Method and system for hemodynamic assessment of aortic coarctation from medical image data

1. A method for non-invasive hemodynamic assessment of aortic coarctation, comprising:

estimating patient-specific lumen anatomy of an aorta and supra-aortic arteries from medical image data of a patient;

estimating patient-specific aortic blood flow rates from the medical image data of the patient;

calculating patient-specific inlet and outlet boundary conditions for a computational model of aortic blood flow based on the patient-specific lumen anatomy, the patient-specific aortic blood flow rates, and non-invasive clinical measurements of the patient; and

computing aortic blood flow and pressure over the patient-specific lumen anatomy using the computational model of aortic blood flow and the patient-specific inlet and outlet boundary conditions.

2. The method of claim 1 , further comprising:

calculating a pressure drop across a coarctation region of the aorta using computed pressure values resulting from computing the aortic blood flow and pressure.

3. The method of claim 2 , wherein calculating a pressure drop across a coarctation region of the aorta using computed pressure values resulting from computing the aortic blood flow and pressure comprises:

calculating a difference between computed pressure values before and after the coarctation region of the aorta at a time instance when a computed flow-rate in the descending aorta is highest.

4. The method of claim 1 , wherein estimating patient-specific lumen anatomy of the aorta and supra-aortic arteries from medical image data of a patient comprises:

detecting a respective pose in the medical image data for each of a plurality of parts of the aorta and supra-aortic arteries;

initializing a respective model for each of the plurality of parts by fitting a corresponding learned mean shape model to the respective pose in the medical image data;

deforming the respective model for each of the plurality of parts using a trained lumen detector; and

merging the respective models for the plurality of parts to generate a final model of lumen anatomy in the aorta and supra-aortic arteries.

5. The method of claim 1 , wherein estimating patient-specific lumen anatomy of the aorta and supra-aortic arteries from medical image data of a patient comprises:

estimating the patient-specific lumen anatomy of the aorta and supra-aortic arteries in a contrast-enhanced magnetic resonance imaging (CE-MRI) image.

6. The method of claim 1 , wherein estimating patient-specific aortic blood flow rates from the medical image data of the patient comprises:

estimating patient-specific aortic blood flow rates from a sequence of velocity encoded 2D phase-contrasted magnetic resonance imaging (PC-MRI) images.

7. The method of claim 6 , wherein estimating patient-specific aortic blood flow rates from a sequence of velocity encoded 2D phase-contrasted magnetic resonance imaging (PC-MRI) images comprises:

overlaying each of the 2D PC-MRI images with the estimated lumen anatomy of the aorta and supra-aortic arteries, wherein the overlayed 2D PC-MRI intersects with the estimated lumen anatomy at a first intersection corresponding to an ascending aorta and a second intersection corresponding to a descending aorta;

detecting a first contour representing the estimated lumen anatomy at the first intersection and a second contour representing the estimated lumen anatomy at the second intersection in each 2D PC-MRI image;

calculating a time-varying flow rate at the ascending aorta by integrating velocity values over an area of the first contour for each of the 2D PC-MRI images; and

calculating a time-varying flow rate at the descending aorta by integrating velocity values over an area of the second contour for each of the 2D PC-MRI images.

8. The method of claim 1 , wherein calculating patient-specific inlet and outlet boundary conditions for a computational model of aortic blood flow based on the patient-specific lumen anatomy, the patient-specific aortic blood flow rates, and non-invasive clinical measurements of the patient comprises:

imposing a time-varying blood flow rate estimated at the ascending aorta as the inlet boundary condition; and

estimating patient-specific parameters of a 3-element Windkessel model at each of four outlets corresponding to a brachiocephalic trunk, left common carotid artery, left subclavian artery, and descending aorta based on the patient-specific lumen anatomy, the patient-specific aortic blood flow rates, and the non-invasive clinical measurements of the patient.

