119 results about "3d segmentation" patented technology

Disclosed are a system, apparatus, and method for 3D object segmentation within an environment. Image frames are obtained from one or more depth cameras or at different times and planar segments are extracted from data obtained from the image frames. Candidate segments that comprise a non-planar object surface are identified from the extracted planar segments. In one aspect, certain extracted planar segments are identified as comprising a non-planar object surface, and are referred to as candidate segments. Confidence of preexisting candidate segments are adjusted in response to determining correspondence with a candidate segment. In one aspect, one or more preexisting candidate segments are determined to comprise a surface of a preexisting non-planar object hypothesis. Confidence in the non-planar object hypothesis is updated in response to determining correspondence with a candidate segment.

An apparatus for generating 3D geographical models includes a processor and memory storing executable computer program code causing the apparatus to at least perform operations including removing points of a cloud depicting vertical structures in meshes of triangles detected in an area corresponding to a set of 3D points responsive to identifying triangles on vertical structures. The triangles include vertices corresponding to geocoordinates. The program code further causes the apparatus to interpolate non-vertical structures of triangles to generate a dense cloud of points. The program code further causes the apparatus to downward project delineated points of segmented rooftops to closest points of a ground to generate 3D polygonal models depicting geographical objects. The program code further causes the apparatus to generate texturized objects responsive to mapping triangle locations to the 3D polygonal models and assign texture to vertical structures and rooftops. Corresponding methods and computer program products are also provided.

A method and system for analyzing tubular structures, such as vascular bodies, with respect to their geometrical properties and mechanical loading conditions is disclosed. To this end, geometrical and structural models of vascular bodies are generated from standard sets of image data. The method or system works automatically and the tubular structure is analyzed within clinical relevant times by users without engineering background. Most critical in that sense is the integration of novel volume meshing and 3D segmentation techniques. The derived geometrical and structural models distinguish between structural relevant types of tissue, e.g., for abdominal aortic aneurysms the vessel wall and the intra-luminal thrombus are considered separately. The structural investigation of the vascular body is based on a detailed nonlinear Finite Element analysis. Here, the derived geometrical model, in-vivo boundary / loading conditions and finite deformation constitutive descriptions of the vascular tissues render the structural biomechanical problem. Different visualization concepts are provided and allow an efficient and detailed investigation of the derived geometrical and mechanical data. In addition, this information is pooled and statistical properties, derived from it, can be used to analyze vascular bodies of interest.

Disclosed are a system, apparatus, and method for 3D object segmentation within an environment. Image frames are obtained from one or more depth cameras or at different times and planar segments are extracted from data obtained from the image frames. Candidate segments that comprise a non-planar object surface are identified from the extracted planar segments. In one aspect, certain extracted planar segments are identified as comprising a non-planar object surface, and are referred to as candidate segments. Confidence of preexisting candidate segments are adjusted in response to determining correspondence with a candidate segment. In one aspect, one or more preexisting candidate segments are determined to comprise a surface of a preexisting non-planar object hypothesis. Confidence in the non-planar object hypothesis is updated in response to determining correspondence with a candidate segment.

A method for navigating a three-dimensional (3D) image includes accessing a 3D image dataset, generating a 3D mesh corresponding to a 3D segmentation result using the 3D image dataset, displaying a 3D surface rendering of the 3D surface mesh, and navigating the 3D image based on a manual input received from a user indicated on the rendered 3D mesh.

The present invention refers to an angiographic image acquisition system and method which can beneficially be used in the scope of minimally invasive image-guided interventions. In particular, the present invention relates to a system and method for graphically visualizing a pre-interventionally virtual 3D representation of a patient's coronary artery tree's vessel segments in a region of interest of a patient's cardiovascular system to be three-dimensionally reconstructed. Optionally, this 3D representation can then be fused with an intraoperatively acquired fluoroscopic 2D live image of an interventional tool. According to the present invention, said method comprises the steps of subjecting the image data set of the 3D representation associated with the precalculated optimal viewing angle to a 3D segmentation algorithm (S4) in order to find the contours of a target structure or lesion to be examined and interventionally treated within a region of interest and automatically adjusting (S5) a collimator wedge position and / or aperture of a shutter mechanism used for collimating an X-ray beam emitted by an X-ray source of a C-arm-based 3D rotational angiography device or rotational gantry-based CT imaging system to which the patient is exposed during an image-guided radiographic examination procedure based on data obtained as a result of said segmentation which indicate the contour and size of said target structure or lesion. The aim is to reduce the region of interest to a field of view that covers said target structure or lesion together with a user-definable portion of the surrounding vasculature.

