The General Hospital Corporation

Siemens Aktiengesellschaft

Siemens Healthcare Gmbh

National Institutes of Health, a Component of the US Dept. of Health & Human Services

The Brigham and Women's Hospital, Inc.

Siemens Corporation

Jonathan R Polimeni

Kawin Setsompop

Lawrence Leroy Wald

Leo John Grady

Thomas Witzel

William Scott Hoge

Himanshu Bhat

Stephen Farman Cauley

Bruce Fischl

Yan Zhang

Kun Zhou

William Allyn Grissom

Vijayanand Alagappan

Uvo Hölscher

Vivek K Goyal

Berkin Bilgic

Keith Aaron Heberlein

Andre Van Der Kouwe

Matthew Tisdall

Qiyuan Tian

Susie Yi Huang

Daniel Weller

Herve Jun Alfred Lombaert

Farida Cheriet

Yulin Chang

Olivia M Viessmann

Avery J L Berman

Described here are systems and methods for producing images with a magnetic resonance imaging ('MRI') system using a high-resolution, motion-robust, artifact-free segmented echo planar imaging (“EPI”) technique. In particular, a fast low angle excitation echo planar imaging technique (“FLEET”) using variable flip angle (“VFA”) radio frequency (“RF”) excitation pulses that are recursively designed to have a flat magnitude and phase profile across a slice for a range of different flip angles by accounting for longitudinal magnetization remaining after each preceding RF pulse is applied.

Multi-shot echo planar imaging using reordered segments and recursive radio frequency pulse design giving matched slice profiles across segments

Anatomical, physiological, instrumental, and other related biases are removed from functional magnetic resonance imaging ('fMRI') signal data using deep learning algorithms and/or models, such as a neural network. Bias characterization data are used as an auxiliary input to the neural network. The bias characterization data can be subject-specific bias characterization data (e.g., cortical thickness maps, cortical orientation angle maps, vasculature maps), hardware-specific bias characterization data (e.g., coil sensitivity maps, coil transmission profiles), or both. The subject-specific bias characterization data can be extracted from the fMRI signal data using a second neural network. The bias-reduced fMRI signal data can include time-series signals, functional activation maps, functional connectivity maps, or combinations thereof.

Subject-specific and hardware-specific bias removal from functional magnetic resonance imaging signals using deep learning

Super-resolution images are generated from standard-resolution images acquired with a magnetic resonance imaging ('MRI') system. More particularly, super-resolution (e.g., sub-millimeter isotropic resolution) images are generated from standard-resolution images (e.g., images with 1 mm or coarser isotropic resolution) using a deep learning algorithm, from which accurate cortical surface reconstructions can be generated.

Super-resolution anatomical magnetic resonance imaging using deep learning for cerebral cortex segmentation

Systems and methods for performing fast multi-contrast magnetic resonance imaging ('MRI') are provided. In general, data are acquired from both multiple echo times ('TEs') and at multiple effective inversion times (&#x201c;TIs&#x201d;). Following the application of a magnetization preparation radio frequency (&#x201c;RF&#x201d;) pulse, a plurality of different multi-echo acquisitions are performed, thereby acquiring data from multiple different TEs during different portions of the longitudinal magnetization recovery curve. Data acquisition in these inner encoding loops (i.e., during each multi-echo acquisition) can be accelerated to efficiently provide for the acquisition of multiple contrasts in the time normally required to acquire a single contrast.

System and methods for fast multi-contrast magnetic resonance imaging

Systems and methods include conversion of a first frame of k-space data acquired using a first initial readout polarity to first hybrid (k, y)-space data, conversion of a second frame of k-space data acquired using a second initial readout polarity to second hybrid (k, y)-space data, determination of a relationship between phase difference and y-position based on phase differences between a plurality of pixels located at k=a of first hybrid (k, y)-space data and a plurality of pixels at k=b of second hybrid (k, y)-space data, where a and b are constants, modification of the second hybrid (k, y)-space data based on the relationship, conversion of the modified second hybrid (k, y)-space data to a modified second frame of k-space data, generation of two single-polarity readout k-space frames based on the first frame of k-space data and the modified second frame of k-space data, and correction of a third frame of EPI image data based on the two single-readout polarity k-space frames.

Phase correction for echo-planar imaging

A method and imaging system is provided that can control a magnetic gradient system and an RF system of an MRI system according to a calibration pulse sequence to acquire positive readout gradient (RO+) data and negative readout gradient (RO&#x2212;) data. The RO+ data and the RQ&#x2212; data are assembled to form complete image data sets for the RO+ data and the RQ&#x2122; data and the RO+ data and the RO&#x2212; data are combined to generate the calibration data that is ghost-corrected, substantially free of ghost artifacts, or having reduced ghost artifacts compared to traditionally-acquired calibration data. Reconstruction coefficients are derived from the calibration data. The magnetic gradient system and the RF system are controlled according to an imaging pulse sequence to acquire image data and the image data is reconstructed into an image of the subject using the reconstruction coefficients.

System and method for dual-kernel image reconstruction

A method for maximizing the signal-to-noise ratio ('SNR') in a combined image produced using a parallel magnetic resonance imaging ('MRI') technique is provided. The image combination used in such techniques require an accurate estimate of the noise covariance. Typically, the thermal noise covariance matrix is used as this estimate; however, in several applications, including accelerated parallel imaging and functional MRI, the noise covariance across the coil channels differs substantially from the thermal noise covariance. By combining the individual channels with more accurate estimates of the channel noise covariance, SNR in the combined data is significantly increased. This improved combination employs a regularization of noise covariance on a per-voxel basis.