9. The method of claim 8 , wherein estimating patient-specific parameters of a 3-element Windkessel model at each of four outlets corresponding to a brachiocephalic trunk, left common carotid artery, left subclavian artery, and descending aorta based on the patient-specific lumen anatomy, the patient-specific aortic blood flow rates, and the non-invasive clinical measurements of the patient comprises:

calculating mean arterial pressure (MAP) based on diastolic and systolic blood pressures of the patient and a heart rate of the patient;

calculating a total resistance for each outlet based on an average flow at the outlet and the MAP; and

calculating proximal and distal resistances for each outlet based on the total resistance calculated for each outlet.

10. The method of claim 9 , wherein calculating a total resistance for each outlet based on an average flow at the outlet and the MAP comprises:

calculating a total flow remaining in the brachiocephalic trunk, left common carotid artery, and left subclavian artery as Q up =Q asc −Q desc , where Q asc is an average flow at the ascending aorta and Q desc is an average flow at the descending aorta;

for each of the brachiocephalic trunk, left common carotid artery, and left subclavian artery, calculating the average flow at that outlet as

Q

i

=

Q

up

·

r

i

2

∑

i

=

1

3

⁢

r

i

2

,

where r i is the vessel radius at the outflow of upper branch i; and

for each of the brachiocephalic trunk, left common carotid artery, left subclavian artery, calculating the total resistance as:

(

R

t

)

i

=

M

⁢

⁢

A

⁢

⁢

P

Q

i

.

11. The method of claim 10 , wherein calculating a total resistance for each outlet based on an average flow at the outlet and the MAP comprises further comprises:

calculating the total resistance for the descending aorta outlet (R t ) desc using the equation: (R t ) desc +R s (Q)=MAP/Q desc , where R S (Q) is a flow-dependent resistance of a coarctation region that is estimated with a semi-empirical model:

R

s

⁡

(

Q

)

=

μ

·

K

v

2

·

π

·

r

0

3

+

ρ

·

K

t

2

·

A

0

2

⁢

(

A

0

A

s

-

1

)

2

·

Q

desc

,

where r 0 is the proximal radius of the coarctation region, A 0 and A s are proximal and minimal cross-sectional areas of the coarctation region, respectively, μ is a dynamic viscosity, ρ is blood density, and K v and K t are two constants that represent the viscous and turbulent losses of energy, respectively.

12. The method of claim 9 , wherein calculating proximal and distal resistances for each outlet based on the total resistance calculated for each outlet comprises:

calculating the proximal resistance for each outlet as a characteristic resistance of that vessel; and

calculating the distal resistance for each outlet as a difference between the total resistance for that outlet and the proximal resistance for that outlet.

13. The method of claim 1 , wherein simulating aortic blood flow and pressure over the patient-specific lumen anatomy using the computational model of aortic blood flow and the patient-specific inlet and outlet boundary conditions comprises:

simulating aortic blood flow using a 3D computational fluid dynamics (CFD) simulation.

14. An apparatus for non-invasive hemodynamic assessment of aortic coarctation, comprising:

means for estimating patient-specific lumen anatomy of an aorta and supra-aortic arteries from medical image data of a patient;

means for estimating patient-specific aortic blood flow rates from the medical image data of the patient;

means for calculating patient-specific inlet and outlet boundary conditions for a computational model of aortic blood flow based on the patient-specific lumen anatomy, the patient-specific aortic blood flow rates, and non-invasive clinical measurements of the patient; and

means for computing aortic blood flow and pressure over the patient-specific lumen anatomy using the computational model of aortic blood flow and the patient-specific inlet and outlet boundary conditions.

15. The apparatus of claim 14 , further comprising:

means for calculating a pressure drop across a coarctation region of the aorta using computed pressure values resulting from computing the aortic blood flow and pressure.

16. The apparatus of claim 14 , wherein the means for estimating patient-specific aortic blood flow rates from the medical image data of the patient comprises:

means for estimating patient-specific aortic blood flow rates from a sequence of velocity encoded 2D phase-contrasted magnetic resonance imaging (PC-MRI) images.

