IP Library Granted Patent US 12670999
Granted Patent B2
US 12670999 · App. 17/864,559 · Granted Jun 30, 2026

Systems and methods for processing electronic images to determine flow using flow ratio

Inventors: Timothy A. Fonte (San Francisco, CA); Leo Grady (Darien, CT)
Assignee: Heartflow, Inc.
G16H50/50A61B5/0044A61B5/02007A61B5/021A61B5/0263A61B5/029A61B5/1073A61B5/7267A61B5/7275A61B5/7278A61B6/032A61B6/503A61B6/507A61B6/5217A61B8/02A61B8/04A61B8/065A61B8/5223A61B34/10G06F17/10G06F17/11G06F30/20G06T19/00G16Z99/00A61B5/0037A61B5/02028A61B5/026A61B2034/105G06T2210/41Y02A90/10
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12670999
App. No.
17/864,559
Granted
Jun 30, 2026
Kind
B2
Abstract

Embodiments include a system for determining patient cardiovascular information which includes at least one computer system configured to receive patient-specific data regarding a geometry of an anatomical structure of a patient; create a model representing at least a portion of the anatomical structure of the patient based on the patient-specific data; determine a first blood flow rate at at least one point of interest in the model by using relations of individual-specific anatomic data to functional estimates of blood flow characteristics generated from a plurality of individuals; modify the model; determine a second blood flow rate at a point in the modified model corresponding to the at least one point of interest by using the relations of individual-specific anatomic data to functional estimates of blood flow characteristics; and determine a fractional flow reserve value as a ratio of the second blood flow rate to the first blood flow rate.

Claims (50)

1 . A method for processing electronic images to determine blood flow information, comprising:

segmenting computed tomography (CT) imaging data of a patient in a diseased condition to identify a list of voxels belonging to at least a portion of anatomy of the patient that includes a diseased vessel, the list of voxels defining, for each voxel, a spatial location of the voxel and an identification of any neighboring voxels;

generating, based on the list of voxels, a patient-specific anatomical model representative of the patient in the diseased condition, the patient-specific anatomical model including a representation of the diseased vessel;

generating a second patient-specific anatomical model of the patient in a hypothetical healthy condition, the second patient-specific anatomical model including a hypothetical healthy condition version of the diseased vessel, wherein generating the second patient-specific anatomical model includes modifying a geometry of the representation of the diseased vessel to simulate removal of a narrowing or obstruction associated with the diseased vessel;

generating a first blood flow for the patient using the patient-specific anatomical model and generating a second blood flow for the patient using the second patient-specific anatomical model wherein the generating includes, in each case:

determining feature vectors for a plurality of locations within the anatomy of a respective model;

estimating a Fractional Flow Reserve (FFR) value for each feature vector at a corresponding location within the respective model, and appending the estimated FFR value to the feature vector; and

applying a trained machine-learning model to the appended feature vectors for the respective model, the trained machine-learning model trained based on training feature vectors and training blood flow data from one or more individuals to output the first and second blood flow values, respectively; and

calculating a hemodynamic index using the first blood flow and the second blood flow.

2 . The method of claim 1 , wherein the hemodynamic index is based on a ratio of the first blood flow to the second blood flow.

3 . The method of claim 1 , further comprising:

determining a functional significance of the diseased vessel based on the hemodynamic index.

4 . The method of claim 1 , wherein generating the first and second blood flows for the patient each include simulating a state of hyperemia for the patient.

5 . The method of claim 1 , wherein each feature vector includes a numerical description of geometry of the anatomy at the corresponding location.

6 . The method of claim 5 , wherein estimating the FFR value includes using a further trained machine-learning model that has been trained based on training geometry and training FFR values to output the estimated FFR value.

7 . The method of claim 5 , further comprising, for each of the plurality of locations, extracting one or more geometric features of the anatomy from the respective model, wherein the numerical description of geometry includes a numerical description of the one or more geometric features.

