cycle CT image manually sketch the contours of frame by frame the vascular tree profile and manually mark the surgical target point according to clinical demand based on clinical experience, obtain segmentation result, and according to segmentation result reconstruction acquisition vascular tree model, construct a three-dimensional virtual scene that comprises the dynamic vascular tree-model, this three-dimensional virtual scene has consisted of a virtual endoscopic images;

The CT view data was at the transition matrix of image coordinate system coordinate with the volume coordinate of true surgical object before the 3rd step, body surface initial registration module utilized the body surface registration to obtain art, i.e. coordinate points under two different spaces coordinate systems, by mutually mapping and realize one-to-one relationship of feature, reach the corresponding of CT image and surgical object Real-time Two-dimensional ultrasonoscopy before the final art;

The 4th goes on foot, will be complementary with corresponding several phase places and current heartbeat phase place in the Three-Dimensional Dynamic coronary artery tree of surgical target point by the input ECG signal;

The 5th goes on foot, by external optical alignment true endoscope and virtual endoscopic images is merged, namely by two width of cloth scenes are merged mutually with different transparencies, just formed one and strengthened reality environment, so that the real scene that the virtual scene that comprises the heartbeat model that virtual endoscope is seen and true endoscope are seen is consistent; The transition matrix Ti corresponding according to current heartbeat Selecting phasing from one group of transition matrix that step 4.3 obtains simultaneously, thereby set up the Mapping and Converting relation between image space and the real space, this Mapping and Converting relation is used for demonstrating in real time distance and the relative position relation between apparatus and the impact point the most at last, thereby guided surgery is accurately implemented.

8. method according to claim 7, it is characterized in that, described the 3rd step is specifically by choosing several register mark points in the CT image in the preoperative, find the point corresponding with register mark point in the image and utilize the optical guidance instrument to obtain these at the coordinate of real space in real space, utilize this two groups of different coordinates, but the position coordinates of point set is tried to achieve two transition matrixes between the space one to one.

9. method according to claim 7, it is characterized in that, described the 4th the step specifically refer to: with transition matrix T as the initial conversion matrix, if Ti is for the transition matrix that obtains behind the CT correct image before the art of each phase place on the basis of transition matrix T, namely for CT image i before the art of a phase place, Ti * T can further proofread and correct the registration error between this phase place CT image and the surgical object Real-time Two-dimensional ultrasonoscopy, and wherein N is the phase place number of image before the art in the CT view data before the art of a cardiac cycle.

10. method according to claim 7 is characterized in that, described the 4th step comprises:

4.1) by ultrasonic probe heart is gathered the surgical object realtime graphic of a series of surgical objects, wherein the surgical object realtime graphic that collects of each width of cloth all before corresponding some corresponding arts before the art of CT image, each phase place the CT image corresponding a series of two-dimensional ultrasonic image, for CT view data before the art of phase place i, extract the surface profile of wall of the heart;

4.2) corresponding each width of cloth two-dimensional ultrasonic image of phase place i extracted the profile of wall of the heart, the inwall that extracts from this a series of two-dimensional ultrasonic image has formed one group of point set, extract the unique point point set from this point is concentrated, the wall of the heart surface profile that extracts the CT image before art is another group point set;

4.3) by iterative closest point algorithms the point set on the two-dimensional ultrasonic image is registrated to the point set on the CT image before the art, obtain the transition matrix Ti between these two point sets, be used for the initial matrix of follow-up smart registration process.

Images