Systems and methods for treatment planning based on plaque progression and regression curves
Systems and methods are disclosed for evaluating a patient with vascular disease. One method includes receiving patient-specific data regarding a geometry of the patient's vasculature; creating an anatomic model representing at least a portion of a location of disease in the patient's vasculature based on the received patient-specific data; identifying one or more changes in geometry of the anatomic model based on a modeled progression or regression of disease at the location; calculating one or more values of a blood flow characteristic within the patient's vasculature using a computational model based on the identified one or more changes in geometry of the anatomic model; and generating an electronic graphical display of a relationship between the one or more values of the calculated blood flow characteristic and the identified one or more changes in geometry of the anatomic model.
1 . A method for non-invasive assessment and therapy planning for coronary artery disease from patient-specific medical image data of a specific patient, comprising:
extracting geometric features from the patient-specific medical image data representing a geometry of the specific patient;
detecting one or more lesions in the geometry of the specific patient using the extracted geometric features, wherein detecting the one or more lesions in the geometry of the specific patient comprises analyzing a plaque composition at the one or more lesions to quantify a functional significance of the one or more lesions based on a stochastic collocation algorithm that correlates an impact of the functional significance of the one or more lesions in the geometry of the specific patient with a predicted impact on hemodynamics for the specific patient;
determining a plurality of pressures at one or more points of interest along the geometry of the specific patient including multiple points within each of the one or more lesions based on the extracted geometric features using a first machine learning model;
predicting post-treatment values for the plurality of pressures at the one or more points of interest along the geometry of the specific patient for each of one or more candidate treatment options for the specific patient based on the functional significance of the one or more lesions and based on the plurality of pressures, wherein predicting the post-treatment values includes evaluating a risk score for the specific patient based on the functional significance of the one or more lesions and the plurality of pressures, and wherein the one or more candidate treatment options are selected from the group consisting of a surgical treatment, medical therapy, continuous monitoring, and a combination thereof; and
displaying a visualization of a treatment prediction for at least one of the candidate treatment options for the specific patient to determine an optimized treatment plan for the specific patient.
2 . The method of claim 1 , wherein the first machine learning model comprises a plurality of machine learning algorithms, and determining the plurality of pressures at the one or more points of interest along the geometry of the specific patient based on the extracted geometric features using a first machine learning algorithm of the plurality of machine learning algorithms comprises:
determining the plurality of pressures at points along healthy segments of the geometry of the specific patient using the plurality of machine learning algorithms; and
determining the plurality of pressures at the multiple points within each of the one or more lesions using the plurality of machine learning algorithms.
3 . The method of claim 2 , wherein determining the plurality of pressures at the multiple points within each of the one or more lesions using the plurality of machine learning algorithms comprises, for each of the one or more lesions:
receiving a plurality of feature vectors of physiological parameters associated with a plurality of blood flow characteristics of the specific patient; and
computing a plurality of predicted blood flow characteristics of the one or more lesions based on the plurality of feature vectors associated with the plurality of blood flow characteristics.
4 . The method of claim 1 , wherein the first machine learning model comprises a support vector machine (SVM).
5 . The method of claim 1 , wherein the first machine learning model comprises a multi-layer perceptron (MLP).
6 . The method of claim 1 , wherein the first machine learning model comprises a multivariate regression (MVR).
7 . The method of claim 1 , wherein at least one of the one or more candidate treatment options corresponds to one or more candidate percutaneous coronary intervention (PCI) treatments, and predicting post-treatment values for the plurality of pressures at the one or more points of interest along the geometry of the specific patient including the multiple points within each of the one or more lesions for the one or more candidate treatment options for the specific patient, comprises:
inputting the extracted geometric features to a second trained machine learning model;
predicting patient-specific post-PCI geometric features for the one or more candidate PCI treatments based on the inputted extracted geometric features using the second trained machine learning model; and
determining, for the one or more candidate PCI treatments, post-PCI values for the plurality of pressures at the one or more points of interest along the geometry of the specific patient including the multiple points within each of the one or more lesions based on the predicted patient-specific post-PCI geometric features using the first machine learning model.
