CN102346811B - For heart being carried out the method and system of the comprehensive modeling specific to patient - Google Patents
For heart being carried out the method and system of the comprehensive modeling specific to patient Download PDFInfo
- Publication number
- CN102346811B CN102346811B CN201110204933.9A CN201110204933A CN102346811B CN 102346811 B CN102346811 B CN 102346811B CN 201110204933 A CN201110204933 A CN 201110204933A CN 102346811 B CN102346811 B CN 102346811B
- Authority
- CN
- China
- Prior art keywords
- heart
- patient
- specific
- ingredient
- model
- 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.)
- Active
Links
- 210000002216 Heart Anatomy 0.000 title claims abstract description 301
- 230000017531 blood circulation Effects 0.000 claims abstract description 85
- 230000000747 cardiac effect Effects 0.000 claims abstract description 50
- 239000004615 ingredient Substances 0.000 claims description 99
- 210000000709 Aorta Anatomy 0.000 claims description 57
- 210000004369 Blood Anatomy 0.000 claims description 27
- 239000008280 blood Substances 0.000 claims description 27
- 210000001308 Heart Ventricles Anatomy 0.000 claims description 26
- 239000012530 fluid Substances 0.000 claims description 20
- 238000002059 diagnostic imaging Methods 0.000 claims description 16
- 238000002224 dissection Methods 0.000 claims description 14
- 230000000875 corresponding Effects 0.000 claims description 13
- 230000000004 hemodynamic Effects 0.000 claims description 11
- 230000001105 regulatory Effects 0.000 claims description 10
- 238000002560 therapeutic procedure Methods 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 9
- 238000003745 diagnosis Methods 0.000 claims description 9
- 201000010099 disease Diseases 0.000 claims description 9
- 238000000034 method Methods 0.000 claims description 9
- 206010002329 Aneurysm Diseases 0.000 claims description 7
- 239000007943 implant Substances 0.000 claims description 7
- 210000001519 tissues Anatomy 0.000 claims description 6
- 238000011068 load Methods 0.000 claims description 4
- 239000000203 mixture Substances 0.000 claims description 4
- 210000000988 Bone and Bones Anatomy 0.000 claims description 3
- 210000003540 Papillary Muscles Anatomy 0.000 claims description 3
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 3
- 241000208340 Araliaceae Species 0.000 claims 2
- 235000003140 Panax quinquefolius Nutrition 0.000 claims 2
- 235000005035 ginseng Nutrition 0.000 claims 2
- 235000008434 ginseng Nutrition 0.000 claims 2
- 230000001172 regenerating Effects 0.000 claims 1
- 210000003484 anatomy Anatomy 0.000 abstract description 18
- 210000001765 Aortic Valve Anatomy 0.000 description 31
- 210000004115 Mitral Valve Anatomy 0.000 description 31
- 230000010247 heart contraction Effects 0.000 description 28
- 210000002837 Heart Atria Anatomy 0.000 description 26
- 238000004088 simulation Methods 0.000 description 21
- 230000002685 pulmonary Effects 0.000 description 15
- 230000004907 flux Effects 0.000 description 12
- 210000000591 Tricuspid Valve Anatomy 0.000 description 11
- 238000005516 engineering process Methods 0.000 description 10
- 238000005259 measurement Methods 0.000 description 10
- 206010020565 Hyperaemia Diseases 0.000 description 9
- 238000010992 reflux Methods 0.000 description 8
- 239000000243 solution Substances 0.000 description 7
- 230000001131 transforming Effects 0.000 description 7
- 210000003462 Veins Anatomy 0.000 description 6
- 210000001174 Endocardium Anatomy 0.000 description 5
- 210000003492 Pulmonary Veins Anatomy 0.000 description 5
- 238000004590 computer program Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 238000011156 evaluation Methods 0.000 description 5
- 239000000463 material Substances 0.000 description 5
- 238000001356 surgical procedure Methods 0.000 description 5
- 210000001367 Arteries Anatomy 0.000 description 4
- 210000003516 Pericardium Anatomy 0.000 description 4
- 210000001147 Pulmonary Artery Anatomy 0.000 description 4
- 238000009877 rendering Methods 0.000 description 4
- 239000007787 solid Substances 0.000 description 4
- 210000003291 Sinus of Valsalva Anatomy 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 3
- 230000006399 behavior Effects 0.000 description 3
- 230000015572 biosynthetic process Effects 0.000 description 3
- 238000005755 formation reaction Methods 0.000 description 3
- 239000007788 liquid Substances 0.000 description 3
- 230000002980 postoperative Effects 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 238000001228 spectrum Methods 0.000 description 3
- 206010004552 Bicuspid aortic valve Diseases 0.000 description 2
- 208000008787 Cardiovascular Disease Diseases 0.000 description 2
- 210000004072 Lung Anatomy 0.000 description 2
- 230000000712 assembly Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000034994 death Effects 0.000 description 2
- 231100000517 death Toxicity 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000001595 flow curve Methods 0.000 description 2
- 201000010238 heart disease Diseases 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000002969 morbid Effects 0.000 description 2
- 230000035479 physiological effects, processes and functions Effects 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 210000004872 soft tissue Anatomy 0.000 description 2
- 238000003325 tomography Methods 0.000 description 2
- 230000002792 vascular Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 206010002855 Anxiety Diseases 0.000 description 1
- 206010057666 Anxiety disease Diseases 0.000 description 1
- 210000002376 Aorta, Thoracic Anatomy 0.000 description 1
- 208000007474 Aortic Aneurysm Diseases 0.000 description 1
- 206010002915 Aortic valve incompetence Diseases 0.000 description 1
- 238000010989 Bland-Altman Methods 0.000 description 1
- 210000004204 Blood Vessels Anatomy 0.000 description 1
- 210000002808 Connective Tissue Anatomy 0.000 description 1
- 210000004351 Coronary Vessels Anatomy 0.000 description 1
- 206010061818 Disease progression Diseases 0.000 description 1
- 241000196324 Embryophyta Species 0.000 description 1
- 206010016193 Fallot's tetralogy Diseases 0.000 description 1
- 208000002907 Heart Valve Disease Diseases 0.000 description 1
- 210000003709 Heart Valves Anatomy 0.000 description 1
- 238000005481 NMR spectroscopy Methods 0.000 description 1
- 206010058046 Post procedural complication Diseases 0.000 description 1
- 230000036506 anxiety Effects 0.000 description 1
- 201000002064 aortic valve insufficiency Diseases 0.000 description 1
- 230000036772 blood pressure Effects 0.000 description 1
- 210000004903 cardiac system Anatomy 0.000 description 1
- 230000000876 cardiodynamic Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000004087 circulation Effects 0.000 description 1
- 230000015271 coagulation Effects 0.000 description 1
- 238000005345 coagulation Methods 0.000 description 1
- 230000001427 coherent Effects 0.000 description 1
- 238000002591 computed tomography Methods 0.000 description 1
- 201000003137 congenital heart disease Diseases 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000001808 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000003111 delayed Effects 0.000 description 1
- 230000001419 dependent Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000002001 electrophysiology Methods 0.000 description 1
- 230000002708 enhancing Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000002513 implantation Methods 0.000 description 1
- 238000000126 in silico method Methods 0.000 description 1
- 238000002347 injection Methods 0.000 description 1
- 239000007924 injection Substances 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000002452 interceptive Effects 0.000 description 1
- 239000010410 layer Substances 0.000 description 1
- 230000036651 mood Effects 0.000 description 1
- 230000000877 morphologic Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000003534 oscillatory Effects 0.000 description 1
- 239000003973 paint Substances 0.000 description 1
- 230000000803 paradoxical Effects 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 230000036285 pathological change Effects 0.000 description 1
- 231100000915 pathological change Toxicity 0.000 description 1
- 229920001707 polybutylene terephthalate Polymers 0.000 description 1
- 238000005086 pumping Methods 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000010008 shearing Methods 0.000 description 1
- 239000002356 single layer Substances 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 201000003005 tetralogy of Fallot Diseases 0.000 description 1
- 238000004642 transportation engineering Methods 0.000 description 1
- 238000002604 ultrasonography Methods 0.000 description 1
- 230000002861 ventricular Effects 0.000 description 1
Abstract
The present invention relates to the method and system for heart being carried out the comprehensive modeling specific to patient.Disclose a kind of method and system for whole cardioanatomy, kinetics, hematodinamics and fluidic structures being carried out alternately according to 4D medical image the modeling specific to patient.Anatomy and the kinetics of heart is determined by the parameter specific to patient estimating the physiological models of heart according to the 4D medical image for patient.The described anatomy specific to patient and kinetics are used as the input for 3D Navier Stokes solver, and described 3D Navier Stokes solver derives the actual hematodinamics retrained by topology along whole Cardiac cycle.In described Cardiac cycle inner iteration, the deformation being calculated cardiac structure by the blood flow at emulation step preset time and blood flow based on described emulation determines that fluidic structures is mutual, so that using the described deformation of cardiac structure in the blood flow emulation at next time step.
Description
This application claims the U.S. Provisional Application No. JIUYUE 17 in 61/366,294,2010 submitted on July 21st, 2010
The U.S. Provisional Application No. 61/409,633 that the U.S. Provisional Application No. 61/383,942 of day submission and on November 3rd, 2010 submit to
Rights and interests, by with reference to the disclosure of which is incorporated in this.
Technical field
The present invention relates to the use of medical image heart is modeled, and systems based on 4D medical science figure
As data carry out the comprehensive modeling specific to patient to heart.
Background technology
In the U.S., heart disease is the major causes of death of masculinity and femininity, and accounts for the ratio of deaths worldwide not
Less than 30%.Although the progress of medical science in the last few years for Complex Congenital Heart Disease (such as valvular heart disease, aneurysm of thoracic aorta and
Tetralogy of Fallot) diagnosis and treatment aspect provide important improvement, but still can occur in a large number too early morbidity and
Dead (morality).Each medical imaging modalities (such as computer tomography (CT), magnetic resonance (MR), rotation X can be used
Ray and ultrasonic) with the high a large amount of form of time-space resolution acquisition and functional images data.But owing to data understand
Ability delayed, doctor can be forced the measurement that is restricted based on scope and method makes particularly important judgement.These limits
System be at least partially due to lack for describe heart-aortic dissection, physiology and hemodynamic specific to
Efficiently and accurately the estimating and lack progression of disease model and cause of the parameter of patient.
Summary of the invention
The present invention provides a kind of side that heart carries out the comprehensive modeling specific to patient according to 4D medical image
Method and system.Specifically, embodiments of the invention provide for whole cardioanatomy, power according to 4D medical image
The modeling specific to patient that, hematodinamics and fluidic structures are mutual.
By estimating the parameter specific to patient of the physiological models of heart according to the 4D medical image for patient
Determine anatomy and the kinetics of heart.The described anatomy specific to patient and kinetics are used as 3D Navier-
The input of Stokes solver, described 3D Navier-Stokes solver is derived by topography along whole Cardiac cycle
Learn the actual hematodinamics of constraint.The heart is calculated by the blood flow at emulation step preset time and blood flow based on described emulation
In described Cardiac cycle inner iteration, the deformation of coagulation of YIN-cold in ZANG-organ structure determines that fluidic structures is mutual, thus the blood flow at next time step
Emulation uses the described deformation of cardiac structure.Represent the heart that anatomy, kinetics, hematodinamics and fluidic structures are mutual
The dirty comprehensive model specific to patient can be used for the evaluation of the non-intrusion type to heart and diagnosis, and for virtual treatment
Method planning and cardiovascular disease manage.
In one embodiment of the invention, the 4D dissection specific to patient of heart is generated according to 4D medical imaging data
Model.Ask at each time step in the middle of the level set framework multiple time steps in a Cardiac cycle by utilizing subsequently
Solution by the described Navier-Stokes equation retrained specific to the 4D anatomical model of patient to the blood flow emulating in heart.
