Abstract: Systems and methods for determining rendering parameters based on authored snapshots or templates. In one aspect, clinically relevant snapshots of patient medical data are created by experts to support educational or clinical workflows. Alternatively, the snapshots are created by automation processes from AI-based organ and disease segmentations. In another aspect, clinically relevant templates are generated. Rendering parameters are derived from the snapshots or templates, stored, and then applied for either rendering new data or interactive viewing of existing data.

Abstract: A convolution neural network, CNN, is trained to determine illumination parameters for image rendering of medical data. For training, a set of images are rendered using a set of rendering configurations, selected from the group of: a set of different camera parameters, a set of different transfer functions for assigning optical properties, like for example color and opacity, to original values of the raw data to be rendered, and a set of different illumination parameters. An evaluation score is computed for representing an amount of image information for each of the rendered images. The computed evaluation score for each rendered image and the rendering configurations applied for rendering the image are used to train the CNN.

Abstract: Systems and methods for automatically determining an image quality assessment of a rendered medical image are provided. A rendered medical image is received. One or more measures of interest are extracted from the rendered medical image. An image quality assessment of the rendered medical image is determined using a machine learning based image quality assessment network based on the one or more measures of interest. The image quality assessment of the rendered medical image is output.

Abstract: Distance estimation is optimized in virtual or augmented reality. A distance map of a surgical instrument to a region of interest is determined, at least at the beginning and when a position of the surgical instrument has changed. A render-image is rendered based on a medical 3D image and the position of the surgical instrument, at least at the beginning and when the position of the surgical instrument has changed. At least the region of interest and those parts of the surgical instrument positioned in the volume of the render-image are shown in the render-image. Based on the distance map, at least for a predefined area of the region of interest, visible, acoustic, and/or haptic distance-information is added.

Abstract: An automatic light arrangement for medical visualization includes: providing a medical 3D image (D), providing spatial information about a region of interest (R) in this 3D-image (D) and spatial information about a virtual camera (C), determining a plurality of possible arrangements for light sources (L) by using depth information based on the 3D image (DI) together with the spatial information about the region of interest (R) in this image and the spatial information about the virtual camera (C), wherein valid arrangements are those where shadows (S) on the region of interest (R) are below a predefined threshold, and/or wherein the determination or arrangements is based on a number of predefined perceptual metrics applied specifically to the regions of interest, prioritizing the determined arrangements, and choosing the arrangement with the best prioritization.

Abstract: For three-dimensional rendering, a machine-learnt model is trained to generate representation vectors for rendered images formed with different rendering parameter settings. The distances between representation vectors of the images to a reference are used to select the rendered image and corresponding rendering parameters that provides a consistency with the reference. In an additional or different embodiment, optimized pseudo-random sequences are used for physically-based rendering. The random number generator seed is selected to improve the convergence speed of the renderer and to provide higher quality images, such as providing images more rapidly for training compared to using non-optimized seed selection.

Abstract: A method includes presenting an image in a head-mounted display (HMD) controlled by a user. The image is rendered with a transfer function that defines one or more presentation characteristics of voxels in the image. Eye sensors in the HMD are used to determine a region of interest being viewed by the user for longer than a predetermined amount of time. The region of interest comprises a subset of voxels included in the image. The transfer function is automatically adjusted to modify one or more of the presentation characteristics of at least the subset of voxels in response to the region of interest. The presentation of the image in the HMD is modified to present the subset of voxels with modified presentation characteristics.

Abstract: A method includes presenting an image in a head-mounted display (HMD) controlled by a user. The image is rendered with a transfer function that defines one or more presentation characteristics of voxels in the image. Eye sensors in the HMD are used to determine a region of interest being viewed by the user for longer than a predetermined amount of time. The region of interest comprises a subset of voxels included in the image. The transfer function is automatically adjusted to modify one or more of the presentation characteristics of at least the subset of voxels in response to the region of interest. The presentation of the image in the HMD is modified to present the subset of voxels with modified presentation characteristics.

Abstract: For three-dimensional rendering, a machine-learnt model is trained to generate representation vectors for rendered images formed with different rendering parameter settings. The distances between representation vectors of the images to a reference are used to select the rendered image and corresponding rendering parameters that provides a consistency with the reference. In an additional or different embodiment, optimized pseudo-random sequences are used for physically-based rendering. The random number generator seed is selected to improve the convergence speed of the renderer and to provide higher quality images, such as providing images more rapidly for training compared to using non-optimized seed selection.

