SYSTEMS AND METHODS FOR ESTIMATING ISCHEMIA AND BLOOD FLOW CHARACTERISTICS FROM VESSEL GEOMETRY AND PHYSIOLOGY
Systems and methods are disclosed for determining individual-specific blood flow characteristics. One method includes acquiring, for each of a plurality of individuals, individual-specific anatomic data and blood flow characteristics of at least part of the individual's vascular system; executing a machine learning algorithm on the individual-specific anatomic data and blood flow characteristics for each of the plurality of individuals; relating, based on the executed machine learning algorithm, each individual's individual-specific anatomic data to functional estimates of blood flow characteristics; acquiring, for an individual and individual-specific anatomic data of at least part of the individual's vascular system; and for at least one point in the individual's individual-specific anatomic data, determining a blood flow characteristic of the individual, using relations from the step of relating individual-specific anatomic data to functional estimates of blood flow characteristics.
1 - 31 . (canceled)
32 . A method for determining fractional flow reserve (FFR) for a plaque of interest for a patient, comprising:
receiving a medical image of the patient including the plaque of interest;
detecting image regions corresponding to the plaque of interest and a coronary artery tree of the patient; and
determining an FFR value associated with the plaque of interest using a trained machine learning algorithm applied directly to the detected image regions.
33 . The method of claim 32 , wherein detecting image regions corresponding to the plaque of interest and a coronary tree of the patient comprises:
detecting image regions corresponding to the plaque of interest, coronary ostia, coronary vessels, and coronary branching.
34 . The method of claim 32 , wherein detecting image regions corresponding to the plaque of interest and a coronary tree of the patient comprises:
detecting the image regions in one or more feature vector spaces using a respective trained machine learning algorithm for each of the feature vector spaces.
35 . The method of claim 34 , wherein the one or more feature vector spaces comprises a position parameter space, and a position-curvature parameter space.
36 . The method of claim 34 , wherein the respective trained machine learning algorithm for each of the feature vector spaces comprises a machine learning algorithm trained that inputs parameters, and for each parameter outputs a prediction for an image region in the respective parameter space.
37 . The method of claim 34 , wherein the respective trained machine learning algorithm for each of the feature vector spaces comprises a discriminative machine learning algorithm that inputs parameters, and for each parameter outputs a confidence for the image region corresponding to the prediction.
38 . The method of claim 34 , wherein the respective machine learning algorithm for each of the feature vector spaces is a machine learning algorithm trained using one of a support vector machine (SVM), multi-layer perceptron (MLPs), and multivariate regression (MVR).
39 . The method of claim 32 , wherein the trained machine learning algorithm is a machine learning algorithm with a plurality of layers and a final layer calculates a plaque specific FFR value.
40 . The method of claim 39 , wherein the trained machine learning algorithm is trained by tuning weights for each layer other than the final layer using a first set of training images without corresponding FFR values and then refining the weights for each layer including the final layer based on a second set of training images with corresponding FFR values.
41 . An apparatus for determining fractional flow reserve (FFR) for a plaque of interest for a patient, comprising:
means for receiving a medical image of the patient including the plaque of interest;
means for detecting image regions corresponding to the plaque of interest and a coronary tree of the patient; and
means for determining an FFR value for the plaque of interest using a trained machine learning algorithm applied directly to the detected image regions.
42 . The apparatus of claim 41 , wherein the means for detecting image regions corresponding to the plaque of interest and a coronary tree of the patient comprises:
means for detecting image regions corresponding to the plaque of interest, coronary ostia, coronary vessels, and coronary branching.
43 . The apparatus of claim 41 , wherein the means for detecting image regions corresponding to the plaque of interest and a coronary tree of the patient comprises:
means for detecting the image regions in a series of feature vector spaces using a respective trained machine learning algorithm for each of the feature vector spaces.
44 . The apparatus of claim 43 , wherein the series of feature vector spaces comprises a position parameter space, a position-curvature parameter space, and a position-curvature-scale parameter space.
45 . The apparatus of claim 43 , wherein the respective trained machine learning algorithm for each of the feature vector spaces comprises a machine learning algorithm trained that inputs hypotheses in the respective parameter space and for each hypothesis, outputs a prediction for an image region in the respective parameter space.
46 . The apparatus of claim 43 , wherein the respective trained machine learning algorithm for each of the feature vector spaces comprises a discriminative machine learning algorithm that inputs hypotheses in the respective parameters space, and for each hypothesis outputs a confidence for the image region corresponding to the hypothesis.
47 . The apparatus of claim 43 , wherein the respective machine learning algorithm for each of the feature vector spaces is a machine learning algorithm trained using one of a convolutional neural network (CNN), a stacked restricted Boltzmann machine (RBM), or a stacked auto-encoder (AE).
48 . The apparatus of claim 41 , wherein the trained machine learning algorithm is a machine learning algorithm with a plurality of layers and a final layer calculates a plaque specific FFR value.
49 . The apparatus of claim 48 , wherein the trained machine learning algorithm is trained by tuning weights for each layer other than the final layer using a first set of training images without corresponding FFR values and then refining the weights for each layer including the final layer based on a second set of training images with corresponding FFR values.
50 . A non-transitory computer readable medium storing computer program instructions for determining fractional flow reserve (FFR) for a plaque of interest for a patient, the computer program instructions when executed by a processor cause the processor to perform operations comprising:
receiving a medical image of the patient including the plaque of interest;
detecting image regions corresponding to the plaque of interest and a coronary tree of the patient; and
determining an FFR value for the plaque of interest using a trained machine learning algorithm applied directly to the detected image regions.
51 . The non-transitory computer readable medium of claim 50 , wherein detecting image regions corresponding to the plaque of interest and a coronary tree of the patient comprises:
detecting image regions corresponding to the plaque of interest, coronary ostia, coronary vessels, and coronary branching.
52 . The non-transitory computer readable medium of claim 50 , wherein detecting image regions corresponding to the plaque of interest and a coronary tree of the patient comprises:
detecting the image regions in a series of feature vector spaces using a respective trained machine learning algorithm for each of the feature vector spaces.
53 . The non-transitory computer readable medium of claim 52 , wherein the series of feature vector spaces comprises a position parameter space, a position-curvature parameter space, and a position-curvature-scale parameter space.
54 . The non-transitory computer readable medium of claim 52 , wherein the respective trained machine learning algorithm for each of the feature vector spaces comprises a machine learning algorithm trained that inputs hypotheses in the respective parameter space and for each hypothesis, outputs a prediction for an image region in the respective parameter space.
55 . The non-transitory computer readable medium of claim 52 , wherein the respective trained machine learning algorithm for each of the feature vector spaces comprises a discriminative machine learning algorithm that inputs hypotheses in the respective parameters space, and for each hypothesis outputs a confidence for the image region corresponding to the hypothesis.
56 . The non-transitory computer readable medium of claim 52 , wherein the respective machine learning algorithm for each of the feature vector spaces is a machine learning algorithm trained using one of a convolutional neural network (CNN), a stacked restricted Boltzmann machine (RBM), or a stacked auto-encoder (AE).
57 . The non-transitory computer readable medium of claim 50 , wherein the trained machine learning algorithm is a machine learning algorithm with a plurality of layers and a final layer calculates a plaque specific FFR value.
58 . The non-transitory computer readable medium of claim 57 , wherein the trained machine learning algorithm is trained by tuning weights for each layer other than the final layer using a first set of training images without corresponding FFR values and then refining the weights for each layer including the final layer based on a second set of training images with corresponding FFR values.