In another embodiment of the present invention, the 4D solution specific to patient of heart is generated according to 4D medical imaging data
Cut open model.It is in described specific to emulation at least one heart ingredient of the 4D anatomical model of patient in current time step
Blood flow, this is by utilizing level set framework to solve by least one heart ingredient position at current time step described
Put the Navier-Stokes equation of constraint and realize.Count at current time step based on the emulation blood flow at current time step
Calculate the deformation of at least one heart ingredient described.Described emulation and calculation procedure repeats for multiple time steps, and extremely
It is at least partly based at previous time step the deformation of at least one heart ingredient described in calculating to determine current time
The current location of at least one the heart ingredient described at step.
In another embodiment of the present invention, comprehensive specific to patient of heart is generated according to 4D medical imaging data
Property 4D model.A described part specific to the comprehensive 4D model of patient is adjusted, in order to emulate such as disease or treatment
The situation of method etc.Regenerate the comprehensive 4D model specific to patient of heart subsequently to emulate described through overregulating
Part is for the impact of the described comprehensive 4D model specific to patient.
With reference to following detailed description and drawings, these and other advantages of the present invention are for those of ordinary skill in the art
Will become clear from.
Accompanying drawing explanation
Fig. 1 graphic extension is according to an embodiment of the invention a kind of comprehensive for carry out heart specific to patient
Property modeling method；
Fig. 2 graphic extension heart model for each heart ingredient according to an embodiment of the invention；
Fig. 3 graphic extension a kind of 4D specific to patient for generating heart according to an embodiment of the invention solves
The method cuing open model；
Fig. 4 graphic extension is from the exemplary Hearts model specific to patient of CT of many phases sequence estimation；
The Hearts model specific to patient that Fig. 5 graphic extension is embedded in rectangular domain as level set；
Fig. 6 graphic extension is according to an embodiment of the invention a kind of imitative based on the Hearts model specific to patient
The method of the blood flow during sincerity is dirty；
Fig. 7 graphic extension utilizes the exemplary blood flow simulation result of the method for Fig. 6；
Fig. 8 graphic extension a kind of fluidic structures for estimating aorta according to an embodiment of the invention is mutual
(FSI) alternative manner；
The example results mutual for the fluidic structures of described aorta is estimated in Fig. 9 graphic extension；
The multiple dimensioned anatomical model of Figure 10 graphic extension LV；
Figure 11 graphic extension is for aortic area and the certainty of measurement of mitral area；
Figure 12 graphic extension various types of pulmonary trunk form；
Figure 13 graphic extension emulates for the hematodinamics of heart contraction event and structure；
Figure 14 graphic extension utilizes the formation of the whirlpool in the aortic valve hole district that blood flow emulation obtains；
The hematodinamics emulation of Figure 15 graphic extension diastole event and structure；
Figure 16 graphic extension is shown in the curve chart of the time flux of the interior blood flow through Ge Ban district of a Cardiac cycle；
Figure 17 and 18 graphic extensions are used to measure the site of the slice position of stream；
Figure 19 graphic extension is for having the blood flow emulation of two point backflow aortic valve and the mitral heart of morbid state；
Figure 20 graphic extension is for a healthy heart and the hemodynamic comparison of the emulation of two diseased hearts；
Figure 21 graphic extension has and does not have the blood flow simulation result of left atrium；
Figure 22 graphic extension a kind of utilization according to an embodiment of the invention is specific to the comprehensive 4D model of patient
The method of being predicted property planning；
Figure 23 acts on the power on implant model during being illustrated in virtual deployment；
Figure 24 graphic extension is for the emulation of the stent deployment at endaortic aneurysm；
The virtual aneurysm of Figure 25 graphic extension is excised；And
Figure 26 is the high level block diagram of the computer that can implement the present invention.
Detailed description of the invention
The present invention relates to according to volume data sequence (such as computer tomography (CT), nuclear magnetic resonance (MRI) and super
Sound cardiographic data) heart carried out the comprehensive modeling specific to patient.Such volume data sequence (is the most also claimed
Make 4D view data or 4D image) it is the sequence obtaining to contain one or more cardiac cycle in certain period of time, wherein
Each frame is a width 3D rendering (body).It is described herein as embodiments of the invention to provide for described heart modeling method
Visual analysis.One width digital picture is usually made up of the numeral expression of one or more objects (or shape).The digital table of object
Show the most usually according to identifying and handling described object and describe.Such manipulation be the memorizer of computer system or its
The virtual manipulation realized in his circuit/hardware.It will therefore be appreciated that can utilize in computer system be stored in described
Data in computer system implement embodiments of the invention.
Embodiments of the invention provide one for according to 4D medical image to whole cardioanatomy, kinetics,
Hematodinamics and fluidic structures carry out the method and system of the modeling specific to patient alternately.By according to the 4D for patient
Medical image estimates that the parameter specific to patient of the physiological models of heart determines anatomy and the kinetics of heart.Institute
State anatomy and kinetics specific to patient and be used as the input for 3D Navier-Stokes solver, described 3D
Navier-Stokes solver derives the actual hematodinamics retrained by topology along whole Cardiac cycle.Logical
Cross the blood flow at emulation step preset time and blood flow based on described emulation calculates the deformation of cardiac structure in described heartbeat week
Determine to phase inner iteration that fluidic structures is mutual, thus the blood flow emulation at next time step uses described in cardiac structure
Deformation.The comprehensive mould specific to patient of the heart that expression anatomy, kinetics, hematodinamics and fluidic structures are mutual
Type can be used for the evaluation of the non-intrusion type to heart and diagnosis, and manages for virtual therapy planning and cardiovascular disease.
Fig. 1 illustrates according to an embodiment of the invention a kind of for carrying out combining specific to patient to heart
The method of conjunction property modeling.The method of Fig. 1 would indicate that the image data transformation in the coronary artery district of patient become heart specific to trouble
The anatomical model of person, and use this specific to the heart model of patient to emulate blood flow and the fluidic structures friendship of described heart
Mutually.
With reference to Fig. 1, in step 102 place, receive 4D medical image.Specifically, at least one body picture number is received
According to sequence.Described volumetric image data sequence can be 3D rendering (body) sequence gathered in special time period.For example, may be used
To gather such 4D view data (3D+ time) in a complete cardiac cycle.Various medical imaging modalities can be utilized
Receive one or more sequence.For example, according to various embodiments of the present invention, 4D CT data, the ultrasonic heart of 4D can be received
Move and trace and/or 4D magnetic resonance (MR) view data, and other kinds of view data.Can adopt from one or more images
Collection equipment (such as CT scanner, ultrasonic device or MR scanning device) directly receives described view data.It is also possible to such as from meter
The memorizer of calculation machine system or storage device or certain other computer-readable recording medium load previously stored picture number
According to.
In step 104 place, generate the 4D anatomical model specific to patient of heart according to the 4D view data received.Tool
For body, described 4D anatomical model is the multicomponent model with multiple heart ingredient, and it includes such as each chamber
Room (left ventricle, left atrium, right ventricle and right atrium), each cardiac valve (aortic valve, Bicuspid valve, Tricuspid valve and pulmonary artery
Lobe) and aorta.The comprehensive model of such heart is used to capture various form, function and pathological change.Can
Dissect to use a kind of modularity and layered approach to reduce complexity and promote for independent anatomical structure effective and clever
The estimation lived.Embodiments of the invention use a kind of heart model, and this heart model meets anatomy and for whole aroused in interest
Cycle keeps consistent parametrization with different patients, and this is the constraint by utilizing physiology to drive and sampling plan and realizes.
The Global Dynamics change of each heart chamber and valve is parameterized is the similarity transformation with time correlation,
Its definition translation (translation), the quaternary number rotated represent, time in similarity transformation scale factor and cardiac cycle
Position.In an advantageous embodiment, use for 152 anatomic landmarks of heart chamber with for valvular 33 solutions
Cut open complexity and the tandem pattern of the collection all cardiac anatomy of incompatible parametrization of mark.Thus, each mark is by three
A track in dimension space describes, and by described and time correlation similarity transformation normalization.By in order to represent chamber
9 dense surface mesh set and 13 structured set that valve is added to complete final model.Each
Grid is the dissection grid sampling along the summit by described tag definition.
Fig. 2 illustrates heart model for each heart ingredient according to an embodiment of the invention.Specifically
For, image (a)-(f) of Fig. 2 shows the anatomical definition of each ingredient for described Hearts model.Based on
4D view data estimates the parameter of these anatomical models in a cardiac cycle for particular patient.
The image (a) of Fig. 2 shows for left ventricle 200 and the model of left atrium 202.Left ventricle 200 is from 78 marks
Will (in control point, 16 Bicuspid valve sides, 15 Bicuspid valve every control point, 16 left ventricle output channel control point and 32 masters
Arterial valve control point) and four surface geometries (LV visceral pericardium, LV endocardium and LV output channel) construct.Zuo Xin
Surface, room 202 is connected to left ventricle 200 by aortic valve control point.
The image (b) of Fig. 2 shows for right ventricle 204 and the model of right atrium 206.Right ventricle 204 is from 74 marks
Will (in control point, 16 Tricuspid valve sides, 15 Tricuspid valvies every control point, 28 Tricuspid valve control point and 18 valve of pulmonary trunk controls
Point processed) and four surface geometries (RV point (apex), RV output channel and RV flow ipe) construct.Right atrium table
Face 206 is by 28 Tricuspid valve control point constraints and is linked to right ventricle 204.
The image (c) of Fig. 2 shows the model for aortic valve 208.Aortic valve model 208 is from 11 marks (3
Individual commissure, 3 hinges, 3 distal lobular and 2 ostiums) and four surface textures (aortic root, N lobule, L are little
Leaf and R lobule) construct.Aortic root is by described hinge and commissure plane restriction, and each leaflet spans exists
Between two commissure and a hinge.
The image (d) of Fig. 2 shows the model for Bicuspid valve 210.Mitral 210 is from seven marks (3 three
Angular region, 2 commissure and 2 distal lobular) construct.Front lobule is by two trigonums, a distal lobular and two companies
Conjunction portion defines, and rear lobule is defined by three trigonums, a distal lobular and a commissure.
The image (e) of Fig. 2 shows the model of valve of pulmonary trunk 212.Valve of pulmonary trunk model 212 is from nine mark (3 companies
Conjunction portion, 3 hinges and 3 distal lobular) and four planar structure (pulmonary artery root, N lobule, L lobule and R lobule) structures
's.
The image (f) of Fig. 2 shows the model of Tricuspid valve 214.Tricuspid valve model 214 is from four surface geometries
(circulus, in every lobule, front lobule and rear lobule) and six anatomic landmarks (three commissure and three distal lobular)
Structure.
The described 4D anatomical model specific to patient provides the form of the heart of patient, and can be used to determine for
The form (size) of any ingredient of heart and kinetic parameter.Fig. 3 illustrates according to one embodiment of present invention
A kind of method of 4D anatomical model specific to patient for generating heart.The method of Fig. 3 is to representing that patient's is crown dynamic
The view data in arteries and veins district carries out converting to generate the anatomical model specific to patient of the heart for this patient.The method of Fig. 3
Can be used to implement the step 104 of the method for Fig. 2.Number of patent application 2010/0280352 and 2011/ in U.S. Publication
The method that described in more detail the anatomical model specific to patient for generating heart of Fig. 3 in 0060576, described specially
The disclosure of profit application is by with reference to being incorporated in this.
