Method and system for fast patient-specific cardiac electrophysiology simulations for therapy planning and guidance
View Patent ↗A method and system for patient-specific cardiac electrophysiology is disclosed. Particularly, a patient-specific anatomical model of a heart is generated from medical image data of a patient, a level-set representation of the patient-specific anatomical model is generated of the heart on a Cartesian grid; and a transmembrane action potential at each node of the level-set representation of the of the patient-specific anatomical model of the heart is computed on a Cartesian grid.
1. A method for patient-specific cardiac electrophysiology computations, comprising:
generating a patient-specific anatomical model of a heart from medical image data of a patient;
identifying at least one key electrophysiological structures of interest from the patient-specific anatomical model and medical image data;
generating a level-set representation of the patient-specific anatomical model of the heart on a Cartesian grid; and
computing a transmembrane action potential on at least one node of the level-set representation of the patient-specific anatomical model of the heart, wherein computing the transmembrane action potential is based at least in part on a stimulus current input as an initial condition.
2. The method of claim 1 , wherein the step of generating a patient-specific anatomical model of a heart from medical image data of a patient comprises:
detecting a patient-specific left ventricle endocardium model, a patient-specific right ventricle endocardium model, and a patient-specific epicardium model in the medical image data; and
fusing the left ventricle endocardium model, the right ventricle endocardium model and the epicardium model into a single bi-ventricular volumetric mesh.
3. The method of claim 2 , wherein the step of generating a patient-specific anatomical model of a heart from medical image data of a patient further comprises:
mapping spatial information corresponding to at least one of scars, grey zones, or fibrosis identified from medical-image data and other clinical measurements onto the bi-ventricular volumetric mesh.
4. The method of claim 2 , wherein the step of generating a patient-specific anatomical model of a heart from medical image data of a patient further comprises:
registering a tensor field of an in-vivo diffusion tensor magnetic resonance image of cardiac fibers to the bi-ventricular volumetric mesh.
5. The method of claim 2 , wherein the step of generating a patient-specific anatomical model of a heart from medical image data of a patient further comprises:
generating a global model of fiber architecture based on the bi-ventricular volumetric mesh and nominal values for fiber elevation angle distribution.
6. The method of claim 1 , wherein the step of computing the transmembrane action potential on at least one node of the level-set representation of the patient-specific anatomical model of the heart further comprising:
solving at least one monodomain electrophysiology model using a Lattice-Boltzmann method on at least one node of the level-set representation on a Cartesian grid.
7. The method of claim 1 , wherein the step of generating a level-set representation of the patient-specific anatomical model of the heart on a Cartesian grid comprises:
discretizing a domain of the patient-specific model of the heart using a Cartesian lattice in which each node is connected via edges to a predetermined number of neighboring nodes.
8. The method of claim 1 , wherein the step of computing the transmembrane action potential on at least one node of the level-set representation of the patient-specific anatomical model of the heart further comprises:
at each of a plurality of time steps:
for at least one node of the level-set representation of the patient-specific anatomical model of the heart on a Cartesian grid,
calculating collisions of particles for each edge connected to the node based on a distribution function calculated for each edge;
calculating the transmembrane action potential at the node based on the collisions of particles calculated at the node;
updating all cellular model-dependent internal variables at the node;
updating the distribution function for each edge of the node to represent streaming of a particle traveling along each of the edges to a neighboring node; and
applying optional Dirichlet boundary conditions.
9. The method of claim 8 , wherein the step of calculating collisions of particles for each edge connected to the node based on a distribution function calculated for each edge comprises:
calculating collisions on each edge connected to the node as:
f i =f i −A ij ( f j −ω j υ)+δ tω i R (υ, { h })
where f i is the distribution function that represents a probability of finding a particle traveling along the i th edge e i connected to the node, A=(A ij ) i,j ∈ is a collision matrix that relaxes the distribution function f i towards a local potential value, υ, R(v, {h}) is a model-dependent term for electrophysiological currents, which depends on a model-dependent set of internal variables {h}, and ω i and ω j are weighting factors for edges e i and e j , respectively.
10. The method of claim 9 , wherein the collision matrix A is defined as A=M −1 SM,
where M is a matrix designed to transform the distribution functions into a vector of moments and S −1 is a matrix of relaxation times corresponding to the moments, and M and S −1 are respectively defined as:
M
=
(
1
1
1
1
1
1
1
0
-
1
0
0
0
0
0
0
0
1
-
1
0
0
0
0
0
0
0
1
-
1
0
1
1
1
1
1
1
-
6
1
1
-
1
-
1
0
0
0
1
1
1
1
-
2
-
2
0
)
S
-
1
=
(
τ
1
0
0
0
0
0
0
0
τ
22
τ
23
τ
24
0
0
0
0
τ
32
τ
33
τ
34
0
0
0
0
τ
42
τ
43
τ
44
0
0
0
0
0
0
0
τ
5
0
0
0
0
0
0
0
τ
6
0
0
0
0
0
0
0
τ
7
)
.
