Siemens Aktiengesellschaft

Sas Institute Inc.

Other

Siemens Healthcare Gmbh

Siemens Corporation

Computer Sciences Corporation

Jun Liu

Mariappan S Nadar

Alban Lefebvre

Michael Zenge

Qiu Wang

Edgar Mueller

Ti-chiun Chang

Nirmal Janardhanan

Zhili Yang

Joshua David Griffin

Yan Xu

Zheng Zhao

Ruiwen Zhang

Jeremy Rapin

Axel Loewe

Marcel Dominik Nickel

James Allen Cox

Michaela Schmidt

Hui Xue

Ya Xue

Edgar Müller

Manoj Keshavmurthi Chari

Nicolas Chesneau

Steven Martin

Marc Oglesby

Nick Tamburro

Betty Lau

An exemplary method, for preventing an outage in a network, includes collecting network metrics and producing statistical features of the network during a predetermined time window based on statistical analysis of the collected metrics. The method further includes selecting, from the produced statistical features, a first data set of relevant statistical features. Further, the method includes collecting information about a past event, extracting event data from the past event, and vectorizing the extracted event data to form a second data set. The method further includes concatenating the first data set and the second data set to form a third data set and classifying the third data set relative to a model of historical network performance to determine a probability of the outage. Further, the method includes modifying, in response to the probability being above a threshold, the network/activity within the network to avoid onset of the outage.

System and method for outage prediction

A method for estimating a coil sensitivity map for a magnetic resonance (MR) image includes providing a matrix A of sliding blocks of a 3D image of coil calibration data, calculating a left singular matrix Vfrom a singular value decomposition of A corresponding to &#x3c4; leading singular values, calculating P=VV, calculating a matrix that is an inverse Fourier transform of a zero-padded matrix P, and solving Mc=(S)cfor c, where cis a vector of coil sensitivity maps for all coils at spatial location r, and 

Eigen-vector approach for coil sensitivity maps estimation

Processing speeds for generating a model can be enhanced. For example, the model can be generated by using regression coefficient values as weights for independent variables in the model. The regression coefficient values can be determined using a coordinate descent method to find a minimum value of a least absolute shrinkage and selection operator cost function. Each iteration of the coordinate descent method can include determining a starting coordinate based on (i) a previous starting coordinate or a previous regression coefficient value from an immediately prior iteration of the coordinate descent method; (ii) a current regression coefficient value associated with a current iteration of the coordinate descent method; and (iii) a refinement factor configured to minimize a result of a univariate algorithm. Each iteration can also include performing a coordinate descent using the starting coordinate to determine a next regression coefficient value for a next iteration of the coordinate descent method.

Enhancing processing speeds for generating a model on an electronic device

A computing device trains models for streaming classification. A baseline penalty value is computed that is inversely proportional to a square of a maximum explanatory variable value. A set of penalty values is computed based on the baseline penalty value. For each penalty value of the set of penalty values, a classification type model is trained using the respective penalty value and the observation vectors to compute parameters that define a trained model, the classification type model is validated using the respective penalty value and the observation vectors to compute a validation criterion value that quantifies a validation error, and the validation criterion value, the respective penalty value, and the parameters that define a trained model are stored to the computer-readable medium. The classification type model is trained to predict the response variable value of each observation vector based on the respective explanatory variable value of each observation vector.

Classification system training

A power method can be enhanced. For example, an electronic communication indicating a job to be performed can be received. A best rank-1 approximation of a matrix associated with the job can be determined using the power method. Each iteration of the power method can include determining a point that lies on a line passing through (i) a first value for a first singular vector from an immediately prior iteration of the power method; and (ii) a second value for the first singular vector from another prior iteration of the power method. Each iteration of the power method can also include determining, by performing the power method using the point, a current value for the first singular vector and a current value for a second singular vector for a current iteration of the power method. The job can then be performed using the best rank-1 approximation of the matrix.

Enhanced power method on an electronic device

A method for parallel magnetic resonance imaging (MRI) reconstruction of digital images includes providing a set of acquired k-space MR image data v, a redundant Haar wavelet matrix W satisfying WW=I, wherein I is an identity matrix, a regularization parameter &#x3bb;&#x2267;0, and a counter limit k, initializing a variable z=Wv, and intermediate quantities p=q=0, calculating y=arg min&#xbd;&#x2225;z&#x2212;(p+z)&#x2225;+&#x3bb;&#x2225;z&#x2225;for 0&#x2266;i&#x2266;k, wherein z denotes values of an MR image sought to be reconstructed, updating p=(p+z)&#x2212;y, updating z=arg min&#xbd;&#x2225;z&#x2212;(q+z)&#x2225;+g(z), wherein and updating q=(q+y)&#x2212;z, wherein x=Wz is a solution of that specifies a reconstruction of the MR image.

Efficient redundant haar minimization for parallel MRI reconstruction

A system for machine training can comprise one or more data processors and a non-transitory computer-readable storage medium containing instructions which, when executed on the one or more data processors, cause the one or more data processors to perform operations including: accessing a dataset comprising data tracking a plurality of features; determining a series of values for a regularization parameter of a sparse support vector machine model, the series including an initial regularization value and a next regularization value; computing an initial solution to the sparse support vector machine model for the initial regularization value; identifying, using the initial solution, inactive features of the sparse support vector machine model for the next regularization value; and computing a next solution to the sparse support vector machine model for the next regularization value, wherein computing the next solution includes excluding the inactive features.

