Abstract: An asynchronous distributed generative adversarial network (AsynDGAN) can include a central computing system and at least two discriminator nodes. The central computing system can include a generator neural network, an aggregator, and a network interface. Each discriminator node can have its own corresponding training data set. In addition, different discriminator nodes can use different data modalities. The central computing system communicates with each of the at least two discriminator nodes via the network interface and aggregates data received from the at least two discriminator nodes, via the aggregator, to update a model for the generator neural network during training of the generator neural network. The central computing system can further include a data access system that supports third party access to synthetic data generated by the generator neural network.

Abstract: Systems and methods for joint reconstruction and segmentation of organs from magnetic resonance imaging (MRI) data are provided. Sparse MRI data is received at a computer system, which jointly processes the MRI data using a plurality of reconstruction and segmentation processes. The MRI data is processed using a joint reconstruction and segmentation process to identify an organ from the MRI data. Additionally, the MRI data is processed using a channel-wise attention network to perform static reconstruction of the organ from the MRI data. Further, the MRI data can is processed using a motion-guided network to perform dynamic reconstruction of the organ from the MRI data. The joint processing allows for rapid static and dynamic reconstruction and segmentation of organs from sparse MRI data, with particular advantage in clinical settings.

Abstract: An image processing neural network system includes a base net of at least one convolutional layer and at least one pooling layer; and a scenario block layer. The scenario block layer performs scene classification and generates a dictionary of scenarios and a vector of scenario encoding coefficients to output a probabilistic scene class assignment and the vector of scenario encoding coefficients. The vector of scenario encoding coefficients corresponds to reasoning for the scene classification.

Abstract: An image processing neural network system includes a base net of at least one convolutional layer and at least one pooling layer; and a scenario block layer. The scenario block layer performs scene classification and generates a dictionary of scenarios and a vector of scenario encoding coefficients to output a probabilistic scene class assignment and the vector of scenario encoding coefficients. The vector of scenario encoding coefficients corresponds to reasoning for the scene classification.

Abstract: Systems and methods for joint reconstruction and segmentation of organs from magnetic resonance imaging (MRI) data are provided. Sparse MRI data is received at a computer system, which jointly processes the MRI data using a plurality of reconstruction and segmentation processes. The MRI data is processed using a joint reconstruction and segmentation process to identify an organ from the MRI data. Additionally, the MRI data is processed using a channel-wise attention network to perform static reconstruction of the organ from the MRI data. Further, the MRI data can is processed using a motion-guided network to perform dynamic reconstruction of the organ from the MRI data. The joint processing allows for rapid static and dynamic reconstruction and segmentation of organs from sparse MRI data, with particular advantage in clinical settings.

Abstract: A system and methods for generating 3D images from 2D bioluminescent images and visualizing tumor locations are provided. A plurality of 2D bioluminescent images of a subject are acquired using any suitable bioluminescent imaging system. The 2D images are registered to align each image and to compensate for differences between adjacent images. After registration, corresponding features are identified between consecutive sets of 2D images. For each corresponding feature identified in each set of 2D images, an orthographic projection model is applied, such that rays are projected through each point in the feature. The intersection points of the rays are plotted in a 3D image space. All of the 2D images are processed in the same manner, such that a resulting 3D image of a tumor is generated.

Abstract: This invention presents a method to simulate how blood flows through the heart, using the reconstructed 4D motion of the endocardial surface of the left ventricle. The reconstruction, utilizing a computing device, captures the motion of the full 3D surfaces of the complex features, such as the papillary muscles and the ventricular trabeculae. By visualizing the flow field, the capability of viewing the interactions between the blood and the trabeculae in far more detail has been achieved.