11. The method of claim 8 , wherein the step of calculating the transmembrane action potential at the node based on the collisions of particles calculated at the node comprises:
calculating the transmembrane action potential v(x, t) for a node x at time step t as:
υ(x, t)=Σ i f i (x,t)
where f i (x, t) represents a probability of finding a particle traveling along the i th edge e i connected to node x at time step t.
12. The method of claim 8 , wherein the step of updating all cellular model-dependent internal variables at the node comprises:
using an integration scheme to solve a system of model-specific differential-algebraic equations modeling each internal variables.
13. The method of claim 8 , wherein the step of updating the distribution function for each edge of the node to represent streaming of a particle traveling along each of the edges to a neighboring node comprises:
calculating the distribution functions at the next time step as:
f i (x+e i δt,t+δt)=f i (x,t).
where f i (x, t) represents a probability of finding a particle traveling along the i th edge e i connected to node x at time step t.
14. The method of claim 1 , wherein the step of identifying key electrophysiology structures of interest from the patient-specific anatomical model and medical image data comprises:
identifying and tagging surface zones representing electrophysiology structures of the heart using machine learning algorithms; and
assigning spatially variable electrophysiological properties of myocardial tissue.
15. The method of claim 14 , wherein the step of assigning spatially variable properties of the myocardial tissue comprises:
assigning tissue electrical conductivity and other model-dependent internal parameters estimated from medical image data and other clinical measurements.
16. An apparatus for patient-specific electrocardiography simulation, comprising:
a processor; and
a memory storing computer program instructions, which when executed by the processor cause the processor to perform operations comprising:
generating a patient-specific anatomical model of a heart from medical image data of a patient;
identifying at least one key electrophysiological structures of interest from the patient-specific anatomical model and medical image data;
generating a level-set representation of the patient-specific anatomical model of the heart on a Cartesian grid; and
computing a transmembrane action potential on at least one node of the level-set representation of the patient-specific anatomical model of the heart, wherein computing the transmembrane action potential is based at least in part on a stimulus current input as an initial condition.
17. The apparatus of claim 16 , wherein generating a patient-specific anatomical model of a heart from medical image data of a patient comprises:
detecting a patient-specific left ventricle endocardium model, a patient-specific right ventricle endocardium model, and a patient-specific epicardium model in the medical image data; and
fusing the left ventricle endocardium model, the right ventricle endocardium model and the epicardium model into a single bi-ventricular volumetric mesh.
18. The apparatus of claim 17 , wherein generating a patient-specific anatomical model of a heart from medical image data of a patient further comprises:
mapping spatial information corresponding to at least one of scars, grey zones, or fibrosis identified from an MR sequences onto a tetrahedral representation of bi-ventricular volumetric mesh.
19. The apparatus of claim 17 , wherein generating a patient-specific anatomical model of a heart from medical image data of a patient further comprises:
registering a tensor field of an in-vivo diffusion tensor magnetic resonance image of cardiac fibers to the bi-ventricular volumetric mesh.
20. The apparatus of claim 17 , wherein generating a patient-specific anatomical model of a heart from medical image data of a patient further comprises:
generating a global model of fiber architecture based on the bi-ventricular volumetric mesh and nominal values for fiber elevation angle distribution.
21. The apparatus of claim 16 , wherein computing the transmembrane action potential on at least one node of the level-set representation of the patient-specific anatomical model of the heart further comprises:
solving at least one monodomain electrophysiology model using a Lattice-Boltzmann method on at least one node of the level-set representation on a Cartesian grid.
22. The apparatus of claim 16 , wherein generating a level-set representation of the patient-specific anatomical model of the heart on an isotropic Cartesian grid comprises:
discretizing a domain of the patient-specific model of the heart using a Cartesian lattice in which each node is connected via edges to a predetermined number of neighboring nodes.
23. The apparatus of claim 16 , wherein computing the transmembrane action potential on at least one node of the level-set representation of the patient-specific anatomical model of the heart further comprises:
at each of a plurality of time steps:
for at least one node of the level-set representation of the patient-specific anatomical model of the heart on a Cartesian grid,
calculating collisions of particles for each edge connected to the node based on a distribution function calculated for each edge;
calculating the transmembrane action potential at the node based on the collisions of particles calculated at the node;
updating all cellular model-dependent internal variables at the node;
updating the distribution function for each edge of the node to represent streaming of a particle traveling along each of the edges to a neighboring node; and
applying optional Dirichlet boundary conditions.