The invention provides a surgical puncture path intelligent automatic planning method and system based on machine learning and a medical system. The surgical puncture path intelligent automatic planning method and system based on machine learning and the medical system can rapidly determine a surgical puncture path and a needle inserting point position and can allow a brain stereotaxic apparatus or a medical mechanical arm to achieve automatic puncture surgery operation. The planning method comprises the following steps that (1) sample image data of an existing case are acquired, and trainingdata and test data are made; (2) a three-dimensional segmentation deep neural network model is designed, and the three-dimensional segmentation deep neural network model is trained; (3) the sample image data of a patient are segmented and identified by using the trained three-dimensional segmentation deep neural network model; (4) a three-dimensional model is constructed based on a segmentation identification result and the sample image data; (5) a safe needle insertion constraint area is determined based on the target point position and medical prior information; and (6), in the safe needle insertion constraint area, a three-dimensional space trajectory planning algorithm is utilized to complete surgical puncture path planning.

A method and apparatus for automated reconstruction of a 3D computed tomography (CT) volume from a small number of X-ray images is disclosed. A sparse 3D volume is generated from a small number of x-ray images using a tomographic reconstruction algorithm. A final reconstructed 3D CT volume is generated from the sparse 3D volume using a trained deep neural network. A 3D segmentation mask can also be generated from the sparse 3D volume using the trained deep neural network.

The disclosure relates to a method of automatically producing a three-dimensional (3D) segmentation of a heart chamber, the method comprising: obtaining data sets from cardiac magnetic resonance imaging (MRI) or ultrasound, generating a 3D segmentation of the heart chamber from the data sets using an active contour method, modifying the 3D segmentation by adding a plurality of intra-chamber structures; and identifying an enclosing myocardium using the 3D segmentation generated by the method.

The embodiment of the invention discloses a hip joint segmentation method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring each sectional image of a hip joint; inputting each tomographic image into a target hip joint segmentation model to generate a two-dimensional segmentation result corresponding to each tomographic image, the target hipjoint segmentation model being obtained by pre-training based on a convolutional neural network model; and generating a target three-dimensional segmentation result according to each two-dimensional segmentation result, and taking the target three-dimensional segmentation result as the hip joint segmentation result. According to the technical scheme, the hip joint can be segmented quickly and precisely, and the effects of considering the precision and efficiency of hip joint segmentation and meeting clinical requirements to a greater extent are achieved.

A method is provided for improving a segmentation of a 3D image and / or validating a segmentation of a 3D image includes rendering an acquired 3D image and a segmentation of the acquired 3D image on a segmentation display that has at least one spatially fixed slice and an interactive slice with a reference mark corresponding to the cursor location in the spatially fixed slice or slices on the display. The method further includes utilizing an interactive user input to update image data of the interactive slice and the reference mark to coincide with the cursor in the spatially fixed slice or slices. The method further includes using the cursor and the reference mark to verify that cursor locations on the boundaries of the segmentation of the acquired 3D image correspond to object boundaries in the image data of the interactive slice.

An apparatus and a method for post processing 2D image slices (110a-c) defining a 3D image volume. The apparatus comprises a graphical user interface controller (160), a 2D segmenter (170) and a 3D segmenter (180). The apparatus allows a user to effect calculation and display of a 2D segmentation of a cross section of an object shown in a slice (110a) and calculation and display of a 3D segmentation of the object across the 3D image volume, the 3D segmentation based on the object's previously calculated 2D segmentation.

An image processing system comprising 3D image data processing means (10) of automatic mapping a 3-D Surface Model onto the surface of an object of interest in a 3-D image, for estimating a model-based 3-D segmentation surface, comprising visualizing means (60) and further comprising means of interactive adaptation (20) of the segmentation surface to the actual surface of the object of interest including means of interactive selection (40) of a 2D data plane (DP) that intersects the 3-D segmentation surface along a 2-D Model Curve (MC), said Data Plane having a user-selected orientation with respect to said surface, which is appropriate for the user to visualize a 2-D portion called Aberrant Curve (AC)of said Model Curve to be modified; means of interactive definition of a Guiding Curve (GC) in the 2-D Data Plane; means of interactive adaptation of said Aberrant Curve (AC) to said Guiding Curve (GC); and means of further automatically adapting the 3D segmentation surface within a neighborhood of the interactively adapted Aberrant Curve. The surface Model is favorably a Mesh Model.

