The present disclosure relates to training one or more neural networks for vascular vessel assessment using synthetic image data for which ground-truth data is known. In certain implementations, the synthetic image data may be based in part, or derived from, clinical image data for which ground-truth data is not known or available. Neural networks trained in this manner may be used to perform one or more of vessel segmentation, decalcification, Hounsfield unit scoring, and/or estimation of a hemodynamic parameter.

A medical image processing apparatus of an embodiment includes a processing circuitry. The processing circuitry extracts a plurality of valve leaflets of a heart valve from image data of a subject. The processing circuitry measures, with respect to at least one valve leaflet of the valve leaflets, a length of a region at which the valve leaflet is in contact with another valve leaflet, in a predetermined reference direction. The processing circuitry controls a display to display a distribution of the length at each of a plurality of positions on the valve leaflet.

A system and method are provided for determining vascular velocity using non-invasively acquired medical images. The method includes reconstructing CT angiography (CTA) data into a plurality of images of the subject by producing a composite image using the CTA data corresponding to a set of the plurality of view angles, backprojecting each view angle in the CTA data and weighting a value backprojected into at image pixel by an attenuation value of a corresponding pixel in the composite image, and summing backprojected values for each image pixel to produce a CT image of the subject. The method also includes determining a flow direction or a velocity of flow within a vessel, calculating, using the flow direction or velocity, a pressure in the vessel, and generating a quantitative map of the subject indicating the flow direction, velocity, or pressure in the vessel against an image of the subject including the vessel.

Method for functional imaging of the brain, comprising the following steps: (a) a brain is imaged by ultrasound imaging in order to obtain a vascular image to be studied (IVO), (b) the vascular image to be studied (IVO) is compared automatically, by shape recognition, with a cerebral vascular atlas (AV), and the vascular image to be studied (IVO) is thus located in the cerebral vascular atlas (AV), (c) a cerebral functional atlas (AF) corresponding to said cerebral vascular atlas (AV) and comprising cerebral functional zones (1c) located in this cerebral vascular atlas (AV) is used in such a way as to identify cerebral functional zones (1e) on the vascular image to be studied (IVO).

The brain appears to have organized cardiac frequency angiographic phenomena with such coherence as to qualify as vascular pulse waves. Separate arterial and venous vascular pulse waves may be resolved. This disclosure states the method of extracting a spatiotemporal reconstruction of the cardiac frequency phenomena present in an angiogram obtained at faster than cardiac frequency. A wavelet transform is applied to each of the pixel-wise time signals of the angiogram. If there is motion alias then instead a high frequency resolution wavelet transform of the overall angiographic time intensity curve is cross-correlated to high temporal resolution wavelet transforms of the pixel-wise time signals. The result is filtered for cardiac wavelet scale then pixel-wise inverse wavelet transformed. This gives a complex-valued spatiotemporal grid of cardiac frequency angiographic phenomena. It may be rendered with a brightness-hue color model or subjected to further analysis.

Image rotation in an endoscopic hyperspectral imaging system is described. A system includes an emitter for emitting pulses of electromagnetic radiation and an image sensor comprising a pixel array for sensing reflected electromagnetic radiation. The system includes a rotation sensor for detecting an angle of rotation of a lumen relative to a handpiece of an endoscope. The system is such that at least a portion of the pulses of electromagnetic radiation emitted by the emitter comprises electromagnetic radiation having a wavelength from about 513 nm to about 545 nm, from about 565 nm to about 585 nm, or from about 900 nm to about 1000 nm.

Data is computed in order to visualize the velocity of blood fluid flowing through a blood vessel at a fundus. An image processing method includes a step of performing registration of each frame in a moving image configured by plural frames of an imaged fundus, and a step of computing visualization data enabling visualization of a state of blood fluid flowing through a blood vessel at the fundus based on a pixel value in each of the registered frames.

Systems and methods for producing velocity data associated with three-dimensional (3D) flow field images. In some embodiments, the method includes receiving data associated with a plurality of frames of a flow field relating to image data acquired by a medical imaging device, in which the data includes information corresponding to measurements of the flow field over time within a chamber; performing, for each of the plurality of frames, the following operations including: generating, for a respective frame, a data correction based on an interaction of the flow field with the chamber, applying the data correction to a velocity field corresponding to the respective frame, and imposing an incompressibility constraint for the flow field on one or more data points of the respective frame; and generating, subsequent to imposing the incompressibility constraint, a plurality of corrected velocity fields each of which corresponds to one of the plurality of frames.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for jointly training an encoder neural network and a decoder neural network. In one aspect, a method comprises: updating current values of a set of encoder parameters and current values of a set of decoder parameters using gradients of a reconstruction loss function that measures an error in a reconstruction of multi-modal data from a training example, wherein: the reconstruction loss function comprises a plurality of scaling factors that each scale a respective term in the reconstruction loss function that measures an error in the reconstruction of a corresponding proper subset of feature dimensions of the multi-modal data from the training example.

An ultrasound system (100) includes an ultrasound transducer, a processing circuit (210, 300), and a display. The ultrasound transducer is configured to detect ultrasound information regarding a patient and output the ultrasound information as an ultrasound data sample. The processing circuit (210, 300) is configured to segment the ultrasound data sample into a binary image including at least one first region and at least one second region, obtain a first location of a first vascular feature of the binary image based on a boundary between the at least one first region and the at least one second region, and modify the binary image based on the first location of the first vascular feature. The first vascular feature is associated with an intima media thickness.The display is configured to display the modified image.