24. The apparatus of claim 23 , wherein calculating collisions of particles for each edge connected to the node based on a distribution function calculated for each edge comprises:
calculating collisions on each edge connected to the node as:
f i =f i −A ij (f j −ω j υ)+δtω i R(υ,{h})
where f i is the distribution function that represents a probability of finding a particle traveling along the i th edge e i connected to the node, A=(A ij ) i,j ∈ is a collision matrix that relaxes the distribution function f i towards a local potential value, υ, R(v, {h}) is a model-dependent term for electrophysiological currents, which depends on a model-dependent set of internal variables {h}, and ω i and ω j are weighting factors for edges e i and e j , respectively.
25. The apparatus of claim 24 , wherein the collision matrix A is defined as A=M −1 SM, where M is a matrix designed to transform the distribution functions into a vector of moments and S −1 is a matrix of relaxation times corresponding to the moments, and M and S −1 are respectively defined as:
M
=
(
1
1
1
1
1
1
1
0
-
1
0
0
0
0
0
0
0
1
-
1
0
0
0
0
0
0
0
1
-
1
0
1
1
1
1
1
1
-
6
1
1
-
1
-
1
0
0
0
1
1
1
1
-
2
-
2
0
)
S
-
1
=
(
τ
1
0
0
0
0
0
0
0
τ
22
τ
23
τ
24
0
0
0
0
τ
32
τ
33
τ
34
0
0
0
0
τ
42
τ
43
τ
44
0
0
0
0
0
0
0
τ
5
0
0
0
0
0
0
0
τ
6
0
0
0
0
0
0
0
τ
7
)
.
26. The apparatus of claim 23 , wherein calculating the transmembrane action potential at the node based on the collisions of particles calculated at the node comprises:
means for calculating the transmembrane action potential for a node x as:
υ(x,t)=Σ i f i (x,t)
where f i (x, t) represents a probability of finding a particle traveling along the i th edge e i connected to node x at time step t.
27. The apparatus of claim 23 , wherein updating all cellular model-dependent internal variables at the node comprises:
using an integration scheme to solve a system of model-specific differential-algebraic equations modeling each of the internal variables.
28. The apparatus of claim 23 , wherein updating the distribution function for each edge of the node to represent streaming of a particle traveling along each of the edges to a neighboring node comprises:
calculating the distribution functions at the next time step as:
f i (x+e i δt,t+δt) =f i (x,t)
where f i (x,t) represents a probability of finding a particle traveling along the i th edge e i connected to node x at time step t.
29. The apparatus of claim 17 , wherein identifying key electrophysiology structures of interest from the patient-specific anatomical model and medical image data comprises:
identifying and tagging surface zones representing electrophysiology structures of the heart using machine learning algorithms; and
assigning spatially variable electrophysiological properties of the myocardial tissue.
30. The apparatus of claim 29 , wherein assigning spatially variable properties of the myocardial tissue comprises:
assigning tissue electrical conductivity and other model-dependent internal parameters estimated from medical image data and other clinical measurements.
31. A non-transitory computer readable medium storing computer program instructions for patient-specific electrocardiography simulation, the computer program instructions, when executed, cause a processor to perform a method comprising:
generating a patient-specific anatomical model of a heart from medical image data of a patient;
identifying at least one key electrophysiological structures of interest from the patient-specific anatomical model and medical image data;
generating a level-set representation of the patient-specific anatomical model of the heart on a Cartesian grid; and
computing a transmembrane action potential on at least one node of the level-set representation of the patient-specific anatomical model of the heart, wherein computing the transmembrane action potential is based at least in part on a stimulus current input as an initial condition.
32. The non-transitory computer readable medium of claim 31 , wherein the step of generating a patient-specific anatomical model of a heart from medical image data of a patient comprises:
detecting a patient-specific left ventricle endocardium model, a patient-specific right ventricle endocardium model, and a patient-specific epicardium model in the medical image data; and
fusing the left ventricle endocardium model, the right ventricle endocardium model and the epicardium model into a single bi-ventricular volumetric mesh.
33. The non-transitory computer readable medium of claim 32 , wherein the step of generating a patient-specific anatomical model of a heart from medical image data of a patient further comprises:
mapping spatial information corresponding to at least one of scars, grey zones, or fibrosis identified from medical-image data and other clinical measurements onto the bi-ventricular volumetric mesh.
34. The non-transitory computer readable medium of claim 32 , wherein the step of generating a patient-specific anatomical model of a heart from medical image data of a patient further comprises:
registering a tensor field of an in-vivo diffusion tensor magnetic resonance image of cardiac fibers to the bi-ventricular volumetric mesh.
