CN108648548A - A kind of neuro-surgery virtual operation training system - Google Patents

A kind of neuro-surgery virtual operation training system Download PDF

Info

Publication number
CN108648548A
CN108648548A CN201810352058.0A CN201810352058A CN108648548A CN 108648548 A CN108648548 A CN 108648548A CN 201810352058 A CN201810352058 A CN 201810352058A CN 108648548 A CN108648548 A CN 108648548A
Authority
CN
China
Prior art keywords
point
particle
model
module
coordinate
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201810352058.0A
Other languages
Chinese (zh)
Inventor
冯远静
张驰
谭志豪
朱珂权
曾庆润
曹喆文
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University of Technology ZJUT
Original Assignee
Zhejiang University of Technology ZJUT
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhejiang University of Technology ZJUT filed Critical Zhejiang University of Technology ZJUT
Priority to CN201810352058.0A priority Critical patent/CN108648548A/en
Publication of CN108648548A publication Critical patent/CN108648548A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G09EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
    • G09BEDUCATIONAL OR DEMONSTRATION APPLIANCES; APPLIANCES FOR TEACHING, OR COMMUNICATING WITH, THE BLIND, DEAF OR MUTE; MODELS; PLANETARIA; GLOBES; MAPS; DIAGRAMS
    • G09B9/00Simulators for teaching or training purposes

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Educational Administration (AREA)
  • Educational Technology (AREA)
  • General Physics & Mathematics (AREA)
  • Apparatus For Radiation Diagnosis (AREA)

Abstract

A kind of neuro-surgery virtual operation training system, using Geomagic Touch force feedback equipments as surgical hardware operating platform, the force feedback situation after simulating surgical instrument variation in space by six direction steering engine of equipment and being contacted with brain tissue;Pass through the three-dimensional modeling of the true brain image data such as CT, MRI, on Unity3D platforms design realize match with force feedback equipment, and reality brain similar in surgical scene, the system comprises for realizing the tissue cutting module of surgical procedure, force feedback module, the supplementary module for the visualization model for visual effect display of performing the operation and for simulating operation common tool.The present invention can carry out preferable real-time tracking to surgical procedure, obtain more true, accurate touch feedback and visual feedback.Processing speed by GPU accelerate to obtain it is very big must improve, can be trained for cerebral surgery operation provide stablize, Training scene efficiently, repeatable.

