Methods for reconstructing images from undersampled k-space data using a machine learning approach to learn non-linear mapping functions from acquired k-space lines to generate unacquired target points across multiple coils are described.

A method includes generating first contrast significance data for a first computer vision model generated from a first training set of medical scans. First significant contrast parameters are identified based on the first contrast significance data. A first re-contrasted training set is generated based on performing a first intensity transformation function on the first training set of medical scans, where the first intensity transformation function utilizes the first significant contrast parameters. A first re-trained model is generated from the first re-contrasted training set, which is associated with corresponding output labels based on abnormality data for the first training set of medical scans. Re-contrasted image data of a new medical scan is generated based on performing the first intensity transformation function. Inference data indicating at least one abnormality detected in the new medical scan is generated based on utilizing the first re-trained model on the re-contrasted image data.

An image acquisition unit acquires a plurality of projection images corresponding to a plurality of radiation source positions at the time of tomosynthesis imaging, the plurality of projection images being generated by causing an imaging apparatus to perform tomosynthesis imaging. A positional shift amount derivation unit derives a positional shift amount between the plurality of projection images based on body movement of the subject with a reference projection image generated at a radiation source position where an optical axis of the radiation emitted from the radiation source is perpendicular to a detection surface of the detection unit, among the plurality of projection images, as a reference. A reconstruction unit generates a tomographic image of at least one tomographic plane of the subject by reconstructing the plurality of projection images while correcting the positional shift amount.

Disclosed are a method and apparatus for processing a low-dose X-ray computed tomography image based on efficient unsupervised learning by using an invertible neural network. The method of processing a low-dose X-ray computed tomography image based on unsupervised learning by using an invertible neural network performed by a computer device includes providing an invertible generator for restoring an image, and training the invertible generator to restore a low-dose computed tomography image to a normal computed tomography image.

Image reconstruction is an inverse problem that solves for a computational image based on sampled sensor measurement. Sparsely sampled image reconstruction poses addition challenges due to limited measurements. In this work, we propose an implicit Neural Representation learning methodology with Prior embedding (NeRP) to reconstruct a computational image from sparsely sampled measurements. The method differs fundamentally from previous deep learning-based image reconstruction approaches in that NeRP exploits the internal information in an image prior, and the physics of the sparsely sampled measurements to produce a representation of the unknown subject. No large-scale data is required to train the NeRP except for a prior image and sparsely sampled measurements. In addition, we demonstrate that NeRP is a general methodology that generalizes to different imaging modalities such as CT and MRI. We also show that NeRP can robustly capture the subtle yet significant image changes required for assessing tumor progression.

Methods and systems are provided for increasing a quality of computed tomography (CT) images reconstructed from high helical pitch scans. In one embodiment, the current disclosure provides for a method comprising generating a first computed tomography (CT) image from projection data acquired at a high helical pitch; using a trained multidimensional statistical regression model to generate a second CT image from the first CT image, the multidimensional statistical regression model trained with a plurality of target CT images reconstructed from projection data acquired at a lower helical pitch; and performing an iterative correction of the second CT image to generate a final CT image.

An X-ray imaging apparatus includes an X-ray generator including a plurality of X-ray sources, an X-ray detector configured to detect X-rays radiated from the plurality of X-ray sources and generate a plurality of pieces of projection data, and a processor configured to apply log projection to each of the plurality of pieces of projection data, to apply weighted projection to the log-projected projection data, to apply a bidirectional ramp filter to the weighted-projected projection data, and to generate a tomographic image reconstructed based on each of the projection data to which the bidirectional ramp filter is applied.

Disclosed are computer implemented techniques for conducting a simulation of physical properties of a porous medium. The features include receiving a micro-CT 3D image that captures a representative elemental volume of the porous medium, the porous medium defined as having mineral types and fluid types with individual grains and grain to grain contacts, labeling the micro-CT 3D image as individual voxels according to mineral and fluid types and labeling the mineral type voxels as belonging to separated and fixed individual grains. The features also include transforming the labeled voxels into an unstructured conformal mesh representation for all grains and applying the unstructured conformal mesh representation to a parametric cohesive contact engine, with the parametric cohesive contact engine executing a parametric cohesive contact model that has an adjustable parameter, a critical separation δ.sup.0 conditioned according to consolidation level.

Embodiments of the present invention provide systems and methods to detect a moving anatomic feature during a treatment sequence based on a computed and/or a measured shortest distance between the anatomic feature and at least a portion of an imaging system.

A system and method for generating a parametric map of a subject's brain includes receiving non-contrast computed tomography (NCCT) imaging data and receiving computed tomography perfusion (CTP) data. The method further includes creating a baseline image by utilizing the NCCT data and generating a parametric map using the CTP data and the baseline image.