Acceleration of sparse support vector machine training through safe feature screening

A method of image reconstruction for a magnetic resonance imaging (MRI) system includes obtaining k-space scan data captured by the MRI system, the k-space scan data being representative of an undersampled region over time, iteratively reconstructing preliminary dynamic images for the undersampled region from the k-space scan data via optimization of a first instance of a minimization problem, the minimization problem including a regularization term weighted by a weighting parameter array, generating a motion determination indicative of an extent to which each location of the undersampled region exhibits motion over time based on the preliminary dynamic images, and iteratively reconstructing motion-compensated dynamic images for the region from the k-space scan data via optimization of a second instance of the minimization problem, the second instance having the weighting parameter array altered as a function of the motion determination.

MRI reconstruction with motion-dependent regularization

A system for performing image reconstruction in a multi-threaded computing environment includes one or more central processing units executing a plurality of k-space components and a plurality of graphic processing units executing a reconstruction component. The k-space components executing on the central processing units include a k-space sample data component operating in a first thread and configured to receive k-space sample data from a first file interface; a k-space sample coordinate data component operating in a second thread and configured to receive k-space sample coordinate data from a second file interface; and a k-space sample weight data component operating in a third thread and configured to retrieve k-space sample weight data from a third file interface. The reconstruction component is configured to receive one or more k-space input data buffers comprising the k-space sample data, the k-space sample coordinate data, and the k-space sample weight data from the one or more central processing units, and reconstruct an image based on the input data buffers using an iterative reconstruction algorithm.

Multi-GPU FISTA implementation for MR reconstruction with non-uniform K-space sampling

A computer-implemented method for learning a tight frame includes acquiring undersampled k-space data over a time period using an interleaved process. An average of the undersampled k-space data is determined and a reference image is generated based on the average of the undersampled k-space data. Next, a tight frame operator is determined based on the reference image. Then, a reconstructed image data is generated from the undersampled k-space data via a sparse reconstruction which utilizes the tight frame operator.

Dynamic image reconstruction with tight frame learning

A method of image reconstruction for a magnetic resonance imaging (MRI) system having a plurality of coils includes obtaining k-space scan data captured by the MRI system, the k-space scan data being representative of an undersampled region over time, determining a respective coil sensitivity profile for the region for each coil of the plurality of coils, and iteratively reconstructing dynamic images for the region from the k-space scan data via an optimization of a minimization problem. The minimization problem is based on the determined coil sensitivity profiles and redundant Haar wavelet transforms of the dynamic images.

MRI reconstruction with incoherent sampling and redundant haar wavelets

A computer-implemented method for calculating a multi-dimensional wavelet transform in an image processing system comprising a plurality of computation units includes receiving multi-dimensional image data. An overlap value corresponding to a number of non-zero filter coefficients associated with the multi-dimensional wavelet transform is identified. Then the multi-dimensional image data is divided into a plurality of multi-dimensional arrays, wherein the multi-dimensional arrays overlap in each dimension by a number of pixels equal to the overlap value. A multi-dimensional wavelet transform is calculated for each multi-dimensional array, in parallel, across the plurality of computation units.

Zero communication block partitioning

A computer-implemented method for reconstruction of a magnetic resonance image includes acquiring a first incomplete k-space data set comprising a plurality of first k-space lines spaced according to an acceleration factor and one or more calibration lines. A parallel imaging reconstruction technique is applied to the first incomplete k-space data to determine a plurality of second k-space lines not included in the first incomplete k-space data set, thereby yielding a second incomplete k-space data set. Then, the parallel imaging reconstruction technique is applied to the second incomplete k-space data to determine a plurality of third k-space lines not included in the second incomplete k-space data, thereby yielding a complete k-space data set.

Multi-stage magnetic resonance reconstruction for parallel imaging applications

A method for image reconstruction includes receiving under-sampled k-space data, determining a data fidelity term of a first image of the under-sampled k-space data in view of a second image of the under-sampled k-space data, wherein a time component separated the first image and the second image, determining a spatial penalization on redundant Haar wavelet coefficients of the first image in view of the second image, and optimizing the first image according the data fidelity term and the spatial penalization, wherein the spatial penalization selectively penalizes temporal coefficients and an optimized image of the first image is output.

Image reconstruction using redundant Haar wavelets

A method for reconstructing parallel magnetic resonance images includes providing a set of acquired k-space MR image data y, and finding a target MR image x that minimizes &#xbd;&#x2225;Fv&#x2212;y&#x2225;+&#x3bb;&#x2225;z&#x2225;where v=Sx and z=Wx where S is a diagonal matrix containing sensitivity maps of coil elements in an MR receiver array, F is an FFT matrix, W is a redundant Haar wavelet matrix, and &#x3bb;&#x2267;0 is a regularization parameter, by updating where k is an iteration counter, &#x3bc;and &#x3bc;are parameters of an augmented Lagrangian function, and band bare dual variables of the augmented Lagrangian.

Growing community of inventors

Cary, NC, United States of America

Jun Liu