Abstract: A method and system for non-invasive assessment and therapy planning for coronary artery disease from medical image data of a patient is disclosed. Geometric features representing at least a portion of a coronary artery tree of the patient are extracted from medical image data. Lesions are detected in coronary artery tree of the patient and a hemodynamic quantity of interest is computed at a plurality of points along the coronary artery tree including multiple points within the lesions based on the extracted geometric features using a machine learning model, resulting in an estimated pullback curve for the hemodynamic quantity of interest.

Abstract: Systems and methods for retraining a trained machine learning model are provided. One or more input medical images are received. Measures of interest for a primary task and a secondary task are predicted from the one or more input medical images using a trained machine learning model. The predicted measures of interest for the primary task and the secondary task are output. User feedback on the predicted measure of interest for the secondary task is received. The trained machine learning model is retrained for predicting the measures of interest for the primary task and the secondary task based on the user feedback on the output for the secondary task.

Abstract: For segmentation in medical imaging, a shape generative adversarial network (shape GAN) is used in training. By including shape information in a lower dimensional space than the pixels or voxels of the image space, the network may be trained with a shape loss or optimization. The adversarial loss and the shape loss are used to train the network, so the resulting generator may segment complex shapes in 2D or 3D. Other optimization may be used, such as using a loss in image space.

Abstract: A method and system for non-invasive medical image based assessment of coronary artery disease (CAD) for clinical decision support using on-site and off-site processing is disclosed. Medical image data of a patient is received. A processing strategy for assessing CAD of the patient using one of on-site processing, off-site processing, or joint on-site and off-site processing is automatically selected based on clinical requirements for a current clinical scenario. Non-invasive assessment of CAD of the patient is performed based on the medical image data of the patient using one of on-site processing, off-site-processing, or joint on-site and off-site processing according to the selected processing strategy. A final assessment of CAD of the patient is output based on the non-invasive assessment of CAD.

Abstract: A method and system for simulating blood flow in a vessel of a patient to estimate hemodynamic quantities of interest using enhanced blood flow computations based on invasive physiological measurements of the patient is disclosed. Non-invasive patient data including medical image data is received and a patient-specific anatomical model the patient's vessels is generated. Invasive physiological measurements of the patient are received and a computational blood flow model is personalized using the invasive physiological measurements. Blood flow is simulated in the patient-specific anatomical model and one or more hemodynamic quantities of interest are computed using the personalized computational blood flow model.

Abstract: Machine learning is used to assess data for a patient in an emergency, providing rapid diagnosis based on a large amount of information. Assistance in triage may be provided. Given the large variety of patients and conditions that may occur, the machine learning may rely on synthetically generated images for more accurate prediction. The machine learning may accurately predict even with missing information and may be used to determine what missing information for a given patient is more or less important to obtain.

Abstract: Systems and methods are provided for automatic segmentation of a vessel. A sequence of image slices containing a vessel is acquired. Features maps are generated for each of the image slices using a trained fully convolutional neural network. A trained bi-directional recurrent neural network generates a segmented image based on the feature maps.

Abstract: A method for providing a personalized evaluation of CAD for a patient includes acquiring one or more non-invasive images depicting a patient's coronary arteries and extracting a first set of features of interest from the one or more non-invasive images. A machine learning model is applied to the first set of features of interest to yield a prediction of one or more coronary measures of interest. One or more invasive images depicting the patient's coronary arteries are acquired and a second set of features of interest are extracted from the one or more invasive images. The first set of features of interest and the second set of features of interest are combined to yield a combined set of features of interest. Then, the machine learning model may be applied to the combined set of features of interest to yield an enhanced prediction of the coronary measures of interest.

Abstract: A method and a corresponding system for assessing a haemodynamic parameter for a vascular region of interest of a patient based on angiographic images are provided. After acquiring multiple angiographic images, a three dimensional (3D) representation of at least a first portion of the respective region of interest is performed, and geometric features are extracted from complete or partial views. Additional geometric features are extracted from partial incomplete views. A complete set of 3D geometric features for an anatomical structure, such as a vessel tree, is then generated by combining the extracted geometric features and estimating any missing geometric features. Using the complete set of 3D geometric features, a feature-based assessment of the haemodynamic parameter, such as a fractional flow reserve, is then performed.

Abstract: Systems and methods for determining one or more measures of interest for optimizing throughput of a catheterization laboratory are provided. A priori medical procedure data relating to a medical procedure to be performed on a patient in a catheterization laboratory is received. One or more measures of interest are predicted based on the received a priori medical procedure data using a trained machine learning model. The one or more measures of interest include an overall time for performing the medical procedure on the patient in the catheterization laboratory. The one or more predicted measures of interest are output.