8 . The method of claim 1 , wherein at least one of the one or more candidate treatment options corresponds to a candidate percutaneous coronary intervention (PCI) treatment, and determining the plurality of pressures at the one or more points of interest along the geometry of the specific patient including the multiple points within each of the one or more lesions for the one or more candidate treatment options for the specific patient, comprises:
inputting the extracted geometric features, a number of the lesions detected in the geometry of the specific patient, and locations of the lesions detected in the geometry of the specific patient to a second trained machine learning model; and
determining, for the candidate PCI treatment, post-PCI values for the plurality of pressures at the one or more points of interest along the geometry of the specific patient including the multiple points within each of the one or more lesions based on the input extracted geometric features using the second trained machine learning model, wherein the candidate PCI treatment corresponds to a possible combination of stenting at the lesions in the geometry of the specific patient.
9 . The method of claim 1 , wherein at least one of the one or more candidate treatment options corresponds to a candidate percutaneous coronary intervention (PCI) treatment, and the method further comprises:
determining, for the candidate PCI treatment, a plaque remodeling index using a second trained machine learning model based on the geometric features corresponding to post-PCI anatomy for the candidate PCI treatment and other features including one or more of demographic features or blood biomarkers.
10 . The method of claim 9 , wherein predicting, for the candidate PCI treatment, a plaque remodeling index using a second trained machine learning model based on the geometric features corresponding to post-PCI anatomy for the candidate PCI treatment and other features including one or more of demographic features or blood biomarkers comprises:
determining, for the candidate PCI treatment, the plaque remodeling index at the plurality of points along the geometry including the multiple points within each of the one or more lesions using the second trained machine learning model, resulting in a respective sensitivity curve for the candidate PCI treatment, wherein the plaque remodeling index at each point corresponds to a likelihood that the one or more lesions are functionally significant.
11 . The method of claim 10 , wherein the candidate PCI treatment includes one or more stenting locations, and displaying a visualization of a treatment prediction for the candidate treatment option for the specific patient comprises:
displaying, for the candidate PCI treatment, an image showing at least a portion of the geometry of the specific patient with a visual representation of a treatment option overlaid on the geometry at the one or more stenting locations for the candidate PCI treatment, the respective sensitivity curve for the candidate PCI treatment, and the determined plurality of pressures for the candidate PCI treatment.
12 . The method of claim 9 , further comprising:
outputting a risk score for the candidate PCI treatment based on the predicted post-treatment values of the determined plurality of pressures and the determined plaque remodeling index for the candidate PCI treatment.
13 . A system for non-invasive assessment and therapy planning for coronary artery disease from patient-specific medical image data of a specific patient, comprising:
at least one data storage device storing instructions for controlling image segmentation; and
at least one processor configured to execute the instruction to perform operations comprising:
extracting geometric features from the patient-specific medical image data representing a geometry of the specific patient;
detecting one or more lesions in the geometry of the specific patient, wherein detecting the one or more lesions in the geometry of the specific patient comprises analyzing a plaque composition at the one or more lesions to quantify a functional significance of the one or more lesions based on a stochastic collocation algorithm that correlates an impact of the functional significance of the one or more lesions in the geometry of the specific patient with a predicted impact on hemodynamics for the specific patient;
determining a plurality of pressures at one or more points of interest along the geometry of the specific patient based on the extracted geometric features using a first machine learning model;
predicting post-treatment values for the plurality of pressures at the one or more points of interest along the geometry of the specific patient for each of one or more candidate treatment options for the specific patient based on the functional significance of the one or more lesions and based on the plurality of pressures, wherein predicting the post-treatment values includes evaluating a risk score for the specific patient based on the functional significance of the one or more lesions and the plurality of pressures, and wherein the one or more candidate treatment options is selected from the group consisting of a surgical treatment, medical therapy, continuous monitoring, and a combination thereof; and
displaying a visualization of a treatment prediction for at least one of the candidate treatment options for the specific patient to determine an optimized treatment plan for the specific patient.
