Methods and systems for PET image reconstruction are provided. A method may include obtaining an image sequence associated with a subject. The image sequence may include one or more images generated via scanning the subject at one or more consecutive time periods. The method may also include obtaining a target machine learning model. The method may further include generating at least one target image using the target machine learning model based on the image sequence. The at least one target image may present a dynamic parameter associated with the subject. The target machine learning model may provide a mapping between the image sequence and the at least one target image.

Model-based image reconstruction (MBIR) methods using convolutional neural networks (CNNs) as priors have demonstrated superior image quality and robustness compared to conventional methods. Studies have explored MBIR combined with supervised and unsupervised denoising techniques for image reconstruction in magnetic resonance imaging (MRI) and positron emission tomography (PET). Unsupervised methods like the deep image prior (DIP) have shown promising results and are less prone to hallucinations. However, since the noisy image is used as a reference, strategies to prevent overfitting are unclear. Recently, Bayesian DIP (BDIP) networks that model uncertainty tend to prevent overfitting without requiring early stopping. However, BDIP has not been studied with data-fidelity term for image reconstruction. Present disclosure provides systems and method that implement a MBIR framework with a modified BDIP. Specifically, an uncertainty-based penalty is included to the BDIP to improve reconstruction across iterations.

A method of training an image decomposition model, a method of decomposing an image, an electronic device, and a storage medium are provided. The method of training the image decomposition model includes: acquiring a training set; inputting first and second training images into first and second adversarial neural networks respectively, so as to determine a first loss function value; inputting a third training image into the first and second adversarial neural networks respectively, so as to determine a second loss function value; determining a third loss function value according to a comparison result between an acquired fusion image and the third training image, where the fusion image is generated by fusing generated images of the first and second adversarial neural networks, and adjusting a parameter of the image decomposition model according to at least one of the first to third loss function values.

Systems and methods for training a machine-learning model to generate denoised and dealiased image data are provided. The present disclosure provides techniques for training a machine-learning (ML) model to generate denoised and dealiased imaging data. A method includes (1) training a first ML model using a first training dataset comprising first image data to obtain a second ML model; and (2) training (a) the second ML model or (b) a third ML model using a second training dataset to obtain a fourth ML model. The second training dataset includes (i) the first image data and (ii) training image data obtained by applying at least one of the second ML model or the third ML model to second image data. The denoising and dealiasing ML model may be either the fourth ML model or derived from the fourth ML model.

Disclosed are methods, systems, and other implementations, including a unified complex-valued deep learning framework (AFT-Net), which determines the k-space domain to image domain mapping for MRI reconstruction and allows incorporation of existing deep learning models. Embodiments include a computer-implemented method for reconstructing images that includes obtaining resonance (MR) k-space data resulting from a scan performed by an MRI scanner on tissue of a patient, with the MR k-space data including complex-valued data, and processing, by a complex-valued machine learning image reconstruction system, the complex-valued data of the MR k-space data to generate image data representing features of the MR k-space data. The processing may include performing data filtering operations, by one or more machine learning filter blocks implemented according to a CU-Net architecture realized using one or more convolutional neural networks (CNN) configured for complex data processing, on data that is based on the k-space data.

The present disclosure is directed to systems and/or methods that may be used for determining scene information (for example, 3D scene information) using data obtained at least in part from a camera array. Certain embodiments may be used to create scene measurements of depth (and the probability of accuracy of that depth) using an array of cameras. One purpose of certain embodiments may be to determine the depths of elements of a scene, where the scene is observed from a camera array that may be moving through the scene. Certain embodiments may be used to determine open navigable space and to calculate the trajectories of objects that may be occupying portions of that space. In certain embodiments, the scene information may be used to generate a virtual space of voxels where the method then determines the occupancy of the voxel space by comparing a variety of measurements, including spectral response.

A method for performing real-time magnetic resonance (MR) imaging on a subject is disclosed. A prep pulse sequence is applied to the subject to obtain a high-quality special subspace, and a direct linear mapping from k-space training data to subspace coordinates. A live pulse sequence is then applied to the subject. During the live pulse sequence, real-time images are constructed using a fast matrix multiplication procedure on a single instance of the k-space training readout (e.g., a single k-space line or trajectory), which can be acquired at a high temporal rate.

A system and method for reducing scan time of periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER) imaging include acquiring a plurality of blades of k-space data of a region of interest in a rotational manner around a center of k-space via a magnetic resonance imaging (MRI) scanner during a PROPELLER sequence, wherein each blade of the plurality of blades of k-space data includes a plurality of parallel phase encoding lines sampled in a phase encoding order. Each blade of the plurality of blades of k-space data is undersampled. The system and method include utilizing a deep learning-based Cartesian-like reconstruction network to individually and separately reconstruct each blade of the plurality of blades of k-space data to generate a plurality of fully sampled blades. The system and method include utilizing a PROPELLER reconstruction algorithm to generate a complex image from the plurality of fully sampled blades.

The present invention relates to a method (1), resp. a device, system and computer-program product, for material decomposition of spectral imaging projection data. The method comprises receiving (2) projection data acquired by a spectral imaging system and reducing (3) noise in the projection data by combining corresponding spectral values for different projection rays to obtain noise-reduced projection data. The method comprises applying (6) a first projection-domain material decomposition algorithm to the noise-reduced projection data to obtain a first set of material path length estimates, and applying (7) a second projection-domain material decomposition algorithm to the projection data to obtain a second set of material path length estimates. The second projection-domain material decomposition algorithm comprises an optimization that penalizes a deviation between the second set of material path length estimates being optimized and the first set of material path length estimates.

Described herein is a medical system (100, 300) comprising a memory (110) storing machine executable instructions (120) and a motion estimating neural network (122, 700, 800, 900, 1000) configured for outputting trajectory data (130) in response to receiving a trial motion trajectory (128) as input. The execution of the machine executable instructions causes a computational system (104) to: receive (200) measured k-space data (124) descriptive of a subject (318); perform (202) motion estimation of the subject between the sequence of discrete acquisitions by solving an optimization problem to determine a calculated motion trajectory of the subject in the predefined coordinate system, wherein the optimization problem is modified using the trajectory data; and reconstruct (204) a final motion corrected magnetic resonance image (136) from the measured k-space data and the calculated motion trajectory in the predefined coordinate system.