Abstract: Methods and apparatuses for performing automated detection of lung slide using a computing device (e.g., an ultrasound system, etc.) are disclosed. In some embodiments, the techniques determine lung sliding using one or more neural networks. In some embodiments, the neural networks are part of a process that determines probabilities of the lung sliding at one or more M-lines. In some embodiments, the techniques display one or more probabilities of lung sliding in a B-mode ultrasound image.

Abstract: Systems and methods are provided for a deep learning-based digital breast tomosynthesis (DBT) image reconstruction that mitigates limited angular artifacts and improves in-depth resolution of the resulting images. The systems and methods may reduce the sparse-view artifacts in DBT via deep learning without losing image sharpness and contrast. A deep neural network may be trained in a way to reduce training-time computational cost. An ROI loss method may be used for further improvement on the resolution and contrast of the images.

Abstract: Techniques are described for computer-implemented techniques for managing various aspects of the cardiac care pathway using machine learning. According to an embodiment, a method can include training an outcomes forecasting model to predict patient outcomes resulting from undergoing a cardiac valve procedure using multi-modal training data for a plurality of different patients, wherein the training comprising separately training different machine learning sub-models of the forecasting model to predict preliminary patient outcome data and mapping the preliminary patient outcome data to the patient outcomes, resulting in a trained version of the outcome forecasting model. The method further includes applying the trained version of the outcomes forecasting model to new multi-modal data for a new patient to predict the patient outcomes for the new patient resulting from undergoing the cardiac value procedure.

Abstract: An image correction system and method takes an initial volume reconstructed from a number of projections obtained by the imaging system as the sole input to the image correction system. In a first step, the image correction system reconstructs a number of reconstructed or forward projections from the initial volume. In a second and optionally concurrent step, the image correction system forms a corrected volume by applying an artificial intelligence/deep learning/convolutional neural network model to the initial volume. In a third step, the image correction system employs the forward projections and the corrected volume as inputs to an iterative reconstruction process to achieve an optimized volume as an output from the image correction system. The use of the initial volume as the only input to the image correction system simplifies the computational processes of the image processing system while providing an optimized image having improved detail and image quality.

Abstract: Methods and apparatuses for performing automated detection of lung slide using a computing device (e.g., an ultrasound system, etc.) are disclosed. In some embodiments, the techniques determine lung sliding using one or more neural networks. In some embodiments, the neural networks are part of a process that determines probabilities of the lung sliding at one or more M-lines. In some embodiments, the techniques display one or more probabilities of lung sliding in a B-mode ultrasound image.

Abstract: Systems and methods are provided for a deep learning-based digital breast tomosynthesis (DBT) image reconstruction that mitigates limited angular artifacts and improves in-depth resolution of the resulting images. The systems and methods may reduce the sparse-view artifacts in DBT via deep learning without losing image sharpness and contrast. A deep neural network may be trained in a way to reduce training-time computational cost. An ROI loss method may be used for further improvement on the resolution and contrast of the images.

Abstract: Systems and methods are provided for a multi-scale deep learning-based digital breast tomosynthesis (DBT) image reconstruction that mitigates the superposition of breast tissue along with the limited angular artifacts, and improves in-depth resolution of the resulting images. A multi-scale deep neural network may be used where a first network may focus on a first parameter, such as limited angular artifacts reduction, and a second network may focus on a second parameter, such as image detail refinement. The output from the first neural network may be used as the input for the second neural network. The systems and methods may reduce the sparse-view artifacts in DBT via deep learning without losing image sharpness and contrast. A deep neural network may be trained in a way to reduce training-time computational cost. An ROI loss method may be used for further improvement on the resolution and contrast of the images.

Abstract: In accordance with one aspect of the disclosure, an image generation system is provided. The system includes at least one processor and at least one non-transitory, computer- readable memory accessible by the processor and having instructions that, when executed by the processor, cause the processor to receive a first patient image associated with a patient, receive a second patient image associated with the patient, train an untrained model based on the first patient image and the second patient image, provide the first patient image to the model, receive a third patient image from the model, and output the third patient image to at least one of a storage system or a display.

Abstract: Systems and techniques for monitoring, predicting and/or alerting for short-term oxygen support needs of patients are presented. A system can include a data collection component that receives multimodal patient data for a patient having a respiratory condition in association with monitoring and treating the respiratory condition in real-time, the multimodal patient data comprising at least physiological data regarding physiological parameters tracked for the patient over a period of time, and current oxygen support data regarding a current oxygen support mechanism of the patient.

Abstract: The present disclosure a system and method for generating a tomographic image of a subject. In some aspects, the method includes receiving an initial image acquired from a subject using the tomographic imaging system, and performing, using the initial image and a cost function model, a penalty calculation based on a spatially variant hyper-parameter. The method also includes generating an updated image using the penalty calculation, and generating a finalized image by iteratively updating the updated image until a stopping criterion is met.