Description

A kind of neuro-surgery virtual operation training system
Technical field
The present invention relates to automatic technology, information technology computer graphics and medical imaging, Nervous System Anatomy field, Especially a kind of neuro-surgery virtual operation training system.
Background technology
With the progress of modern medicine condition, people's health requirement is higher and higher.In all kinds of medical procedures, brain god Demand surgically is also increasing.But neuro-surgery operation is that current operation difficulty highest, required precision are most stringent One of surgical operation, the brain for being used as training material has the characteristics that not re-usable, and quantity lacks, thus most of The training of neuro-surgery doctor depends on the imagination of trained doctor and instructs the level of education of doctor.Trained lacks The weary training difficulty for leading to qualified brain surgery doctor is very big, has no small limitation.
The development of computer technology so that virtual reality technology gradually provides possibility for neuro-surgery operative training. Virtual operation training system can combining information technology, automatic technology, computer graphics, computer vision, computer graphic As the mature technology of the multiple directions such as processing, is rendered using three-dimensional modeling, visualization and show the similar brain tissue structure with reality And symptom, the true vision of doctor, touch feedback are given using soft tissue deformation technology and force feedback equipment, from multiple sides Face is that doctor reappears surgical scene true to nature.By building Training scene appropriate, the scoring of design standard, Yi Shike Operation experiment is repeated and finds deficiency, then Effective Way by Using Project is horizontal, while also solving tradition and training and can not weigh The problem of utilizing training resource again.
Neuro-surgery operation require doctor can in surgical procedure can timely and effectively handle art the problem, because The training of this virtual operation is an instant, dynamic process.Most of neuro-surgery virtual operation training systems is at present When property is poor, in training the operation of trainer cannot accurately feed back.For example, due to brain soft tissue deformation than human body its His position is increasingly complex, and computer is influenced by program operational efficiency, and the operation of trainer can not be handled and be shown immediately Into visualization interface;Or since complicated stressing conditions cause brain soft tissue stress model to establish difficulty so that operation Haptic feedback effect it is undesirable etc..These problems lead to operative training, and there are larger gap, nothings with the performance in actual operation Method tempers the ability that surgical doctor is handled when participating in the cintest, has obscured training goal, has reduced training effectiveness, has been unfavorable for doctor's technical merit Promotion.
Invention content
To overcome the above problem present in existing neuro-surgery virtual operation training system, the present invention to devise one kind Realize the neuro-surgery virtual operation training system of instant, stable, true visualization and accurate touch feedback.
The technical solution adopted by the present invention to solve the technical problems is:
A kind of neuro-surgery virtual operation training system include for realizing surgical procedure effect tissue cutting module, Force feedback matching module, the auxiliary for the visualization model for visual effect display of performing the operation and for simulating operation common tool Module;
The tissue cutting module includes tissue deformation module based on internal point domain set domain restricted model and for performing the operation The real-time rendering module of cutting;
The tissue deformation module uses a kind of tissue deformation model based on internal point collection region constraint, in mass spring mould Internal point collection region constraint is added on the basis of type shows the deformation of soft tissue for simulating the active force inside brain soft tissue;
Plane S indicates that brain tissue model surface, plane T indicate that organization internal stress surface, plane T are parallel to plane S, plane Point set P on S projects to the correspondence point set P' on plane T, restricted spring be point point corresponding with plane T on plane S it Between virtual spring, be used for analog force transmission;
As a part of point P on brain tissue model surfaceiWhen being deformed upon by outer force effect, each point passes through restricted spring Point P' into corresponding internal point set P'iStress is transmitted, so that it is generated strain, and by internal strain distribution function, by P'i The power that point is subject to passes to the consecutive points in internal point set;The deformation effect of these points is on restricted spring, by internal stress Feedback of status is to the corresponding points on plane S;After removing the external force being applied on plane S, the point of point set P no longer calculates adaptability to changes Size and distribution, the point of plane S gradually return to initial position under the action of rigid core, and the point in internal point set P' also by The effect of restricted spring, follows P to be restored to initial position;
Designing rigid core constraint, it is ensured that the point on surface model can be restored to initial position after deformation, when initialization, Rigid core records the initial position of each particle, and establishes virtual spring between corresponding 2 points;When external force F acts on particle When, particle is subjected to displacement, and surrounding particle also generates displacement by spring-force driven dual;Rigid core is by virtual spring to occurring The particle application one of displacement limits it and leaves the power of initial position;After external force F is removed, particle can be restored to initial position;
One update particle domain of setting, dynamically to store the particle that will carry out deformation calculation in iterative process, not in domain Interior particle is not involved in operation, realizes as follows:
As a certain particle P of plane SiWhen by outer force effect, which is stored in update particle domain, and according to pre-processing The information arrived scans for the field of the point by certain search rule to obtain adjacent particle, these particles is also stored in update In particle domain;The adaptability to changes that the particle and field particle generate passes to the inside point set P' of plane T by restricted spring, makes It also obtains corresponding update;Given threshold h will table corresponding with the point when the offset that certain in internal point set P' is put is more than h In the particle deposit update particle domain of face, the attribute that holding is updated by the particle of deformation more neofield is stored in a deformation process Until this deformation restores.
The real-time rendering module uses a kind of real-time rendering algorithms for surgical cut, is operated for scalpel tool When computation organization cutting after newly-generated model vertices texture coordinate, and carry out notch and redraw and render, comprise the steps of:
(1) when scalpel is with brain tissue model collision, point of a knife stress intensity is persistently obtained, is recorded when more than a certain threshold value Contact point is feed coordinate, and the threshold value indicates that tissue surface ruptures dynamics, and when collision releasing and surface stress is less than threshold value When, it is considered as cutting and terminates, it is coordinate of returning sword to record the point;
(2) after determining access site, the path point of all processes on dynamic access cutting path, path point is that collision is passed through Tri patch vertex;
(3) using feed coordinate as starting point, point of a knife is that end point draws notch using Bezier;
(4) the tri patch vertex that notch depth and newly-generated surface are calculated according to the point of a knife stress size of acquisition is sat Mark, these coordinates is compared with original surface coordinate, if new vertex inside the tri patch of script, records these triangles Dough sheet;If not in tri patch, the relationship on the vertex and field Triangle ID dough sheet is calculated;
(5) according to above-mentioned two situations, the operation deleted tri patch or add tri patch is executed respectively, and newly to push up Point is that texture coordinate carries out texture mapping;
(6) amount of force and direction, adjustment notch generated according to scalpel and the collision of surface tri patch is rebuild Width, depth and angle.
The force feedback module includes collision detection module and base based on AABB bounding volume hierarchy (BVH)s Yu Mesh Collider In the Hardware match module of Geomagic Touch force feedback equipments;
The collision detection module being based on AABB (Aligned Axis Bounding Box) bounding volume hierarchy (BVH) using a kind of With the collision detection algorithm of Mesh Collider, for carrying out collision detection to operation tool, AABB bounding boxs are cuboid, It surrounds geometric object and each side is parallel to three dimensional coordinate space, vertex is with geometric object local coordinate on three dimensional space coordinate axis The most value expression of projection, i.e.,:AABB{(xmin,ymin,zmin),(xmax,ymax,zmax), include the following steps:
Step 1 establishes bounding volume hierarchy (BVH) tree, and it is similar to bounding box to improve operation tool by multilayer bounding box nesting Degree increases collision detection precision, using top-down approach, by the primitive geometric element set S of given node VvIt is divided into two A disjoint subset, until all basic elements in subset:
(1) the longest axis method is used to determine the split axis for dividing primitive geometric element;
(2) median method is used to determine the split point for dividing primitive geometric element:In the geometry for calculating each basic element Then heart point calculates the intermediate value that they are projected in split axis;
(3) splitting plane is obtained according to split axis and split point, the element in space is divided into two sons using splitting plane Collection;