Abstract: A system and method for tracking features is provided which allows for the tracking of features that move in a series of images. A training set of images is processed to produce clustered shape subspaces corresponding to the set of images, such that non-linear shape manifolds in the images are represented as piecewise, overlapping linear surfaces that are clustered according to similarities in perspectives. A landmark-based training algorithm (e.g., ASM) is applied to the clustered shape subspaces to train a model of the clustered shape subspaces and to create training data. A subsequent image is processed using the training data to identify features in the target image by creating an initial shape, superimposing the initial shape on the target image, and then iteratively deforming the shape in accordance with the model until a final shape is produced corresponding to a feature in the target image.

Abstract: A system and method of reconstructing an image through solution of a data fitting problem, wherein the data fitting problem is not susceptible to efficient solution as a whole, is disclosed, which may comprise gathering, via a computing device, k-space image data, selecting a data fitting problem solution algorithm for the k-space data, decomposing the data fitting problem solution into a plurality of sub-problem solutions each susceptible to efficient solution separately for the k-space data, obtaining, via the computing device, the plurality of sub-problem solutions for the k-space data, and reconstructing, via the computing device, the image based upon a weighted average of the plurality of sub-problem solutions for the k-space data. The image may be a magnetic resonance image. The data fitting problem may comprise a minimization of a plurality of linear terms of a least square data fitting solution. The image may comprise under-sampled k-space image data.

Abstract: A system and methods for generating 3D images from 2D bioluminescent images and visualizing tumor locations are provided. A plurality of 2D bioluminescent images of a subject are acquired using any suitable bioluminescent imaging system. The 2D images are registered to align each image and to compensate for differences between adjacent images. After registration, corresponding features are identified between consecutive sets of 2D images. For each corresponding feature identified in each set of 2D images, an orthographic projection model is applied, such that rays are projected through each point in the feature. The intersection points of the rays are plotted in a 3D image space. All of the 2D images are processed in the same manner, such that a resulting 3D image of a tumor is generated.

Abstract: A system and method for tracking features is provided which allows for the tracking of features that move in a series of images A training set of images is processed to produce clustered shape subspaces corresponding to the set of images, such that non-linear shape manifolds in the images are represented as piecewise, overlapping linear surfaces that are clustered according to similarities in perspectives. A landmark-based training algorithm (e.g., ASM) is applied to the clustered shape subspaces to train a model of the clustered shape subspaces and to create training data. A subsequent image is processed using the training data to identify features in the target image by creating an initial shape, superimposing the initial shape on the target image, and then iteratively deforming the shape in accordance with the model until a final shape is produced corresponding to a feature in the target image.

Abstract: This invention presents a method to simulate how blood flows through the heart, using the reconstructed 4D motion of the endocardial surface of the left ventricle. The reconstruction, utilizing a computing device, captures the motion of the full 3D surfaces of the complex features, such as the papillary muscles and the ventricular trabeculae. By visualizing the flow field, the capability of viewing the interactions between the blood and the trabeculae in far more detail has been achieved.

Abstract: A system and methods for generating 3D images (24) from 2D bioluminescent images (22) and visualizing tumor locations are provided. A plurality of 2D bioluminescent images of a subject are acquired during a complete revolution of an imaging system about a subject, using any suitable bioluminescent imaging system. After imaging, the 2D images are registered (20) according to the rotation axis to align each image and to compensate for differences between adjacent images. After registration (20), corresponding features are identified between consecutive sets of 2D image (22). For each corresponding feature identified in each set of 2D images an orthographic projection model (24) is applied, such that rays are projected through each point in the feature. The intersection point of the rays are plotted in a 3D image of a tumor is generated. The 3D image can be registered with a reference image of the subject, so that the shape and location of the tumor can be precisely visualized with respect to the subject.

Abstract: A system and method of reconstructing an image through solution of a data fitting problem, wherein the data fitting problem is not susceptible to efficient solution as a whole, is disclosed, which may comprise gathering, via a computing device, k-space image data, selecting a data fitting problem solution algorithm for the k-space data, decomposing the data fitting problem solution into a plurality of sub-problem solutions each susceptible to efficient solution separately for the k-space data, obtaining, via the computing device, the plurality of sub-problem solutions for the k-space data, and reconstructing, via the computing device, the image based upon a weighted average of the plurality of sub-problem solutions for the k-space data. The image may be a magnetic resonance image. The data fitting problem may comprise a minimization of a plurality of linear terms of a least square data fitting solution. The image may comprise under-sampled k-space image data.