Abstract: Systems and methods are provided for acquiring a series of angiographic images and identifying the anatomical structures represented in the series of images using a machine learnt classifier. Additional series of images that would yield the optimal visualization of the structure of interest may be suggested.

Abstract: For soft tissue deformation prediction, a biomechanical or other tissue-related physics model is used to find an instantaneous state of the soft tissue. A machine-learned artificial neural network is applied to predict the position of volumetric elements (e.g., mesh node) from the instantaneous state. Since the machine-learned artificial neural network may predict quickly (e.g., in a second or less), the soft tissue position at different times or a further time given the instantaneous state is provided in real-time without the minutes of physics model computation. For example, a real-time, biomechanical solver is provided, allowing interaction with the soft tissue model, while still getting accurate results. The accuracy allows for generating images of a soft tissue with greater accuracy and/or the benefit of user interaction in real-time.

Abstract: A method and system for personalized non-invasive assessment of renal artery stenosis for a patient is disclosed. Medical image data of a patient is received. Patient-specific renal arterial geometry of the patient is extracted from the medical image data. Features are extracted from the patient-specific renal arterial geometry of the patient. A hemodynamic index is computed for one or more locations of interest in the patient-specific renal arterial geometry based on the extracted features using a trained machine-learning based surrogate model. The machine-learning based surrogate model is trained based on features extracted from synthetically generated renal arterial geometries.

Abstract: A first sequence of cardiac image frames are received by a first neural network of the neural network system. The first neural network outputs a first set of feature values. The first set of feature values includes a plurality of data subsets, each corresponding to a respective image frame and relating to spatial features of the respective image frame. The first set of feature values are received at a second neural network of the neural network system. The second neural network outputs a second set of feature values relating to temporal features of the spatial features. Based on the second set of feature values, a cardiac phase value relating to a cardiac phase associated with a first image frame is determined.

Abstract: The uncertainty, sensitivity, and/or standard deviation for a patient-specific hemodynamic quantification is determined. The contribution of different information, such as the fit of the geometry at different locations, to the uncertainty or sensitivity is determined. Alternatively or additionally, the amount of contribution of information at one location (e.g., geometric fit at the one location) to uncertainty or sensitivity at other locations is determined. Rather than relying on time consuming statistical analysis for each patient, a machine-learnt classifier is trained to determine the uncertainty, sensitivity, and/or standard deviation for the patient.

Abstract: A processor acquires image data from a medical imaging system. The processor generates a first model from the image data. The processor generates a computational model which includes cardiac electrophysiology and cardiac mechanics estimated from the first model. The processor performs tests on the computational model to determine outcomes for therapies. The processor overlays the outcome on an interventional image. Using interventional imaging, the first heart model may be updated/overlaid during the therapy to visualize its effect on a patient's heart.

Abstract: Methods and systems for estimating patient-specific cardiac electrical properties from medical image data and non-invasive electrocardiography measurements of a patient are disclosed. A patient-specific anatomical heart model is generated from medical image data of a patient. Patient-specific cardiac electrical properties are estimated by simulating cardiac electrophysiology over time in the patient-specific anatomical heart model using a computational cardiac electrophysiology model and adjusting cardiac electrical parameters based on the simulation results and the non-invasive electrocardiography measurements. A patient-specific cardiac electrophysiology model with the patient-specific cardiac electrical parameters can then be used to perform virtual cardiac electrophysiology interventions for planning and guidance of cardiac electrophysiology interventions.

Abstract: A computer-implemented method for providing a personalized evaluation of assessment of atherosclerotic plaques for a patient acquiring patient data comprising non-invasive patient data, medical images of the patient, and blood biomarkers. Features of interest are extracted from the patient data and one or more machine learning models are applied to the features of interest to predict one or more measures of interest related to atherosclerotic plaque.

Abstract: A computer-implemented method for editing image processing results includes performing one or more image processing tasks on an input image using an iterative editing process. The iterative editing process is executed until receiving a user exit request. Each iteration of the iterative editing process comprises using a first machine learning model to generate a plurality of processed images. Each processed image corresponds to a distinct set of processing parameters. The iterative editing process further comprises presenting the plurality of processed images to a user on a display and receiving a user response comprising (i) an indication of acceptance of one or more of the processed images, (ii) an indication of rejection of all of the processed images, or (iii) the user exit request. Following the iterative editing process clinical tasks are performed using at least one of the processed images generated immediately prior to receiving the user exit request.

Abstract: A method and system for fast non-invasive computer-based computation of a hemodynamic index, such as fractional flow reserve (FFR) from medical image data of a patient is disclosed. A patient-specific anatomical model of one or more arteries of a patient is automatically generated based on medical image data of the patient. Regions in the automatically generated patient-specific anatomical model for which user feedback is required for accurate computation of a hemodynamic index are predicted using one or more trained machine learning models.