(4) consider that a part of special geometric element may be located just on splitting plane, it can be according in these elements Space residing for heart point is allocated, and if central point is just on splitting plane, then distributes to the less subset of element;
Bounding box is applied to operation tool by step 2
The action of operation tool includes three-dimensional rotation, translation totally 6 degree of freedom, defines operation tool object translation vector T =(x, y, z)T, around x, y, z-axis rotation angle isCorrespondent transform matrix is:
The spin matrix of three axis is synthesized into R:
The state transformation of tool is described with x → Rx+T, when user is by external equipment operation tool, is surrounded Each coordinate of box will be converted by the transformation for mula, you can realized the update of bounding box, ensured the phase of bounding box and operation tool It is constant to position;In addition, traversal bounding volumetree carry out Intersection detection process in, on each time sampling point only detection and more The descendent node of new crossed node;
Step 3 will be bound by operation tool and Mesh Collider
Whether detection tissue model contacts with bounding box, if the two contacts, obtains contact point list and establishes contact point Set C={ c0c1...cn-1cn, using recursion method, contact point is successively detected whether in bounding box, if contact point c (x, y, Z), AABB { (xmin,ymin,zmin),(xmax,ymax,zmax), it needs to meet
Just it is considered as in bounding box, when contact point is not in bounding box, eliminates set C, ifEnd is passed Return;Otherwise recurrence is to bounding volumetree end leaf node, using recurrence after the set C that obtains as finally determining contact point Set is used for the processing of follow-up collision effect;
Hardware match module based on Geomagic Touch force feedback equipments refers to that Geomagic Touch force feedbacks are set The standby data interactive method with operative training platform, matching step are as follows:
(1) communication initialization of PC machine and the force feedback equipment is realized by TCP/IP Handshake Protocols, while statement is set Standby name character string, the basic function of initialization apparatus, including create current device context starter, opening force output, It opens dynamic model, open equipment thread;
(2) working space size is arranged, respectively in the visual space in the tactile physical space and PC machine on initialization apparatus Establish coordinates matrix;
(3) by tactile physical space coordinates system and visual space coordinate system matching, it is that work is empty that equipment, which gives tacit consent to cursor position, Between center;
(4) agent point position when cursor contact model in the corresponding scene of mechanical arm tail end, and cursor at this time are obtained Location information is fed back to the tactile physical space of equipment by position.
(5) the power generating algorithm for using GeoMagic Touch equipment, is counted according to agent point position and haptic apparatus position It calculates reaction force and feeds back to user by six direction steering engine simulated resistances of equipment;
The visualization model includes that brain data three-dimensional modeling module and smoothed particle method bleed and simulate mould Block.
The brain data three-dimensional modeling includes data prediction, registration, three-dimensional reconstruction, grid correspondence, and steps are as follows:
(1) anisotropy and non local average filter method is used to carry out image filter to CT, MRI and the fMRI data of input Wave;
(2) when input data is CT, MRI, using Rigid Registration, two images are aligned by rotation, translation, scaling Space coordinate;When input data is fMRI, MRI, to fMRI into row interpolation and resampling, obtain identical with MRI resolution ratio Image reuses the method for registering registration based on mutual information;
(3) registration data is made by three-dimensional reconstruction by VTK and exported, textures and rendering are carried out using Maya.
The analog module of bleeding is bled simulation algorithm using a kind of smoothed particle method, including particle configuration, Resurfacing and GPU parallelizations handle three parts;
Particle configuration includes particle property setting and particle search, particle property can not press SPH algorithms to obtain by Predictor Corrector , its step are as follows:
(1) respond cutting event, create particle flux, according to cutting event transmit information assign all particle initial positions, Initial velocity, acceleration, pressure p suffered by particleiWith pressure FiIt is determined by cutting stress situation;
(2) according to the current pressure F of particlei, calculate the acceleration a at lower a momenti, speed viWith position xiAnd the position The particle density ρ setiWith neighborhood density contrast Δ ρ;
(3) according to neighborhood density contrast to pressure piIt is modified, calculates new pressure Fi′;
(4) (2) are repeated, according to pressure FiEvery attribute of ' calculating next step particle;
(5) when Particles Moving is beyond limiting spatial dimension or colliding with cotton piece, particle physics attribute is reset simultaneously Delete the index of the particle;
Particle search traverses particle using nearest neighbor method, and interim mesh generation, whole size are built in aiming field Equal to the size of support region, inside includes the equal sub-grid of many length, width and height;When the nearest neighbor particle for searching for some particle When, the adjacent mesh of grid and it where need to only traversing the particle;In the grid in Construct question region, boundary is leaned in search Nearest particle finds out minimum and maximum position coordinates, the point using these coordinates as boundary reference in x, y, z on tri- directions Build grid;
Resurfacing uses Marching Cubes algorithms, goes out cube near liquid surface by calculating isosurface extraction Body finds out the vertex position for the tri patch for needing to draw by linear interpolation method;Cloud will all be put by often carrying out a step SPH simulations It is converted into colorable surface, while the triangular mesh on surface is finely divided using PN deltic methods, is obtained more thin Close triangular mesh;
GPU parallelizations processing corresponds to the thinking of a particle using a thread, orderly spatial index is established, using simultaneously Row sort algorithm accelerates the process searched for recently every time, for the particle array A being loaded into video memoryp, establish spatial index Steps are as follows:
(1) space is divided into uniform grid, a call number, the space bit of call number and grid is specified for each grid Set correlation;
(2) grid and manipulative indexing number belonging to it are calculated according to each particle position, by particle in ApIn subscript set In first half, by call number be placed in it is later half be spliced into a new subscript, storage obtains array of indexes AIdx
(3) to array of indexes AIdxCarry out multithreading radix sorting.The A of all particles after sequence in same gridp Subscript will be continuously stored in video memory;
(4) judge A after sortingIdxEach particle in array whether be a grid beginning or end, work as by comparing Whether the subscript high position of a subscript high position for preceding particle and a front or rear particle unanimously judges, inconsistent, illustrates current particle It is the beginning or end of a grid, by current particle in array AIdxIn subscript protect to a record space and divide size In 3D texture arrays.
The supplementary module includes three kinds of scalpel, bipolar electrocoagulator and cotton piece operation tools;
Scalpel is provided with rigid body, quality, skin friction resistance Fd, surgical cut power critical value m attributes, and add aforementioned AABB bounding boxs and Mesh Collider collision detection algorithms, the notch that scalpel can initialize Bezier control point are wide Spend the number num of coefficient W and composition point;When scalpel works, using the real-time rendering algorithms of aforementioned surgical cut to notch It is drawn, collision detection provides contact point coordinates and cutting force degree information for Rendering algorithms;
Bipolar electrocoagulator shows cauterization effect of the electric coagulating apparatus to tissue by changing pixel color and the texture of image, and It is added to AABB bounding boxs and Mesh Collider collision detection algorithms, realizes that steps are as follows:
(1) detection electric coagulating apparatus and tissue model collision situation traverse tri patch around the point of impingement after colliding generation, Judge the point of impingement whether in model tri patch;
(2) if in tri patch, contact point space coordinate, texture coordinate and the pixel value when collision are calculated;
(3) texture pixel is read, texture mapping is replaced, using the pixel as the center of circle, is drawn using Bresenham algorithms solid Circle is as the effect after electric coagulating apparatus calcination;
Cotton piece devises the double-deck collision detection, act as sucking the particle that simulation is bled, and realizes that steps are as follows:
(1) first layer, i.e. outer layer use the OnParticleCollision methods of ParticleSystem in Unity3D, Box Collider detection particles are added in cotton piece and whether cotton piece collides;
(2) second layer, i.e. internal layer add single layer AABB bounding box collision detection algorithms;
(3) particle coordinate is world coordinates, and cotton piece coordinate is local coordinate, therefore with transform interfaces in Unity3D Transform.TransformPoint () method the local coordinate of cotton piece is converted into world coordinates, then calculate particle seat Mark and bounding box whether geometrical intersection;
(4) particle intersected makees extinction processing, and it is red that the position of collision of cotton piece, which changes pixel color,.
The invention adopts the above technical scheme, which has the following advantages:Surgical procedure can be carried out preferable real When track, obtain more true, accurate touch feedback and visual feedback.Processing speed accelerates to obtain so very big that carry by GPU It is high.The Training scene stablizing, is efficient, repeating can be provided for neuro-surgery operative training.