8 . A system for processing images to determine blood flow information, comprising:

at least one memory storing instructions and computed tomography (CT) imaging data of a patient in a diseased condition; and

at least one processor operatively connected to the at least one memory and configured to perform operations, including:

segmenting the CT imaging data to identify a list of voxels belonging to at least a portion of anatomy of the patient that includes a diseased vessel, the list of voxels defining, for each voxel, a spatial location of the voxel and an identification of any neighboring voxels;

generating, based on the list of voxels, a patient-specific anatomical model representative of the patient in the diseased condition, the patient-specific anatomical model including a representation of the diseased vessel;

generating a second patient-specific anatomical model of the patient in a hypothetical healthy condition, the second patient-specific anatomical model including a hypothetical healthy condition version of the diseased vessel, wherein generating the second patient-specific anatomical model includes modifying a geometry of the representation of the diseased vessel to simulate removal of a narrowing or obstruction associated with the diseased vessel;

generating a first blood flow for the patient using the patient-specific anatomical model and generating a second blood flow for the patient using the second patient-specific anatomical model wherein the generating includes, in each case:

determining feature vectors for a plurality of locations within the anatomy of a respective model;

estimating a Fractional Flow Reserve (FFR) value for each feature vector at a corresponding location within the respective model, and appending the estimated FFR value to the feature vector; and

applying a trained machine-learning model to the appended feature vectors for the respective model, the trained machine-learning model trained based on training feature vectors and training blood flow data from one or more individuals to output the first and second blood flow values, respectively; and

calculating a hemodynamic index using the first blood flow and the second blood flow.

9 . The system of claim 8 , wherein the hemodynamic index is based on a ratio of the first blood flow second blood flow.

10 . The system of claim 8 , wherein the operations further include:

determining a functional significance of the diseased vessel based on the hemodynamic index.

11 . The system of claim 8 , wherein generating each blood flow for the patient includes simulating a state of hyperemia for the patient.

12 . The system of claim 8 , wherein each feature vector includes a numerical description of geometry of the anatomy at the corresponding location.

13 . The system of claim 12 , wherein estimating the FFR value includes using a further trained machine-learning model that has been trained based on training geometry and training FFR values to output the estimated FFR value.

14 . The system of claim 12 , wherein:

the operations further include, for each of the plurality of locations, extracting one or more geometric features of the anatomy from the respective model; and

the numerical description of geometry includes a numerical description of the one or more geometric features.

15 . A non-transitory computer-readable medium comprising instructions that are executable by at least one processor to perform operations, including:

segmenting computed tomography (CT) imaging data of a patient in a diseased condition to identify a list of voxels belonging to at least a portion of anatomy of the patient that includes a diseased vessel, the list of voxels defining, for each voxel, a spatial location of the voxel and an identification of any neighboring voxels;

generating, based on the list of voxels, a patient-specific anatomical model representative of the patient in the diseased condition, the patient-specific anatomical model including a representations of the diseased vessel;

generating a second patient-specific anatomical model of the patient in a hypothetical healthy condition, the second patient-specific anatomical model including a hypothetical healthy condition version of the diseased vessel, wherein generating the second patient-specific anatomical model includes modifying a geometry of the representation of the diseased vessel to simulate removal of a narrowing or obstruction associated with the diseased vessel;

generating a first blood flow for the patient using the patient-specific anatomical model and generating a second blood flow for the patient using the second patient-specific anatomical model wherein the generating includes, in each case:

determining feature vectors for a plurality of locations within the anatomy of a respective model;

estimating a Fractional Flow Reserve (FFR) value for each feature vector at a corresponding location within the respective model, and appending the estimated FFR value to the feature vector; and

applying a trained machine-learning model to the appended feature vectors for the respective model, the trained machine-learning model trained based on training feature vectors and training blood flow data from one or more individuals to output the first and second blood flow values, respectively; and

calculating a hemodynamic index using the first blood flow and the second blood flow.

16 . The non-transitory computer-readable medium of claim 15 , wherein each feature vector includes a numerical description of geometry of the anatomy at the corresponding location.

17 . The non-transitory computer-readable medium of claim 16 , wherein estimating the FFR value includes using a further trained machine-learning model that has been trained based on training geometry and training FFR values to output the estimated FFR value.

18 . The non-transitory computer-readable medium of claim 16 , wherein:

the operations further include, for each of the plurality of locations, extracting one or more geometric features of the anatomy from the respective model; and

the numerical description of geometry includes a numerical description of the one or more geometric features.