14 . The system of claim 13 , wherein the processor is further configured to determine, for each of one or more candidate percutaneous coronary intervention (PCI) treatments, a plaque remodeling index using a second trained machine learning model based on the geometric features corresponding to post-PCI anatomy for each of the one or more candidate PCI treatments and other features including one or more of demographic features or blood biomarkers.
15 . The system of claim 14 , wherein the processor is further configured to output a risk score for the one or more candidate PCI treatments based on the predicted post-treatment values of the determined plurality of pressures and the determined plaque remodeling index for each of the one or more candidate PCI treatments.
16 . The system of claim 13 , the operations further comprising:
inputting the extracted geometric features, a number of the lesions detected in the geometry of the specific patient, and locations of the lesions detected in the geometry of the specific patient to a second trained machine learning model; and
determining, for each of one or more candidate percutaneous coronary intervention (PCI) treatments corresponding to respective possible combinations of stenting at the detected lesions in the geometry, post-PCI values for the plurality of pressures at the one or more points of interest along the geometry of the specific patient including multiple points within each of the one or more lesions based on the input extracted geometric features using the second trained machine learning model.
17 . A non-transitory computer readable medium for use on a computer system containing computer-executable programming instructions for non-invasive assessment and therapy planning for coronary artery disease from patient-specific medical image data of a specific patient, the instructions being executable by the computer system for:
extracting geometric features from the patient-specific medical image data representing a geometry of the specific patient;
detecting one or more lesions in the geometry of the specific patient, wherein detecting the one or more lesions in the geometry of the specific patient comprises analyzing a plaque composition at the one or more lesions to quantify a functional significance of the one or more lesions based on a stochastic collocation algorithm that correlates an impact of the functional significance of the one or more lesions in the geometry of the specific patient with a predict impact on hemodynamics for the specific patient;
determining a plurality of pressures at one or more points of interest along the geometry of the specific patient including multiple points within each of the one or more lesions based on the extracted geometric features using a first machine learning model;
predicting post-treatment values for the plurality of pressures at the one or more points of interest along the geometry of the specific patient including the multiple points within each of the one or more lesions for each of one or more candidate treatment options for the specific patient based on the functional significance of the one or more lesions and based on the plurality of pressures, wherein predicting the post-treatment values includes evaluating a risk score for the specific patient based on the functional significance of the one or more lesions and the plurality of pressures, and wherein the one or more candidate treatment options is selected from the group consisting of a surgical treatment, medical therapy, continuous monitoring, and a combination thereof; and
displaying a visualization of a treatment prediction for at least one of the candidate treatment options for the specific patient to determine an optimized treatment plan for the specific patient.
18 . The non-transitory computer readable medium of claim 17 , the instructions further comprising:
determining, for each of one or more percutaneous coronary intervention (PCI) candidate treatments, a plaque remodeling index using a second trained machine learning model based on the geometric features corresponding to post-PCI anatomy for each of the one or more candidate PCI treatments and other features including one or more of demographic features or blood biomarkers.
19 . The non-transitory computer readable medium of claim 18 , the instructions further comprising:
outputting a risk score for the one or more candidate PCI treatments based on the predicted post-treatment values of the determined plurality of pressures and the determined plaque remodeling index for each of the one or more candidate PCI treatments.
20 . The non-transitory computer readable medium of claim 17 , the instructions further comprising:
inputting the extracted geometric features, a number of the lesions detected in the geometry of the specific patient, and locations of the lesions detected in the geometry of the specific patient to a second trained machine learning model; and
determining, for each of one or more candidate percutaneous coronary intervention (PCI) treatments corresponding to respective possible combinations of stenting at the detected lesions in the geometry, post-PCI values for the plurality of pressures at the one or more points of interest along the geometry of the specific patient including the multiple points within each of the one or more lesions based on the input extracted geometric features using the second trained machine learning model.