In step 302 place, come according to for each the received view data in the middle of multiple heart ingredients
Generate single model.According to one embodiment of present invention, for the following generation model: each heart chamber, the most left heart
Room (LV) (endocardium and visceral pericardium), right ventricle (RV), left atrium (LA) and right atrium (RA)；Each valve, i.e. Bicuspid valve, master
Arterial valve, valve of pulmonary trunk and Tricuspid valve；And great vessels, i.e. aorta and pulmonary trunk.All these parts of heart exist
Here " heart ingredient " it is referred to collectively as.For each heart ingredient, the discriminant utilizing data base to guide is estimated
Meter/detection technique estimates the physiological models of described heart ingredient in each frame of described 4D view data.
Before generating for the individualized heart model of particular patient, off-line constructs each anatomical structure (heart composition
Part) physiological models.Each physiological models is based on the corresponding heart ingredient in the training data set having note
Mathematical notation and generate.For example, it is possible to use have putting down of the heart ingredient in the training data set of note
All shapes generate the physiological models for each heart ingredient.For example, by special with reference to merging the U.S. in this
Profit application publication number 2008/0101676 describes generation four chamber electrophysiology cardiac models and is fitted to by described heart model
View data.As described herein, described heart model is 3D grid, and for the original net of each chamber
Lattice are that the average shape utilizing each chamber in the training data having note generates.Additionally, by with reference to merging U.S. in this
State's number of patent application 2009/0123050 describes a kind of 4D physiological models of aortic valve.Note can be similarly based on
Training dataset is combined into each described heart ingredient off-line and generates physiological models.
In order to estimate the physiological models of the specific heart ingredient in 3D rendering (i.e. the frame of 4D image sequence), based on relatively
The parameter of described physiological models is estimated by the data base manipulation discriminant machine learning techniques of the big training image having note
Meter is so that image described in matching.According to an embodiment, use limit space learning (marginal space learning,
MSL) in every piece image, described physiological models is positioned.
The idea of MSL is not direct Study strategies and methods in complete similarity transformation parameter space, but based on there being note
Training data learns discriminant grader in the dimension increased with going forward one by one.Along with dimension increases, effectively (just) area of space becomes
Must be by the more multiple constraint of previous marginal spatial classification device.In order to estimate anatomical structure (such as specific heart ingredient)
Physiological models, in the picture, can be similarity transformation (i.e. position, the orientation of the position corresponding to described heart ingredient
And yardstick) estimation be divided into three phases: location estimation, position-orientation estimation, and completely similarity transformation estimate.Based on institute
State training data for one discriminant grader of each stage-training.Can be probability all discriminant classifier trainings
Advance tree (PBT).In addition to reducing the size of search volume, the another advantage of MSL is possible in each limit empty
Between level use different features (such as 3D Haar feature or controllable feature) train described grader.
The life utilizing MSL to estimate each heart ingredient in 3D rendering data is described in following publication
The example of reason model, by with reference to the disclosure of which is incorporated in this: U.S. Patent Application Publication No. 2008/0101676, it is retouched
State and estimated the model for each chamber in 3D CT view data；U.S. Patent Application No. 2009/0123050, it is retouched
State and the physiological models of aortic valve has been fitted to 4D CT data；And " the 3D Ultrasound Tracking of Yang et al.
of the Left Ventricles Using One-Step Forward Prediction and Data Fusion of
Collaborative Trackers ", CVPR 2008, which depict the models fitting of left ventricle to 3D ultrasonic image sequence.
It should be appreciated that each heart can be formed by utilizing the discriminant machine learning techniques similar to above example
The physiological models of part is fitted to view data to estimate described heart ingredient.
MSL is the most such as utilized to have estimated each single heart ingredient mould in each frame of 4D view data
The parameter of type, it is possible in every piece image, described single heart ingredient model is performed border based on study inspection
Survey, in order to the model parameter estimated by refinement.Specifically, it is possible to use described border detection based on study refines each
The border of individual estimated model, in order to improve the precision that the physiological models for each heart ingredient is estimated.
In step 304 place, generate heart by being integrated into each independent model of each heart ingredient generation
4D specific to patient individualizes anatomical model.Each the single heart ingredient model obtained from step 402 is one
The individual grid being made up of certain number of point.According to a kind of advantageous embodiment, for integrated LV(endocardium and visceral pericardium),
The independent model of RV, LA, RA, Bicuspid valve, aortic valve, aorta and pulmonary trunk, connect or between the model of overlap
Set up mesh point corresponding.Described mesh point correspondence allows each model correctly with respect to aligned with each other.Possibly through right
Model carries out resampling and sets up mesh point correspondence between each model.For example, by with reference to merging the U.S. in this
Patent application publication number 2008/0262814 describes in order to set up mesh point between the model of four heart chambers corresponding
Various resampling methods, in order to be properly aligned with each heart chamber model.It should be appreciated that in U.S. Patent Application Publication
Technology described in number 2008/0262814 can be extended to described here at each single heart ingredient mould
Mesh point is set up corresponding between type.
In step 306 place, export the 4D Hearts model specific to patient.Can be by the 4D specific to patient be solved
Cut open heart model store memorizer, storage device or computer-readable medium in export the 4D Hearts specific to patient
Model.Specific to the 4D Hearts model of patient or the 4D Hearts mould specific to patient can also be printed by display
The image of type exports the 4D Hearts model specific to patient.The 4D Hearts model specific to patient exported can
To be used for further Medical Image Processing.For example, personalized 4D Hearts model can be used to estimate the heart
Dirty various forms and vital measuring.Personalized 4D Hearts model can be utilized to emulate blood flow or blood-tissue
Alternately, as described in the subsequent step in the method for Fig. 1.
As it has been described above, the method for Fig. 3 illustrate according to an embodiment a kind of for generate specific to patient's
The method of 4D anatomical model.In another embodiment, natural level of detail based on basic anatomical structure one can be followed
Strategy from coarse to fine estimates the parameter of described 4D anatomical model for patient.In the first step, from the 4D image received
Data recover the posture of each model component and corresponding kinematic parameter, and this is to utilize one by MSL and stochastical sampling
Concordance (RANSAC) technology is combined to obtain robust and the method for time upper coherent object location and realizing.?
" the Patient-Specific Modeling and Quantification of the Aortic and of Ionasec et al.
Mitral Valves from 4D Cardiac CT and TEE ", IEEE Transactions on Medical
Imaging, describes this method in detail in 2010, by with reference to being incorporated into this.In the second step, track is utilized to compose
Study (trajectory spectrum learning, TSL) algorithm estimates position and the motion of each anatomic landmark simultaneously, should
Algorithm uses feature based on track and strong track spectrum grader." Robust Motion at Ionasec et al.
Estimation Using Trajectory Spectrum Learning: Application to Aortic and
Mitral Valve Modeling from 4D TEE”Proceedings of 12th IEEE International
Conference on Computer Vision, 2008, the 1601-1608 page described in more detail described TSL and calculate
Method, by with reference to being incorporated into this.In a final step, the border performing complete heart surface in whole cardiac cycle is retouched
Paint.The regulation of this method (leverage) edge detector of robust together with cooperation tracker and motion manifold.
Fig. 4 illustrates from one of CT of the many phases sequence estimation exemplary Hearts model specific to patient.
As illustrated in fig. 4, image 402,404 and 406 shows left ventricle, left atrium, right ventricle, right atrium, master
Arterial valve (AV), Bicuspid valve (MV), valve of pulmonary trunk (PV) and Tricuspid valve (TV) and ascending aorta and pulmonary artery.Although Fig. 4 does not has
Whole aorta is shown it should be appreciated that be, it is also possible to as the 4D anatomical model specific to patient a part come
Estimate aorta.For example, it is possible to use the method described in the number of patent application 2010/0239148 of U.S. Publication is come
Estimate aorta, by with reference to the disclosure of which is incorporated in this.
Returning to Fig. 1, in step 106 place, (blood moves the blood flow emulated in heart based on the 4D anatomical model specific to patient
Mechanics).In order to emulate blood flow, it is defeated that the described geometry specific to patient serves as 3D Navier-Stokes solver
Entering, described solver derives the actual hematodinamics retrained by topology along whole Cardiac cycle.
Hematodinamics as described herein calculates and uses classical continuous model for blood.Level set formula utilizes
Directly numerical simulation solves incompressible with viscosity term as the standard continuity mechanics model for fluid stream
Navier-Stokes equation (equation below (1)):
(1)
Navier-Stokes is to describe the momentum for fluid stream and the partial differential equation of the conservation of mass, and it depends on stream
Speed u of body and pressure p and fluid density ρ and dynamic viscosity μ.Density of blood and dynamic viscosity can be configured to the most strong
The general average that health is individual, i.e.With.The numerical discretization on uniform lattice is utilized to ask
Solve described equation, use both finite difference and limited bulk technology.Specifically, embodiments of the invention make use of with for
The approximation of pressure projects the combined method of fractional steps.
According to a kind of advantageous embodiment, blood is modeled as Newtonian liquid.Previous numerically modeling has been found that
In bigger tremulous pulse, the non-newtonian behaviour of blood is inessential in the period of major part cardiac cycle.Non newtonian importance factors (its
It is defined as the non-Newtonian viscosity of blood and the ratio of Newtonian viscosity) become during the sub-time period of the 15-20% of cardiac cycle
Important, during deceleration time section speed close to zero time reach peak value.Hematodinamics in heart mainly by high speed (or
The high shearing that person is more accurate) determine.The rheological behavior that this point adds blood is (a kind of close to Bingham plastic fluid
As rigid body performance but at the heavily stressed lower material as viscous fluid flow under low stress) the thing of rheological behavior
Real, support following qualitative conclusions: during Cardiac cycle, the kinetics of blood is mainly Newtonianism.Possible exception is: time
Between aspect have a diastasis, and space aspect has the apex of the heart and injection stagnant wake.
Calculate mobile in general geometry and/or relate to the hydrodynamics of stream of many phases and constitute and challenge, spy
When not being robust and accurate computational methods conception is simple especially true.Level Set Method is at the such complicated meter of reply
Obtaining certain success during calculation, it can capture between different materials (between such as water and air, or the deformation pair of complexity
Between liquid) the fine kinetics of the complexity at interface.Navier-Stokes equation is designated as retouching in view of such
The form stated.For example, for having two kinds of fluids of different densities and viscosity, Navier-Stokes can be expressed as
The level set formula of equation:
(2)
In equation (2), H is the sharp-pointed difference numerically being used to produce between first fluid and second fluid
Heaviside function, wherein first fluid is by level setOn the occasion of sign, second fluid is by level setNegative value characterize.
Especially when the advanced method utilizing such as Level Set method method etc is implemented, described level set formula is more several than classical formulas has
Individual advantage.For example, the advantage of level set formula includes: be prone to calculated free surface stream and the deformation material mutual with fluid
Material, for the simple enforcement of tensor extrapolation technique, and for such as normal direction field () or average curvature
(such as the level set function of unit gradient) etc the simple formula of various geometric parameters.
In another important advantage of cardiodynamics described in level set framework is that likely reply is mobile automatically
/ wall of serious deformation, without grid previously become excessively deflection thus when robustly cannot tackle numerical computations
(being exactly such as this situation in FEM (finite element) model (FEM)), generates new computation grid at every several time steps.This just permits
Permitted the heart grid specific to patient to be automatically integrating in blood flow simulation engine, and provided for for obtaining specific to trouble
A kind of Clinical practice framework of the hemodynamic heart model of person.
An advantageous embodiment according to the present invention, utilizes level set to be modeled heart wall/blood interface.More accurate
Ground is said, utilizes the spline function heart with certain number of frame (such as 10 frame) to obtaining the most at step 104
Dirty grid sequence carries out interpolation, in order to derive the described grid at given simulation time.By means of level set functionBy this net
Lattice are embedded in computational fields, described level set functionIt is defined as, wherein dx is between grid
Away from.Heart/the blood interface used in code is defined as the zero level of described level set function, thus actually each
On side, original triangle gridding " is thickeied " dx.Fig. 5 illustrate as level set be embedded in rectangular domain 504 specific to
The Hearts model 502 of patient.Fig. 5 shows the transparent zero level corresponding to the heart wall thickeied.