Description of the drawings
Fig. 1 is the implementation method of each module of this programme.
Fig. 2 is internal point domain set domain restricted model schematic diagram.Wherein plane S indicates that model surface, plane T are parallel to surface S, Point set P in S planes projects to the corresponding internal point set P' in T planes.
Fig. 3 is rigid core fundamental diagram.N in figure1,n2,n3It is particle in soft tissue surfaces, n1′,n′2,n3' it is internal Correspond to n in point set plane1,n2,n3Particle.Six points record corresponding fixed coordinates position before the collision.
Fig. 4 is the schematic diagram of dynamic deformation model and the internal point domain set domain rule of correspondence.Plane S is soft tissue surfaces in figure, Plane T is internal point set plane, and h is the particle offset threshold for updating particle domain.When particle deviant is more than h in plane T, Update particle domain is added in the particle.
Fig. 5 is surface cutting rendering effect schematic diagram.A in figure0,a1,a2,a3,a4It is scalpel cutting track, b0,b1,b2, b3,b4,b5,b6,b7It is the edge vertices that surface cutting Vee formation dough sheet is redrawn.
Fig. 6 is the provisional grid schematic diagram built on the basis of problem area central point.Provisional grid entirety size in figure Equal to the size of support region, inside includes the equal sub-grid of many length, width and height.When building grid, problem area central point corresponds to Grid element center point, x, y, the minimum and maximum position coordinates on tri- directions z are as boundary reference point.
Fig. 7 is the relational graph of operation tool and agent location and haptic position after soft tissue collision.
Fig. 8 is the unrelated trilinear interpolation voxel cell model in direction.
Specific implementation process
The invention will be further described below.
Virtual operation training system referring to Fig.1~Fig. 8, a kind of neuro-surgery virtual operation training system, use Geomagic Touch force feedback equipments simulate surgical instrument as surgical hardware operating platform, by six direction steering engine of equipment and exist Variation in space and the force feedback situation after being contacted with brain tissue;Pass through the three-dimensional of the true brain image data such as CT, MRI Modeling, on Unity3D platforms design realize match with force feedback equipment, and reality brain similar in surgical scene, the system System includes the visualization for visual effect display of performing the operation for realizing the tissue cutting module of surgical procedure, force feedback module Module and for simulate operation common tool supplementary module.
The implementing procedure of this programme is:Image filtering, three-dimensional reconstruction are carried out to the brain data that outside imports, model is led Enter scene, mesh generation is carried out to it.After the completion of pretreatment, the image of brain is shown in the display, and user is prompted to pass through Force feedback equipment operates operation tool.Operation tool and tissue model surface carry out real-time collision during surgical simulation Detection.When a collision occurs, according to the dynamic change of tissue deformation model template surface notch and internal distortions power, and to operation Tool feedback force.The power of feedback is exported by equipment, passes to user.It is painted using simulation algorithm of bleeding while drawing notch Effect of bleeding is made, and realizes notch and bleed particle grain textures and rendering, is updated in visualization window.
The tissue cutting module includes tissue deformation module based on internal point domain set domain restricted model and for performing the operation The real-time rendering module of cutting;
The tissue deformation module uses a kind of tissue deformation model based on internal point collection region constraint, including following step Suddenly:
Step 1 obtains brain soft tissue surfaces mesh coordinate, and rigid core is added for all tri patch vertex
Rigid core records the initial position of each particle, and establishes virtual spring between corresponding 2 points, can simulate The process of deformation recovery after stress disappears limits it to the particle application one being subjected to displacement by virtual spring and leaves initial bit The power set is as constraint, it is ensured that the point on surface model can be restored to initial position after deformation;
Step 2 establishes the kinetics equation of Mass-spring Model, solves equation initial value
If the network of the Mass-spring Model of soft tissue is by being connected between the abstract particle after n discretization and particle Virtual spring damper constitute, it is discrete after particle Ni(i=1,2...n) is indicated, the spring-damper connected between particle Use lij(i, j ∈ [1, n], i ≠ j) is indicated.Arbitrary particle NiThe resultant force f born is by internal force finWith external force fexTwo parts form, Wherein internal force be by with particle NiConnected spring-damper lijThe spring force f of generationsWith damping force fdComposition, has:
miai+fd+fs=fex
I.e.
M in formulaiFor particle NiQuality, kijFor particle NiAnd NjBetween be connected spring coefficient of elasticity, dijFor particle NiWith NjBetween damper damped coefficient, P (i) be and particle NiThe set of adjacent all particles, xiAnd xjRespectively NiAnd Nj's Displacement,For particle NiAnd NjBetween be connected spring initial length.Since the displacement x of particle is unknown, the speed v of particle is introduced This intermediate quantity converts differential equation of higher order to One first-order ordinary differential equation and solves, the moving displacement and speed of each particle Degree is expressed as
Elastic deformation model is substituted the above to, the initial value of each attribute of particle when deformation can be obtained in equation is solved.
Step 3, the Strain Distribution function of selected point domain set domain
Accurate internal point domain set domain Strain Distribution function is selected, carries out simplifying processing, chooses exponential function Strain Distribution inside simulated tissue.Connecting inner point domain set domain establish expression formula be:
H in formulaiIndicate deviating for the point i in deformation process with surface in by corresponding domain set domain of the particle of outer force effect Amount, hjIndicate the offset of point domain set domain midpoint j, dijFor the distance of domain set domain midpoint j to point i, a is the exponential function truth of a matter, and s is contracting Put the factor.The motion state that each particle is constantly updated using the Strain Distribution function, can simulate brain tissue after opening cranium to The power of outer expansion and the gather effect after being squeezed by operation tool.
Step 4 establishes dynamic deformation model
By the way that a update particle domain is arranged, dynamic stores the particle that deformation calculation will be carried out in iterative process.Work as table The a certain particle P in face (plane S)iWhen by outer force effect, which is stored in update particle domain, and the letter obtained according to pretreatment Adjacent particle is also stored in update particle domain by breath by certain search rule.Given threshold h is put when certain in internal point set P' When offset is more than h, surface particle deposit corresponding with the point is updated in particle domain.It is stored in deformation in a deformation process The particle of more neofield will be retained in more neofield until this deformation restores.
The real-time rendering module uses a kind of real-time rendering algorithms for surgical cut, comprises the steps of:
(1) when scalpel is collided with tissue model, point of a knife stress intensity is persistently obtained, records and connects when more than a certain threshold value Contact is feed coordinate, and the threshold value is the threshold value that tissue surface ruptures dynamics, and when collision releasing and surface stress is less than threshold When value, it is considered as cutting and terminates, it is coordinate of returning sword to record the point;
(2) after determining access site, the path point of all processes on dynamic access cutting path, path point is that collision is passed through Tri patch vertex;
(3) using feed coordinate as starting point, point of a knife is that end point draws notch using Cubic kolmogorov's differential system;
(4) the tri patch vertex that notch depth and newly-generated surface are calculated according to the point of a knife stress size of acquisition is sat These coordinates and original surface coordinate are carried out traversal comparison, if new vertex records these inside the tri patch of script by mark Tri patch;If not in tri patch, the relationship on the vertex and field Triangle ID dough sheet is calculated;
(5) according to above-mentioned two situations, the operation deleted tri patch or add tri patch is executed respectively, and newly to push up Point is that texture coordinate carries out texture mapping;
(6) amount of force and direction, adjustment notch generated according to scalpel and the collision of surface tri patch is rebuild Width, depth and angle.
The force feedback module includes collision detection module and base based on AABB bounding volume hierarchy (BVH)s Yu Mesh Collider In the Hardware match module of Geomagic Touch force feedback equipments;
The collision detection module being based on AABB (Aligned Axis Bounding Box) bounding volume hierarchy (BVH) using a kind of With the collision detection algorithm of Mesh Collider, comprise the steps of:
Step 1 establishes bounding volume hierarchy (BVH) tree to AABB bounding boxs, by multilayer bounding box nesting improve operation tool with The similarity of bounding box increases collision detection precision.Using top-down approach, by the primitive geometric element collection of given node V Close SvTwo disjoint subsets are divided into, until all basic elements in subset:
(1) the longest axis method is used to determine the split axis for dividing primitive geometric element.
(2) median method is used to determine the split point for dividing primitive geometric element:In the geometry for calculating each basic element Then heart point calculates the intermediate value that they are projected in split axis.