17. The apparatus of claim 14 , wherein the means for calculating patient-specific inlet and outlet boundary conditions for a computational model of aortic blood flow based on the patient-specific lumen anatomy, the patient-specific aortic blood flow rates, and non-invasive clinical measurements of the patient comprises:

means for imposing a time-varying blood flow rate estimated at the ascending aorta as the inlet boundary condition; and

means for estimating patient-specific parameters of a 3-element Windkessel model at each of four outlets corresponding to a brachiocephalic trunk, left common carotid artery, left subclavian artery, and descending aorta based on the patient-specific lumen anatomy, the patient-specific aortic blood flow rates, and the non-invasive clinical measurements of the patient.

18. The apparatus of claim 17 , wherein the means for estimating patient-specific parameters of a 3-element Windkessel model at each of four outlets corresponding to a brachiocephalic trunk, left common carotid artery, left subclavian artery, and descending aorta based on the patient-specific lumen anatomy, the patient-specific aortic blood flow rates, and the non-invasive clinical measurements of the patient comprises:

means for calculating mean arterial pressure (MAP) based on diastolic and systolic blood pressures of the patient and a heart rate of the patient;

means for calculating a total resistance for each outlet based on an average flow at the outlet and the MAP; and

means for calculating proximal and distal resistances for each outlet based on the total resistance calculated for each outlet.

19. The apparatus of claim 1 , wherein the means for simulating aortic blood flow and pressure over the patient-specific lumen anatomy using the computational model of aortic blood flow and the patient-specific inlet and outlet boundary conditions comprises:

means for simulating aortic blood flow using a 3D computational fluid dynamics (CFD) simulation.

20. A non-transitory computer readable medium storing computer program instructions for non-invasive hemodynamic assessment of aortic coarctation, the computer program instructions when executed by a processor cause the processor to perform operations comprising:

estimating patient-specific lumen anatomy of an aorta and supra-aortic arteries from medical image data of a patient;

estimating patient-specific aortic blood flow rates from the medical image data of the patient;

calculating patient-specific inlet and outlet boundary conditions for a computational model of aortic blood flow based on the patient-specific lumen anatomy, the patient-specific aortic blood flow rates, and non-invasive clinical measurements of the patient; and

computing aortic blood flow and pressure over the patient-specific lumen anatomy using the computational model of aortic blood flow and the patient-specific inlet and outlet boundary conditions.

21. The non-transitory computer readable medium of claim 20 , wherein the operations further comprise:

calculating a pressure drop across a coarctation region of the aorta using computed pressure values resulting from computing the aortic blood flow and pressure.

22. The non-transitory computer readable medium of claim 20 , wherein estimating patient-specific lumen anatomy of the aorta and supra-aortic arteries from medical image data of a patient comprises:

detecting a respective pose in the medical image data for each of a plurality of parts of the aorta and supra-aortic arteries;

initializing a respective model for each of the plurality of parts by fitting a corresponding learned mean shape model to the respective pose in the medical image data;

deforming the respective model for each of the plurality of parts using a trained lumen detector; and

merging the respective models for the plurality of parts to generate a final model of lumen anatomy in the aorta and supra-aortic arteries.

23. The non-transitory computer readable medium of claim 20 , wherein estimating patient-specific aortic blood flow rates from the medical image data of the patient comprises:

estimating patient-specific aortic blood flow rates from a sequence of velocity encoded 2D phase-contrasted magnetic resonance imaging (PC-MRI) images.