Described interface location is used to apply without slip boundary condition to fluid zone.Meter can be easy to by following operation
Calculate the Grid Velocity at each time step: the grid position at adjacent time step carries out temporal interpolation, followed by outward
Slotting kernel carries out extrapolation.It should be mentioned that, can effectively implement from solid-state by applying boundary condition in this manner
Heart grid transmits to the unidirectional momentum of fluid.Heart is substantially modeled as pushing to the pumping of territory wall by this method, and it is permissible
The resistance of approximation cycle system.
Fig. 6 illustrates according to an embodiment of the invention a kind of based on the Hearts model specific to patient
The method that blood flow in heart is emulated.The method of Fig. 6 can be used to implement the step 106 of Fig. 1.The method of Fig. 6 is detailed
Carefully describe the solution of Navier-Stokes equation (2) in described level set framework.The method of Fig. 6 can be used to solve for the heart
All stages in dynamic cycle, including each isovolumetric phase (IP).It should be mentioned that, IP is not quiescence periods, but intracavity stream
The period of dynamic change.In the method for Fig. 6 can be used to emulate whole cardiac anatomy or one or more heart
Blood flow in ingredient.In the description of Fig. 6.
In step 602 place, at initial time (n=0) place based on heart mould specific to patient described at this initial time
The position of type (i.e. grid) determines initial level collection condition.Described above is for level set functionZero level and speed u
Determination.Can determine that by solving Poisson equation initial pressure p, wherein said Poisson equation have based on grid
The Neumann boundary condition that position applies.
In step 604 place, it is incremented by the time step n from described level set, so that n=n+1.In step 606 place, meter
Calculate for level setConvection current (convective) with speed u updates.In this step, the grid at step preset time is utilized
Location updating level set numerical value.In middle geometry step, calculate each ingredient being connected defined by new level set.This
Be subsequently used in implicit expression viscosity and pressure Possion linear system are carried out robust be connected to form one by one part invert
(inversion).Then three rank accurate ENO(essence non-oscillatories are utilized) technology calculating convection current power item.
In step 608 place, speed is carried out half implicit expression renewal to contribute in view of viscous force.Specifically, two are utilized
Speed is updated by rank half implicit expression decomposition, and utilize efficient multi-grid preconditioning conjugate gradient solver to described system
System is inverted.
In step 610 place, update pressure by utilizing Neumann force boundary condition to solve described Poisson equation.This
Outward also by utilize efficient multi-grid preconditioning conjugate gradient solver that described system is inverted and in described discretization
Each of territory is connected to form in part and solves described Poisson equation.Carrying out before system inverts, utilizing solid speed weight
Write the speed in solid area.
In step 612 place, calculate the new speed at current time step and update.Specifically, new level set is utilized more
New density, and in a liquid the renewal of described speed is calculated as, or in solid (solid)
It is calculated as.Interface location be used to convection cell (fluid) district apply without slip boundary condition, i.e. for
The convection current (step 606) of the Navier-Stokes equation of momentum and viscosity (step 608) ingredient are
, and for pressure Poisson equations (step 610) be.Here,It is as described above
The normal vector field calculated from level set.The global precision of described method is second order inside described territory, and at boundary
This global precision drops to single order.
In step 614 place, it is determined whether complete emulation for whole Cardiac cycle.If not yet for whole heartbeat
Cycle completes this emulation, and the most described method returns step 604, and is incremented by another time step.Subsequently for the next time
Step performs step 606-612 again.If completing emulation for whole Cardiac cycle, the most described method proceeds to step
616.In step 616 place, Output simulation result.
The heart of the mankind serves as a part for cardiac system, thus can be loaded by from Venous system and Arterial system
The impact of pressure.Simulation frame described above solves the Navier-Stokes equation of (wall) Boundary motion with regulation, from
And guarantee to calculate pressure field, but instead can calculate its gradient, this is because pressure be one relatively rather than absolutely
To variable.In other words, as long as its gradient does not changes, first Navier-Stokes equation (conservation of momentum) in (2) is right
Metering movement in pressure is exactly constant.Mathematically, the pressure Poisson equations with Neumann boundary condition has one
Individual one-parameter solution race, just can be fixed once after introducing Dirichlet boundary condition.Therefore, pressure is being solved
During Poisson equation, described solution can be fixed by selecting the basic pressure (equal to zero) at the point outside aorta, and
It to apply the similar fashion effect of p=0 in aorta outflow surface.
It should be mentioned that, any numerical value can be selected for described basic pressure, such as, can utilize from general raw
Manage the numerical value of curve, and this does not interferes with the hematodinamics in regulation moving frame given above.But this will shadow
Ring the kinetics in FSI problem described below, wherein the change of pressure side dividing value can be responded deformed microstructure, or this will
Can affect the kinetics in dual reflux situation, wherein aortic valve and Bicuspid valve (or its eight sides homologue) both of which are let out
Leakage.For such situation, it is necessary to described model is coupled with one-dimensional vascular model.
Fig. 7 illustrates the exemplary blood flow simulation result of the method utilizing Fig. 6.The image 702-716 of Fig. 7 is profit
With the emulation generation of a series of blood flows, the emulation of wherein said blood flow utilizes the Hearts model specific to patient.In this embodiment,
Left side is connected not over system circulation with right side, this is because lack the material data for complete vascular tree and geometry knot
Structure.Therefore, the every side for heart individually performs described emulation, and this is computationally probably efficiently, and not by heart
Slightly different kinemic impact in the left side of model and right side.The image 702-716 of Fig. 7 shows at Cardiac cycle
Beginning heart contraction 702, heart contraction mid-term 704, heart contraction later stage 706, it illustrates that lung moves to start diastole 708(
Arteries and veins lobe refluxes), diastole in early days 710, diastole mid-term 712, diastasis 714 and Bicuspid valve hyperemia later stage 716
The vorticity (vorticity magnitude) of the left and right cardiac flow in the stage.
Returning to Fig. 1, in step 108 place, blood flow estimating of fluid structure based on emulation is mutual.Can be mutual by fluidic structures
It is used in conjunction with (step 106), in order to emulation is specific to each heart in the 4D model of patient with hematodinamics emulation
The deformation of ingredient.Specifically, can by describe the most in step 106 be used for hematodinamics emulation framework with
Model for the biomechanics characteristic of ad hoc structure is coupled.Each heart ingredient can be modeled as one passively
Tissue, its motion is determined by this structure law.FEM (finite element) model (FEM) can be used to solve the partial differential relevant with described law
Equation.The wall motion of ad hoc structure (such as aorta) is driven by following two power:
1, the internal force that the passive characteristic of described tissue is modeled.It is for instance possible that by one linear, respectively to
The structure of such as aorta etc is modeled by the single layer elastomeric model of the same sex, and wherein said elastic model has corotation school
Just to tackle relatively large deformation.It is also possible to utilize more detailed model to emulate the heterogeneous ingredient of aorta (respectively to different
Property, three main stor(e)ies (inner membrance, middle film and adventitia), non-linear etc.).
2, the external force that the loading produced by the blood flow of described inside configuration is modeled.It is converted into described in pressure and adds
Carry the internal layer being applied to described structure.
It is mutual that Fig. 8 illustrates a kind of fluidic structures for estimating aorta according to an embodiment of the invention
(FSI) alternative manner.It should be appreciated that other ingredients of described heart model can be applied similar method with
Estimating of fluid structure is mutual.In step 802 place, current location based on aorta wall emulates blood flow in preset time at step.Can
In order to by the method for above-mentioned Fig. 6 based in the described aorta wall 4D anatomical model specific to patient at current time step
Position emulate blood flow.In step 804 place, at current time step, calculate the pressure of wall interface.It should be mentioned that,
In the method for Fig. 6, at each time step, calculate the pressure at blood/structural interface.In step 806 place, based on wall interface
The calculation of pressure at place is for the deformation of aorta.Described pressure on aorta wall, and causes aorta as External Force Acting
The wall motoricity being based upon the modeling of this aorta deforms upon.In step 808 place, described time step is incremented by (n=n+1), and institute
Method of stating returns step 802.Correspondingly, the deformation of the aorta calculated is used at next time step emulate blood flow.
Repeat described method, until emulation terminates.For example, described method can be repeated, until one complete cardiac of emulation
Till cycle.
Fig. 9 illustrates and estimates the example results mutual for the fluidic structures of aorta.Image 902 shows spy
The regularly emulation of the blood flow in the aorta at spacer step.Image 904 shows that the fluidic structures that blood flow based on image 902 emulates is handed over
Mutually, it causes the aorta wall 906 deformed upon.
Can use FSI emulation come by by FSI framework described above and inverse problem strategy (such as Kalman filter or
Confidence region technology) it is coupled and estimates the inherent character of structure, such as organize hardness.In this way it is possible to regulation for
The model parameter of the motion of described structure, so that the simulating sports of the parameter in FSI emulation is observed with in medical image
The movement which matches arrived.A kind of possible embodiment is to minimize the cost letter that assessment simulating sports differs with observation
Number.So can obtain for the biomechanical parameter estimated by described structure, such as wall hardness.Patent in U.S. Publication
Application number 2011/0060576 is described in detail a kind of mutual and estimate the Biological Strength of aorta for implementing fluidic structures
The method learning parameter, by with reference to being incorporated into this.
The anatomical model specific to patient as described previously for each heart ingredient is smooth grid, and
Smooth grid based on each heart ingredient performs hematodinamics emulation and fluidic structures is mutual.According to the present invention one
Individual advantageous embodiment, the described anatomical model specific to patient may be implemented as multiple dimensioned anatomical model, is wherein described above
Smooth heart model be in rougher yardstick, and under finer yardstick, there is more detailed anatomical model.Lift
For example, the anatomical model specific to patient of LV can be used to extract the endocardial mould of LV under finer resolution
Type, including papillary muscles and bone trabecula.Figure 10 illustrates the multiple dimensioned anatomical model of LV.Image 1000 He such as Figure 10
Shown in 1010, the multiple dimensioned anatomical model of LV includes LV visceral pericardium 1002, coarse scale LV endocardium 1004 and fine chi
Degree LV endocardium 1006, it includes papillary muscles and bone trabecula.In the number of patent application 2009/0080745 of U.S. Publication in detail
Describe a kind of method for extracting such multiple dimensioned LV model, by with reference to being incorporated into this.Can based on for
The position of the fine dimension anatomical model of one or more heart ingredients retrains hematodinamics emulation (step 106).
Furthermore, it is possible to it is mutual to carry out estimating of fluid structure based on the fine dimension anatomical model for one or more heart ingredients
(step 108).
Return to Fig. 1, in step 110 place, export described comprehensive modeling result.For example, result can be exported to obtain
The 4D anatomical model specific to patient, blood flow emulation and/or fluidic structures interactive simulation, this can represent these by display
The image of result realizes.Additionally, the comprehensive modeling described in step 104,106 and 108 produces dissection and morphological parameters
(step 104), hemodynamic parameter (step 106) and biomechanical parameter (step 108), described parameter provides the heart of patient
The comprehensive observation of dirty function.These parameters such as can be stored in memorizer or the storage device of computer system, or
Person is stored on computer-readable medium.These parameters (or model) may be utilized for the such as step being described below
In 112 and 114, patient is further assessed.
In step 112 place, it is possible to use described comprehensive modeling result performs to evaluate for the non-intrusion type of patient's heart
And diagnosis.The modeling of step 104,106 and 108 provides the comprehensive observation of the current state for patient's heart.This can be by
For evaluating current structure or the hematodinamics of patient's heart, the heart disease of evaluation previous diagnosis, or for heart is asked
Topic carries out non-invasive diagnosing.