35. The non-transitory computer readable medium of claim 32 , wherein the step of generating a patient-specific anatomical model of a heart from medical image data of a patient further comprises:
generating a global model of fiber architecture based on the bi-ventricular volumetric mesh and nominal values for fiber elevation angle distribution.
36. The non-transitory computer readable medium of claim 31 , wherein the step of computing the transmembrane action potential on at least one node of the level-set representation of the patient-specific anatomical model of the heart further comprising:
solving at least one monodomain electrophysiology model using a Lattice-Boltzmann method on at least one node of the level-set representation on a Cartesian grid.
37. The non-transitory computer readable medium of claim 31 , wherein the step of generating a level-set representation of the patient-specific anatomical model of the heart on a Cartesian grid comprises:
discretizing a domain of the patient-specific model of the heart using a Cartesian lattice in which each node is connected via edges to a predetermined number of neighboring nodes.
38. The non-transitory computer readable medium of claim 31 , wherein the step of computing the transmembrane action potential on at least one node of the level-set representation of the patient-specific anatomical model of the heart further comprises:
at each of a plurality of time steps:
for at least one node of the level-set representation of the patient- specific anatomical model of the heart on a Cartesian grid,
calculating collisions of particles for each edge connected to the node based on a distribution function calculated for each edge;
calculating the transmembrane action potential at the node based on the collisions of particles calculated at the node;
updating all cellular model-dependent internal variables at the node;
updating the distribution function for each edge of the node to represent streaming of a particle traveling along each of the edges to a neighboring node; and
applying optional Dirichlet boundary conditions.
39. The non-transitory computer readable medium of claim 38 , wherein the step of calculating collisions of particles for each edge connected to the node based on a distribution function calculated for each edge comprises:
calculating collisions on each edge connected to the node as:
f i =f i −A ij (f j −ω j υ)+Δtω i R(υ,{h})
where f i is the distribution function that represents a probability of finding a particle traveling along the i th edge e i connected to the node, A=(A ij ) i,j ∈ is a collision matrix that relaxes the distribution function f i towards a local potential value, υ, R(υ, {h}) is a model-dependent term for electrophysiological currents, which depends on a model-dependent set of internal variables {h}, and ω i and ω j are weighting factors for edges e i and e j , respectively.
40. The non-transitory computer readable medium of claim 39 , wherein the collision matrix A is defined as A=M −1 SM ,
where M is a matrix designed to transform the distribution functions into a vector of moments and S −1 is a matrix of relaxation times corresponding to the moments, and M and S −1 are respectively defined as:
M
=
(
1
1
1
1
1
1
1
0
-
1
0
0
0
0
0
0
0
1
-
1
0
0
0
0
0
0
0
1
-
1
0
1
1
1
1
1
1
-
6
1
1
-
1
-
1
0
0
0
1
1
1
1
-
2
-
2
0
)
S
-
1
=
(
τ
1
0
0
0
0
0
0
0
τ
22
τ
23
τ
24
0
0
0
0
τ
32
τ
33
τ
34
0
0
0
0
τ
42
τ
43
τ
44
0
0
0
0
0
0
0
τ
5
0
0
0
0
0
0
0
τ
6
0
0
0
0
0
0
0
τ
7
)
.
41. The non-transitory computer readable medium of claim 38 , wherein the step of calculating the transmembrane action potential at the node based on the collisions of particles calculated at the node comprises:
calculating the transmembrane action potential for a node x as:
υ(x,t)=Σ i f i (x,t) where f i (x, t) represents a probability of finding a particle traveling along the i th edge e i connected to node x at time step t.
42. The non-transitory computer readable medium of claim 38 , wherein the step of updating all cellular model-dependent internal variables at the node comprises:
using an integration scheme to solve a system of model-specific differential-algebraic equations modeling each internal variables.
43. The non-transitory computer readable medium of claim 38 , wherein the step of updating the distribution function for each edge of the node to represent streaming of a particle traveling along each of the edges to a neighboring node comprises:
calculating the distribution functions at the next time step as:
f i (x+e i Δt,t+Δt)=f i (x,t)
where f i (x,t) represents a probability of finding a particle traveling along the i th edge e i connected to node x at time step t.
44. The non-transitory computer readable medium of claim 31 , wherein the step of identifying key electrophysiology structures of interest from the patient-specific anatomical model and medical image data comprises:
identifying and tagging surface zones representing electrophysiology structures of the heart using machine learning algorithms; and
assigning spatially variable electrophysiological properties of myocardial tissue.
45. The non-transitory computer readable medium of claim 44 , wherein the step of assigning spatially variable properties of the myocardial tissue comprises:
assigning tissue electrical conductivity and other model-dependent internal parameters estimated from medical image data and other clinical measurements.