Abstract: For three-dimensional rendering, a machine-learnt model is trained to generate representation vectors for rendered images formed with different rendering parameter settings. The distances between representation vectors of the images to a reference are used to select the rendered image and corresponding rendering parameters that provides a consistency with the reference. In an additional or different embodiment, optimized pseudo-random sequences are used for physically-based rendering. The random number generator seed is selected to improve the convergence speed of the renderer and to provide higher quality images, such as providing images more rapidly for training compared to using non-optimized seed selection.

Abstract: For rendering in medical imaging, a transfer function is determined. A simple approach to setting the transfer function uses a combination of a rendered image and the voxel data, providing a hybrid of both image and data-driven approaches. A region of interest on a rendered image is used to select some of the voxel data. A characteristic of the selected voxel data is used to determine the transfer function for rendering another image. Both the visual aspect of the rendered image and the voxel data from the scan are used to set the transfer function.

Abstract: For rendering in medical imaging, a transfer function is determined. A simple approach to setting the transfer function uses a combination of a rendered image and the voxel data, providing a hybrid of both image and data-driven approaches. A region of interest on a rendered image is used to select some of the voxel data. A characteristic of the selected voxel data is used to determine the transfer function for rendering another image. Both the visual aspect of the rendered image and the voxel data from the scan are used to set the transfer function.

Abstract: For three-dimensional rendering, a machine-learnt model is trained to generate representation vectors for rendered images formed with different rendering parameter settings. The distances between representation vectors of the images to a reference are used to select the rendered image and corresponding rendering parameters that provides a consistency with the reference. In an additional or different embodiment, optimized pseudo-random sequences are used for physically-based rendering. The random number generator seed is selected to improve the convergence speed of the renderer and to provide higher quality images, such as providing images more rapidly for training compared to using non-optimized seed selection.

Abstract: A method (100) of aneurysm analysis (110) and virtual stent simulation (120) for endovascular treatment of sidewall intracranial aneurysms.

Abstract: A method for visualizing brain connectivity includes receiving image data including molecular diffusion of brain tissue, constructing a tree data structure from the image data, wherein the tree data structure comprises a plurality of network nodes, wherein each network node is connected to a root of the tree data structure, rendering a ring of a radial layout depicting the tree data structure, wherein a plurality of vertices may be traversed from the top to the bottom, duplicating at least one control point for spline edges sharing a common ancestor, and bundling spline edges by applying a global strength parameter ?.

Abstract: An object or haptic device is registered with a holograph. The position of the object or haptic device relative to the projector or holographic image is sensed. An eye tracking system acts as an additional source of information about the position. As a viewer interacts with the holograph, their eyes focus on the location of interaction. The eye tracking, such as the focal location, provides an additional source of position information to reduce or avoid misregistration.

Abstract: Methods for computed tomography data-based cycle estimation and four-dimensional reconstruction are provided. A gated reconstruction is derived from CT data acquired without gating using an added artificial trigger. The resulting images for different slices are used to determine local or slice variations over time. The local variations over time for the various slices are combined to create a respiratory cycle signal. This respiratory cycle signal is used to bin the images for different phases, allowing four-dimensional CT reconstruction.

Abstract: Methods for computed tomography data-based cycle estimation and four-dimensional reconstruction are provided. A gated reconstruction is derived from CT data acquired without gating using an added artificial trigger. The resulting images for different slices are used to determine local or slice variations over time. The local variations over time for the various slices are combined to create a respiratory cycle signal. This respiratory cycle signal is used to bin the images for different phases, allowing four-dimensional CT reconstruction.

Abstract: A method for extracting a colonic centerline includes segmenting a colon from a digital image of a patient's abdomen, selecting one extreme point of the colon as a source point, calculating a first distance transform of every point in the colon that is a distance of a point to the source point, and calculating a second distance transform of every point in the colon, that is a shortest distance of a point to a wall point of the colon. A centerline path is generated through the colon using the first and second distance transforms, starting from a point with a greatest distance to the source point as determined by the first distance transform, and adding points to the centerline path by selecting points with a greatest distance to the source point that are farthest from the wall of the colon using the second distance transform.

Abstract: A method for visualizing brain connectivity includes receiving image data including molecular diffusion of brain tissue, constructing a tree data structure from the image data, wherein the tree data structure comprises a plurality of network nodes, wherein each network node is connected to a root of the tree data structure, rendering a ring of a radial layout depicting the tree data structure, wherein a plurality of vertices may be traversed from the top to the bottom, duplicating at least one control point for spline edges sharing a common ancestor, and bundling spline edges by applying a global strength parameter ?.