According to one embodiment of present invention, the described comprehensive heart model specific to patient can be used for accurately
Quantify anatomical structure and the function of heart.Clinical goldstandard still processes 2D image and performs manual measurement, described artificial survey
Amount acquisition is got up the most loaded down with trivial details and may inaccuracy.The present invention proposes model's transfer of the clinical assessment for heart, its
Middle substitute manual analysis based on 2D image with automatic based on model the quantization from 4D data.It is described below permissible
The various examples of the clinical measurement automatically extracted.
Table 1 shows the aorta-Bicuspid valve coupling for the various sizes measured from the 4D anatomical model specific to patient
The measurement result closed, the diameter of have a common boundary including ventricle-tremulous pulse boundary (VAJ), sinus of Valsalva (SV) and hole pipe (SJ),
Annulus of mitral valve girth (AC), front and back diameter (APD), and front outside-rear inside diameter (AL-PM-D).Figure 11 illustrates
For aortic area (AV district) and the certainty of measurement in mitral area (MV district).Specifically, Figure 11 shows for AV district
The Bland-Altman figure in 1102 and MV districts 1104.The experiment of described aortic valve is to perform in the CT data of 36 patients
, and described Bicuspid valve be in the tee data of 10 patients assessment.
Table 1
| Average | STD | |
| VAJ(cm) | 0.137 | 0.017 |
| SV(cm) | 0.166 | 0.043 |
| STJ(cm) | 0.098 | 0.029 |
| AC(cm) | 0.846 | 0.3 |
| APD(cm) | 0.325 | 0.219 |
| AL-PM-D(cm) | 0.509 | 0.37 |
Chamber motor pattern during cardiac cycle provides many important clinical of its function to measure, such as Ve
Asynchronism in mark, MWT and chamber or between different chamber.Some benefits of analysis based on model
Including precision, efficiency and comprehensive.
Alternatively possible application for the comprehensive 4D model specific to patient is for automated diagnostic and case
Retrieval.Clinical appraisal is to a great extent based on from the general information of clinical guidelines and publication and rule set and from facing
The personal experience of bed doctor.In addition to quantitative characteristic discussed above, described comprehensive heart model can be utilized to
Discriminant distance function based on study is utilized to be derived automatically from advanced clinical information.Can formulate in comprehensive feature space and push away
Disconnected, this incorporates form and the function information of complexity.In the exemplary embodiment, this is used to perform two general tasks:
Utilizing the distance function retrieval similar cases learnt, described distance function measures the similarity of two specific heart shape；With
And based on geometric model and the binary classification problems of derivation feature.
Distance study is here discussed to two kinds of technology, i.e. from equivalence constraint and intrinsic random forest distance
Practise, but the invention is not restricted to this.Utilizing tlv triple to represent that equivalence retrains, described tlv triple is the spy of two model instances
Levy vector and show that said two example is a similar or dissimilar label.Carry out usually learning from these tlv triple
It is referred to as learning in product space, and is proved to for having many relevant, the weak relevant and height of uncorrelated features
It is effective for dimension data.The signed surplus utilizing the model of propelling or random forest structure is used as required
Distance function.
Described conventional method realizes study and depends on any user-defined similarity concept of application.This point can be led to
Cross two kinds of exemplary application to prove: 1) for the diagnosis of aortic valve and seriousness evaluation；And 2) for percutaneous pulmonary
Lobe is implanted the patient of (PPVI) and is selected.Selecting about the patient for PPVI, the form of pulmonary trunk is the PPVI for patient
One main determining factor of fitness.Patient can be made inserted by unnecessary intrusive mood in unaccommodated intervention in the patient
Guan Shu.Figure 12 illustrates various types of pulmonary trunk form.As shown in Figure 12, Class1 is pyramidal, type
2 is constant diameter, and type 3 is reverse pyramid, but type 4 is narrow-minded the widest in being, and type 5 is
In but not lend oneself to worry and anxiety the narrowest.Patient from Class1 is considered to be not suitable for PPVI, this is because narrow tremulous pulse relatively and
The high probability of equipment transportation.Correspondingly, from estimated pulmonary trunk, (it can be as the described anatomical model specific to patient
A part estimate) shape facility that extracts can be used to learn discriminant distance function, in order to by the dissection of Class1
Structure distinguishes with other classification, in order to automatically determines patient and is appropriate for PPVI.
Except in addition to the Hearts model of patient, the hematodinamics specific to patient of heart can also be by
For non-intrusion type evaluation and diagnosis.Specifically, the hematodinamics of cardiac cycle event can be emulated, as above at figure
As described in the step 106 of 1.It is described herein as the example simulation for cardiac cycle event.Utilize 1283Rule
Grid carries out discretization to emulation territory, and these are 128 years old3Regular grids is corresponding to the physical resolution of 1mm, and time step is then 0.001
Second.In order to ensure stability, when maximal rate will be greater than dx/0.001, i.e. as Ke Lang-Friedrich Si-Lie Wei
When (Courant-Friedrichs-Lewy, CFL) number max (u) * 0.001/dx goes above 1, just use subcycle.Heart
Bounding box occupies the 95% of described territory.
Figure 13 illustrates the hematodinamics for heart contraction event and structure and emulates.As shown in Figure 13, figure
As in 1302,1304 and 1306 hearts respectively illustrating heart contraction early stage, heart contraction mid-term and end-systolic
Emulation blood flow rate.Image 1312,1314 and 1316 respectively illustrates based on heart contraction early stage, heart contraction mid-term and heart
The faces such as the whirlpool that end-systolic simulation velocity generates.It is performed the specific heart heart at 925ms of described emulation
The dynamic cycle has the heart contraction of 290ms.As shown in image 1302 and 1312, shrink (IVC) from diastole with waiting to hold
The initial flow condition that phase inherits has the residual of the ring-type whirlpool from the Bicuspid valve hyperemia later stage.This feature has dominated heart receipts
The ventricle bottom pattern of contracting (ES) in early days, so that apex of the heart blood pool major part is by described whirlpool recirculation.Ironically, institute
Stating initial condition and be further characterized in that as viewed from the apex of the heart it is to rotate counterclockwise, it is refunded during heart contraction and turns clockwise.
As shown in image 1304 and 1314, heart contraction is characterised by powerful aorta flux mid-term, and it is usually
Having the multiply whirlpool being directed at aorta axle, its guide blood enters the right-handed helix motion of aorta.Blood this inverse
The downstream that hour hands are rotated in ascending aorta bottom is continued, and is a well-known characteristic of healthy aorta stream.These meters
Calculation result is based on the grid obtained from CT data, and does not the most include twist motion.Described result of calculation support is to draw a conclusion:
Aorta right hand spiral rotating mainly by the aorta longitudinal axis about aortic valve base plane and aortic valve geometry self
Geometry is disposed rather than LV twist motion determines.As shown in image 1306 and 1316, at end-systolic, described vortex
Stock is more weak, and the fluid particles entering aorta defines broader spiral path.
Described flux is the most powerful at the heart contraction early and middle portion of right side heart, and it has similar with described above
Feature, but due to backflow valve of pulmonary trunk, at the end of paradoxical expansion it can be seen that backflow flux.
The whirlpool that Figure 14 illustrates in the aortic valve hole district utilizing blood flow emulation to obtain is formed.As Figure 14 schemes
Solving as explanation, image 1400 shows the position in the aortic valve hole district 1402 in the anatomical model specific to patient of heart
Put, and image 1410 shows that the whirlpool in hole district 1402 is formed.As shown in Figure 14, the emulation of described blood flow recovers aorta
The known formation of the stream mode of lobe far-end, the whirlpool at each the lobule rear being i.e. formed in hole district 1402.
Figure 15 illustrates the hematodinamics emulation of diastole event and structure.As shown in Figure 15, image
1502,1504 and 1506 respectively illustrate in the heart in diastole early stage, diastole mid-term and Bicuspid valve hyperemia later stage
Emulation blood flow rate.Image 1512,1514 and 1516 respectively illustrates based on heart contraction early stage, heart contraction mid-term and heart
The faces such as the whirlpool that end-systolic simulation velocity generates.As shown in image 1502 and 1504, diastole (ED) flux in early days
Starting from opening along with Bicuspid valve and forming asymmetric ring-type whirlpool, described ring-type whirlpool is about in the link apex of the heart and annulus of mitral valve
The axle of the heart (AMR) becomes about 30 degree of angles to advance further towards the apex of the heart.After the rear wall striking LV, described whirlpool is by one
Fixed dissipation and its axle change over and become about 75 degree about AMR.As shown in image 1504 and 1514, this develops into have water
The swirl pattern compared with the inswept apex of the heart of Maelstrom of flat rotary shaft, it is to be located exactly at the rotation in the opposite direction below aortic valve
The less whirlpool turned terminates.The formation of this less whirlpool is strengthened by (opening) mitral front lobule.This swirl pattern is at the heart
Gradually weaken during dirty mid-diastolic, become the most visible subsequently.As shown in image 1506 and 1516, it is successfully acquired two
The cusp hyperemia later stage.It produces extra ring-type whirlpool, described extra ring-type whirlpool be travel downwardly and with as positive ordinary affair
The less pulmonary vein reflux of part occurs simultaneously.
Described flux is the most similar in right side heart.Compared with Bicuspid valve flow curve, the emulation contracting captured
Short Tricuspid valve flow curve diastasis, as shown in the curve chart 1610 in Figure 16.
Blood flow described above emulation is supported to draw a conclusion: tip whirlpool is to flow out from Bicuspid valve during diastole
's.This is not partly inconsistent with following observation: lobule self is directed at stream, and thus does not flow out whirlpool and do not manipulate described
Stream.This difference is likely due to include that the enhancing heart model of the present invention of accurate leaflet of mitral valve causes.Described
Emulation stream shows, Bicuspid valve plays manipulation effect and also functions to whirlpool outflow effect.
In order to assess the precision of hematodinamics emulation, each valve is estimated blood flux.Typical PC-MRI flow
Changing agreement utilization is made the MR imaging plane being directed at anatomical structure gather blood flow over time by operator.In order to perform and document
The comparison of observation of report, is emulated this agreement, and utilize the plane being directed at each valve according to for PC-
The identical mode that MRI is completed quantifies the blood flow calculated with time correlation.Calculate amassing of the normal velocity in this plane
Point, and depict the curve that the case checked is obtained in figure 16.Figure 16 illustrate illustrate one aroused in interest
The curve chart of the time flux of the blood flow through valve area in the cycle.Curve chart 1600 illustrates through aortic valve 1602 and two
Stream in the left cardiac in cusp 1604 district.Curve chart 1610 shows through valve of pulmonary trunk 1612 and the right side in Tricuspid valve 1614 district
Stream in the heart of side.Curve chart 1600 is the most similar to the normal stream curve for normal heart with the stream shown in 1610,
Exception therein is pulmonary artery stream, and it demonstrates backflow above-mentioned.Figure 17 and 18 illustrates and is used to described in measurement
The site of the slice position of stream.The image 1702,1704 and 1706 of Figure 17 illustrates the cardiac cycle at 0.92 second respectively
0.18,0.3 and 0.42 second time velocity at the cross-sectional slice 1608 of aorta.Image 1802 and 1804 illustrates
The slightly below cross-sectional slice 1806 in mitral district.
Use the most not it should be mentioned that, measure average or peak velocity clinical practice for (such as in ultrasonic)
Same measuring point (before aorta position is slightly in valve area, and the Bicuspid valve visual field is slightly under Bicuspid valve).But
Here measure blood flux (integration of the normal velocity on given surface), therefore select with clinical measurement position meaningfully
The each position of skew (but the most close), so that they: described heart model is divided by (1) along the direction of described stream
Become the district of two topological separation；And (2) do not span across valve area, calculate (or at least make it because this may pollute flux
More complicated).