(3) splitting plane is obtained according to split axis and split point, the element in space is divided into two sons using splitting plane Collection.
(4) consider that a part of special geometric element may be located just on splitting plane, it can be according in these elements Space residing for heart point is allocated.If central point is just on splitting plane, then the less subset of element is distributed to.
Bounding box is applied to operation tool by step 2
The general three-dimensional rotation for considering operation tool translates totally 6 degree of freedom.If operation tool object translation vector T= (x,y,z)T, around x, y, z-axis rotation angle isCorrespondent transform matrix is:
Three is synthesized into a spin matrix R:
The state transformation of tool is described with x → Rx+T.Bounding box is converted by the transformation for mula, you can realize and surround The update of box.In addition, in traversal bounding volumetree carries out Intersection detection process, only detects and update on each time sampling point The descendent node of crossed node does not update entire bounding volumetree, greatly improves renewal speed.
Step 3 binds operation tool and Mesh Collider
Whether detection tissue model contacts with bounding box.If the two contacts, obtains contact point list and establish contact point Set C={ c0c1...cn-1cn}.Recursion method is reused, successively detects contact point whether in bounding box.If contact point c (x, Y, z), AABB { (xmin,ymin,zmin),(xmax,ymax,zmax), it needs to meet
Just it is considered as in bounding box.When contact point is not in bounding box, set C is eliminated, ifTerminate recurrence. Otherwise recurrence is to bounding volumetree end leaf node.The set C obtained after using recurrence is as finally determining contact point set It closes, is used for the processing of follow-up collision effect.
Hardware match scheme based on Geomagic Touch force feedback equipments refers to that Geomagic Touch force feedbacks are set The standby data interaction with operative training platform, matching step are as follows:
(1) communication initialization of PC machine and the force feedback equipment is realized by TCP/IP Handshake Protocols, while statement is set Standby name character string.The basic function of initialization apparatus, including create current device context starter, opening force output, Open dynamic model, open equipment thread (multiple devices work at the same time) etc..
(2) working space size is arranged, respectively in the visual space in the tactile physical space and PC machine on initialization apparatus Establish coordinates matrix.
(3) by tactile physical space coordinates system and visual space coordinate system matching, it is that work is empty that equipment, which gives tacit consent to cursor position, Between center.
(4) agent point position when cursor contact model in the corresponding scene of mechanical arm tail end, and cursor at this time are obtained Location information is fed back to the tactile physical space of equipment by position.
(5) the power generating algorithm for using GeoMagic Touch equipment, is counted according to agent point position and haptic apparatus position It calculates reaction force and feeds back to user by six direction steering engine simulated resistances of equipment.
Simulation that the visualization model includes brain data three-dimensional modeling module and smoothed particle method is bled Module.
The brain data three-dimensional modeling module includes data prediction, registration, three-dimensional reconstruction, grid correspondence etc., step It is as follows:
(1) anisotropy and non local average filter method is used to carry out image filter to CT, MRI and the fMRI data of input Wave reduces the influence of noise on image quality.
(2) when input data is CT, MRI, using Rigid Registration, two images are aligned by rotation, translation, scaling Space coordinate;When input data is fMRI, MRI, to fMRI into row interpolation and resampling, obtain identical with MRI resolution ratio Image reuses the method for registering registration based on mutual information.
(3) registration data is made by three-dimensional reconstruction by VTK and exported, textures and rendering are carried out using Maya.
The analog module of bleeding is bled simulation algorithm using a kind of smoothed particle method accelerated based on GPU, It is characterized in that establishing contour surface using marching cubes algorithm, and subdivision surfaces is carried out using PN deltic methods, use is bilateral Smoothing processing is made in normal filtering, and improves treatment effeciency by GPU parallel processings, comprises the steps of:
Step 1 establishes voxel model
In the algorithm, it is basic data cell to define voxel, and every eight neighbouring sample points define a cube area Domain, each cubical area constitute a voxel.Meanwhile this eight sampled points are defined as the angle point of voxel.Appointing in voxel It anticipates the value of a bit, can be estimated further according to the relative position of the point and eight angle points by sampling the values of eight angle points of voxel It arrives.Voxel model is using the unrelated Tri linear interpolation model in direction.
Step 2 approaches contour surface in three dimensional dispersed data field by linear interpolation establishment tri patch
Using in three dimensional dispersed data field, each grid cell is as a voxel, and there are corresponding on each vertex of voxel Scalar value.If the value on voxel vertex is greater than or equal to equivalent face amount, defines the vertex and be located at except contour surface;If voxel Value on vertex is less than equivalent face amount, defines the vertex and is located within contour surface.According to the relativeness on each vertex and contour surface Tri patch is built, nearly through the contour surface of the voxel.
The normal vector direction for providing the tri patch to form contour surface utilizes the contour surface figure of graphic hardware display extraction Picture.For the every bit on contour surface, the gradient component along the tangential direction in face is zero, the direction generation of the gradient vector of the point Table contour surface is in the normal vector direction of the point.The gradient that each apex of voxel is calculated using centered difference, by working as precursor The gradient on two vertex on plain side carries out linear interpolation, obtains the gradient on each vertex of tri patch.
Step 3 segments triangular surface patch grid using PN deltic methods
The normal on given tri patch vertex and each point, calculates its corresponding bezier surface control point:
(1) by bijkIt is evenly distributed on triangle, three vertex correspondences, three control points of triangle.
(2) by the nearest spot projection to the plane vertical with normal in distance angle on two sides at each angle, it is known as cutting Point.
(3) equalization point at central point towards six point of contacts is moved, and crosses mobile direction and moves 1/2 again.
Step 4 makees smoothing processing with bilateral normal filtering method
By in the raster grating of subdivision to two textures, respectively with floating point vector storage location and normal information.It examines The surface curvature for considering fluid is not usually high, right angle and acute angle seldom occurs, therefore the similar pixel of normal vector is considered as belonging to together One plane, makees smoothing processing;Dissmilarity is considered as belonging to Different Plane, does not make smooth.Pixel exists similar in the spatial position of pixel It is closely located in flow surface, make smoothing processing;Mutually from pixel farther out, its surface distance is usually farther for spatial position, does not make Smoothing processing.
Aforementioned four step is accelerated with aforementioned GPU parallel processing techniques, to reach preferable real-time.
The supplementary module includes three kinds of scalpel, bipolar electrocoagulator and cotton piece operation tools;
The scalpel is provided with rigid body, quality, skin friction resistance Fd, the attributes such as surgical cut power critical value m, and add Add aforementioned AABB bounding boxs and Mesh Collider collision detection algorithms.Scalpel can initialize Bezier control point The number num of kerf width coefficient W and composition point, number is more, and curve is more smooth.When scalpel works, aforementioned operation is used The real-time rendering algorithms of cutting draw notch, and collision detection provides contact point coordinates and cutting force degree letter for Rendering algorithms Breath.
The bipolar electrocoagulator burns effect to show electric coagulating apparatus by the pixel color and texture for changing image to tissue Fruit, and it is added to AABB bounding boxs and Mesh Collider collision detection algorithms.Implementation step is as follows:
(1) detection electric coagulating apparatus and tissue model collision situation traverse tri patch around the point of impingement after colliding generation, Judge the point of impingement whether in model tri patch.
(2) if in tri patch, contact point space coordinate, texture coordinate and the pixel value when collision are calculated.
(3) texture pixel is read, texture mapping is replaced.Using the pixel as the center of circle, drawn using Bresenham algorithms solid Circle is as the effect after electric coagulating apparatus calcination.
The cotton piece devises the double-deck collision detection, act as sucking the particle that simulation is bled.Implementation step is as follows:
(1) first layer (outer layer) uses the OnParticleCollision methods of ParticleSystem in Unity3D, Box Collider detection particles are added in cotton piece and whether cotton piece collides.
(2) second layer (internal layer) adds single layer AABB bounding box collision detection algorithms.
(3) particle coordinate is world coordinates, and cotton piece coordinate is local coordinate, therefore with transform interfaces in Unity3D Transform.TransformPoint () method the local coordinate of cotton piece is converted into world coordinates, then calculate particle seat Mark and bounding box whether geometrical intersection.
(4) particle intersected makees extinction processing, and it is red that the position of collision of cotton piece, which changes pixel color,.

