System and method for machine learning based mapping function of patient data from one physiological state to another based on computational fluid dynamics simulation of blood via offline training

1. A method for determining a quantity of interest of a patient, comprising:

receiving, using a processor, patient data of the patient at a first physiological state, wherein the patient data comprises medical image data of the patient;

determining, using the processor, a value of a quantity of interest of the patient at the first physiological state based on a patient-specific computational fluid dynamics simulation of blood flow, the patient-specific computational fluid dynamics simulation performed using boundary conditions corresponding to the first physiological state determined based on the medical image data of the patient, the quantity of interest representing a medical characteristic of the patient;

extracting, using the processor, features from the patient data, wherein the features which are extracted are based on the quantity of interest to be determined for the patient at a second physiological state; and

mapping, using the processor, the value of the quantity of interest of the patient at the first physiological state to a value of the same quantity of interest of the patient at the second physiological state using a machine learning based mapping function based on the extracted features, the machine learning based mapping function receiving as input 1) the value of the quantity of interest of the patient at the first physiological state and 2) the extracted features and outputting the value of the same quantity of interest of the patient at the second physiological state,

wherein the machine learning based mapping function is trained during an offline step using training data comprising values of the quantity of interest for a set of patients at the first physiological state and ground truth values of the same quantity of interest for the set of patients at the second physiological state.

2. The method as recited in claim 1 , wherein mapping the value of the quantity of interest of the patient at the first physiological state to the value of the same quantity of interest of the patient at the second physiological state further comprises:

mapping the value of the quantity of interest of the patient at the first physiological state to the value of the same quantity of interest of the patient at the second physiological state without using data of the patient at the second physiological state.

3. The method as recited in claim 1 , wherein the machine learning based mapping function is trained to learn a relationship between the quantity of interest of a set of patients at the first physiological state and the same quantity of interest of the set of patients at the second physiological state.

4. The method as recited in claim 1 , wherein the training data further comprises the quantities of interest of the set of patients simulated at the first physiological state and the same quantities of interest simulated at the second physiological state.

5. The method as recited in claim 1 , the patient data comprises medical image data of the patient, and extracting features from the patient data comprises:

processing the medical image data of the patient to determine measurements of the patient.

6. An apparatus for determining a quantity of interest of a patient, comprising:

a memory storing computer program instructions; and

at least one processor configured to execute the computer program instructions, the computer program instructions configured to cause the at least one processor to perform operations of:

receiving patient data of the patient at a first physiological state;

determining a value of a quantity of interest of the patient at the first physiological state based on a patient-specific computational fluid dynamics simulation of blood flow, the patient-specific computational fluid dynamics simulation performed using boundary conditions corresponding to the first physiological state determined based on the medical image data of the patient, the quantity of interest representing a medical characteristic of the patient;

extracting features from the patient data, wherein the features which are extracted are based on the quantity of interest to be determined for the patient at a second physiological state; and

mapping the value of the quantity of interest of the patient at the first physiological state to a value of the same quantity of interest of the patient at the second physiological state using a machine learning based mapping function based on the extracted features, the machine learning based mapping function receiving as input 1) the value of the quantity of interest of the patient at the first physiological state and 2) the extracted features and outputting the value of the same quantity of interest of the patient at the second physiological state,

wherein the machine learning based mapping function is trained during an offline step using training data comprising values of the quantity of interest for a set of patients at the first physiological state and ground truth values of the same quantity of interest for the set of patients at the second physiological state.

7. The apparatus as recited in claim 6 , wherein mapping the value of the quantity of interest of the patient at the first physiological state to the value of the same quantity of interest of the patient at the second physiological state further comprises:

mapping the value of the quantity of interest of the patient at the first physiological state to the value of the same quantity of interest of the patient at the second physiological state without using data of the patient at the second physiological state.

8. The apparatus as recited in claim 6 , wherein the machine learning based mapping function is trained to learn a relationship between the quantity of interest of a set of patients at the first physiological state and the same quantity of interest of the set of patients at the second physiological state.

9. The apparatus as recited in claim 6 , wherein the training data further comprises the quantities of interest of the set of patients simulated at the first physiological state and the same quantities of interest simulated at the second physiological state.

10. A non-transitory computer readable medium storing computer program instructions for determining a quantity of interest of a patient, the computer program instructions when executed by a processor cause the processor to perform operations comprising:

receiving patient data of the patient at a first physiological state;

determining a value of a quantity of interest of the patient at the first physiological state based on a patient-specific computational fluid dynamics simulation of blood flow the patient-specific computational fluid dynamics simulation performed using boundary conditions corresponding to the first physiological state determined based on the medical image data of the patient, the quantity of interest representing a medical characteristic of the patient;

extracting features from the patient data, wherein the features which are extracted are based on the quantity of interest to be determined for the patient at a second physiological state; and

mapping the value of the quantity of interest of the patient at the first physiological state to a value of the same quantity of interest of the patient at the second physiological state using a machine learning based mapping function based on the extracted features, the machine learning based mapping function receiving as input 1) the value of the quantity of interest of the patient at the first physiological state and 2) the extracted features and outputting the value of the same quantity of interest of the patient at the second physiological state,

wherein the machine learning based mapping function is trained during an offline step using training data comprising values of the quantity of interest for a set of patients at the first physiological state and ground truth values of the same quantity of interest for the set of patients at the second physiological state.

11. The non-transitory computer readable medium as recited in claim 10 , wherein mapping the value of the quantity of interest of the patient at the first physiological state to the value of the same quantity of interest of the patient at the second physiological state further comprises:

mapping the value of the quantity of interest of the patient at the first physiological state to the value of the same quantity of interest of the patient at the second physiological state without using data of the patient at the second physiological state.

12. The non-transitory computer readable medium as recited in claim 10 , the patient data comprises medical image data of the patient, and extracting features from the patient data comprises:

processing the medical image data of the patient to determine measurements of the patient.