Due to just at the externally measured flux of LV, the most described flux be likely to be due to aorta or the radial motion in atrium and
The ingredient changed including volume.Although suitable is little, but these ingredients are shown as going out the most really simultaneously
Existing aorta-Bicuspid valve (or pulmonary artery-Tricuspid valve) stream district.For the same reason, isovolumetric phase not may utilize this agreement
Accurately quantify, and the most invisible (Isovolumic Relaxation in particularly LV).
Velocity mode in the measurements is extremely complex, as in Figure 17 and 18 it can be seen that as, and this address
Use the importance of correct geography information especially for each valve, thus anyone can minimize for flowing in or out
The use of imperfect model.Geometric model including each valve can be easy to calculate complicated stream mode.
Figure 19 illustrates the blood flow emulation for having two point backflow aortic valve and the mitral heart of morbid state, its
Experience narrow and reflux the two.Image 1902-1920 illustrates and utilizes the left cardiac model derived from 4D CT data
The whirlpool for a Cardiac cycle obtained is emulated from blood flow.In image 1902 and 1904, left ventricle is relaxed, and Bicuspid valve is beaten
Open, and blood enters left ventricle, due to narrow and strike wall.Relatively placidity in image 1906, and image 1908 shows
Go out diastasis.In image 1910 and 1912, Bicuspid valve second time is opened, and more blood enters left ventricle,
Again strike wall.In image 1914,1916 and 1918, left ventricular contraction, aortic valve opens, and blood enters aorta,
And left atrium starts hyperemia simultaneously.Stronger spiral is formed due to two point structures of aortic valve.In image 1920, main
Arterial valve close, and reversely penetrate blood show backflow.As shown in image 1916 and 1918, blood flow patterns with under ascending aorta
In portion, the blood flow patterns of normal heart is very different.The heart contraction pointed to generally along the centrage of aorta is penetrated blood and is biased
Aorta wall, thus increase partial wall stress.This can explain why the patient with double small leaf aortic valve can suffer from
Aortic root is expanded.Additionally, as shown in image 1902,1904 and 1912, experience narrow and the two described ill two that reflux
Described stream is guided LV wall backward by cusp.As shown in image 1914 and 1918, can be respectively at heart contraction and diastole
Period observation aortic valve and mitral backflow.
Figure 20 illustrates the hemodynamic comparison of emulation for a healthy heart and two diseased hearts.As
Shown in Figure 20, the first row image shows that healthy heart is relaxed at heart contraction early stage 2002, heart contraction later stage 2004, heart
Open the simulation velocity field of the left side of heart in early stage 2006 and diastole hyperemia later stage 2008.Second row image shows have expansion
Open the heart of aorta at heart contraction 2012, heart contraction later stage 2014, diastole early stage 2016 and diastole in early days
The simulation velocity field of the left side of heart in congested later stage 2018.The third line image shows that the heart with bicuspid aortic valve is at the heart
Dirty contraction 2022, heart contraction later stage 2024, diastole early stage 2026 and the heart in diastole hyperemia later stage 2028 in early days
The simulation velocity field in left side.
Described diseased heart is characterised by, has reflux phenomenon in both each valve and pulmonary vein district.For health
Only reflux of heart is the less pulmonary vein stream reflux during the Bicuspid valve hyperemia later stage, and it's a normal phenomenon.With
Diseased heart is compared, and is as short as almost 1/2nd for the heart contraction of healthy heart.This is phenomenon known to one, wherein suffers from
Sick heart develops longer systolic cycle, in order to offset the anatomic defect causing backflow and poor efficiency pump blood.
Healthy heart (image 2002-2008) has extremely short heart contraction in the cardiac cycle of 923ms
(190ms).During described heart contraction, aortic flow is powerful, as the stream in left atrium.Diastole starts from
Through mitral high current, it is centrally formed main rotation whirlpool in left ventricle in the meantime, and in the porch of aortic valve
Form less whirlpool.The Bicuspid valve hyperemia later stage occurs with the less pulmonary vein reflux as normal phenomenon simultaneously.
The heart (image 2012-2018) with expansion aorta is characterised by the aorta that seriously expands.Therefore,
Aortic valve from not completely closed, thus causes a large amount of aortic regurgitation for this specific heart during diastole.At lung
But the backflow of certain less exception also can be there is in the level of vein and mitral area.One significant fact is, at the heart
During dirty contraction, described stream is directed directly and expands district to the exception of aorta, thus which proposes and results in another
Individual problem.It is to say, be stream described in weak valve inclined lead thus cause expanding aorta, or aorta dies down and pulls
Aortic valve and change the direction of described stream.
For having the heart (image 2022-2028) of bicuspid aortic valve, can observe that output penetrates blood towards actively
The deflection of astillen, this can explain that the patient with double small leaf aortic valve can suffer from the fact that aortic root is expanded.Additionally, institute
It is also not enough for stating Bicuspid valve, thus causes the backflow towards left ventricle observed when diastole starts to penetrate blood and (see
Image 2024).Simulation result is it is also shown that aortic valve and Bicuspid valve asynchronous open and close, and this facilitates from ventricle
AR.
By utilizing framework described in Fig. 1 and 6 above, utilize the left side heart and do not have with atrium geometry
Dirty model performs emulation (every other parameter is the most equal).Figure 21 illustrates to be had and not to have left atrium (LA)
Blood flow simulation result.As shown in Figure 21, image 2102,2104,2106 and 2108 respectively illustrate for heart contraction mid-term,
Diastole early stage, diastole mid-term and the simulation result with LA in Bicuspid valve congestive heart stage.Image 2112,
2114,2116 and 2118 respectively illustrate for heart contraction mid-term, diastole in early days, diastole mid-term and Bicuspid valve fill
The simulation result without LA of blood heart phase.As shown in image 2102 and 2112, heart contraction stream has LA and not
In the case of there is LA nearly identical.Described emulation illustrates, mainly generates from pulmonary vein and is transported downstream to LV
Whirlpool for emulation blood flow tool have a significant impact.This point is clearly shown that in image 2104 and 2114, and it is also shown in
In the case of not having the conveying of LA to LV whirlpool, the emulation stream within LV is likely not to have bright between with or without the model of LA
Significant difference is other.This is the most significant, because this suggests that a kind of for actual boundary condition being entered in the case of there is no LA
Possible the solution of row modeling, this is by including for the whirlpool generation within LA and the suitable model to the conveying in LV thereof
And realize.
It should be appreciated that can be combined with fluidic structures to perform hematodinamics described above emulation alternately, with
Just simulation accuracy is improved.For example, model can consider that the actual geometry of the wall for heart ingredient recovers,
For the soft tissue model of the wall of heart ingredient, and the elastic characteristic for valve.As it has been described above, fluidic structures is mutual
Can be used to estimate the biomechanics characteristic of structure, such as organize hardness.So estimate biomechanics characteristic can by with
In diagnostic purpose.For example, by assessing the hardness of the structure of such as aorta etc, anyone can evaluate rupture, high
The risk of blood pressure etc..This information can also open the passage leading to more accurate Aneurysmformation model, wherein said life
Thing mechanical characteristic is revised by pathology.
Return to Fig. 1, in step 114 place, it is possible to use described comprehensive modeling result performs virtual therapy planning and/or disease
Disease progression is predicted.Described comprehensive model gives the comprehensive observation of the heart body for patient.This can be used to pre-
Survey specified disease and how can affect the heart of patient, or prediction patient is to the response of certain types of therapy or treatment such as
What.Figure 22 illustrates a kind of utilization according to an embodiment of the invention to be carried out specific to the comprehensive 4D model of patient
The method of predictability planning.The method of Figure 22 can be used to implement the step 114 of Fig. 1.With reference to Figure 22, in step 2202 place,
Generate the comprehensive 4D model specific to patient.Specifically, the described comprehensive model specific to patient is based on 4D medical science
View data generates, as above as described in step 104,106 and 108 of Fig. 1.Institute for particular patient
State comprehensive model and can include the dissection specific to patient, form, hematodinamics and biomechanical parameter.
In step 2204 place, regulate at least some of so that emulation is a kind of of the described comprehensive 4D model specific to patient
Situation.Described situation can be disease, therapy or treatment, or can affect any other shape of the parameter of described heart model
Condition.In order to emulate a kind of situation, in dissection, form, hematodinamics or biomechanical parameter or multiple can be regulated
?.For example, the anatomic parameter of size that can regulate such as aorta etc, in order to represent the progress of disease.As above institute
State, perform the emulation of various blood flow for having the heart of various disease, thus obtain the hematodinamics of diseased heart
Parameter.Likely regulate some hemodynamic parameter to emulate existence or the seriousness of the particular pathologies of heart.Additionally
The biomechanical parameter of the such as tissue hardness that can regulate specific heart ingredient etc.For example, master can be regulated
The tissue hardness of tremulous pulse, in order to represent the hardening of aorta.Described model can also be regulated to utilize described model to difference
Therapy or treatment carry out virtual test, thus predict patient's reaction to described therapy.It should be appreciated that these examples are not
It is intended to limit the present invention.
In step 2206 place, regenerate the comprehensive 4D model specific to patient based on the described part through overregulating,
To predict the impact on patient of the described situation.The method of Figure 22 can be repeated so that predicted conditions shadow within multiple time periods
Ring, or predict the multiple different situation impact on patient.For example, the method for Figure 22 can be repeated to utilize described spy
Due to the treatment alternative that the model measurement of patient is different, thus select the treatment of optimum for patient.Utilization is described below
Described comprehensive model carries out the example of therapy planning.It should be appreciated that these examples are not intended to limit the present invention.
In one embodiment, the comprehensive model specific to patient is used to come for percutaneous surgery (procedure)
Calculating judge support.It is reduced due to postoperative complication and the follow-up rate of patient is reduced, therefore percutaneous surgery
Just become to be becoming increasingly popular.In such operation, utilize through vein, through tremulous pulse or false by catheter delivery through apex of the heart technology
Body implant, this can hinder clinicist directly to observe and be immediately adjacent to affected anatomical structure.Therefore, the one-tenth of described intervention
Merit is largely dependent upon in art image and the experience of operator and technology, the suboptimal deployment of the most described prosthese
Position may cause poor hemodynamic performance together with serious lobe week leakage and/or high gradient and suboptimal to have
Effect mouth.
According to one embodiment of present invention, it is possible to use the described comprehensive model specific to patient is implemented for percutaneous
The preoperative planning of operation.Described anatomical model includes the dissection of aortic valve complex (it includes aortic valve and ascending aorta)
Structure, it is used to based on the simple form grid of deformation and geometrical constraint performing being carried out in computer of valve implant
(in-silico) (virtual) delivers.Described equipment is modeled as support grid and calculates grid, described support grid accurately mould
Intending the geometry of prosthese, described calculating grid is the 2 simple form grids being used to guide the superposition of the extension of described equipment.Pass through
The extension of described equipment is modeled by the external force utilizing iterative approximation method balance to run in actual surgery with internal force.Logical
Cross the finite discrete of second order differential equation to describe the deformation of described equipment.Figure 23 illustrates and makees during virtual deployment
It is used in the power on implant model.As illustrated in Figure 23, the arrow of image (a) represents, it is pole
Joint portion applies characteristic angle 2302.The arrow of image (b) represents, it keeps strut lengths.Image (c) shows
The short shaft section of described support grid.The arrow of image (c) represents, it applies girth 2304, simultaneouslySuppress and eliminate
Strong with the institute along described support grid normal direction effect of the ratio weighting of distance 2310 by distance 2306, wherein distance 2306
Being from described pole joint portion to support barycenter 2308, distance 2310 is to support barycenter 2308 from blood vessel wall 2312.
Based on the power shown in Figure 23, the implantation to prosthese carries out virtual emulation.It is each that this technology can be used to prediction
Plant optimal implant type, size and the deployed position and orientation treated under assuming.Except prediction prosthese dissection matching it
Outward, it is also possible to regenerate described comprehensive model, and the prosthese execution hematodinamics being virtually implanted described in utilization is imitative
True and fluidic structures is mutual.
