Personalized assessment of patients with acute coronary syndrome

1. A computer-implemented method for personalized assessment of patients with acute coronary syndrome (ACS), the method comprising:

extracting patient-specific coronary geometry data from one or more medical images of a patient;

extracting a plurality of features of a patient-specific coronary arterial tree based on the patient-specific coronary geometry data, wherein the patient-specific coronary arterial tree is generated by estimating a total coronary resistance index (TCRI ACS ) value of each artery of the patient-specific coronary arterial tree; wherein each TCRI ACS is a function of a distance between a myocardial bed of each artery of the patient-specific coronary arterial tree and a myocardial area affected by the ACS;

extracting a plurality of ACS-related features from additional patient measurement data including blood biomarkers acquired at different time points and contrast propagation information; and

using a surrogate model to predict patient-specific hemodynamic measures of interest related to ACS based on the plurality of features of the patient-specific coronary arterial tree and the plurality of ACS-related features from the additional patient measurement data.

2. The method of claim 1 , wherein the additional patient measurement data further comprises one or more of perfusion imaging data, invasive measurements, and ECG signals.

3. The method of claim 1 , wherein the patient-specific hemodynamic measures of interest comprise one or more of Fractional Flow Reserve and Coronary Flow Reserve.

4. The method of claim 1 , wherein the patient-specific hemodynamic measures of interest comprise wall shear stress.

5. The method of claim 1 , wherein the patient-specific hemodynamic measures of interest comprise risk of plaque rupture.

6. The method of claim 1 , further comprising training the surrogate model using a process comprising:

generating a database of coronary arterial trees representative of ACS conditions;

performing flow computations on each artery included in the database of coronary arterial trees to extract hemodynamic measures of interest;

extracting features of coronary arterial trees and AC S related features from the database of coronary arterial trees;

applying one or more machine learning methods to train the surrogate model to predict the hemodynamic measures of interest based on the features of coronary arterial trees and the ACS related features.

7. The method of claim 6 , wherein the database of coronary arterial trees comprises a plurality of synthetic coronary arterial trees.

8. The method of claim 7 , wherein the database of coronary arterial trees comprises a plurality of in silico models and the flow computations comprise computational fluid dynamics (CFD) computations.

9. The method of claim 7 , wherein the database of coronary arterial trees comprises a plurality of in vitro models and the flow computations comprise flow experiments.

10. The method of claim 7 , wherein the database of coronary arterial trees further comprises a plurality of non-synthetic coronary arterial trees.

11. The method of claim 1 , further comprising:

predicting a risk of future events for patients with ACS based on the patient-specific hemodynamic measures of interest.

12. The method of claim 1 , further comprising:

determining a confidence interval or a measure of uncertainty for the patient-specific hemodynamic measures of interest related to ACS.

13. The method of claim 12 , wherein the confidence interval or the measure of uncertainty is determined by comparing the predictions of the surrogate model based on medical images of the patient-specific coronary geometry acquired with at least two different imaging modalities.

14. A computer-implemented method for personalized assessment of patients with acute coronary syndrome (ACS), the method comprising:

extracting patient-specific coronary geometry data from a plurality of medical images;

extracting geometric features of a patient-specific vessel tree based on the patient-specific coronary geometry data;

training a first machine learning model on a database of synthetic coronary arterial trees; wherein the database of synthetic coronary arterial trees representative of ACS conditions are generated by estimating a total coronary resistance index (TCRI ACS ) value of each artery of each synthetic coronary arterial tree; wherein each TCRI ACS is a function of a distance between a myocardial bed of each artery of each synthetic coronary arterial tree and a myocardial area affected by the ACS;

using the first machine learning model to determine one or more patient-specific hemodynamic measures of interest under stable conditions based on the patient-specific vessel tree;

extracting a plurality of ACS-related features from additional patient measurement data including blood biomarkers acquired at different time points and contrast propagation information; and

using a second machine learning model to refine the one or more patient-specific hemodynamic measures of interest based on the plurality of ACS-related features.

15. The method of claim 14 , wherein the additional patient measurement data further comprises one or more of perfusion imaging data, invasive measurements, and ECG signals.

16. The method of claim 14 , wherein the one or more patient-specific hemodynamic measures of interest comprise one or more of Fractional Flow Reserve and Coronary Flow Reserve.

17. The method of claim 14 , wherein the one or more patient-specific hemodynamic measures of interest comprise wall shear stress.

18. The method of claim 14 , wherein the one or more patient-specific hemodynamic measures of interest comprise risk of plaque rupture.

19. The method of claim 14 , further comprising:

predicting a risk of future events for patients with ACS based on the one or more patient-specific hemodynamic measures of interest.

20. The method of claim 14 , further comprising:

predicting evolution in time of the hemodynamic measures of interest after onset of ACS.

21. A parallel processing computing system for personalized assessment of patients with acute coronary syndrome (ACS), comprising:

a host computer configured to:

extract patient-specific coronary geometry data from one or more medical images of a patient,

extract features of a patient-specific coronary arterial tree based on the patient-specific coronary geometry data, wherein the patient-specific coronary arterial tree representative of a ACS condition is generated by estimating a total coronary resistance index (TCRI ACS ) value of each artery of the patient-specific coronary arterial tree; wherein each TCRI ACS is a function of a distance between a myocardial bed of each artery of the patient-specific coronary arterial tree and a myocardial area affected by the ACS, and

extract a plurality of ACS-related features from additional patient measurement data including blood biomarkers acquired at different time points and contrast propagation information; and

a device computer configured to predict patient-specific hemodynamic measures of interest related to ACS based on the features of the patient-specific coronary arterial tree and the plurality of AC S-related features from the additional patient measurement data by applying one or more machine learning models in parallel across a plurality of computation units.