Systems and methods for risk assessment and treatment planning of arterio-venous malformation

1. A computer implemented method for assessing an arterio-venous malformation (AVM), the method comprising:

receiving a patient-specific three-dimensional anatomic model of a portion of an anatomy of a patient, including one or more blood vessels having one or more vessel walls;

identifying, using a computer processor, features of the one or more blood vessels of the patient-specific model associated with the AVM;

calculating one or more vessel wall properties of the one or more vessel walls by inputting the identified features into a trained machine learning algorithm;

estimating a risk of an undesirable outcome caused by the AVM based on the calculated vessel wall properties;

modifying the patient-specific three-dimensional anatomic model to evaluate one or more treatments of the AVM by first identifying larger blood vessels for intervention;

identifying, using a computer processor, features of the one or more blood vessels of the modified patient-specific three-dimensional anatomic model associated with a treatment of the AVM; and

calculating one or more vessel wall properties at one or more points through the modified patient-specific three-dimensional anatomic model, by inputting the identified features into a trained machine learning algorithm.

2. The method of claim 1 , further comprising:

determining a blood flow characteristic at one or more points of the patient-specific model associated with the AVM;

creating feature vectors comprising the identified features and the determined blood flow characteristics; and

calculating one or more vessel wall properties of the one or more vessel walls, by inputting the identified features into a trained machine learning algorithm.

3. The method of claim 2 , wherein the blood flow characteristic is determined by performing a blood flow simulation through the one or more identified blood vessels having the one or more vessel walls of the patient-specific three-dimensional anatomic model.

4. The method of claim 3 , wherein the blood flow simulation includes simulating or predicting progression, regression, or remodeling of the AVM.

5. The method of claim 1 , further comprising:

determining a second blood flow characteristic at one or more points of the modified patient-specific model associated with one or more treatments of the AVM;

creating feature vectors comprising the identified features and the determined second blood flow characteristic; and

calculating one or more vessel wall properties of the one or more vessel walls, by inputting the created feature vector into a trained machine learning algorithm.

6. The method of claim 1 , wherein the machine learning algorithm is trained by:

receiving data from a plurality of individuals comprising: (1) features at one or more points of a three-dimensional model associated with an AVM of one or more blood vessels having one or more vessel walls, and (2) one or more vessel wall properties at points corresponding to the one or more points of the three-dimensional model;

associating the features at the one or more points of a three-dimensional model associated with an AVM with the one or more vessel wall properties at points corresponding the one or more points; and

training a machine learning algorithm that can predict one or more vessel wall properties at points of a three-dimensional model of one or more blood vessels having one or more vessel walls from an input of features at one or more points of a three-dimensional model, based on learning from the associated features.

7. The method of claim 1 , wherein the patient-specific model is based on images of at least a portion of a vascular system of the patient.

8. The method of claim 1 , wherein the undesirable outcome includes one or more of rupture of a blood vessel, a mass effect, an increase in venous pressure, and a change in tissue perfusion.

9. The method of claim 1 , wherein evaluating the one or more treatments includes:

identifying one or more blood vessels for treatment;

determining an effect on blood flow caused by treatment of the one or more blood vessels; and

assessing a change the risk of the undesirable outcome.

10. The method of claim 9 , wherein the treatment includes one or more of embolization of the one or more blood vessels, ablation of the one or more blood vessels, surgical removal of all or part of the AVM, or radiosurgery.

11. The method of claim 1 , further comprising estimating the risk of the undesirable outcome based on a homeostatic vessel state.