Utilize preoperative and postoperative 3D heart scanning to have evaluated with 20 patients and described based on model be carried out at computer
In the predictive ability that substitutes of valve, this is by will predict the outcome with artificial fitting to post-procedure data for each patient
In the ground truth model of real equipment imaging be compared to assessment.Under the precision less than 2mm of ring level, for
This method demonstrates the potential supporting preoperative planning, and this is by treating under hypothesis by planting of being carried out in computer various
Enter to find optimal implant type, size and deployed position and orientation until observing that optimum prediction performance realizes
's.
The alternatively possible application utilizing the virtual therapy planning of the described comprehensive 4D model specific to patient is actively
The emulation of the stent deployment at aneurysm in arteries and veins.Figure 24 illustrates the imitative of the stent deployment at the aneurysm in aorta
Very.As shown in Figure 24, it is virtually implanted at the aneurysm in aorta 2404 and represents the grid 2402 of support.By disposing institute
The power stating support and generate makes aorta wall at the anchor portion generation local deformation of described support.Described fluidic structures is mutual
It is used to this deformation is modeled.This deformation is modeled for evaluating described stent anchors and intensity and the shadow to blood flow thereof
For sound it is critical that.Utilizing such framework, cardiologist can test different support Design and be patient
Select optimum support.
The alternatively possible application utilizing the virtual therapy planning of the described comprehensive 4D model specific to patient is virtual
Excision aortic aneurysm.Specifically, the excision aneurysm that this application simulation is carried out in the computer impact on blood flow.Can be
Aorta model (it includes personalized geometry and biomechanical parameter) specific to patient is loaded into real-time soft tissue
In Interference service platform.Utilizing this platform, user can perform virtual intervention, including excision, closes and sews up.Figure 25 illustrates
Illustrate that virtual aneurysm is excised.As illustrated in figure 25, image 2502 shows have aneurysmal active
Arteries and veins, image 2504 shows described aneurysmal excision, and image 2506 shows the stitching of described excision, and image 2508 shows
Go out the described excision of Guan Bi.FSI model is performed the most after surgery to emulate postoperative blood flow on geometry.By according to postoperative
Test described model in the data described model of regulation and after surgery data and verify proposed method.This frame proposed
Frame can help to prepare surgical intervention by the exact part that definition will be removed.
Described above for heart is carried out comprehensive modeling, generate heart the 4D anatomical model specific to patient,
Blood flow in emulation heart, emulation fluidic structures mutual and utilize comprehensive being predicted property of heart model to plan method all
May be implemented within computer, described computer utilizes known computer processor, memory cell, storage device, meter
Calculation machine software and other assemblies.Illustrate the high level block diagram of such computer in fig. 26.Computer 2602 comprises place
Reason device 2604, it controls described operation by performing the computer program instructions of the overall operation of definition computer 2602.Institute
State computer program instructions and can be stored in storage device 2612(such as disk) in, and perform described computer in expectation
It is loaded into during programmed instruction in memorizer 2610.Therefore, the method step of Fig. 1,3,6,8 and 22 can be by being stored in memorizer
2610 and/or storage device 2612 in computer program instructions define, and can by perform described computer program refer to
The processor 2604 of order controls.Can by image capture device 2620(such as CT scan equipment, MR scanning device, ultrasonic set
Standby etc.) it is connected to computer 2602, in order to view data is input to computer 2602.Likely by image capture device
2620 and computer 2602 be embodied as an equipment.It is also possible to make image capture device 2620 and computer 2602 pass through network
Carry out radio communication.Computer 2602 also includes that one or more network interface 2606 is carried out with other equipment will pass through network
Communication.Computer 2602 also includes other input-output apparatus 2608(example allowing user to interact with this computer 2602
Such as display, keyboard, mouse, speaker, button etc.).Can be by such input-output apparatus 2608 and computer program
Set is used in conjunction with, using as the annotation tool for explaining the body received from image capture device 2620.This
Skilled person is it will be recognized that the enforcement of actual computer also comprises other assemblies, and Figure 26 is such computer
The senior expression of some of them assembly is for descriptive purpose.
Detailed description of the invention above is to be understood as being the most all illustrative and exemplary and non-limiting
, and the scope of the present invention disclosed herein do not determines by described detailed description of the invention, but permitted by according to Patent Law
Claims that the complete range permitted is explained determine.It should be appreciated that shown here as and the embodiment that describes be only
The principle of the explanation present invention, and in the case of without departing substantially from scope and spirit of the present invention, those skilled in the art are permissible
Implement various amendment.In the case of without departing substantially from scope and spirit of the present invention, those skilled in the art can implement various its
He combines feature.
Claims (37)
1., for the method emulating the blood flow in heart based on 4D medical image, wherein said method is not used to
Diagnose or the purpose for the treatment of, and described method include:
The 4D anatomical model specific to patient of heart is generated according to described 4D medical imaging data；And
It is subject to by utilizing to solve at each time step in the middle of the level set framework multiple time steps in a Cardiac cycle
The Navier-Stokes equation retrained to the described 4D anatomical model specific to patient emulates the blood flow in heart,
Wherein, the 4D anatomical model specific to patient of described heart includes multiple heart ingredient, and by utilizing water
Solve by described specific to trouble at each time step in the middle of the flat collection framework multiple time steps in a Cardiac cycle
The Navier-Stokes equation of the 4D anatomical model constraint of person emulates the step of the blood flow in heart and includes for the plurality of
Each in the middle of time step operates below performing:
Based on the position of 4D anatomical model specific to patient described at current time step, calculate for level set function and speed
The convection current of degree updates；
Half implicit expression calculating the speed for representing the viscous force contribution at current time step updates；
Update by solving the pressure at the Poisson Equation for Calculating current time step with Neumann boundary condition；And
Update based on described half implicit expression speed and pressure updates, calculate the new speed for current time step and update.
Described method the most according to claim 1, wherein, the 4D anatomical model specific to patient of described heart includes multiple heart
Dirty ingredient, and by utilizing each time in the middle of the level set framework multiple time steps in a Cardiac cycle
Solve at step by the described Navier-Stokes equation retrained specific to the 4D anatomical model of patient to the blood emulating in heart
The step of stream includes:
One or more heart composition portions in the middle of multiple heart ingredients of the described 4D anatomical model specific to patient
Blood flow is individually emulated in Fen.
Described method the most according to claim 1, wherein, the 4D anatomical model specific to patient of described heart includes multiple heart
Dirty ingredient, and by utilizing each time in the middle of the level set framework multiple time steps in a Cardiac cycle
Solve at step by the described Navier-Stokes equation retrained specific to the 4D anatomical model of patient to the blood emulating in heart
The step of stream includes:
In each heart ingredient in the middle of multiple heart ingredients of the described 4D anatomical model specific to patient
Emulate blood flow simultaneously.
Described method the most according to claim 1, wherein, during by utilizing multiple in a Cardiac cycle of level set framework
Solve at each time step in the middle of spacer step by the described Navier-Stokes retrained specific to the 4D anatomical model of patient
Equation emulates the step of the blood flow in heart and includes:
Based on being used to the zero level of the described level set function being embedded in computational fields specific to the 4D anatomical model of patient
Position, the fluid zone of described anatomical model is applied without slip boundary condition.
Described method the most according to claim 1, wherein, the described 4D anatomical model specific to patient includes described Cardiac cycle
The 3D anatomical model sequence specific to patient of interior heart, and by utilizing level set framework in a Cardiac cycle
Solve at each time step in the middle of multiple time steps by the described Navier-retrained specific to the 4D anatomical model of patient
Stokes equation emulates the step of the blood flow in heart and includes:
The described 3D anatomical model sequence specific to patient is carried out interpolation, in order to derive in the middle of the plurality of time step at least
The position of the described 4D anatomical model specific to patient at one time step.
Described method the most according to claim 1, wherein, according to described 4D medical imaging data generate heart specific to patient
The step of 4D anatomical model include:
Generate the multiple dimensioned anatomical model specific to patient of at least one heart ingredient, described multiple dimensioned anatomical model bag
Include coarse anatomical model and the fine dissection of at least one heart ingredient described of at least one heart ingredient described
Model.
Described method the most according to claim 6, wherein, during by utilizing multiple in a Cardiac cycle of level set framework
Solve at each time step in the middle of spacer step by the described Navier-Stokes retrained specific to the 4D anatomical model of patient
Equation emulates the step of the blood flow in heart and includes:
By solving the Navier-Stokes equation retrained by the fine dissection model of at least one heart ingredient described
Emulate the blood flow at least one heart ingredient described.
Described method the most according to claim 7, wherein, at least one heart ingredient described includes left ventricle, and institute
The fine dissection model stating left ventricle includes papillary muscles and bone trabecula.
9. for the method that based on 4D medical image, heart is carried out the comprehensive modeling specific to patient, Qi Zhongsuo
The method of stating is not used to diagnosis or the purpose for the treatment of, and described method includes:
The 4D anatomical model specific to patient of heart is generated according to described 4D medical imaging data；
By utilizing level set framework to solve at least one the heart composition portion by the described 4D anatomical model specific to patient
The Navier-Stokes equation of the position constraint point at current time step and at current time step emulation described at least one
Blood flow in heart ingredient；
Based on the emulation blood flow at current time step calculate at current time step described in the shape of at least one heart ingredient
Become；And
Repeating described emulation and calculation procedure for multiple time steps, at least a part of which is based in part on and calculates at previous time step
The deformation of at least one heart ingredient described determine at least one the heart ingredient described at current time step
Current location.
Described method the most according to claim 9, it also includes:
Described in the deformation of at least one heart ingredient described in being calculated and described 4D medical imaging data at least
The deformation that the observation of one heart ingredient obtains compares, in order to determine the life of at least one heart ingredient described
Thing mechanics parameter.
11. described methods according to claim 9, wherein, at least one heart ingredient described is aorta.
12. described methods according to claim 9, wherein, by utilizing level set framework to solve by described specific to patient
The Navier-Stokes equation of at least one heart ingredient position constraint at current time step of 4D anatomical model
And the step of the blood flow in emulation at least one heart ingredient described includes at current time step:
The institute at current time step calculating the blood flow at least one heart ingredient described in being emulated and cause
State the pressure of the wall interface of at least one heart ingredient.
13. described methods according to claim 12, wherein, calculate current time step based on the emulation blood flow at current time step
The step of the deformation of at least one heart ingredient described at place includes:
Calculate the pressure of wall interface due at least one heart ingredient described and at least one heart described in causing
The deformation of the wall of ingredient.
14. described methods according to claim 9, wherein, calculate current time step based on the emulation blood flow at current time step
The step of the deformation of at least one heart ingredient described at place includes:
Internal force that passive characteristic based on the tissue at least one heart ingredient described is modeled and to by described extremely
The external force that the loading that few blood flow within a heart ingredient generates is modeled, calculates described at least one heart composition
The deformation of part.
15. 1 kinds of methods utilizing comprehensive 4D being predicted property of the heart model planning specific to patient, wherein said method
It is not used to diagnosis or the purpose for the treatment of, and described method includes:
The comprehensive 4D model specific to patient of heart is generated according to 4D medical imaging data；
A part for the described comprehensive 4D model specific to patient of regulation is to emulate a kind of situation；And
Regenerate the comprehensive 4D model specific to patient of described heart, in order to emulate the described part through overregulating to institute
State the impact of the comprehensive 4D model specific to patient.
16. described methods according to claim 15, wherein, a part for the described comprehensive 4D model specific to patient of regulation
So that the step emulating a kind of situation includes:
The described comprehensive 4D solution to model specific to patient of regulation cuts open parameter.
17. described methods according to claim 15, wherein, a part for the described comprehensive 4D model specific to patient of regulation
So that the step emulating a kind of situation includes:
The hemodynamic parameter of the described comprehensive 4D model specific to patient of regulation.
