Systems and method can be provided to transform input data (e.g., CT imaging data) into structured representations to create interpretable models. Another aspect of the current invention can be generating labels synthetically to apply to real data according to a biologically-based labelling technique to guide the model training with a priori mechanistic knowledge.

There is provided a method for determining a regional rupture potential (RRP) indicative of the state of local weakening of a blood vessel based on parameters that correlate with the expansion and local weakening of the vessel. The method comprises: receiving a plurality of images of the blood vessel into a multiphase stack. A geometrical model of the lumen and the outer wall of the vessel are generated and smoothed to obtain a volume mesh and surface mesh respectively. An ILT thickness distribution, a local deformation at each phase and a wall strain distribution indicative of a maximal principal strain at the outer wall are determined. Blood flow values in the lumen are obtained and a wall shear stress distribution indicative of wall shear disturbances in the lumen is calculated. The RRP is determined based on the ILT thickness distribution, the wall shear stress, and the wall strain.

Systems and methods for automatic processing of input medical images are provided. A set of input medical images acquired at a plurality of locations on a patient is received. For each respective location of the plurality of locations, an image quality score is determined for each input medical image of the set of input medical images acquired at the respective location and one of the input medical images acquired at the respective location is selected based on the image quality scores. The selected input medical images are processed to correct for errors. One or more regions of interest are segmented from the processed selected input medical images. One or more hemodynamic measures are calculated from the processed selected input medical images based on the segmented one or more regions of interest. The calculated one or more hemodynamic measures are output.

A framework for registering anatomical to functional images using deep learning. In accordance with one aspect, the framework extracts features by applying an anatomical image and a corresponding functional image as input to a first trained convolutional neural network. A deformation field is estimated by applying the extracted features as input to a second trained convolutional neural network. The deformation field may then be applied to the anatomical image to generate a registered anatomical image.

There is provided a method of generating a dataset of synthetic images, comprising: for each real image each depicting a real human anatomical structure: extracting and preserving a real anatomical structure region(s) from the real image, generating a synthetic image comprising a synthetic human anatomical structure region and the preserved real anatomical structure region(s), designating pairs of images, each including the real image and the synthetic image, feeding the pair into a machine learning model trained to recognize anatomical structure parts to obtain an outcome of a similarity value denoting an amount of similarity between the real image and the synthetic image, verifying that the synthetic image does not depict the real human anatomical structure when the similarity value is below a threshold, wherein an identity of the real human anatomical structure is non-determinable from the synthetic image, and including the verified synthetic image in the dataset.

Apparatus and methods are described for use with an endoluminal device that includes one or more radiopaque portions and that moves through a lumen of a subject. A sequence of radiographic images of a portion of the subject's body, in which the lumen is disposed, is acquired, during movement of the endoluminal device through the lumen. Locations at which the one or more radiopaque portions of the endoluminal device were imaged during the movement of the endoluminal device through the lumen are identified, by analyzing the sequence of radiographic images. A set of locations at which the one or more radiopaque portions were disposed during the movement of the endoluminal device through the lumen is defined, and an endoluminal path of the device through the lumen is estimated based upon the set of locations. Other applications are also described.

Systems and methods for image processing are provided. The systems may obtain a blood vessel image of a target subject. The systems may generate, based on the blood vessel image, a point cloud including a plurality of data points representing a plurality of blood vessel points of the target subject. Each of the plurality of data points may include values of one or more reference features of the corresponding blood vessel point. The systems may further for each of the plurality of blood vessel points, determine values of one or more target features of the blood vessel point based on the point cloud using a determination model. The determination model may be a trained deep learning model.

A method of performing 3D vessel tree reconstruction includes providing segmented binary angiography images, applying a distance transform to the images, and generating distance transformed binary angiography images. The set of distance transformed binary angiography images are provided to a trained 3D vessel reconstruction machine learning model capable of reconstructing 3D vessels. The 3D vessel tree reconstruction machine learning model includes a multi-stage convolutional neural network comprising a multi-stage architecture with (i) a vessel centerline stage, and (ii) a radius reconstruction stage. Resultant 3D reconstructed vessel trees may be used in performing clinical assessment of coronary vessel health, and occlusion.

Various embodiments described herein relate to systems, devices, and methods for non-invasive image-based plaque analysis and risk determination. In particular, in some embodiments, the systems, devices, and methods described herein are related to analysis of one or more regions of plaque, such as for example coronary plaque, using non-invasively obtained images that can be analyzed using computer vision or machine learning to identify, diagnose, characterize, treat and/or track coronary artery disease.

The invention provides a technique capable of effectively and appropriately removing noise from various kinds of images including noise and artifacts and images in which a noise pattern changes due to a difference in imaging conditions. Based on a noise removal technique using AI, noise characteristics including artifacts are analyzed for each image, the image is classified based on an analysis result, an optimal neural network for a noise processing is applied for each classification, and the noise and the artifacts are reduced.