12. A system for assessing an arterio-venous malformation (AVM), the system comprising:

a data storage device storing instructions for assessing an arterio-venous malformation (AVM); and

a processor configured to execute the instructions to perform a method comprising the steps:

receiving a patient-specific three-dimensional anatomic model of a portion of an anatomy of a patient, including one or more blood vessels having one or more vessel walls;

identifying, using a computer processor, features of the one or more blood vessels of the patient-specific model associated with the AVM;

calculating one or more vessel wall properties of the one or more vessel wall by inputting the identified features into a trained machine learning algorithm;

estimating a risk of an undesirable outcome caused by the AVM based on the calculated vessel wall properties;

modifying the patient-specific three-dimensional anatomic model to evaluate one or more treatments of the AVM by first identifying larger blood vessels for intervention;

identifying, using a computer processor, features of the one or more blood vessels of the modified patient-specific three-dimensional anatomic model associated with a treatment of the AVM; and

calculating one or more vessel wall properties at one or more points through the modified patient-specific three-dimensional anatomic model, by inputting the identified features into a trained machine learning algorithm.

13. The system of claim 12 , wherein the machine learning algorithm is trained by

receiving data from a plurality of individuals comprising: (1) features at one or more points of a three-dimensional model associated with an AVM of one or more blood vessels having one or more vessel walls, and (2) one or more vessel wall properties at points corresponding to the one or more points of the three-dimensional model;

associating the features at the one or more points of a three-dimensional model associated with an AVM with the one or more vessel wall properties at points corresponding the one or more points; and

training a machine learning algorithm that can predict one or more vessel wall properties at points of a three-dimensional model of one or more blood vessels having one or more vessel walls from an input of features at one or more points of a three-dimensional model, based on learning from the associated features.

14. The system of claim 12 , wherein the patient-specific model is based on images of at least a portion of a vascular system of the patient.

15. The system of claim 12 , wherein the undesirable outcome includes one or more of rupture of a blood vessel, a mass effect, an increase in venous pressure, and a change in tissue perfusion.

16. The system of claim 12 , wherein the blood flow simulation includes simulating or predicting progression, regression, or remodeling of the AVM.

17. The system of claim 12 , wherein evaluating the one or more treatments includes:

identifying one or more blood vessels for treatment; and

determining an effect on blood flow caused by treatment of the one or more blood vessels.

18. The system of claim 17 , wherein the treatment includes one or more of embolization of the one or more blood vessels, ablation of the one or more blood vessels, surgical removal of all or part of the AVM, or radiosurgery.

19. A non-transitory computer-readable medium storing instructions that, when executed by a computer, cause the computer to perform a method for assessing an arterio-venous malformation (AVM), the method comprising:

receiving a patient-specific three-dimensional anatomic model of a portion of an anatomy of a patient, including one or more blood vessels having one or more vessel walls;

identifying, using a computer processor, features of the one or more blood vessels of the patient-specific model associated with the AVM;

calculating one or more vessel wall properties of the one or more vessel walls by inputting the identified features into a trained machine learning algorithm;

estimating a risk of an undesirable outcome caused by the AVM based on the calculated vessel wall properties;

modifying the patient-specific three-dimensional anatomic model to evaluate one or more treatments of the AVM by first identifying larger blood vessels for intervention;

identifying, using a computer processor, features of the one or more blood vessels of the modified patient-specific three-dimensional anatomic model associated with a treatment of the AVM; and

calculating one or more vessel wall properties at one or more points through the modified patient-specific three-dimensional anatomic model, by inputting the identified features into a trained machine learning algorithm.

20. The computer-readable medium of claim 19 , wherein the machine learning algorithm is trained by

receiving data from a plurality of individuals comprising: (1) features at one or more points of a three-dimensional model associated with an AVM of one or more blood vessels having one or more vessel walls, and (2) one or more vessel wall properties at points corresponding to the one or more points of the three-dimensional model;

associating the features at the one or more points of a three-dimensional model associated with an AVM with the one or more vessel wall properties at points corresponding the one or more points; and

training a machine learning algorithm that can predict one or more vessel wall properties at points of a three-dimensional model of one or more blood vessels having one or more vessel walls from an input of features at one or more points of a three-dimensional model, based on learning from the associated features.