Claims (10)

1. a kind of neuro-surgery virtual operation training system, which is characterized in that use Geomagic Touch force feedback equipments As surgical hardware operating platform, surgical instrument variation in space and and brain tissue are simulated by six direction steering engine of equipment Force feedback situation after contact;By the three-dimensional modeling of true brain image data, design realization and power on Unity3D platforms Surgical scene similar in feedback device matching and real brain, the system comprises the tissue cuttings for realizing surgical procedure Module, force feedback module, for the visualization model for visual effect display of performing the operation and for simulating the auxiliary of operation common tool Help module;
The tissue cutting module is improved based on the tissue deformation of internal point domain set domain restricted model by Mass-spring Model, Internal constraint is added on the basis of Mass-spring Model, for simulating the active force inside brain soft tissue, shows soft tissue Deformation;The real-time rendering of surgical cut is mainly used for scalpel work using a kind of real-time rendering algorithms for surgical cut Newly-generated model vertices texture coordinate after computation organization's cutting when tool operation, and carry out notch and redraw and render;
Hardware match based on Geomagic Touch force feedback equipments refers to Geomagic Touch force feedback equipments and operation The data interaction of training platform;
The visualization model for visual effect display of performing the operation, the three-dimensional modeling of external input brain data, including brain Data prediction, be registrated respectively according to CT, MRI and fMRI data type, three-dimensional reconstruction, grid correspond to, for obtaining The brain threedimensional model close with practical brain;Smoothed particle method is bled simulation, including particle configuration, resurfacing Three parts are handled with GPU parallelizations, for realizing the simulation bled to notch;
It is described for simulate perform the operation common tool supplementary module include:Scalpel tool contacts the space of points for obtaining point of a knife Coordinate position draws notch;Bipolar electrocoagulator tool shows electric coagulating apparatus to group for changing the pixel color and texture of image The cauterization effect knitted;Cotton piece tool, the particle bled for sucking simulation.
2. a kind of neuro-surgery virtual operation training system as described in claim 1, which is characterized in that the tissue is cut It includes the tissue deformation module based on internal point domain set domain restricted model to cut module, and the tissue deformation module is using a kind of based on interior The tissue deformation model of portion's point set region constraint, comprises the steps of:
Step 1 obtains brain soft tissue surfaces mesh coordinate, and rigid core is added for all tri patch vertex
Rigid core records the initial position of each particle, and establishes virtual spring between corresponding 2 points, can simulate stress The process of deformation recovery after disappearance limits it to the particle application one being subjected to displacement by virtual spring and leaves initial position Power is as constraint, it is ensured that the point on surface model can be restored to initial position after deformation;
Step 2 establishes the kinetics equation of Mass-spring Model, solves equation initial value
In the Mass-spring Model of soft tissue it is discrete after particle Ni(i=1,2...n) is indicated, the spring resistance connected between particle Buddhist nun's device lij(i, j ∈ [1, n], i ≠ j) is indicated, arbitrary particle NiThe resultant force f born is by internal force fintWith external force fextTwo parts Composition, wherein internal force be by with particle NiConnected spring-damper lijThe spring force f of generationsWith damping force fdComposition, has:
Wherein miFor particle NiQuality, kijFor particle NiAnd NjBetween be connected spring coefficient of elasticity, cijFor particle NiAnd NjIt Between damper damped coefficient, P (i) be and particle NiThe set of adjacent all particles, xiAnd xjRespectively NiAnd NjDisplacement,For particle NiAnd NjBetween be connected spring initial length, due to the displacement of particleIt is unknown, introduce particle speed v this A intermediate quantity converts differential equation of higher order to One first-order ordinary differential equation and solves, the moving displacement and speed of each particle It is expressed as
Elastic deformation model is substituted the above to, equation is solved and obtains each attribute initial value of particle when deformation;
Step 3, the Strain Distribution function of selected point domain set domain
Accurately internal point domain set domain Strain Distribution function is selected, carries out simplifying processing, chooses exponential function to simulate The Strain Distribution of organization internal, the expression formula that connecting inner point domain set domain is established are:
Wherein hiIndicate the offset of the point i in deformation process with surface in by corresponding domain set domain of the particle of outer force effect, hj Indicate the offset of point domain set domain midpoint j, dijFor the distance of domain set domain midpoint j to point i, a is the exponential function truth of a matter, s be scaling because Son;
Step 4 establishes dynamic deformation model, and by the way that a update particle domain is arranged, dynamic is stored will carry out in iterative process The particle of deformation calculation, as surface particle PiWhen by outer force effect, the particle and field particle are deposited by certain search rule Enter to update particle domain, the adaptability to changes that the particle and field particle generate passes to the inside point set of plane T by restricted spring P' updates surface particle deposit corresponding with the point in particle domain when the offset that certain in internal point set P' is put is more than h, The particle of deformation more neofield is stored in a deformation process will keep the attribute being updated until this deformation restores.
3. a kind of neuro-surgery virtual operation training system as claimed in claim 2, which is characterized in that the tissue is cut It further includes real-time rendering mould to cut module, and the real-time rendering module includes using a kind of real-time rendering algorithms for surgical cut Following steps:
1) when scalpel is collided with tissue model, point of a knife stress intensity is persistently obtained, contact point is recorded when more than a certain threshold value For feed coordinate, the threshold value is the threshold value that tissue surface ruptures dynamics, when collision releases and surface stress is less than threshold value, It is considered as cutting to terminate, it is coordinate of returning sword to record the point;
2) after determining access site, the path point of all processes on dynamic access cutting path, path point is three that collision is passed through The vertex of edged surface piece;
3) using feed coordinate as starting point, point of a knife is that end point draws notch using Cubic kolmogorov's differential system;
4) the tri patch apex coordinate that notch depth and newly-generated surface are calculated according to the point of a knife stress size of acquisition, by this A little coordinates carry out traversal comparison with original surface coordinate, if new vertex inside the tri patch of script, records these triangular facets Piece;If not in tri patch, the relationship on the vertex and field Triangle ID dough sheet is calculated;
5) according to above-mentioned two situations, the operation deleted tri patch or add tri patch is executed respectively, and be with new summit Texture coordinate carries out texture mapping;
6) in addition, colliding the amount of force generated and direction also according to scalpel and surface tri patch, adjustment notch is rebuild Width, depth and angle.
4. a kind of neuro-surgery virtual operation training system as described in one of claims 1 to 3, which is characterized in that described Force feedback module includes the collision detection module based on AABB bounding volume hierarchy (BVH)s Yu Mesh Collider, the collision detection mould Block uses a kind of collision detection algorithm based on AABB bounding volume hierarchy (BVH)s Yu Mesh Collider, comprises the steps of:
Step 1 establishes bounding volume hierarchy (BVH) tree to AABB bounding boxs, improves operation tool by multilayer bounding box nesting and surrounds The similarity of box, using top-down approach, by the primitive geometric element set S of given node VvIt is divided into two non-intersecting sons Collection, until all basic elements in subset:
1) the longest axis method is used to determine the split axis for dividing primitive geometric element;
2) median method is used to determine the split point for dividing primitive geometric element:The geometric center point of each basic element is calculated, Then the intermediate value that they are projected in split axis is calculated;
3) splitting plane is obtained according to split axis and split point, the element in space is divided into two subsets using splitting plane;
4) consider that a part of special geometric element may be located just on splitting plane, it can be according to these element center point institutes The space at place is allocated, and if central point is just on splitting plane, then distributes to the less subset of element;
Bounding box is applied to operation tool by step 2
Consider the three-dimensional rotation of operation tool, translate totally 6 degree of freedom, if operation tool object translation vector T=(x, y, z)T, Around x, y, z-axis rotation angle is θ,φ, correspondent transform matrix are:
Three is synthesized into a spin matrix R:
The state transformation of tool is described with x → Rx+T, bounding box is converted by the transformation for mula, you can realize bounding box Update only detects and updates intersection in addition, in traversal bounding volumetree carries out Intersection detection process on each time sampling point The descendent node of node does not update entire bounding volumetree;
