Siemens Healthcare Gmbh

Siemens Aktiengesellschaft

Siemens Healthineers Ag

The Johns Hopkins University

Siemens Aktiengesellchaft

Robert Grimm

Berthold Kiefer

Heinrich Von Busch

Ali Kamen

Marcel Dominik Nickel

Bin Lou

Mamadou Diallo

David Jean Winkel

Xin Yu

Tongbai Meng

Dorin Comaniciu

Atilla Peter Kiraly

Simon Bauer

Bibo Shi

Nicolas Von Roden

Dominik Paul

Shaohua Kevin Zhou

Mariappan S Nadar

Boris Mailhe

Sasa Grbic

Paul Arthur Bottomley

Benjamin L Odry

David D Liu

Arne Hengerer

Zhoubing Xu

Lars Lauer

Bernd Schweizer

Eva Rothgang

Thomas Benkert

Thomas Pheiffer

Wesley David Gilson

Guo Li

Christian Geppert

Jan Fritz

Ahmet Tuysuzoglu

Li Pan

Nirmal Janardhanan

Guillaume Chabin

Jyotipriya Das

Sajan Goud Lingala

Kareem Abdelrahman

Matteo Barbieri

Manasi Datar

Alin Madalin Draghia

Ioan Marius Popdan

Ihab Kamel

Sebastian Fürst

Isabel Dregely

Clement Jad Abi Nader

Systems and methods are provided for optimizing a deep learning model. A multi-site dataset associated with different clinical sites and a deployment dataset associated with a deployment clinical site are received. A deep learning model is trained based on the multi-site dataset. The trained deep learning model is optimized based on the deployment dataset. The optimized trained deep learning model is output.

Localization and classification of abnormalities in medical images

Systems and methods for performing an assessment of a lesion are provided. A plurality of input medical images of a lesion is received. The plurality of input medical images comprises an initial input medical image and one or more additional input medical images. The initial input medical image comprises a region of interest around the lesion. A mask of the lesion is curated for the initial input medical image based on the region of interest and a set of candidate masks. The region of interest in the initial input medical image is propagated to the one or more additional input medical images based on prior registration transformations. A mask of the lesion is curated for each of the one or more additional input medical images based on the propagated regions of interest and the set of candidate masks. One or more assessments of the lesion are performed based on the mask for the initial input medical image, the masks for the one or more additional input medical images, and prior assessments of lesions. Results of the one or more assessments of the lesion are output.

AI driven longitudinal liver focal lesion analysis

In a method for an actuation of a magnetic resonance device for capturing image data from an examination region of an examination object, at least one first control sequence is provisioned, the magnetic resonance device is actuated according to the at least one first control sequence to capture first data from the examination object, the first data is analyzed with respect to a property to generate a result, and, based on the result, a selective performance of one of: a selection of a further control sequence, and termination of the actuation of the magnetic resonance device, is performed.

Method for an adaptive actuation of a magnetic resonance device

In a method for generating a surrogate marker based on medical image data mapping an image region, the medical image data is detected using a first interface, a first subregion of the image region is selected by segmenting a first structure included in the image region, a first property of the first subregion is extracted, the surrogate marker is determined based on the first property, and the surrogate marker is provided using a second interface.

Methods and systems for generating surrogate marker based on medical image data

In a computer-implemented method of training a machine learning based processor, the processor can be trained to derive image data from signal data sets of multiple spin echo sequences. The trained processor can be configured to perform image processing for Magnetic Resonance Imaging (MRI) to derive the image data.

Trained image processing for diffusion weighted imaging and/or turbo spin echo sequences with focus on body applications

Systems and methods are provided for classifying an abnormality in a medical image. An input medical image depicting a lesion is received. The lesion is localized in the input medical image using a trained localization network to generate a localization map. The lesion is classified based on the input medical image and the localization map using a trained classification network. The classification of the lesion is output. The trained localization network and the trained classification network are jointly trained.

For radial sampling in magnetic resonance imaging (MRI), a rescaling factor is determined from k-space data for each coil. The rescale factor is inversely proportional to the streak energy in the k-space data. The k-space data from the coils is rescaled for reconstruction, such as weighting the k-space data by the rescale factor in a data consistency term of iterative reconstruction. The rescale factor is additionally or alternatively used to determine a correction field for correction of intensity bias applied to intensities in the image-object space after reconstruction. These approaches may result in a diagnostically useful bias-corrected image with reduced streak artifact while benefiting from the efficient computation (i.e., computer operates to reconstruct more quickly).