Abstract: A system and method for tracking features, e.g., facial features, is provided, which allows for the tracking of features which move in a series of images and whose shape changes nonlinearly due to perspective projection and complex 3D movements. A training set of images is processed to produce clustered shape subspaces corresponding to the set of images, such that non-linear shape manifolds in the images are represented as piecewise, overlapping linear surfaces that are clustered according to similarities in perspectives. A landmark-based training algorithm (e.g., ASM) is applied to the clustered shape subspaces to train a model of the clustered shape subspaces and to create training data.

Abstract: A system and methods for generating 3D images (24) from 2D bioluminescent images (22) and visualizing tumor locations are provided. A plurality of 2D bioluminescent images of a subject are acquired during a complete revolution of an imaging system about a subject, using any suitable bioluminescent imaging system. After imaging, the 2D images are registered (20) according to the rotation axis to align each image and to compensate for differences between adjacent images. After registration (20), corresponding features are identified between consecutive sets of 2D image (22). For each corresponding feature identified in each set of 2D images an orthographic projection model (24) is applied, such that rays are projected through each point in the feature. The intersection point of the rays are plotted in a 3D image of a tumor is generated. The 3D image can be registered with a reference image of the subject, so that the shape and location of the tumor can be precisely visualized with respect to the subject.

Abstract: A system and method for tracking features, e.g., facial features, is provided, which allows for the tracking of features which move in a series of images and whose shape changes nonlinearly due to perspective projection and complex 3D movements. A training set of images is processed to produce clustered shape subspaces corresponding to the set of images, such that non-linear shape manifolds in the images are represented as piecewise, overlapping linear surfaces that are clustered according to similarities in perspectives. A landmark-based training algorithm (e.g., ASM) is applied to the clustered shape subspaces to train a model of the clustered shape subspaces and to create training data.

Abstract: A method and apparatus for automatically detecting breast tumors and lesions in images, including ultrasound, digital and analog mammograms, and MRI images, is provided. An image of a breast is acquired. The image is filtered and contrast of the image is enhanced. Intensity and texture classifiers are applied to each pixel in the image, the classifiers indicative of the probability of the pixel corresponding to a tumor. A seed point is identified within the image, and a region of interest is grown around the seed point. Directional gradients are calculated for each pixel of the image. Boundary points of the region of interest are identified. The boundary points are passed as inputs to a deformable model. The deformable model processes the boundary points to indicate the presence or absence of a tumor.

Abstract: A method and apparatus for human face shape and motion estimation based on integrating optical flow and deformable models is disclosed. The optical flow, constraint equation provides a non-holonomic constraint on the motion of the deformable model. Forces computed from edges and optical flow are used simultaneously. When this dynamic system is solved, a model-based least-squares solution for the optical flow is obtained and improved estimation results are achieved. The use of a 3-D model reduces or eliminates problems associated with optical flow computation. This approach instantiates a general methodology for treating visual cues as constraints on deformable models. The model uses a small number of parameters to describe a rich variety of face shapes and facial expressions.

Abstract: A method and apparatus for dynamic modeling of an object having material points. The method includes receiving signals from a sensor which correspond to respective material points; providing a volumetric model, having functions as parameters, representative of the object; and adapting the parameters to fit a changing model shape. This method of dynamic shape modeling dynamic motion modeling or both includes shape estimation and motion analysis. The apparatus includes a signal processor for receiving signals from a sensor, a second signal processor for providing a volumetric model having functions as parameters representative of the object and a third signal processor for receiving sensed signals and adapting the model and providing a dynamic representation.