Step 3 binds operation tool and Mesh Collider
Whether detection tissue model contacts with bounding box, if the two contacts, obtains contact point list and establishes contact point set C ={ c0c1...cn-1cn, recursion method is reused, successively detects contact point whether in bounding box, if contact point c (x, y, z), AABB{(xmin,ymin,zmin),(xmax,ymax,zmax), it needs to meet
Just it is considered as in bounding box, when contact point is not in bounding box, eliminates set C, ifTerminate recurrence;Otherwise Recurrence to bounding volumetree end leaf node, using recurrence after the set C that obtains as finally determining contact point set, use In the processing of follow-up collision effect.
5. a kind of neuro-surgery virtual operation training system as claimed in claim 4, which is characterized in that the force feedback mould Block further includes the Hardware match module based on Geomagic Touch force feedback equipments, Geomagic Touch force feedback equipments with The data interactive method of operative training platform, comprises the steps of:
1) communication initialization of PC machine and the force feedback equipment is realized by TCP/IP Handshake Protocols, while stating the name of equipment Claim character string, the basic function of initialization apparatus, including creates current device context starter, opening force output, opens and move States model opens equipment thread;
2) visual space in the tactile physical space and PC machine on initialization apparatus is arranged working space size, establishes respectively Coordinates matrix;
3) by tactile physical space coordinates system and visual space coordinate system matching, equipment cursor position is defaulted as in working space The heart;
4) agent point position when cursor contact model in the corresponding scene of mechanical arm tail end, and cursor position at this time are obtained, Location information is fed back to the tactile physical space of equipment;
5) the power generating algorithm for using GeoMagic Tocth equipment, calculates instead according to agent point position and haptic apparatus position Active force simultaneously feeds back to user by six direction steering engine simulated resistances of equipment.
6. a kind of neuro-surgery virtual operation training system as described in one of claims 1 to 3, which is characterized in that described Visualization model includes brain data three-dimensional modeling module, and the brain data three-dimensional modeling module includes data prediction, matches Standard, three-dimensional reconstruction, grid correspondence etc., steps are as follows:
(1) anisotropy and non local average filter method is used to carry out image filtering to CT, MRI and the fMRI data of input;
(2) when input data is CT, MRI, using Rigid Registration, the space of two images is aligned by rotation, translation, scaling Coordinate;When input data is fMRI, MRI, to fMRI into row interpolation and resampling, acquisition is identical as MRI resolution ratio, voxel is big Small identical image reuses the method for registering based on mutual information and carries out three-dimensional registration;
(3) registration data is made by three-dimensional reconstruction by VTK and exported, textures and rendering are carried out using Maya.
7. a kind of neuro-surgery virtual operation training system as claimed in claim 6, which is characterized in that the visualization mould Block further includes that smoothed particle method is bled analog module, and the analog module of bleeding is accelerated using a kind of based on GPU Smoothed particle method is bled simulation algorithm, establishes contour surface using marching cubes algorithm, and use the triangle sides PN Method carries out subdivision surfaces, makees smoothing processing with the filtering of bilateral normal, and by GPU parallel processings, comprise the steps of:
Step 1 establishes voxel model
Definition voxel is basic data cell, and every eight neighbouring sample points define a cubical area, each cube area Domain constitutes a voxel;Meanwhile this eight sampled points are defined as the angle point of voxel, the value at any point in voxel, it can be with By sampling the value of eight angle points of voxel, estimate to obtain further according to the relative position of the point and eight angle points, voxel model uses The unrelated Tri linear interpolation model in direction;
Step 2 approaches contour surface in three dimensional dispersed data field by linear interpolation establishment tri patch
Using in three dimensional dispersed data field, each grid cell is as a voxel, and there are corresponding scalars on each vertex of voxel Value defines the vertex and is located at except contour surface if the value on voxel vertex is greater than or equal to equivalent face amount;If voxel vertex On value be less than equivalent face amount, define the vertex and be located within contour surface, according to the relativeness of each vertex and contour surface structure Tri patch, nearly through the contour surface of the voxel;
The normal vector direction for providing the tri patch to form contour surface utilizes the iso-surface images of graphic hardware display extraction; For the every bit on contour surface, the gradient component along the tangential direction in face is zero, and the direction of the gradient vector of the point represents Contour surface is in the normal vector direction of the point;The gradient that each apex of voxel is calculated using centered difference, passes through current voxel The gradient on two vertex on side carries out linear interpolation, obtains the gradient on each vertex of tri patch;
Step 3 segments triangular surface patch grid using PN deltic methods
The normal on given tri patch vertex and each point, calculates its corresponding bezier surface control point:
(1) by bijkIt is evenly distributed on triangle, three vertex correspondences, three control points of triangle;
(2) by the nearest spot projection to the plane vertical with normal in distance angle on two sides at each angle, referred to as point of contact;
(3) equalization point at central point towards six point of contacts is moved, and crosses mobile direction and moves 1/2 again;
Step 4 makees smoothing processing with bilateral normal filtering method
By in the raster grating of subdivision to two textures, respectively with floating point vector storage location and normal information, normal vector Similar pixel is considered as belonging to same plane, makees smoothing processing;Dissmilarity is considered as belonging to Different Plane, does not make smooth;Pixel On the liquid surface closely located of pixel similar in spatial position, makees smoothing processing;Spatial position mutually from pixel farther out its Surface distance is usually farther, does not make smoothing processing.
8. a kind of neuro-surgery virtual operation training system as described in one of claims 1 to 3, which is characterized in that described In supplementary module for simulating operation common tool, scalpel is provided with rigid body, quality, skin friction resistance Fd, operation cut The attributes such as power critical value m are cut, and add AABB bounding boxs and Mesh Collider collision detection algorithms, scalpel can initialize The bigger curve of number num, num of kerf width coefficient W and the composition point at Bezier control point is more smooth;Scalpel work When making, notch is drawn using the real-time rendering algorithms of aforementioned surgical cut, collision detection provides contact for Rendering algorithms Point coordinates and cutting force degree information.
9. a kind of neuro-surgery virtual operation training system as described in one of claims 1 to 3, which is characterized in that described In supplementary module for simulating operation common tool, bipolar electrocoagulator is showed by changing pixel color and the texture of image Electric coagulating apparatus is added to AABB bounding boxs and Mesh Collider collision detection algorithms to the cauterization effect of tissue, realizes step It is as follows:
1) detection electric coagulating apparatus and tissue model collision situation traverse tri patch around the point of impingement, judge to touch after colliding generation It hits a little whether in model tri patch;
If 2) in tri patch, contact point space coordinate, texture coordinate and the pixel value when collision are calculated;
3) texture pixel is read, texture mapping is replaced, using the pixel as the center of circle, solid circles are drawn using Bresenham algorithms As the effect after electric coagulating apparatus calcination.
10. a kind of neuro-surgery virtual operation training system as described in one of claims 1 to 3, which is characterized in that described In supplementary module for simulating operation common tool, cotton piece devises the double-deck collision detection, act as sucking what simulation was bled Particle realizes that steps are as follows:
1) first layer, i.e. outer layer use the OnParticleCollision methods of ParticleSystem engines, add in cotton piece Add whether Box Collider detection particles and cotton piece collide;
2) second layer, i.e. internal layer add single layer AABB bounding box collision detection algorithms;
3) particle coordinate is world coordinates, and cotton piece coordinate is local coordinate, therefore in the transform interfaces of Unity3D The local coordinate of cotton piece is converted to world coordinates by Transform.TransformPoint () method, then calculates particle coordinate With bounding box whether geometrical intersection,
4) particle intersected makees extinction processing, and it is red that the position of collision of cotton piece, which changes pixel color,.
CN201810352058.0A 2018-04-19 2018-04-19 A kind of neuro-surgery virtual operation training system Pending CN108648548A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810352058.0A CN108648548A (en) 2018-04-19 2018-04-19 A kind of neuro-surgery virtual operation training system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810352058.0A CN108648548A (en) 2018-04-19 2018-04-19 A kind of neuro-surgery virtual operation training system

