An ultrasound imaging system computes real time physiological parameters from measurements of anatomical features in ultrasound image data using a neural network to identify the location of the anatomical features. In one embodiment, cardiac parameters are computed from endocardial wall tracings in M-mode ultrasound image data that are identified by the neural network.

The disclosed subject matter related to methods and apparatus for determining brain swelling and brain shifting in a patient as well as predicting a possible resultant increase in intracranial pressure in the patient. The apparatus can include a transducer such as an ultrasound transducer communicatively connected to a controller via wires or via wireless communications device(s). A monitor and/or alarm device can be provided to notify a practitioner when the controller has determined brain swelling is occurring and/or when an imminent increase in intracranial pressure is likely to occur.

Circuitless heart cycle determination includes capturing a video clip of one or more image frames of a target heart muscle through an ultrasound imaging device and submitting the frames to a classifier that has been trained with an annotated set of images, each of a corresponding heart muscle captured at a specified phase of a heart cycle with a ground truth indication of the specified phase of the heart cycle drawn from a separately recorded cycle graph of an electrical signal measured over time for the corresponding heart muscle. In response to the submission, a classification is received of different portions of the submitted frames according to corresponding phases of the heart cycle. Finally, a contemporaneous phase of the heart cycle is determined in the device for the target heart muscle without sensing electrical signals by way of a closed-loop sensor circuit affixed proximately to the target heart muscle.

Periodic displacement occurs in body tissue due to heartbeat. A peak level D of the movement distance of the body tissue is detected (Step 21), and a heartbeat cycle T is calculated from a frequency spectrum (Steps 22 and 23). By dividing twice the peak level D by the heartbeat cycle T, the moving velocity of the body tissue in a unit heartbeat cycle is calculated (Step 24). By dividing the moving velocity by a frame rate r, an average movement distance of the body tissue between frames is calculated (Step 25). In a case where the average movement distance is smaller than a predetermined threshold value, a time interval between the frames used for the calculation of the movement distance is extended (being Step 26 NO, Step 27).

A Doppler waveform generation unit 30 obtains Doppler information from a reception signal collected from a diagnosis region and generates a Doppler waveform. An initial time-phase setting unit 40 sets a beginning initial time-phase and an ending initial time-phase of the Doppler waveform. In the setting, an electrocardiographic waveform signal obtained from a subject using an electrocardiograph or the like and learned data stored in a learned data storage unit 60 are used. A measurement time-phase search unit 50 searches for a beginning time-phase of the Doppler waveform near the beginning initial time-phase, and searches for an ending time-phase of the Doppler waveform near the ending initial time-phase. In the search process, the learned data stored in the learned data storage unit 60 is used.

A method of processing cardiac ultrasound data for determining information about a mechanical wave in the heart. The method comprises receiving data representative of a time series of three-dimensional data frames, generated from ultrasound signals from a human or animal heart, each frame comprising a set of voxels, each voxel value representing an acceleration component of a respective location in the heart at a common time. The method also comprises identifying, for each voxel, a frame of the series in which the voxel value is at a maximum. A three-dimensional time-propagation data set is generated by assigning each voxel a value representative of the time of the respective frame in the time series for which the corresponding voxel is at a maximum. The method then comprises generating data representative of a three-dimensional velocity vector field by calculating time derivatives from the three-dimensional time-propagation data set.

A method includes obtaining, by a processor, a set of ultrasound frames showing a portion of a heart of a subject, identifying a subset of the frames, responsively to the subset having been acquired at one or more predefined phases of at least one physiological cycle of the subject, computing respective image-quality scores for at least the subset of the frames, each of the scores quantifying an image quality with which one or more anatomical portions of interest are shown in a respective one of the frames, and, based on the image-quality scores, selecting, for subsequent use, at least one frame from the subset of the frames. Other embodiments are also described.

Periodic displacement occurs in body tissue due to heartbeat. A peak level D of the movement distance of the body tissue is detected (Step 21), and a heartbeat cycle T is calculated from a frequency spectrum (Steps 22 and 23). By dividing twice the peak level D by the heartbeat cycle T, the moving velocity of the body tissue in a unit heartbeat cycle is calculated (Step 24). By dividing the moving velocity by a frame rate r, an average movement distance of the body tissue between frames is calculated (Step 25). In a case where the average movement distance is smaller than a predetermined threshold value, a time interval between the frames used for the calculation of the movement distance is extended (being Step 26 NO, Step 27).

Systems, methods and devices for providing combined ultrasound, electrocardiography, and auscultation data are provided. One such system includes an ultrasound sensor, an electrocardiogram (EKG) sensor, an auscultation sensor, and a computing device. The computing device includes memory and a processor, and the processor receives signals from the ultrasound sensor, the EKG sensor, and the auscultation sensor. Artificial intelligence techniques may be employed for automatically analyzing the data obtained from the ultrasound sensor, the EKG sensor, and the auscultation sensor and producing a clinically-relevant determination based on a combined analysis of the data.

The present disclosure relates to a method for detecting and/or quantitatively assessing cardiac dyssynchrony of a subject based on at least one medical imaging scan showing at least part of the myocardium of the subject’s heart, in particular mechanical cardiac dyssynchrony. The medical imaging scan may provide a plurality of values of a predefined myocardial deformation parameter of said part of the myocardium. In a preferred embodiment the method comprises the steps of 1) determining a myocardial deformation deviation between pairs of myocardial deformation parameter values selected from the myocardium of substantially opposite parts of a cardiac chamber, and 2) calculating the cardiac dyssynchrony of the subject based on said myocardial deformation deviation.