Abstract: The present disclosure a system and method for generating a tomographic image of a subject. In some aspects, the method includes receiving an initial image acquired from a subject using the tomographic imaging system, and performing, using the initial image and a cost function model, a penalty calculation based on a spatially variant hyper-parameter. The method also includes generating an updated image using the penalty calculation, and generating a finalized image by iteratively updating the updated image until a stopping criterion is met.

Abstract: Systems and methods for data-driven respiratory gating in positron emission tomography (PET) are provided. In some aspects, a provided method for generating motion information from PET imaging includes receiving time-of-flight (TOF) data acquired using a PET system, and selecting, using at least one image reconstructed from the TOF data, a region of interest (ROI) having tissues subject to motion. The method also includes generating a TOF sinogram mask by projecting an image mask corresponding to the ROI into a sinogram space, and applying the TOF sinogram mask to a TOF sinogram, produced using the TOF data, to identify data in the TOF sinogram associated with motion. The method further includes generating motion information using the data identified.

Abstract: A medical imaging system for reconstructing quantitative dynamic nuclear medicine images of a subject may include a three dimensional positron emission tomography (3D PET) scanner and a data processing system. The three dimensional positron emission tomography (3D PET) scanner may: perform scans of the subject at each of multiple bed positions, each scan being of only a sub-portion of the subject that is within an axial length covered by the scanner; and generate dynamic list-mode PET data that: represents a total number of photon pairs arriving at each of multiple detector pairs in the 3D PET scanner; and is organized in a list of elements, each element in the list including a detector pair index and an arrival time for each detected photon pair; and acquires PET data using a dynamic PET protocol with at least two scans at each of the multiple bed positions.

Abstract: A system and method for determining, from registered positron emission tomography (PET) sinogram data and magnetic resonance (MR) image data, an estimated attenuation sinogram uses a data consistency condition to evaluate a gradient of the PET sinogram data using the MR image data. An image of the subject is reconstructed using the estimate attenuation sinogram.

Abstract: Described here are systems and methods for reconstructing images from limited angle positron emission tomography (“PET”) data acquired using a PET system with a partial-ring detector configuration, such as an in-beam PET system. The reconstruction process is specifically designed to account for the limited angular coverage of the partial-ring detector by implementing a reduced angle system matrix in an iterative reconstruction process.

Abstract: Systems and methods for compensating motion artifacts in positron emission tomography (“PET”) imaging based on medical images acquired with a medical imaging system are provided. In some embodiments, the method includes acquiring PET data from a subject with a PET system during which at least a portion of the subject is undergoing motion, and providing medical images acquired from the subject using a medical imaging system, the medical images including regions depicting motion. The method also includes estimating, from the medical images, motion information associated with the motion of the at least a portion of the subject, and reconstructing a motion-corrected PET image using the PET data using a reconstruction algorithm that incorporates the motion information into a system matrix.

Abstract: Described here are systems and methods for reconstructing images from limited angle positron emission tomography (“PET”) data acquired using a PET system with a partial-ring detector configuration, such as an in-beam PET system. The reconstruction process is specifically designed to account for the limited angular coverage of the partial-ring detector by implementing a reduced angle system matrix in an iterative reconstruction process.

Abstract: A medical imaging system for reconstructing quantitative dynamic nuclear medicine images of a subject may include a three dimensional positron emission tomography (3D PET) scanner and a data processing system. The three dimensional positron emission tomography (3D PET) scanner may: perform scans of the subject at each of multiple bed positions, each scan being of only a sub-portion of the subject that is within an axial length covered by the scanner; and generate dynamic list-mode PET data that: represents a total number of photon pairs arriving at each of multiple detector pairs in the 3D PET scanner; and is organized in a list of elements, each element in the list including a detector pair index and an arrival time for each detected photon pair; and acquires PET data using a dynamic PET protocol with at least two scans at each of the multiple bed positions.

Abstract: A system and method for determining, from registered positron emission tomography (PET) sinogram data and magnetic resonance (MR) image data, an estimated attenuation sinogram uses a data consistency condition to evaluate a gradient of the PET sinogram data using the MR image data. An image of the subject is reconstructed using the estimate attenuation sinogram.

Abstract: Systems and methods for compensating motion artifacts in positron emission tomography (“PET”) imaging based on medical images acquired with a medical imaging system are provided. In some embodiments, the method includes acquiring PET data from a subject with a PET system during which at least a portion of the subject is undergoing motion, and providing medical images acquired from the subject using a medical imaging system, the medical images including regions depicting motion. The method also includes estimating, from the medical images, motion information associated with the motion of the at least a portion of the subject, and reconstructing a motion-corrected PET image using the PET data using a reconstruction algorithm that incorporates the motion information into a system matrix.