18. described methods according to claim 15, wherein, a part for the described comprehensive 4D model specific to patient of regulation
So that the step emulating a kind of situation includes:
The biomechanical parameter of the described comprehensive 4D model specific to patient of regulation.
19. described methods according to claim 15, wherein, a part for the described comprehensive 4D model specific to patient of regulation
So that the step emulating a kind of situation includes:
The described comprehensive 4D solution to model specific to patient of regulation cuts open parameter, hemodynamic parameter and biomechanical parameter
At least one, in order to the progression of disease in emulation heart.
20. described methods according to claim 15, wherein, a part for the described comprehensive 4D model specific to patient of regulation
So that the step emulating a kind of situation includes:
A part for the described comprehensive 4D model specific to patient of regulation, in order to the corresponding part of heart is applied by virtual emulation
Therapy.
21. described methods according to claim 20, wherein, a part for the described comprehensive 4D model specific to patient of regulation
So that the step of the corresponding part application therapy of heart is included by virtual emulation:
Implant specific to virtual emulation percutaneous artificial valve in the comprehensive 4D model of patient described.
22. described methods according to claim 20, wherein, a part for the described comprehensive 4D model specific to patient of regulation
So that the step of the corresponding part application therapy of heart is included by virtual emulation:
At described stent deployment at virtual emulation aneurysm in a part for the comprehensive 4D model of patient.
23. described methods according to claim 20, wherein, a part for the described comprehensive 4D model specific to patient of regulation
So that the step of the corresponding part application therapy of heart is included by virtual emulation:
Described specific to the aneurysmal excision of virtual emulation in a part for the comprehensive 4D model of patient.
24. described methods according to claim 23, wherein, regenerate the comprehensive 4D mould specific to patient of described heart
Type is to emulate the described part through overregulating and including the described step specific to the impact of the comprehensive 4D model of patient:
Aneurysmal excision based on described emulation, emulates in the described part of the comprehensive 4D model of patient described
Blood flow and fluidic structures are mutual.
25. 1 kinds for emulating the equipment of blood flow in heart based on 4D medical image, comprising:
For generating the device of the 4D anatomical model specific to patient of heart according to described 4D medical imaging data；And
For asking at each time step in the middle of the level set framework multiple time steps in a Cardiac cycle by utilizing
Solution by the described Navier-Stokes equation retrained specific to the 4D anatomical model of patient to emulate filling of blood flow in heart
Put,
Wherein, the 4D anatomical model specific to patient of described heart includes multiple heart ingredient, and for by profit
Solve by described specific at each time step in the middle of the level set framework multiple time steps in a Cardiac cycle
Navier-Stokes equation in the 4D anatomical model constraint of patient emulates the described device of the blood flow in heart and includes:
For based on described at current time step specific to the position calculation of 4D anatomical model of patient for level set function
The device that convection current with speed updates；
For calculating the device that half implicit expression of the speed for representing the viscous force contribution at current time step updates；
For calculating pressure at current time step and update by solving the Poisson equation with Neumann boundary condition
Device；And
Calculate what the new speed for current time step updated for updating based on described half implicit expression speed more to newly arrive with pressure
Device.
26. described equipment according to claim 25, wherein, for by utilizing level set framework in a Cardiac cycle
Solve at each time step in the middle of multiple time steps by the described Navier-retrained specific to the 4D anatomical model of patient
Stokes equation emulates the described device of the blood flow in heart and includes:
For based on being used to the zero of the described level set function being embedded in computational fields specific to the 4D anatomical model of patient
The position of level applies the device without slip boundary condition to the fluid zone of described anatomical model.
27. described equipment according to claim 25, wherein, for generating the specific of heart according to described 4D medical imaging data
Described device in the 4D anatomical model of patient includes:
For generating the device of the multiple dimensioned anatomical model specific to patient of at least one heart ingredient, described multiple dimensioned
Anatomical model includes the coarse anatomical model of at least one heart ingredient described and at least one heart ingredient described
Fine dissection model.
28. described equipment according to claim 27, wherein, for by utilizing level set framework in a Cardiac cycle
Solve at each time step in the middle of multiple time steps by the described Navier-retrained specific to the 4D anatomical model of patient
Stokes equation emulates the described device of the blood flow in heart and includes:
For by solving the Navier-Stokes retrained by the fine dissection model of at least one heart ingredient described
Equation emulates the device of the blood flow at least one heart ingredient described.
29. 1 kinds of equipment being used for heart carries out the comprehensive modeling specific to patient based on 4D medical image, its bag
Include:
For generating the device of the 4D anatomical model specific to patient of heart according to described 4D medical imaging data；
For by utilizing level set framework to solve at least one the heart group by the described 4D anatomical model specific to patient
Become part position constraint at current time step Navier-Stokes equation and at current time step described in emulation at least
The device of the blood flow in one heart ingredient, at least a part of which be based in part at previous time step calculate described at least
The deformation of one heart ingredient determines the current location of at least one the heart ingredient described at current time step；
And
At least one heart ingredient described in calculate at current time step based on the emulation blood flow at current time step
The device of deformation.
30. described equipment according to claim 29, it also includes:
Described in the deformation of at least one heart ingredient described in being calculated with described 4D medical imaging data
The deformation that the observation of at least one heart ingredient obtains is compared to determine at least one heart ingredient described
The device of biomechanical parameter.
31. described equipment according to claim 29, wherein, for by utilize level set framework solve by described specific to
The Navier-Stokes of at least one heart ingredient position constraint at current time step of the 4D anatomical model of patient
Equation and at current time step the described device of the blood flow in emulation at least one heart ingredient described include:
At the current time step caused for the blood flow that calculates at least one heart ingredient described in being emulated
The device of pressure of wall interface of at least one heart ingredient described.
32. according to the described equipment of claim 31, wherein, during for calculating current based on the emulation blood flow at current time step
The described device of the deformation of at least one the heart ingredient described at spacer step includes:
For calculate the pressure of wall interface due at least one heart ingredient described and described in causing at least one
The device of the deformation of the wall of heart ingredient.
33. described equipment according to claim 29, wherein, during for calculating current based on the emulation blood flow at current time step
The described device of the deformation of at least one the heart ingredient described at spacer step includes:
For calculating the device of the deformation of at least one heart ingredient described, wherein said internal force pair based on internal force and external force
The passive characteristic of the tissue of at least one heart ingredient described is modeled, and described external force is to by least one heart described
The loading that blood flow within ingredient is generated is modeled.
34. 1 kinds of equipment utilizing comprehensive 4D being predicted property of the heart model planning specific to patient, comprising:
For generating the device of the comprehensive 4D model specific to patient of heart according to 4D medical imaging data；
For regulating a part for the described comprehensive 4D model specific to patient to emulate the device of a kind of situation；And
For regenerating the comprehensive 4D model specific to patient of described heart to emulate the described part through overregulating
Device on the impact of the described comprehensive 4D model specific to patient.
35. according to the described equipment of claim 34, wherein, for regulating the one of the described comprehensive 4D model specific to patient
Part is so that the described device emulating a kind of situation includes:
Parameter, hemodynamic parameter and biomechanics ginseng is cutd open for regulating the described comprehensive 4D solution to model specific to patient
The device of at least one in number.
36. according to the described equipment of claim 34, wherein, for regulating the one of the described comprehensive 4D model specific to patient
Part is so that the described device emulating a kind of situation includes:
Parameter, hemodynamic parameter and biomechanics ginseng is cutd open for regulating the described comprehensive 4D solution to model specific to patient
At least one in number is so that the device of progression of disease in emulation heart.
37. according to the described equipment of claim 34, wherein, for regulating the one of the described comprehensive 4D model specific to patient
Part is so that the described device emulating a kind of situation includes:
For regulating a part for the described comprehensive 4D model specific to patient so that the virtual emulation corresponding part to heart
The device of application therapy.
Applications Claiming Priority (6)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US36629410P | 2010-07-21 | 2010-07-21 | |
| US61/366294 | 2010-07-21 | ||
| US38394210P | 2010-09-17 | 2010-09-17 | |
| US61/383942 | 2010-09-17 | ||
| US13/091,076 US8682626B2 (en) | 2010-07-21 | 2011-04-20 | Method and system for comprehensive patient-specific modeling of the heart |
| US13/091076 | 2011-04-20 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102346811A CN102346811A (en) | 2012-02-08 |
| CN102346811B true CN102346811B (en) | 2016-12-14 |
Family
ID=
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11869669B2 (en) | 2013-03-01 | 2024-01-09 | Heartflow, Inc. | Method and system for image processing to model vasculasture |
Non-Patent Citations (1)
| Title |
|---|
| PATIENT-SPECIFIC MODELING OF LEFT HEART ANATOMY, DYNAMICS AND HEMODYNAMICS FROM HIGH RESOLUTION 4D CT;Viorel Mihalef et al;《2010 IEEE International Symposium on Biomedical Imaging》;20100417;504-507 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11869669B2 (en) | 2013-03-01 | 2024-01-09 | Heartflow, Inc. | Method and system for image processing to model vasculasture |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8682626B2 (en) | Method and system for comprehensive patient-specific modeling of the heart | |
| JP5868052B2 (en) | Comprehensive patient-specific heart modeling method and system | |
| US20210241913A1 (en) | Machine learning system for assessing heart valves and surrounding cardiovascular tracts | |
| Votta et al. | Toward patient-specific simulations of cardiac valves: state-of-the-art and future directions | |
| Votta et al. | Mitral valve finite-element modelling from ultrasound data: a pilot study for a new approach to understand mitral function and clinical scenarios | |
| CN110085321B (en) | Method and system for multi-scale anatomic and functional modeling of coronary circulation | |
| Toma et al. | Fluid–structure interaction analysis of papillary muscle forces using a comprehensive mitral valve model with 3D chordal structure | |
| US9245091B2 (en) | Physically-constrained modeling of a heart in medical imaging | |
| US8527251B2 (en) | Method and system for multi-component heart and aorta modeling for decision support in cardiac disease | |
| Ma et al. | Image-based fluid–structure interaction model of the human mitral valve | |
| CN110268478B (en) | Methods and processes for providing subject-specific computational models for decision support and diagnosis of cardiovascular disease | |
| US20180174068A1 (en) | Method and process for providing a subject-specific computational model used for treatment of cardiovascular diseases | |
| Burlina et al. | Patient-specific modeling and analysis of the mitral valve using 3D-TEE | |
| Chandran et al. | Computational mitral valve evaluation and potential clinical applications | |
| Grbic et al. | Personalized mitral valve closure computation and uncertainty analysis from 3D echocardiography | |
| Voigt et al. | Patient-specific model of left heart anatomy, dynamics and hemodynamics from 4D TEE: a first validation study | |
| Burlina et al. | Patient-specific mitral valve closure prediction using 3D echocardiography | |
| Dimasi et al. | Influence of mitral valve anterior leaflet in vivo shape on left ventricular ejection | |
| CN102346811B (en) | For heart being carried out the method and system of the comprehensive modeling specific to patient | |
| Dahl | Numerical simulations of blood flow in the left side of the heart | |
| Verhey et al. | Feasibility of rapid and automated importation of 3D echocardiographic left ventricular (LV) geometry into a finite element (FEM) analysis model | |
| Soni et al. | A comprehensive review on CFD simulations of left ventricle hemodynamics: numerical methods, experimental validation techniques, and emerging trends | |
| CN112991548A (en) | Personalized mitral valve finite element modeling and simulation method, system and equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20140625 Address after: Munich, Germany Applicant after: Siemens AG Address before: new jersey Applicant before: Siemens AG Applicant before: Siemens AG |
|
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20220117 Address after: Erlangen Patentee after: Siemens Healthineers AG Address before: Munich, Germany Patentee before: SIEMENS AG |