A method for deriving one or more hemodynamic parameters based on blood-velocity and arterial diameter measures, each sampled recurrently or continuously over a time period to obtain for each a data series spanning a time window (i.e. a waveform). Additionally, a radial blood velocity profile is computed indicative of blood velocity as a function of radial position across a plane cut perpendicularly across the vessel lumen. This gives an indication of how blood velocity varies for the individual patient across the vessel diameter. This information supplements the standard blood-velocity and arterial diameter measures as inputs to a transfer function which maps the inputs to hemodynamic parameters.

For cardiac flow detection in echocardiography, by detecting one or more valves, sampling planes or flow regions spaced from the valve and/or based on multiple valves are identified. A confidence of the detection may be used to indicate confidence of calculated quantities and/or to place the sampling planes.

A dynamic analysis apparatus performs dynamic analysis on a dynamic image obtained by irradiating a subject with radiation and includes a hardware processor. The hardware processor generates a dynamic analysis image for measuring a blood flow, based on signal values of pixels constituting the dynamic image. When a pixel of the dynamic analysis image has a signal value less than or equal to a first threshold, the hardware processor detects the pixel as a signal decrease region that indicates a decrease in a blood flow value. The hardware processor compares a quantitative value of the detected signal decrease region or a quantitative value of a non-signal decrease region with a second threshold, the non-signal decrease region being other than the signal decrease region, and based on the comparison, determines whether a blood flow defect is present.

A computer-implemented training data preparation method comprises: receiving an input medical image of vessels of a patient; determining a vessel segmentation from the input medical image; identifying and annotating anatomical landmarks in the vessel segmentation to produce an annotated vessel segmentation; and storing the annotated vessel segmentation as training data. A training method for training neural networks based on the training data and a medical diagnostic method applying trained AI models are also provided.

A computer-implemented method for determining a vascular function of a perfusion imaging sequence, includes the steps of: (i) receiving a perfusion imaging sequence including a voxel time series for a plurality of voxels; (ii) applying a trained classifier on the perfusion imaging sequence for receiving voxel-wise weights; (iii) receiving voxel-wise weights from the classifier; and (iv) determining the vascular function as the weighted sum of the voxel time series; wherein the classifier is trained by optimizing over the similarity between a predicted vascular function and a ground truth vascular function using a set of examples.

A medical image processing apparatus according to an embodiment includes processing circuitry. The processing circuitry is configured to obtain a medical image. The processing circuitry is configured to calculate a first blood flow direction on the basis of a structure of a region of interest rendered in the medical image. The processing circuitry is configured to calculate a second blood flow direction on the basis of a structure in the surroundings of the region of interest. The processing circuitry is configured to identify a condition of the region of interest, on the basis of the first blood flow direction and the second blood flow direction.

Disclosed are approaches to non-invasively characterize a tumor or other lesion in a region of interest (ROI) based on various analyses of magnetic resonance imaging (MRI) data. The MRI data may correspond to ultrafast dynamic contrast enhanced MRI (DCE-MRI) and high spatial resolution DCE-MRI scans, and diffusion-weighted MRI (DW-MRI) scans of the ROI. Vasculature metrics may be determined, and tumor-associated blood flow velocity and/or tumor interstitial pressure may be obtained using the vasculature metrics as inputs to a computational fluid dynamics model. A combination of morphological vascular metrics and functional vascular metrics may be used to characterize the tumor. Malignancy, aggressiveness, treatment response, and other features of tumors or other lesions, in the breast or other regions of a patient, may be characterized through disclosed analyses of MRI data.

A system and method to evaluate a patient's brain condition by looking at venous outflow. The system and method may be computerized and automated. The process of the method identifies a paired set of venous structures to analyze, and then selects and identifies a mirrored pair of regions of interest on the structures and calculates the Hounsfield units for each of these mirrored pair of regions of interest of the paired venous structures. The process then calculates a ratio of the Hounsfield units of the mirrored pair of regions of interest. The process uses the calculated ratio to provide the clinician information on the condition on the brain tissue of the patient and to assess whether a large vessel occlusion actually exists.

A computer implemented method for generating a 3D printable model of a patient specific anatomic feature from 2D medical images is provided. A 3D image is automatically generated from a set of 2D medical images. A machine learning based image segmentation technique is used to segment the generated 3D image. A 3D printable model of the patient specific anatomic feature is created from the segmented 3D image.

In an automated detection system for acute ischemic stroke, a preprocessor performs registration on a whole-brain image and a standard-brain spatial template to extract individual brain region masks from the whole-brain image. A deep learning encoder performs feature extraction on the whole-brain image and the individual brain region masks, thereby converting the whole-brain image into 2D whole-brain slice images. A first processor maps the individual brain masks onto the whole-brain slice images for registration, thereby generating sets of brain region slice images. A second processor computes the stroke-related weight values of the slice images of each of the sets of brain region slice images and sums the weight values to obtain the characteristic value of each brain region. A disparity-aware classifier determines whether any brain region has acute ischemic stroke according to the characteristic value of each brain region.