A method for three-dimensional segmentation of a target in multislice images of volumetric data includes determining a center and a spread of the target by a parametric fitting of the volumetric data, and determining a three-dimensional volume by non-parametric segmentation of the volumetric data iteratively refining the center and spread of the target in the volumetric data.

A system and method for using learned discriminative models to segment a border of an anatomical structure in a three dimensional (3D) image is disclosed. A discriminative probability model is computed for each voxel in the 3D image. Thresholding is performed on each discriminative probability model. One or more two dimensional (2D) slices of the thresholded 3D image along X-Y planes are obtained. Seed regions are selected in the 2D slices. Morphological region growing is performed on the selected seed regions. An initial 3D segmentation is obtained. Boundary evolution is performed on the initial 3D segmentation. The segmented anatomical structure is removed. in the original 3D image.

A general purpose method to segment any kind of lesions in 3D images is provided. Based on a click or a stroke inside the lesion from the user, a distribution of intensity level properties is learned. The random walker segmentation method combines multiple 2D segmentation results to produce the final 3D segmentation of the lesion.

A method for three dimensional (3D) segmentation of an object is provided. The method obtains a volumetric data set containing object data and non-object data in proximity to the object data, the object data having a reference axis extending through the object data. The method defines at least one reference slice and multiple object slices within the volumetric data set, the reference slice and the object slices intersecting one another along the reference axis and containing the reference axis. Reference points are determined within the reference slice at edges of the object data. With reference points determined, the method generates an estimated contour extended through the reference and object slices based on the reference points, the estimated contour intersecting the object slices to define estimated contour points. The method then adjusts the estimated contour points until corresponding substantially to actual contour points of the object data.

A system and method for using learned discriminative models to segment a border of an anatomical structure in a three dimensional (3D) image is disclosed. A discriminative probability model is computed for each voxel in the 3D image. Thresholding is performed on each discriminative probability model. One or more two dimensional (2D) slices of the thresholded 3D image along X-Y planes are obtained. Seed regions are selected in the 2D slices. Morphological region growing is performed on the selected seed regions. An initial 3D segmentation is obtained. Boundary evolution is performed on the initial 3D segmentation. The segmented anatomical structure is removed. in the original 3D image.

The embodiment of the invention provides a CT image lung and trachea segmentation method and system based on deep learning, and the method employs a 2D UNet and 3D Unet deep learning network model atthe same time, and comprises the following steps: S1, preprocessing; s2, carrying out two-dimensional resampling; s3, carrying out two-dimensional segmentation; s4, performing three-dimensional sampling; s5, carrying out three-dimensional segmentation; and S6, fusion: carrying out combination operation on the two-dimensional and three-dimensional segmentation results of the lung trachea to obtaina fused segmented lung trachea, and then taking the maximum three-dimensional connected region in the calculation image as a final lung trachea segmentation result. According to the CT image lung trachea segmentation method and system based on deep learning, a 2D UNet network and a 3D Unet network are used at the same time, so that the trachea segmentation result is better, and the segmentation method is more robust. In the 3D UNet network training process, the training efficiency and precision of the network can be improved by a tracheal skeleton point sampling method.

The invention belongs to the technical field of image segmentation, and specifically relates to a 3D full-automatic glioma segmentation method combining a Volume of Interest and a GrowCut algorithm. The method first expands a Bounding box algorithm to 3D, utilizes the algorithm to extract a Volume of Interest (VOI) containing glioma, then utilizes a reflection symmetry algorithm to estimate the VOI and overcomes the defects of Bounding box when the glioma is detected to cross over a median sagittal plane, and finally marks pixel points in an image based on the accurate VOI, and thus a semi-automatic 2D GrowCut algorithm is optimized to a full-automatic 3D segmentation method. While accurately segmenting the glioma, the method is more rapid in theory and reality compared with the 2D segmentation algorithm of the same principle, and is better in convenience and feasibility compared with a manual segmentation method. As an image segmentation method, the 3D full-automatic glioma segmentation method provided by the invention can serve as a powerful auxiliary tool for clinical diagnosis of glioma.

