In a magnetic resonance (MR) apparatus and segmentation method, a region in an MR image, acquired from a scan of a patient with an MR scanner of the apparatus, is provided to a computer for segmentation of the region from the overall image dataset. The segmentation takes place based on a classification of image elements of the image dataset, and the classification is iteratively re-trained in a weakly supervised learning algorithm based on examination-specific information provided to the computer.

For topogram predication from surface data, a sensor captures the outside surface of a patient. A generative adversarial network (GAN) generates the topogram representing an interior organ based on the outside surface of the patient. To further adapt to specific patients, internal landmarks are used in the topogram prediction. The topogram generated by one generator of the GAN may be altered based on landmarks generated by another generator.

Human mesh model recovery may utilize prior knowledge of the hierarchical structural correlation between different parts of a human body. Such structural correlation may be between a root kinematic chain of the human body and a head or limb kinematic chain of the human body. Shape and/or pose parameters relating to the human mesh model may be determined by first determining the parameters associated with the root kinematic chain and then using those parameters to predict the parameters associated with the head or limb kinematic chain. Such a task can be accomplished using a system comprising one or more processors and one or more storage devices storing instructions that, when executed by the one or more processors, cause the one or more processors to implement one or more neural networks trained to perform functions related to the task.

This invention is related to a method to improve the performance of a deep neural network (10) for the identification of a segmentation target (111) in a medical image (12, 110), comprising the steps of performing n training steps on said deep neural network (10) for the identification of said region of interest on two different representations (13, 14) of the same segmentation target (111), said representations (13,14) being definitions of the same segmentation target (111).

A method for detecting bone structure includes allocating at least one bone portion from a bone image composed by pixels each including luminance value relating to bone structural parameter; aligning a major axis of principal axes of moment of inertia of the bone portion to a principal axis of Cartesian coordinate system; a cortical bone area of the bone portion intersecting at least one principal plane perpendicular to the principal axis, and each principal plane forming an outer and inner contour line of the cortical bone area; processing an analytic algorithm for the bone structural parameter; calculating distributed state of the bone structural parameter in each principal plane to obtain a distributed state of the bone structural parameter of the bone portion; and obtaining a distributed state of the bone structural parameter of the bone portion by assembling distributed state of the bone structural parameter of each bone portion.

An electronic device includes a camera, a display, and a processor. The camera provides an image. The display displays a cursor. The processor executes a palm detection algorithm to identify a palm in the image and mark a bounding box around the palm. The processor also executes a hand key-point detection algorithm to mark a plurality of key points on the palm that has been marked in the image to obtain spatial coordinates of key points on the palm. The processor executes a hand motion detection algorithm to control the camera to turn in the corresponding direction, move the cursor in the display in a way that corresponds to the position change of the bounding box around the palm, and trigger an event according to the change of the spatial coordinates of at least one of the key points on the palm within a certain period of time.

A method and system for image processing is provided. The technique includes acquiring data related to the processing, performing a pre-processing of the data, performing a segmentation of a subject, performing a post-processing of the result of the segmentation and managing storage of the data. The post-process includes a calibrating and a rendering of the data.

There provides an apparatus for recognizing a head region in a CT lateral image of a subject, comprising: a deriving unit for deriving a first image representing a bone of the subject from the CT lateral image; an extracting unit for extracting a boundary curve indicating an outer contour of a region comprising at least part of the occipital bone and at least part of the cervical vertebra of the subject in the first image; and a determining unit for determining a first pixel position indicating a bottommost point of the head region of the subject, based on a shape feature parameter of the boundary curve.

A computer system implements a neural network to process raw dental images to detect and number teeth and to diagnose pathological, non-pathological, and post-treatment conditions. Detected teeth, corresponding numbers, and any corresponding detected conditions are correlated to the dental image and presented in a graphical user interface comprising the image and a standard, symbolic dental chart associating the tooth number, detected conditions, and regions of the image to teeth represented in the symbolic chart.

Disclosed are a cranial CT-based grading method and a corresponding system, which relate to the field of medical imaging. The cranial CT-based grading method as disclosed solves the problems of relatively great subjective disparities and poor operability in eye-balling ASPECTS assessment. The grading method includes: determining frames where target image slices are located from to-be-processed multi-frame cranial CT data; extracting target areas; performing infarct judgment on each target area included in the target areas to output an infarct judgment outcome regarding the target area; and outputting a grading outcome based on infarct judgment outcomes regarding all target areas. The grading method and system as disclosed may eliminate or mitigate the diagnosis disparities caused by human factors and imaging deviations due to different imaging devices, and shorten the time taken by human observation, consideration, and bared-eye grading, thereby serving as a computer-aided method to provide reference for medical studies on stoke.