Siemens Healthcare Gmbh

University of California

Siemens Aktiengesellschaft

Siemens Medical Solutions Usa, Inc.

Siemens Healthineers Ag

Duke University

Emory University

University of Virginia Patent Foundation

Xiaodong Zhong

Marcel Dominik Nickel

Stephan Kannengiesser

Vibhas S Deshpande

Berthold Kiefer

Brian Dale

Xiaoming Bi

Li Pan

Holden H Wu

John N Oshinski

Mustafa R Bashir

Fei Han

Tess Armstrong

Deqiang Qiu

Amit Saindane

Chang Gao

Frederick H Epstein

Peng Hu

Stephan Kannengießer

Xiaoying Cai

Mahesh Bharath Keerthivasan

Jin Jin

John Paul Finn

Valid Ghodrati

A system and methods for recreating images from undersampled magnetic resonance imaging (MRI) image data to remove artifacts from the undersampling. Using trained transformer networks, undersampled k-space data can be acquired and the networks can predict additional projected radial MRI k-space data from the undersampled radial MRI k-space data. Images with fewer artifacts can then be generated.

Method and apparatus for accelerated acquisition and artifact reduction of undersampled MRI using a k-space transformer network

Systems and methods for generative adversarial networks (GANs) to remove artifacts from undersampled magnetic resonance (MR) images are described. The process of training the GAN can include providing undersampled 3D MR images to the generator model, providing the generated example and a real example to the discriminator model, applying adversarial loss, L2 loss, and structural similarity index measure loss to the generator model based on a classification output by the discriminator model, and repeating until the generator model has been trained to remove the artifacts from the undersampled 3D MR images. At runtime, the trained generator model of the GAN can be generate artifact-free images or parameter maps from undersampled MRI data of a patient.

Method and system for accelerated acquisition and artifact reduction of undersampled MRI using a deep learning based 3D generative adversarial network

A computer implemented method of processing a medical image is disclosed. The method includes receiving a medical image comprising a first plurality of pixels each having an initial pixel value. For each of the first plurality of pixels, a filtering operation is applied to the pixel to generate a filtered pixel value for the pixel based on the initial pixel values of pixels that surround the pixel in the medical image. For each of the first plurality of pixels, a comparison of the initial pixel value with the filtered pixel value is performed. The method comprises, for each of the first plurality of pixels, determining, based on the comparison, whether or not to categorize the pixel as an erroneous pixel; and for each of the first plurality of pixels for which it is determined to categorize the pixel as an erroneous pixel, categorizing the pixel as an erroneous pixel.

Method and apparatus for processing a medical image

A method for acquiring magnetic resonance imaging data with respiratory motion compensation using one or more motion signals includes acquiring a plurality of gradient-delay-corrected radial readout views of a subject using a free-breathing multi-echo pulse sequence, and sampling a plurality of data points of the gradient-delay-corrected radial readout views to yield a self-gating signal. The self-gating signal is used to determine a plurality of respiratory motion states corresponding to the plurality of gradient-delay-corrected radial readout views. The respiratory motion states are used to correct respiratory motion bias in the gradient-delay-corrected radial readout views, thereby yielding gradient-delay-corrected and motion-compensated multi-echo data. One or more images are reconstructed using the gradient-delay-corrected and motion-compensated multi-echo data.

Free-breathing MRI with motion compensation

A system and method for performing accelerated k-space shift correction calibration scans for non-Cartesian trajectories is provided. The method can include applying an MRI sequence, performing a calibration scan based on the MRI sequence using the non-Cartesian trajectory to acquire k-space shift data, wherein one or more partitions are skipped during the calibration scan, interpolating the skipped one or more partitions using the k-space shift data from adjacent partitions, and calibrating the MRI system using the k-space shift data and the interpolated k-space shift data. In some embodiments, an acceleration factor Acc can be defined and the calibration scan acquires k-space shift data for only one partition in every Acc partitions.

Method and apparatus for improved efficiency of non-cartesian imaging using accelerated calibration scan for k-space shift correction

A method for using a multi-echo magnetic resonance imaging (MRI) simultaneously quantify T1 and fat fraction in an anatomical region of interest includes performing a radial single shot multi-echo acquisition of the anatomical region of interest. The radial single shot multi-echo acquisition comprises applying a preparation pulse to invert longitudinal magnetization of the anatomical region of interest, and acquiring a plurality of radial readouts at different echo times (TE). A magnetization recovery curve is continuously sampled using the plurality of radial readouts to yield a plurality of radial spokes. The radial spokes for each TE are ground together to generate under-sampled k-space data for each TE. The under-sampled k-space data is reconstructed into a plurality of multi-echo images corresponding to the different echo times. One or more fitting algorithms are applied to the multi-echo images to generate a water-only T1 map and a proton density fat fraction (PDFF) measurement.

Multi-echo radial look-locker imaging for simultaneous T1 and fat fraction quantification

A method of generating biomarker parameters includes acquiring imaging data depicting a patient using a MRI system. The imaging data is acquired for a plurality of contrasts resulting from application of a pulse on the patient's anatomy. A process is executed to generate a MoCoAve image for each contrast. This process includes dividing the imaging data for the contrast into bins corresponding to one of a plurality of respiratory motion phases, and reconstructing the imaging data in each bin to yield bin images. The process further includes selecting a reference bin image from the bin images, and warping the bin images based on the reference bin image. The warped bin images and the reference bin image are averaged to generate the MoCoAve image for the contrast. One or more biomarker parameter maps are calculated based on the MoCoAve images generated for the contrasts.