A method for navigating a three-dimensional (3D) image includes accessing a 3D image dataset, generating a 3D mesh corresponding to a 3D segmentation result using the 3D image dataset, displaying a 3D surface rendering of the 3D image intensities on the 3D mesh, and navigating the 3D image based on a manual input received from a user indicated on the rendered 3D image.

The invention is particularly suitable to restitution of complex deformity of skull surface. Based on computer image processing and CAD, emulation of 3D segmentation and reconstruction is carried out for structure of skull surface matched to real body. Qualitative even accurate quantitative analyses are carried out for shape and structure of injured skull (including eye socket). The invention discloses layered osteotomy method of splitting off skull socket and surface along natural curvature in middle. The method includes following steps: in operation, fixing moved position promptly; embedding bone tractor to move position gradually; reconstructing skull surface such as inverted valve with vessel pedicle behind the ear. The invention reconstructs fine figure of face to recover normal function.

An apparatus and method for performing automatic 3D image segmentation and reconstruction of organ structures, which is particularly well-suited for use on cortical surfaces is presented. A brain extraction process removes non-brain image elements, then classifies brain tissue as to type in preparation for a cerebrum segmentation process that determines which portions of the image information belong to specific physiological structures. Ventricle filling is performed on the image data based on information from a ventricle extraction process. A reconstruction process follows in which specific surfaces, such as white matter (WM) and grey matter (GM), are reconstructed.

Presented herein are systems and methods that provide for improved 3D segmentation of nuclear medicine images using an artificial intelligence-based deep learning approach. For example, in certain embodiments, the machine learning module receives both an anatomical image (e.g., a CT image) and a functional image (e.g., a PET or SPECT image) as input, and generates, as output, a segmentation mask that identifies one or more particular target tissue regions of interest. The two images are interpreted by the machine learning module as separate channels representative of the same volume. Following segmentation, additional analysis can be performed (e.g., hotspot detection / risk assessment within the identified region of interest).

A robotic system includes a robot having an associated workspace; a vision sensor constructed to obtain a 3D image of a robot scene including a workpiece located in the workspace; and a control system communicatively coupled to the vision sensor and to the robot. The control system is configured to execute program instructions to filter the image by segmenting the image into a first image portion containing substantially only a region of interest within the robot scene, and a second image portion containing the balance of the robot scene outside the region of interest; and by storing image data associated with the first image portion. The control system is operative to control movement of the robot to perform work, on the workpiece based on the image data associated with the first image portion.

A method for segmenting a tubular structure includes providing a three-dimensional image containing the tubular structure, providing at least one seed point within the tubular structure, fitting an initial cylinder into the tubular structure at the seed point, adding cylinder segments to the initial cylinder in forward and backward directions within the three-dimensional image of the tubular structure by tracking the cylinder model over the tubular structure, modeling each cylinder segment using parameters to account for a three-dimensional orientation, a radius, a length and a curvature, describing an intensity image of the tubular structure using a second model to represent an edge and intensity distribution of the tubular structure, fitting the parameters as the cylinder segment is added to the intensity image, and outputting cylinder segments as a segmentation of the tube.

The invention discloses a 3D vertebra CT image active contour segmentation method fusing weighted random forest, and relates to the field of medical image processing. For the problem of sensitivity ofa vertebra CT image segmentation method to an initial contour, a method for automatically locating a vertebra and segmenting a vertebra CT image is proposed. The method comprises the steps of firstly, proposing a weighted random regression and classification forest algorithm to determine a vertebra center; secondly, putting an initial contour ball of active contour segmentation in the vertebra center, and segmenting out the vertebra in the image by adopting a 3D active contour segmentation method in combination with an energy function; and finally, performing combination output on trained models to obtain a complete vertebra CT image segmentation model. A spinal CT segmentation model proposed in the method can automatically locate the vertebra center and can perform automatic 3D segmentation on the vertebra, so that the spinal CT image segmentation steps and processes are simplified.

A seed point is selected inside a structure that is to be segmented in image data. An adaptive model is defined around the seed point, and a preprocessing filter is applied only within the bounding region. A presegmentation of the preprocessed result is performed, and the bounding region is expanded if necessary to accommodate the presegmentation result. An adaptive model for post-processing may be used. The model is translated, rotated and scaled to find a best fit with the pre-segmented data. Additional models can be grown based on testing performed on points on a perimeter of previous models to improve the segmentation of an object.