Methods and systems for deep learning based image reconstruction are disclosed herein. An example method includes receiving a set of imaging projections data, identifying a voxel to reconstruct, receiving a trained regression model, and reconstructing the voxel. The voxel is reconstructed by: projecting the voxel on each imaging projection in the set of imaging projections according to an acquisition geometry, extracting adjacent pixels around each projected voxel, feeding the regression model with the extracted adjacent pixel data to produce a reconstructed value of the voxel, and repeating the reconstruction for each voxel to be reconstructed to produce a reconstructed image.

A method for image reconstruction from plural copies, the method including receiving a series of measured projections p.sub.i of a target object h and associated background; iteratively reconstructing images h.sub.i(k) of the target object and images g.sub.i(k) of the background of the target object for each member i of the series of the measured projections p.sub.i over plural iterations k; and generating a final image of the target object h, based on the reconstructed images h.sub.i, when a set condition is met. The index i describes how many elements are in the series of projections p.sub.i, and iteration k indicates how many times the reconstruction of the image target is performed.

The systems and method for processing scanning data of a scanning object are provided. The method may include acquiring, in a scanning process, at least two target phases of a motion of the scanning object, wherein the scanning process involves multiple data acquisition time points each of which corresponds to a scanning data set; identifying at least two first time periods during the scanning process, each first time period corresponding to one of the two target phases; determining a second time period that encloses the at least two first time periods; and retrieving once, from the multiple scanning data sets, second scanning data sets for reconstructing phase images each of which corresponds to one target phase, the second scanning data sets being acquired at second data acquisition time points of the multiple data acquisition time points within the second time period.

A method for rendering metal obscured regions in a volume radiographic image reconstructs a first 3D image using a plurality of 2D projection images obtained over a scan angle range relative to the subject and identifies metal in the first 3D image or metal shadows in the plurality of 2D projection images. Then, metal obscured regions are determined in a reconstructed 3D image of the object, and an alternative reconstruction being a limited angle reconstruction is performed for the metal obscured regions and displayed to the user with an indication of the spatial relationship to a corresponding metal obscured region.

A computer-implemented method for providing a result dataset, comprising: acquisition of at least one projection mapping pair of an object under examination by a medical biplane imaging device, wherein the at least one projection mapping pair contains a first and a second projection mapping of the object under examination, that map the object under examination simultaneously in a first and a second detection plane, wherein the first and second detection plane are arranged non-parallel to one another, determination of a correction model for the correction of an artifact and/or a movement, wherein the artifact and/or the movement is mapped simultaneously in the at least one first and the at least one second projection mapping, wherein the at least one projection mapping pair specifies a consistency condition for the determination of the correction model, reconstruction of the result dataset at least from the at least one first projection mapping and on the basis of the correction model, provision of the result dataset.

Devices, systems, and methods receive scan data that were generated by scanning a region of a subject with a computed tomography apparatus; generate multiple partial angle reconstruction (PAR) images based on the scan data; obtain corresponding characteristics of the multiple PAR images; perform correspondence mapping on the multiple PAR images based on the obtained corresponding characteristics and on the multiple PAR images, wherein the correspondence mapping generates correspondence-mapping data; and generate a motion-corrected reconstruction image based on the correspondence-mapping data and on one or both of the scan data and the PAR images.

Systems and methods are provided for improved intra-operative micro-CT imaging of explanted tissue samples and for improved visualization of such samples. These embodiments provide for reduced scan times and the ability for radiologists to quickly receive useful scan imagery and to provide accurately-communicated recommendations to the operating surgeon. Improved scan visualization methods facilitate surgeon and radiologist interaction with the scan data, including of annotation, viewing, and reorientation to accurately reflect the orientation of imaged tissue samples relative to the body prior to explantation. Improved visualization methods include color-coded sample texturing to indicate sample orientation, color-coded tumor visualization to indicate proximity to sample margins, and intuitive methods for adjusting the location and orientation of two-dimensional visualizations relative to the sample.

A method for diagnostic imaging reconstruction uses a prior image x.sup.pr from a scan of a subject to initialize parameters of a neural network which maps coordinates in image space to corresponding intensity values in the prior image. The parameters are initialized by minimizing an objective function representing a difference between intensity values of the prior image and predicted intensity values output from the neural network. The neural network is then trained using subsampled (sparse) measurements of the subject to learn a neural representation of a reconstructed image. The training includes minimizing an objective function representing a difference between the subsampled measurements and a forward model applied to predicted image intensity values output from the neural network. Image intensity values output from the trained neural network from coordinates in image space input to the trained neural network are computed to produce predicted image intensity values.

According to one embodiment, a medical image processing apparatus includes processing circuitry. The processing circuitry designates a region of interest in a first tomogram among multiple tomograms which are based on tomosynthesis imaging performed with a subject compressed in a first direction. The processing circuitry specifies a second tomogram corresponding to the region of interest from among multiple tomograms which are based on tomosynthesis imaging performed with the subject compressed in a second direction different from the first direction.

An X-ray imaging system using multiple puked X-ray sources to perform highly efficient and ultrafast 3D radiography is presented. There are multiple puked X-ray sources mounted on a structure in motion to form an array of sources. The multiple X-ray sources move simultaneously relative to an object on a pre-defined arc track at a constant speed as a group. Electron beam inside each individual X-ray tube is deflected by magnetic or electrical field to move focal spot a small distance. When focal spot of an X-ray tube beam has a speed that is equal to group speed but with opposite moving direction, the X-ray source and X-ray flat panel detector are activated through an external exposure control unit so that source tube stay momentarily standstill equivalently. 3D scan can cover much wider sweep angle in much shorter time and image analysis can also be done in real-time.