24. The non-transitory computer readable medium of claim 23 , wherein estimating patient-specific aortic blood flow rates from a sequence of velocity encoded 2D phase-contrasted magnetic resonance imaging (PC-MRI) images comprises:

overlaying each of the 2D PC-MRI images with the estimated lumen anatomy of the aorta and supra-aortic arteries, wherein the overlayed 2D PC-MRI intersects with the estimated lumen anatomy at a first intersection corresponding to an ascending aorta and a second intersection corresponding to a descending aorta;

detecting a first contour representing the estimated lumen anatomy at the first intersection and a second contour representing the estimated lumen anatomy at the second intersection in each 2D PC-MRI image;

calculating a time-varying flow rate at the ascending aorta by integrating velocity values over an area of the first contour for each of the 2D PC-MRI images; and

calculating a time-varying flow rate at the descending aorta by integrating velocity values over an area of the second contour for each of the 2D PC-MRI images.

25. The non-transitory computer readable medium of claim 20 , wherein calculating patient-specific inlet and outlet boundary conditions for a computational model of aortic blood flow based on the patient-specific lumen anatomy, the patient-specific aortic blood flow rates, and non-invasive clinical measurements of the patient comprises:

imposing a time-varying blood flow rate estimated at the ascending aorta as the inlet boundary condition; and

estimating patient-specific parameters of a 3-element Windkessel model at each of four outlets corresponding to a brachiocephalic trunk, left common carotid artery, left subclavian artery, and descending aorta based on the patient-specific lumen anatomy, the patient-specific aortic blood flow rates, and the non-invasive clinical measurements of the patient.

26. The non-transitory computer readable medium of claim 25 , wherein estimating patient-specific parameters of a 3-element Windkessel model at each of four outlets corresponding to a brachiocephalic trunk, left common carotid artery, left subclavian artery, and descending aorta based on the patient-specific lumen anatomy, the patient-specific aortic blood flow rates, and the non-invasive clinical measurements of the patient comprises:

calculating mean arterial pressure (MAP) based on diastolic and systolic blood pressures of the patient and a heart rate of the patient;

calculating a total resistance for each outlet based on an average flow at the outlet and the MAP; and

calculating proximal and distal resistances for each outlet based on the total resistance calculated for each outlet.

27. The non-transitory computer readable medium of claim 26 , wherein calculating a total resistance for each outlet based on an average flow at the outlet and the MAP comprises:

calculating a total flow remaining in the brachiocephalic trunk, left common carotid artery, and left subclavian artery as Q up =Q asc −Q desc , where Q asc is an average flow at the ascending aorta and Q desc is an average flow at the descending aorta;

for each of the brachiocephalic trunk, left common carotid artery, and left subclavian artery, calculating the average flow at that outlet as

Q

i

=

Q

up

·

r

i

2

∑

i

=

1

3

⁢

r

i

2

,

where r i is the vessel radius at the outflow of upper branch i; and

for each of the brachiocephalic trunk, left common carotid artery, left subclavian artery, calculating the total resistance as:

(

R

t

)

i

=

M

⁢

⁢

A

⁢

⁢

P

Q

i

.

28. The non-transitory computer readable medium of claim 27 , wherein calculating a total resistance for each outlet based on an average flow at the outlet and the MAP comprises further comprises:

calculating the total resistance for the descending aorta outlet (R t ) desc using the equation: (R t ) desc +R s (Q)=MAP/Q desc , where R s (Q) is a flow-dependent resistance of a coarctation region that is estimated with a semi-empirical model:

R

s

⁡

(

Q

)

=

μ

·

K

v

2

·

π

·

r

0

3

+

ρ

·

K

t

2

·

A

0

2

⁢

(

A

0

A

s

-

1

)

2

·

Q

desc

,

where r 0 is the proximal radius of the coarctation region, A 0 and A s are proximal and minimal cross-sectional areas of the coarctation region, respectively, μ is a dynamic viscosity, ρ is blood density, and K v and K t are two constants that represent the viscous and turbulent losses of energy, respectively.

29. The non-transitory computer readable medium of claim 20 , wherein simulating aortic blood flow and pressure over the patient-specific lumen anatomy using the computational model of aortic blood flow and the patient-specific inlet and outlet boundary conditions comprises:

simulating aortic blood flow using a 3D computational fluid dynamics (CFD) simulation.