Streak artifact reduction in magnetic resonance imaging

Tissue is characterized using machine-learnt classification. The prognosis, diagnosis or evidence in the form of a similar case is found by machine-learnt classification from features extracted from frames of medical scan data. The texture features for tissue characterization may be learned using deep learning. Using the features, therapy response is predicted from magnetic resonance functional measures before and after treatment in one example. Using the machine-learnt classification, the number of measures after treatment may be reduced as compared to RECIST for predicting the outcome of the treatment, allowing earlier termination or alteration of the therapy.

Tissue characterization based on machine learning in medical imaging

A method is disclosed for determining result data based upon medical measurement data of an examination object. Within the method, a high-dimensional first parameter space is formed, in which measurement values of the various measurements are represented with the aid of value tuples. The measurement values of the various measurements are assigned to a value tuple based on their spatial arrangement in the examination object and/or on their temporal arrangement relative to one another. In the first parameter space, the value tuples are analyzed, using at least one mapping function to at least one further parameter space including a lower dimension than the first parameter space, in order to obtain result data. Furthermore, the result data is output, preferably visualized. In addition, a corresponding device for determining result data is described.

Determination of result data on the basis of medical measurement data from various measurements

A magnetic resonance method and system are provided for magnetic resonance (MR) image-guided insertion of an object into a biological tissue along a predetermined trajectory. The trajectory provides a path between a starting point and a target site within the tissue. Sufficiently high resolution images can be generated in real time to precisely guide the needle placement. A compressed sensing approach is used to generate the images based on minimization of a cost function, where the cost function is based on the predetermined needle path, artifact effects associated with the needle, the negligible changes in the images away from the trajectory, and the limited differences between successive images. The improved combination of spatial and temporal resolution facilitates an insertion procedure that can be continuously adjusted to accurately follow a predetermined trajectory in the tissue, without interruptions to obtain verification images.

System and method for real-time MRI-guided object navigation

In a method and apparatus for generating a two-dimensional projection image from a three-dimensional image data set of an examination object, a three-dimensional image data set of the examination object having a number of voxels is provided to a computer wherein one scan value is associated with one voxel of the examination object. The computer determines at least one value from voxels respectively located along a viewing direction of the three-dimensional image data set, and their scan values, using at least one selection criterion. The at least one determined value is a value that differs from an upper limit value and a lower limit value of the scan values of the voxels located along the viewing direction. The at least one value is used for generating an image point in the two-dimensional projection image.

Method and apparatus for generating a two-dimensional projection image from a three-dimensional image data set

For the creation of a magnetic resonance (MR) image of a predetermined volume section of an object of investigation, MR data of the volume section are collected along radial trajectories, and each trajectory is assigned a readout direction in which the MR data are collected along the respective trajectory. The MR image is reconstructed on the basis of the collected MR data. In the reconstruction, only MR data are taken into consideration that were collected along trajectories with a readout direction restricted to the field of a solid angle that is defined by a partial surface of a sphere. The partial surface (A) at most corresponds to 75% of the total surface of the sphere.

Method and apparatus for generating a magnetic resonance image with radio collection of magnetic resonance data to avoid image artifacts

A method and apparatus for automated prostate tumor detection and classification in multi-parametric magnetic resonance imaging (MRI) is disclosed. A multi-parametric MRI image set of a patient, including a plurality of different types of MRI images, is received. Simultaneous detection and classification of prostate tumors in the multi-parametric MRI image set of the patient are performed using a trained multi-channel image-to-image convolutional encoder-decoder that inputs multiple MRI images of the multi-parametric MRI image set of the patient and includes a plurality of output channels corresponding to a plurality of different tumor classes. For each output channel, the trained image-to image convolutional encoder-decoder generates a respective response map that provides detected locations of prostate tumors of the corresponding tumor class in the multi-parametric MRI image set of the patient.

Deep convolutional encoder-decoder for prostate cancer detection and classification

In a method and system for generating a magnetic resonance image, a raw data record with non-Cartesian recorded data points is provided to a processor, and the recorded data points are gridded on a Cartesian grid to form transformed data points, wherein the number of transformed data points is selected in at least one k-space direction such that the field of view is reduced in size, and Fourier transformation of the transformed data points is implemented to form a magnetic resonance image with a field of view as the raw data record that is smaller in at least one direction.

Method and system for generating a magnetic resonance image

Growing community of inventors

Nuremberg, Germany

Robert Grimm