Koninklijke Philips Corporation N.v.

Tsinghua University

Beijing Baidu Netcom Science Technology Co., Ltd

Invivo Corporation

Ca, Inc.

Mri Devices Corp.

Iovivo Corporation

Feng Huang

Wei Lin

G Randy Duensing

Hua Guo

George Randall Duensing

Lei Nie

Zhigang Wu

Ming Yang

Jianfa Zou

Zheng Wang

Miha Fuderer

Wenchuan Wu

Arne Reykowski

Max Oskar Köhler

Chun Liang Yuan

Peixue Jiang

Charles Albert Saylor

Ruina Xu

Yunmei Chen

Ye Su

Jinbao Zhu

Chenguang Zhao

Enhao Gong

Yabo Liu

Zhe Zhang

Zhongbiao Wu

Bi Da Bd Zhang

Yu Li Huang

Xiaodong Ma

Jukka Iimari Tanttu

Yunzhen Lin

Nannan Cao

James Akao

Bida Zhang

The present application discloses a method and apparatus for digitizing paper data, an electronic device and a storage medium, relating to fields of image processing and cloud computing, in particular to image recognition technologies. The method includes: determining a standard template according to an image to be processed and mark information corresponding to the image to be processed, wherein the image to be processed is obtained by photographing paper data and the standard template is used to represent a reference coordinate system of the image to be processed; recognizing graphic handwriting information comprised in the image to be processed; and generating digitized data corresponding to the image to be processed according to the graphic handwriting information and the standard template.

Method and apparatus for digitizing paper data, electronic device and storage medium

A product defect detection method and apparatus, an electronic device, and a storage medium are provided. A method includes: acquiring a multi-channel image of a target product; inputting the multi-channel image to a defect detection model, wherein the defect detection model includes a plurality of convolutional branches, a merging module and a convolutional headbranch; performing feature extraction on each channel in the multi-channel image by using the plurality of convolutional branches, to obtain a plurality of first characteristic information; merging the plurality of first characteristic information by using the merging module, to obtain second characteristic information; performing feature extraction on the second characteristic information by using the convolutional headbranch, to obtain third characteristic information to be output by the defect detection model; and determining defect information of the target product based on the third characteristic information.

Product defect detection method and apparatus, electronic device and storage medium

The present invention provides an enhanced oil recovery method by nanofluid-assisted COhuff-n-puff. A nanofluid is injected into an injection well in an oil reservoir formation having hydrophilic wetting characteristics, and soaking is performed; in the soaking stage, after the nanofluid tends to replace crude oil to occupy dead-ends of fractures and pores, and liquid COis injected; and in the process of production, COoriginally dissolved in the formation fluid reaches a supersaturated state. As a result, COpreferentially nucleates at the nanofluid side to generate CObubbles, grows and expands with the continuous reduction of formation pressure, and pushes crude oil near the dead-ends of fractures and pores to move outward, thereby improving the recovery efficiency.

Enhanced oil recovery method by nanofluid-assisted COhuff-n-puff

The present invention provides a magnetic resonance imaging system for imaging a subject by a multi-shot imaging. The magnetic resonance imaging system comprises an acquiring unit for acquiring MR raw data corresponding to a plurality of shots; an imaging unit for generating a plurality of folded images from the MR raw data, wherein each of the plurality of folded images is generated from a subset of the MR raw data; a deriving unit for deriving magnitude of each pixel of each folded image; a detecting unit for detecting a motion of the subject during the multi-shot imaging based on similarity measurements of any two folded images of the plurality of folded images, wherein the detecting unit further comprises a first deriving unit configured to derive the measured similarities; and a reconstructing unit for reconstructing a MR image of the subject based on MR raw data obtained according to a detection result of the detecting unit. Since the partially acquired MR raw data is used for motion detection directly, it would be more rapid and stable.