Publications (1)

Publication Number Publication Date
CN108648548A true CN108648548A (en) 2018-10-12

Family

ID=63746594

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810352058.0A Pending CN108648548A (en) 2018-04-19 2018-04-19 A kind of neuro-surgery virtual operation training system

Country Status (1)

Country Link
CN (1) CN108648548A (en)

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109285225A (en) * 2018-10-15 2019-01-29 东北大学 A kind of method for building up of the virtual reality auxiliary operation based on medical image
CN109637252A (en) * 2019-01-14 2019-04-16 晋城市人民医院 A kind of neurosurgery virtual operation training system
CN110232741A (en) * 2019-06-17 2019-09-13 腾讯科技(深圳)有限公司 Multilayer bounding box determines method, collision detection and motion control method and equipment
CN110288693A (en) * 2019-06-06 2019-09-27 昆明医科大学第二附属医院 A kind of intelligent microtubular Shaping System of intracranial aneurysm intervention operation
CN110335516A (en) * 2019-06-27 2019-10-15 王寅 A kind of VR openheart surgery simulation system and its analogy method
CN110341192A (en) * 2019-07-12 2019-10-18 东北大学 A kind of guide plate 3D printing model method for building up based on VR
CN110376922A (en) * 2019-07-23 2019-10-25 广东工业大学 Operating room scenario simulation system
CN110970134A (en) * 2019-11-05 2020-04-07 华中科技大学 Bone surgery simulation method and application thereof
CN111026269A (en) * 2019-12-04 2020-04-17 上海褚信医学科技有限公司 Haptic feedback method, device and equipment of biological tissue structure based on force feedback
CN111274705A (en) * 2020-01-21 2020-06-12 武汉大学 Multi-rate cooperation method and system for magnetic suspension visual-tactile interaction
CN113288346A (en) * 2021-05-20 2021-08-24 陈磊峰 Positioning and cutting device for treating liver cancer
CN115019591A (en) * 2022-08-05 2022-09-06 上海华模科技有限公司 Operation simulation method, operation simulation device and storage medium
CN115662234A (en) * 2022-11-03 2023-01-31 郑州大学 Thoracoscope surgery teaching system based on virtual reality
US11911120B2 (en) 2020-03-27 2024-02-27 Verb Surgical Inc. Training and feedback for a controller workspace boundary

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
李永强: ""基于Unity3D的神经外科虚拟手术训练软件设计与开发"", 《中国优秀硕士学位论文全文库》 *
肖昌盛: ""基于 GPU 加速的光滑粒子流体动力学流血模拟算法研究"", 《中国优秀硕士学位论文全文库》 *
胡家庆: ""基于内部点集域约束模型的虚拟手术仿真研究"", 《中国优秀硕士学位论文全文库》 *

Cited By (25)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109285225A (en) * 2018-10-15 2019-01-29 东北大学 A kind of method for building up of the virtual reality auxiliary operation based on medical image
CN109637252B (en) * 2019-01-14 2021-06-04 晋城市人民医院 Neurosurgery virtual operation training system
CN109637252A (en) * 2019-01-14 2019-04-16 晋城市人民医院 A kind of neurosurgery virtual operation training system
CN110288693A (en) * 2019-06-06 2019-09-27 昆明医科大学第二附属医院 A kind of intelligent microtubular Shaping System of intracranial aneurysm intervention operation
CN110232741A (en) * 2019-06-17 2019-09-13 腾讯科技(深圳)有限公司 Multilayer bounding box determines method, collision detection and motion control method and equipment
US11922574B2 (en) 2019-06-17 2024-03-05 Tencent Technology (Shenzhen) Company Limited Method and device for determining plurality of layers of bounding boxes, collision detection method and device, and motion control method and device
CN110232741B (en) * 2019-06-17 2022-11-25 腾讯科技(深圳)有限公司 Multilayer bounding box determination method, collision detection and motion control method and device
WO2020253582A1 (en) * 2019-06-17 2020-12-24 腾讯科技(深圳)有限公司 Multi-layer bounding box determination method, and collision detection and motion control methods and devices
CN110335516A (en) * 2019-06-27 2019-10-15 王寅 A kind of VR openheart surgery simulation system and its analogy method
CN110335516B (en) * 2019-06-27 2021-06-25 王寅 Method for performing VR cardiac surgery simulation by adopting VR cardiac surgery simulation system
CN110341192A (en) * 2019-07-12 2019-10-18 东北大学 A kind of guide plate 3D printing model method for building up based on VR
CN110376922A (en) * 2019-07-23 2019-10-25 广东工业大学 Operating room scenario simulation system
CN110970134A (en) * 2019-11-05 2020-04-07 华中科技大学 Bone surgery simulation method and application thereof
CN110970134B (en) * 2019-11-05 2023-08-25 华中科技大学 Simulation method for bone surgery and application thereof
CN111026269B (en) * 2019-12-04 2023-10-10 上海褚信医学科技有限公司 Haptic feedback method, device and equipment for biological tissue structure based on force feedback
CN111026269A (en) * 2019-12-04 2020-04-17 上海褚信医学科技有限公司 Haptic feedback method, device and equipment of biological tissue structure based on force feedback
CN111274705B (en) * 2020-01-21 2021-12-03 武汉大学 Multi-rate cooperation method and system for magnetic suspension visual-tactile interaction
CN111274705A (en) * 2020-01-21 2020-06-12 武汉大学 Multi-rate cooperation method and system for magnetic suspension visual-tactile interaction
US11911120B2 (en) 2020-03-27 2024-02-27 Verb Surgical Inc. Training and feedback for a controller workspace boundary
CN113288346A (en) * 2021-05-20 2021-08-24 陈磊峰 Positioning and cutting device for treating liver cancer
CN113288346B (en) * 2021-05-20 2023-12-29 博思研生物技术(苏州)有限公司 Positioning and cutting device for treating liver cancer
CN115019591A (en) * 2022-08-05 2022-09-06 上海华模科技有限公司 Operation simulation method, operation simulation device and storage medium
CN115019591B (en) * 2022-08-05 2022-11-04 上海华模科技有限公司 Operation simulation method, device and storage medium
CN115662234B (en) * 2022-11-03 2023-09-08 郑州大学 Thoracic surgery teaching system based on virtual reality
CN115662234A (en) * 2022-11-03 2023-01-31 郑州大学 Thoracoscope surgery teaching system based on virtual reality

Similar Documents

Publication Publication Date Title
CN108648548A (en) A kind of neuro-surgery virtual operation training system
CN109285225B (en) Medical image-based virtual reality assisted surgery establishing method
Carr et al. Simplifying flexible isosurfaces using local geometric measures
CN106875462B (en) Real-time digital organ cutting method based on meta-sphere model and hybrid driving method
US8149237B2 (en) Information processing apparatus and program
CN105261062B (en) A kind of personage's segmentation modeling method
CN107392990B (en) Global illumination to render 3D scenes
CN106504320A (en) A kind of based on GPU and the real-time three-dimensional reconstructing method towards depth image
Vilanova et al. VirEn: A virtual endoscopy system
CN110176079A (en) A kind of three-dimensional model deformation algorithm based on quasi- Conformal
CN112966390B (en) Method and apparatus for garment processing based on dual three-dimensional distance fields
CN113379929A (en) Bone tissue repair virtual reality solution method based on physical simulation
Wagner et al. Collision detection and tissue modeling in a vr-simulator for eye surgery
Chen et al. Manipulating, deforming and animating sampled object representations
CN107895398B (en) Relief editing method combined with visual saliency
CN110047145A (en) Metaplasia simulation system and method based on deep learning and finite element modeling
Asgharian et al. Feature-preserving mesh simplification through anisotropic Nyquist-based adaptive sampling of points inside the segmented regions
CN115222894A (en) Image topological deformation method based on laplace optimization
Wang et al. Stable and efficient collision detection scheme for hip-surgery training system
Paiva et al. Fluid-based hatching for tone mapping in line illustrations
JP3724006B2 (en) High speed rendering method and apparatus
Kil et al. 3D warp brush modeling
CN109036567B (en) Soft tissue deformation simulation method based on subspace condensation algorithm
Zeng et al. 3D plants reconstruction based on point cloud
JP3178808B2 (en) Object reconstruction method, object approximation method, and space rendering method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WD01 Invention patent application deemed withdrawn after publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20181012