Quality and signal-to-noise ratio of free-breathing quantitative measurement of magnetic resonance imaging parameters and related biomarkers

A computer-implemented method for using machine learning to suppress fat in acquired MR images includes receiving multi-echo images from an anatomical area of interest acquired using an MRI system. A first subset of the multi-echo images is acquired prior to application of contrast to the anatomical area of interest and a second subset of the multi-echo images is acquired after application of contrast to the anatomical area of interest. Next, data is generated including water images, fat images, and effective R*maps from the multi-echo images. The water images, the fat images, and the effective R*maps are used to create synthetic fat suppressed images. A neural network is trained to use the multi-echo images as input and the synthetic fat suppressed images as ground truth. A plurality of components of the neural network are saved to allow later deployment of the neural network on a computing system.

Artificial intelligence based suppression of chemical species in magnetic resonance imaging

A system for performing magnetic resonance imaging (MRI) of a subject has a pulse sequence system that generates a pulse sequence and has a gradient system, a plurality of gradient coils, a radio-frequency system, and a plurality of RF coils. The pulse sequence system causes the subject to emit MR signals which are captured as k-space data. The system also has a k-space ordering processor that collects first k-space data and second k-space data, an MR signal modeler that generates a signal variation model, and a compensation module that applies the signal variation model to the second k-space data collected to produce compensated k-space data. A display processor reconstructs the compensated k-space data into an image of the subject. The compensated data accounts for variation in magnetization during the pulse sequence and k-space data collection to reduce artifacts in the images.

K-space data correction method for signal variation compensation

A magnetic resonance imaging system and method are provided for improved determination of noise bias effects in calculating fitted parameters for quantitative MRI procedures. The system and method includes selecting a range for the SNR and fitted parameter values, and for each of a plurality of base pairs of these values and for a plurality of b values, adding a random noise term to the real and imaginary components of a plurality of corresponding signal terms, fitting magnitudes of the resulting 'noisy' signals to determine a 'noisy' fitted parameter value, and compare the &#x201c;noisy&#x201d; and base fitted parameter values to determine a noise-based error for each pair of base values. The noise-based errors can be used to generate an error map, modify imaging parameters to reduce such errors, or correct fitted parameters directly.

Systems and methods for predicting errors and optimizing protocols in quantitative magnetic resonance imaging

Disclosed herein is a framework for identifying signal components in image data. In accordance with one aspect, the framework receives multiple measured signal values corresponding to respective quantified signal components in image data. The framework determines at least one first measure of fit map of a signal model based on the measured signal values. The measured signal values may be swapped to generate swapped signal values. At least one second measure of fit map of the signal model may be determined based on the swapped signal values. The multiple signal components may then be identified by comparing the first and second measure of fit maps.

Signal component identification using medical imaging

Embodiments relate to acquiring magnetic resonance (MR) images with suppressed residual blood signal in the early cardiac phases, leading to images with a preferred dark-blood appearance throughout the entire cardiac cycle, which improves accuracy of subsequent post-processing algorithms. The acquisition of the desired blood suppressed tissue images is achieved through a double inversion recovery pulse in DENSE sequences. The double inversion recovery pulse is applied after an electrocardiogram (ECG) trigger at a beginning point of a repetition time period, followed by a displacement encoding module at an inversion time during the repetition time period and a readout module comprised of a plurality of frames during a remainder of the repetition time period. The displacement encoding module applies a labelling process on the tissue, while the readout module applies an un-labelling process. The readout module comprises an imaging sequence adapted to acquire DENSE images.

Magnetic resonance method and apparatus for quantitative time-resolved assessment of tissue displacement and related biomarker parameters with blood suppression in the whole cardiac cycle

Some aspects of the present disclosure relate to systems and methods for free-breathing cine DENSE MRI using self-navigation. In one embodiment, a method includes acquiring magnetic resonance data for an area of interest of a subject, wherein the acquiring comprises performing sampling with phase-cycled, cine displacement encoding with stimulated echoes (DENSE) during free-breathing of the subject; identifying, from the acquired magnetic resonance data, a plurality of phase-cycling data pairs corresponding to matched respiratory phases of the free-breathing of the subject; reconstructing, from the plurality of phase-cycling data pairs, a plurality of intermediate self-navigation images; performing motion correction by estimating, from the plurality of intermediate self-navigation images, the respiratory position associated with the plurality of phase-cycling data pairs; and reconstructing a plurality of motion-corrected cine DENSE images of the area of interest of the subject.

Systems and methods for free-breathing cine DENSE MRI using self-navigation

In a method to correct noise effects in magnetic resonance (MR) images, which is executed in a processor (computer), the processor executes a fitting algorithm in order to calculate initial values for each of selected variables in signal model that models noise effects in a modeled, noise-containing MR image. The processor then iteratively executes the same or a different fitting algorithm, in order to generate final values for each of the selected variables. The processor is provided with an actual, acquired MR image that contains noise, and the processor uses the final values of the selected variables to calculate synthetic signal intensities in the MR image, thereby producing a synthetic MR image with no noise bias effects of errors. This synthetic image is made available in electronic form at an output of the processor, as a data file.

Growing community of inventors

Lilburn, GA, United States of America

Xiaodong Zhong