Mutli-shot magnetic resonance imaging system and method

A method and apparatus for magnetic resonance (MR) diffusion tensor imaging (DTI) with minimizes artifacts caused by motion-induced changes of the diffusion encoding. An object () is subject to at least one shot of a multi-echo imaging sequence to generate a train of echo signals. The multi-echo imaging sequence includes diffusion gradients, acquiring the echo signals for different diffusion gradients, detecting motion of the object () during acquisition of the echo signals, determining a plurality of motional states of the object () based on the detected motion, wherein the object () is stationary in each detected motional state, determining directions of the diffusion gradients relative to a reference frame of the object () for detected motional states of the object (), respectively, and deriving diffusion coefficients from the acquired echo signals for the determined relative directions of the diffusion gradients.

DTI with correction of motion-induced diffusion gradient inconsistency

Embodiments of the present invention provide a method for generating a synthetic diffusion image. The method comprises the steps of acquiring multiple sets of initial diffusion scan data by means of diffusion weighted magnetic resonance scans at multiple initial b-values, deriving an initial diffusion image from at least part of the initial diffusion scan data, acquiring target diffusion scan data by means of a diffusion weighted magnetic resonance scan at a target b-value higher than each of the initial b-values, and generating the synthetic diffusion image by performing a cost function based reconstruction to apply a fidelity term in k space to the synthetic diffusion image based on at least the initial diffusional image, and the target diffusion scan data.

Diffusion MRI method for generating a synthetic diffusion image at a high B-value

The invention provides a device for producing a magnetic resonance image. The device comprises a collecting unit adapted for collecting k space samples that are located on a non-Cartesian grid in a non-Cartesian coordinate space or non-uniformly located on a Cartesian grid in a Cartesian coordinate space, the non-Cartesian/Cartesian coordinate space adapted to be decomposed into a plurality of subspaces, each of which extends along one or more coordinate axes of the non-Cartesian/Cartesian coordinate space; a weight calculation unit () adapted for calculating a decomposed sample density weight function of the k space samples in each individual subspace and determining a density weight function in the non-Cartesian/Cartesian coordinate space via a multiplication of the decomposed sample density weight functions of the k space samples in the plurality of subspaces; a gridding unit adapted for interpolating the k space samples in the non-Cartesian/Cartesian coordinate space onto a uniform Cartesian grid in a k space based on the density weight function of the k space samples in the non-Cartesian/Cartesian coordinate space; and a reconstructing unit adapted for reconstructing the magnetic resonance image based on the interpolated k space samples. By means of the invention, fast, preferably on-demand, calculation of the sampling density weights may be achieved.

MRI with variable density sampling

A magnetic resonance (MR) imaging (MRI) system (), includes at least one controller () configured to: perform a multi-shot image acquisition process to acquire MR information for at least one multi-shot image set; train a convolution kernel including data on at least a portion of the MR information obtained without the use of the gradient or by using a self-training process. The convolution kernel includes convolution data. The MR information obtained with the use of a gradient for at least two of the image shots of the at least one multi-shot image set is iteratively convolved with the trained convolution kernel. The synthetic k-space data for the at least two image shots of the at least one multi-shot image set is projected into image space. The projected synthetic k-space data that are projected into the image space to form image information.

Multi-shot magnetic-resonance (MR) imaging system and method of operation thereof

A magnetic resonance (MR) imaging system may include at least one controller which may acquire echo information of a region of interest (ROI). The echo information may include first image information suitable for reconstructing at least part of a first image at a selected contrast. The MR imaging system can obtain previously-reconstructed image information of one or more previously-reconstructed images having a contrast different than the selected contrast; extract information from the previously-reconstructed image information; determine spatially adaptive regularization weights for regularized reconstruction based upon the extracted information; and/or reconstruct the first image in formation in accordance with the spatially adaptive regularization weights and the echo information.

