Abstract: Abnormality detection within a defined area includes obtaining a plurality of images of the defined area from image-capture devices. An extent of deviation of one or more types of products from an inference of each of the plurality of images is determined using a trained neural network. A localized dimensional representation is generated in a portion of an input image associated with a first location of the plurality of locations, based on gradients computed from the determined extent of deviation. The generated localized dimensional representation provides a visual indication of an abnormality located in the first location within the defined area. An action associated with the first location is executed based on the generated dimensional representation for proactive control or prevention of occurrence of undesired event in the defined area.

Abstract: Systems and methods for image classification include receiving imaging data of in-vivo or excised tissue of a patient during a surgical procedure. Local image features are extracted from the imaging data. A vocabulary histogram for the imaging data is computed based on the extracted local image features. A classification of the in-vivo or excised tissue of the patient in the imaging data is determined based on the vocabulary histogram using a trained classifier, which is trained based on a set of sample images with confirmed tissue types.

Abstract: Methods and systems for enhancing a distributed medical network. For example, a computer-implemented method includes inputting training data corresponding to each local computer into their corresponding machine learning model; generating a plurality of local losses including generating a local loss for each machine learning model based at least in part on the corresponding training data; generating a plurality of local parameter gradients including generating a local parameter gradient for each machine learning model based at least in part on the corresponding local loss; generating a global parameter update based at least in part on the plurality of local parameter gradients; and updating each machine learning model hosted at each local computer of the plurality of local computers by at least updating their corresponding active parameter set based at least in part on the global parameter update.

Abstract: Systems and methods for generating and tracking shapes of a target may be provided. The method may include obtaining at least one first resolution image corresponding to at least one of a sequence of time frames of a medical scan. The method may include determining, according to a predictive model, one or more shape parameters regarding a shape of a target from the at least one first resolution image. The method may include determining, based on the one or more shape parameters and a shape model, at least one shape of the target from the at least one first resolution image. The method may further include generating a second resolution visual representation of the target by rendering the determined shape of the target.

Abstract: A system for reconstructing magnetic resonance images includes a processor that is configured to obtain, from a magnetic resonance scanner, sub-sampled k-space data; apply an inverse fast fourier transform to the sub-sampled k-space data to generate a preliminary image; and process the preliminary image via a trained cascaded recurrent neural network to reconstruct a magnetic resonance image.

Abstract: Described herein are systems, methods and instrumentalities associated with image segmentation. The systems, methods and instrumentalities have a hierarchical structure for producing a coarse segmentation of an anatomical structure and then refining the coarse segmentation based on a shape prior of the anatomical structure. The coarse segmentation may be generated using a multi-task neural network and based on both a segmentation loss and a regression loss. The refined segmentation may be obtained by deforming the shape prior using one or more of a shape-based model or a learning-based model.

Abstract: A trained recurrent neural network having a set of control policies learned from application of a template dataset and one or more corresponding template deep network architectures may generate a deep network architecture for performing a task on an application dataset. The template deep network architectures may have an established level or performance in executing the task. A deep network based on the deep network architecture may trained to perform the task on the application dataset. The control policies of the recurrent neural network may be updated based on the performance of the trained deep network.

Abstract: A method of characterizing a serum and plasma portion of a specimen in regions occluded by one or more labels. The characterization may be used for Hemolysis, Icterus, and/or Lipemia, or Normal detection. The method captures one or more images of a labeled specimen container including a serum or plasma portion, processes the one or more images to provide segmentation data and identification of a label-containing region, and classifying the label-containing region with a convolutional neural network (CNN) to provide a pixel-by-pixel (or patch-by-patch) characterization of the label thickness count, which may be used to adjust intensities of regions of a serum or plasma portion having label occlusion. Optionally, the CNN can characterize the label-containing region as one of multiple pre-defined label configurations. Quality check modules and specimen testing apparatus adapted to carry out the method are described, as are other aspects.

Abstract: Cardiac features captured via an MRI scan may be tracked and analyzed using a system described herein. The system may receive a plurality of MR slices derived via the MRI scan and present the MR slices in a manner that allows a user to navigate through the MR slices. Responsive to the user selecting one of the MR slices, contextual and global cardiac information associated with the selected slice may be determined and displayed. The contextual information may correspond to the selected slice and the global information may encompass information gathered across the plurality of MR slices. A user may have the ability to navigate between the different display areas and evaluate the health of the heart with both local and global perspectives.

