Systems and methods for estimating hemodynamic forces acting on plaque and monitoring risk
Computer-implemented methods are disclosed for estimating values of hemodynamic forces acting on plaque or lesions. One method includes: receiving one or more patient-specific parameters of at least a portion of a patient's vasculature that is prone to plaque progression, rupture, or erosion; constructing a patient-specific geometric model of at least a portion of a patient's vasculature that is prone to plaque progression, rupture, or erosion, using the received one or more patient-specific parameters; estimating, using one or more processors, the values of hemodynamic forces at one or more points on the patient-specific geometric model, using the patient-specific parameters and geometric model by measuring, deriving, or obtaining one or more of a pressure gradient and a radius gradient; and outputting the estimated values of hemodynamic forces to an electronic storage medium. Systems and computer readable media for executing these methods are also disclosed.
1. A computer-implemented method for training a machine learning algorithm comprising:
acquiring a plurality of patient-specific geometric models;
determining at least one patient-specific parameter associated with each of the plurality of patient-specific geometric models;
determine at least one simulated hemodynamic characteristic for one or more points of the plurality of patient-specific geometric models;
for each of the plurality of patient-specific geometric models, associating the at least one patient-specific parameter with the at least one simulated hemodynamic characteristic;
using the association of the at least one patient-specific parameter with the at least one simulated hemodynamic characteristic, training a machine learning algorithm; and
outputting the trained machine learning algorithm.
2. The computer-implemented method of claim 1 , wherein the at least one simulated hemodynamic characteristic includes a wall shear stress and/or an axial plaque stress.
3. The computer-implemented method of claim 1 , wherein the at least one patient-specific parameter includes at least one of a patient characteristic, a physiological characteristic, a geometric characteristic, a plaque characteristic, or a coronary dynamics characteristic.
4. The computer-implemented method of claim 3 , wherein the at least one patient-specific parameter includes a pressure gradient.
5. The computer-implemented method of claim 1 , wherein determining the at least one simulated hemodynamic characteristic includes using computational fluid dynamics.
6. The computer-implemented method of claim 1 , wherein associating the at least one patient-specific parameter with the at least one simulated hemodynamic characteristic includes associating a feature vector of a plurality of patient-specific parameters with the at least one simulated hemodynamic characteristic.
7. The computer-implemented method of claim 6 , wherein the feature vectors are used to train the machine learning algorithm.
8. The computer-implemented method of claim 1 , wherein the at least one patient-specific parameter is obtained non-invasively.
9. The computer-implemented method of claim 1 , wherein the trained machine learning algorithm includes at least one feature weight.
10. A system for monitoring hemodynamic risk in a patient, the system comprising:
at least one memory storing instructions; and
at least one processor operatively connected to the at least one memory and configured to execute the instructions to perform operations, including:
acquiring a plurality of patient-specific geometric models;
determining at least one patient-specific parameter associated with each of the plurality of patient-specific geometric models;
determine at least one simulated hemodynamic characteristic for one or more points of the plurality of patient-specific geometric models;
for each of the plurality of patient-specific geometric models, associating the at least one patient-specific parameter with the at least one simulated hemodynamic characteristic;
using the association of the at least one patient-specific parameter with the at least one simulated hemodynamic characteristic, training a machine learning algorithm; and
outputting the trained machine learning algorithm.
11. The system of claim 10 , wherein the at least one simulated hemodynamic characteristic includes a wall shear stress and/or an axial plaque stress.
12. The system of claim 10 , wherein the at least one patient-specific parameter includes at least one of a patient characteristic, a physiological characteristic, a geometric characteristic, a plaque characteristic, or a coronary dynamics characteristic.
13. The system of claim 10 , wherein the at least one patient-specific parameter includes a pressure gradient.
14. The system of claim 10 , wherein the at least one patient-specific parameter includes at least one of a patient characteristic, a physiological characteristic, a geometric characteristic, a plaque characteristic, or a coronary dynamics characteristic.
15. The system of claim 10 , wherein the at least one patient-specific parameter is obtained non-invasively.
16. A non-transitory computer readable medium for use on a computer system containing computer-executable programming instructions for a method of monitoring hemodynamic risk in a patient, the method comprising:
acquiring a plurality of patient-specific geometric models;
determining at least one patient-specific parameter associated with each of the plurality of patient-specific geometric models;
determine at least one simulated hemodynamic characteristic for one or more points of the plurality of patient-specific geometric models;
for each of the plurality of patient-specific geometric models, associating the at least one patient-specific parameter with the at least one simulated hemodynamic characteristic;
using the association of the at least one patient-specific parameter with the at least one simulated hemodynamic characteristic, training a machine learning algorithm; and
outputting the trained machine learning algorithm.
17. The non-transitory computer readable medium of claim 16 , wherein the at least one simulated hemodynamic characteristic includes a wall shear stress and/or an axial plaque stress.
18. The non-transitory computer readable medium of claim 16 , wherein the at least one patient-specific parameter includes at least one of a patient characteristic, a physiological characteristic, a geometric characteristic, a plaque characteristic, or a coronary dynamics characteristic.
19. The non-transitory computer readable medium of claim 16 , wherein the at least one patient-specific parameter includes a pressure gradient.
20. The non-transitory computer readable medium of claim 16 , wherein the at least one patient-specific parameter includes at least one of a patient characteristic, a physiological characteristic, a geometric characteristic, a plaque characteristic, or a coronary dynamics characteristic.