MRI using spatially adaptive regularization for image reconstruction

This invention provides an image processing method and apparatus for reconstructing an image of a site of a subject. The image processing method comprises: generating a first estimated phase of each shot of multiple shots of the site by sensitivity encoding based on shot data of the corresponding shot obtained for the site; reconstructing a first multiplexed sensitivity encoded image based on the first estimated phase of each shot and shot data of each shot; generating a second estimated phase of each shot of the multiple shots by regularized sensitivity encoding using the first multiplexed sensitivity encoded image; and reconstructing a second multiplexed sensitivity encoded image based on the second estimated phase of each shot and shot data of each shot. In this way, the quality of the finally reconstructed image in magnetic resonance imaging is further improved.

Image reconstruction for MRI using multiplexed sensitivity encoding

An imaging system () includes a plurality of coil channel receivers () and one or more processors or modules (). The plurality of coil channel receivers () demodulate magnetic resonance data from a multi-channel coil () which includes a plurality of coil elements () spatially separated, each element transmitting magnetic resonance data on a corresponding channel (). The one or more processors or modules () are configured to detect () artifacts in the magnetic resonance data on each channel individually. The one or more processors or modules () are further configured to select () the magnetic resonance data from the channels which include detected artifacts at or below a threshold artifact level and reconstruct () one or more images using the selected magnetic resonance data.

Channel by channel artifact reduction in parallel MRI

Magnetic resonance (MR) calibration data are acquired using a plurality of radio frequency receive coils, and both coil sensitivity maps and reference projection vectors are generated based on the MR calibration data. During imaging, extra navigator projection vectors are acquired, or part of the imaging data can be used as navigator projection vectors. Partially parallel imaging (PPI) can performed to enhance the navigation information. The navigator projection vectors and the reference projection vectors are sensitivity weighted using the coil sensitivity maps to generate navigator sensitivity weighted projection vectors (navigator SWPV) and reference sensitivity weighted projection vectors (reference SWPV) respectively, and these are compared to generate subject position information. The subject motions are compensated prospectively or retrospectively using the generated subject position information. The motion compensation may be prospective, performed by adjusting an imaging volume of the PPI based on the subject position information.

Parallel MRI method using calibration scan, coil sensitivity maps and navigators for rigid motion compensation

A magnetic resonance scanner () is configured for thermographic imaging. One or more processors () receive () thermal image data from the magnetic resonance scanner and reconstruct at least one thermal image in which each voxel includes a measure of temperature change. The one or more processors identify () thermally abnormal voxels. A display () displays at least one reconstructed image with the identified abnormal thermal locations.

Magnetic resonance thermography: high resolution imaging for thermal abnormalities

A motion-corrected magnetic resonance imaging method comprises: sequentially acquiring a plurality of interleaved magnetic resonance radial acquisition datasets using a magnetic resonance scanner; reconstructing each magnetic resonance radial acquisition dataset into a corresponding image to generate a set of images, the reconstructing including expanding radial k-space lines of the magnetic resonance radial acquisition dataset into corresponding radial bands in k-space using a generalized auto-calibrating partially parallel acquisition (GRAPPA) operator; selecting a reference image from the set of images; performing three-dimensional spatial registration of each image of the set of images except the reference image with respect to the reference image to generate a spatially registered set of images; and combining the spatially registered set of images to generate a motion corrected image.

MRI method for retrospective motion correction with interleaved radial acquisition

An imaging method comprises acquiring an undersampled magnetic resonance partially parallel imaging (MR-PPI) dataset using a plurality of radio frequency receive coils and reconstructing the undersampled MR-PPI dataset to generate a reconstructed magnetic resonance (MR) image. The reconstructing includes: (i) using a generalized auto-calibrating partially parallel acquisition (GRAPPA) operator or direct convolution to fill in at least some missing data of the undersampled MR-PPI dataset so as to generate an enhanced dataset; and (ii) using an algorithm other than a GRAPPA operator and other than direct convolution to reconstruct the enhanced dataset or to reconstruct the undersampled MR-PPI dataset using the enhanced dataset as an initialization dataset for an iterative reconstruction algorithm. In some embodiments the MR-PPI dataset is a non-Cartesian dataset and a GRAPPA operator for wider radial bands (GROWL) is used in the operation (i).

Growing community of inventors

Gainesville, FL, United States of America

Feng Huang