Method for increasing signal-to-noise ratio in magnetic resonance imaging using per-voxel noise covariance regularization

Systems and methods for a hierarchical mapping framework ('HMF'} for coil compression are provided. The HMF-based coil compression can be applied to existing coil compression algorithms to improve their performance. The receive channels associated with a coil array are divided into subgroups based on the strength of their mutual correlation. In each subgroup, one or more virtual channels are produced based on the channels not in the subgroup. The virtual channels are produced using a coil compression algorithm subject to a hierarchically semiseparable channel mixing across the subgroups. Images are reconstructed for the subgroups and then combined to produce the final image of the subject.

Hierrarchical mapping framework for coil compression in magnetic resonance image reconstruction

Described here are systems and methods for producing images with a magnetic resonance imaging ('MRI') system using a high-resolution, motion-robust, artifact-free segmented echo planar imaging ('EPI') technique. In particular, a fast low angle excitation echo planar imaging technique (&#x201c;FLEET&#x201d;) using variable flip angle (&#x201c;VFA&#x201d;) radio frequency (&#x201c;RF&#x201d;) excitation pulses that are specifically designed to have a flat magnitude and phase profile across a slice for a range of different flip angles.

Multi-shot echo planar imaging using reordered segments and RF excitation pulse phase and slice profiles matched across interleaves

Systems and methods for combined ghost artifact correction and parallel imaging reconstruction of simultaneous multislice ('SMS') magnetic resonance imaging ('MRI') data are provided. Dual-polarity training data are used to generate ghost-free slice data, which are used as target data in a reconstruction kernel training process. The training data are used as source data in the reconstruction kernel training. As a result, reconstruction kernels are computed, which can be used to reconstruct images from SMS data in which slice-specific ghosting artifacts are removed.

System and method for reconstructing ghost-free images from data acquired using simultaneous multislice magnetic resonance imaging

A system and method for medical imaging using a magnetic resonance imaging system includes performing a segmented echo planar imaging (EPI) pulse sequence. The pulse sequence includes performing multiple radio frequency (RF) excitation pulses designed to excite multiple imaging slices across the subject simultaneously. A gradient encoding scheme applied along the slice-encoding direction is implemented to impart controlled phase shifts to the different imaging slices. Additionally, the multiple RF excitation pulses can be designed to further control an overlap of imaging data acquired from adjacent slices of the multiple imaging slices based on a selected offset. The acquired imaging data is reconstructed using a parallel imaging reconstruction method that separates overlapped slices in the imaging data to provide a series of images with respective images for each of the multiple imaging slices across the subject.

System and method for rapid, multi-shot segmented magnetic resonance imaging

A method for accelerated segmented magnetic resonance (MR) image data acquisition includes using a plurality of RF pulses to excite one or more slices of an anatomical area of interest according to a predetermined slice acceleration factor. Next, a collapsed image comprising the slices is acquired using a consecutive segment acquisition process. Then, a parallel image reconstruction method is applied to the collapsed image to separate the collapsed image into a plurality of slice images.

Methods, systems and apparatuses for rapid segmented, accelerated, and simultaneous multi-slice echo planar imaging

A system for accelerated segmented magnetic resonance (MR) image data acquisition includes an RF (Radio Frequency) signal generator and a magnetic field gradient generator. The RF signal generator generates RF excitation pulses in anatomy and enabling subsequent acquisition of associated RF echo data. The magnetic field gradient generator generates magnetic field gradients for anatomical volume selection, phase encoding, and readout RF data acquisition in a three dimensional (3D) anatomical volume. The RF signal generator and the magnetic field gradient generator acquire consecutive segments of k-space line data representative of an individual image slice in a gradient echo method by adaptively varying RF excitation pulse flip angle between acquisition of the consecutive segments.

System for accelerated segmented MR image data acquisition

Methods and systems determine a correspondence of two sets of data, each data set represents an object. A weighted graph is created from each data set, and a Laplacian is determined for each weighted graph, from which spectral components are determined. The spectral components determine a coordinate of a node in a weighted graph. Nodes of a weighted graph are weighted with a quantified feature related to anode. A coordinate related to a quantified feature of a node is added to the coordinate based on the spectral components. Spectral components related to a weighted graph are reordered to a common ordering. Reordered spectral components related to the first and second data set are aligned and a correspondence is determined. An object may be a brain and a feature may be a sulcal depth. Other objects for which a correspondence may be determined include an electrical network, an image and a social network.

Methods and systems for fast automatic brain matching via spectral correspondence

An MR imaging system uses the multiple RF coils for acquiring corresponding multiple image data sets of the slice. An image data processor comprises at least one processing device conditioned for, generating a composite MR image data set representing a single image in a single non-iterative operation by performing a weighted combination of luminance representative data of individual corresponding pixels of the multiple image data sets in providing an individual pixel luminance value of the composite MR image data set. The image data processor reduces noise in the composite MR image data set by generating a reduced set of significant components in a predetermined transform domain representation of data representing the composite image to provide a de-noised composite MR image data set. An image generator comprises at least one processing device conditioned for, generating a composite MR image using the de-noised composite MR image data set.

Growing community of inventors

Cambridge, MA, United States of America

Jonathan R Polimeni