Abstract: A bullseyes plot may be generated based on cardiac magnetic resonance imaging (CMRI) to facilitate the diagnosis and treatment of heart diseases. Described herein are systems, methods, and instrumentalities associated with bullseyes plot generation. A plurality of myocardial segments may be obtained for constructing the bullseye plot based on landmark points detected in short-axis and long-axis magnetic resonance (MR) slices of the heart and by arranging the short-axis MR slices sequentially in accordance with the order in which the slices are generated during the CMRI. The sequential order of the short-axis MR slices may be determined utilizing projected locations of the short-axis MR slices on a long-axis MR slice and respective distances of the projected locations to a landmark point of the long-axis MR slice. The myocardium and/or landmark points may be identified in the short-axis and/or long-axis MR slices using artificial neural networks.

Abstract: A system for generating a bullseye plot of a heart of a subject is provided. The system may obtain multiple slice images in a plurality of groups, wherein each group corresponds to one of a plurality of sections of the heart and includes at least one slice image of the corresponding section, and each slice image includes part of the right ventricle, part of the left ventricle, and part of the myocardium. The system may also identify at least one landmark associated with the left ventricle by applying a landmark detection network in each of the slice images. The system may further generate the bullseye plot of the heart based on the at least one landmark identified in each of the multiple slice images, wherein the bullseye plot includes a plurality of sectors, each of which represents an anatomical region of the myocardium in one of the plurality of sections.

Abstract: Described herein are neural network-based systems, methods and instrumentalities associated with estimating the motion of an anatomical structure. The motion estimation may be performed using a feature pyramid and/or a motion pyramid that correspond to multiple image scales. The motion estimation may be performed using neural networks and parameters that are learned via a training process involving a student network and a teacher network pre-pretrained with abilities to apply progressive motion compensation.

Abstract: 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.

Abstract: A method includes acquiring magnetic resonance imaging (MRI) data with multi-coil dimensions, compressing the coil dimensions to a fixed and predetermined number of virtual coils, and utilizing the fixed and predetermined number of virtual coils by an artificial intelligence engine for artificial intelligence applications.

Abstract: A method of characterizing a serum and plasma portion of a specimen in regions occluded by one or more labels. The characterization may be used for determining Hemolysis (H), Icterus (I), and/or Lipemia (L), or Normal (N) of a serum or plasma portion of a specimen. The method includes capturing one or more images of a labeled specimen container including a serum or plasma portion, processing the one or more images with a convolutional neural network to provide a determination of Hemolysis (H), Icterus (I), and/or Lipemia (L), or Normal (N). In further embodiments, the convolutional neural network can provide N?-Class segmentation information. Quality check modules and testing apparatus adapted to carry out the method are described, as are other aspects.

Abstract: An apparatus for magnetic resonance imaging (MRI) image reconstruction is provided. The apparatus accesses a training set of MRI data for training. The training set can include paired fully sampled data or unpaired fully sampled data. Undersampled MRI data is optimized in an MRI data optimization module to generate reconstructed MRI data. The apparatus builds a discriminative model using the training set and the reconstructed MRI data. During inference, the parameters of the discriminator model are fixed and the discriminator model is used to classify an output of the MRI data optimization model as the reconstructed MRI image.

Abstract: A method includes acquiring MRI data, using an algorithm to predict cardiac cycles from the acquired MRI data, and operating on sections of the acquired MRI data corresponding to selected portions of the predicted cardiac cycles.

Abstract: Described herein are neural network-based systems, methods and instrumentalities associated with estimating the motion of an anatomical structure. The motion estimation may be performed utilizing pre-learned knowledge of the anatomy of the anatomical structure. The anatomical knowledge may be learned via a variational autoencoder, which may then be used to optimize the parameters of a motion estimation neural network system such that, when performing motion estimation for the anatomical structure, the motion estimation neural network system may produce results that conform with the underlying anatomy of anatomical structure.

Abstract: Described herein are neural network-based systems, methods and instrumentalities associated with image segmentation that may be implementing using an encoder neural network and a decoder neural network. The encoder network may be configured to receive a medical image comprising a visual representation of an anatomical structure and generate a latent representation of the medical image indicating a plurality of features of the medical image. The latent representation may be used by the decoder network to generate a mask for segmenting the anatomical structure from the medical image. The decoder network may be pre-trained to learn a shape prior associated with the anatomical structure and once trained, the decoder network may be used to constrain an output of the encoder network during training of the encoder network.

Abstract: An apparatus is configured to receive input image data corresponding to output image data of a first radiology scanner device, translate the input image data into a format corresponding to output image data of a second radiology scanner device and generate an output image corresponding to the translated input image data on a post processing imaging device associated with the first radiology scanner device. Medical images from a new scanner can be translate to look as if they came from a scanner of another vendor.