Koninklijke Philips Corporation N.v.

Tom Brosch

Rafael Wiemker

Axel Saalbach

Tim Philipp Harder

Jochen Peters

Jens Von Berg

Fabian Wenzel

Karsten Sommer

Rolf Jürgen Weese

Hrishikesh Narayanrao Deshpande

Rolf Juergen Weese

Michael Grass

Juergen Weese

Tobias Klinder

Thomas Buelow

Stewart Young

Liran Goshen

Alexandra Groth

Christopher Stephen Hall

Jochen Keupp

Claas Bontus

Thomas Blaffert

Ingmar Graesslin

Hannes Nickisch

Hans Barschdorf

Tanja Nordhoff

André Goossen, Sr

Dominique Siu Li Tio

Cheryl Kimberley Sital

André GOOßEN

A computer implemented method of making a measurement associated with a feature of interest in an image. The method comprises using () a model trained using a machine learning process to take the image as input and predict a pair of points between which to make the measurement of the feature of interest in the image. The method then comprises determining () the measurement, based on the predicted pair of points.

Making measurements in images

The present invention relates to multispectral imaging. In order to improve an identification of relevant multispectral material transitions (in particular caused by injected contrast agent), an apparatus is proposed to use the local maxima of the variances and/or covariances of the intensities of the multi-channel images to locate material transitions. In comparison to gradient vectors, the local variance is not directed and not prone to noise. An alternative apparatus is proposed to use the local covariance deficits of the intensities of the multi-channel images to locate material transitions. The proposed alternative approach is independent of spatial drifts across the image volume.

Local spectral-covariance or local spectral covariance deficits computation and display for highlighting of relevant material transitions in spectral CT and MR

In a method of segmenting a tubular feature in an image, a sequence of overlapping portions of the image are segmented using a trained model. The overlapping portions are positioned along the length of the tubular feature and combined to determine a segmentation of the tubular feature.

Segmentating a tubular feature

In a conventional system for preparing reports on findings in medical images, the actual formulation of the report is not or only insufficiently supported by the computer-based system. Although there are efforts to improve such a system through automatic reporting, however, in previous systems, the error rate is too high and/or the operation of the system too complicated. This application proposes to provide text prediction to a user on the display device. The text prediction is based on prior analyzing the image content of a medical image at is displayed to the user at least when the user activates a text field shown on the display device via the input unit. The displayed text prediction is selected from a pre-defined set of text modules that are associated to the analysis result.

Automatic diagnosis report preparation

A system for reviewing a prior medical image and a related prior medical report, including: a plurality of measurement tools configured to: receive an input medical image and information regarding a medical finding; analyze the input medical image to determine a measurement relating to the medical finding; and output the measurement relating to the medical finding; a report analyzer configured to: receive the prior medical report; analyze the prior medical report to extract first information regarding a first medical finding described in the medical report; select a first measurement tool of the plurality of measurement tools to analyze the prior medical image based upon the first extracted information; and output the first extracted information to the first measurement tool, wherein the first measurement tool analyzes the prior medical image to produce a first updated measurement of the first medical finding and the first measurement tool analyzes a new medical image associated with the prior medical image to produce a new first measurement of the first medical finding; a report generator configured to receive the new first measurement and to produce a new report based upon the new first measurement.

Guided reviews of prior exams and reports

A system and computer-implemented method are provided for annotation of image data. A user is enabled to iteratively annotate the image data. An iteration of said iterative annotation comprises generating labels for a current image data part based on user-verified labels of a previous image data part, and enabling the user to verify and correct said generated labels to obtain user-verified labels for the current image data part. The labels for the current image data part are generated by combining respective outputs of a label propagation algorithm and a machine-learned classifier trained on user-verified labels and image data and applied to image data of the current image data part. The machine-learned classifier is retrained using the user-verified labels and the image data of the current image data part to obtain a retrained machine-learned classifier.

Interactive iterative image annotation

A system and computer-implemented method are provided for preprocessing medical image data for machine learning. Image data is accessed which comprises an anatomical structure. The anatomical structure in the image data is segmented to obtain a segmentation of the anatomical structure as a delineated part of the image data. A grid is assigned to the delineated part of the image data, the grid representing a partitioning of an exterior and interior of the type of anatomical structure using grid lines, wherein said assigning comprises adapting the grid to fit the segmentation of the anatomical structure in the image data. A machine learning algorithm is then provided with an addressing to the image data in the delineated part on the basis of coordinates in the assigned grid. In some embodiments, the image data of the anatomical structure may be resampled using the assigned grid. Advantageous, a standardized addressing to the image data of the anatomical structure is provided, which may reduce the computational overhead of the machine learning, require fewer training data, etc.

Preprocessing medical image data for machine learning

The invention relates to a system for assessing a pulmonary image which allows for an improved assessment with respect to lung nodules detectability. The pulmonary image is smoothed for providing different pulmonary images (20, 21, 22) with different degrees of smoothing, wherein signal values and noise values, which are indicative of the lung vessel detectability and the noise in these images, are determined and used for determining an image quality being indicative of the usability of the pulmonary image to be assessed for detecting lung nodules. Since a pulmonary image shows lung vessels with many different vessel sizes and with many different image values, which cover the respective ranges of potential lung nodules generally very well, the image quality determination based on the different pulmonary images with different degrees of smoothing allows for a reliable assessment of the pulmonary image's usability for detecting lung nodules. The image quality is used to determine a radiation dose level to be applied for generating a next pulmonary image.

System and method for assessing a pulmonary image

The present disclosure relates to a method for medical imaging method for locating anatomical landmarks of a predetermining defined anatomy. The method comprises: a) providing a machine learning model for predicting anatomical landmarks in image data obtained using a set of acquisition parameters and for predicting a subsequent set of acquisition parameters of the set of acquisition parameters for subsequent acquisition of image data; b) determining 5 a current set of acquisition parameters; c) receiving survey image data representing a slice of the anatomy, the survey image data having the current set of current acquisition parameters; d) identifying anatomical landmarks in the received image data using the machine learning model; e) predicting another set of acquisition parameters using the machine learning model and repeating steps c)-e) for a predefined number of repetitions using the predicted set of 10 acquisition parameters as the current set of parameters; and f) providing the identified anatomical landmarks.

Artificial intelligence-enabled localization of anatomical landmarks

The invention relates to a segmentation system for segmenting an object in an image. The segmentation system is configured to place a surface model comprising surface elements within the image, to determine for each surface element a respective sub volume () of the image and to use a neural network () for determining respective distances between the surface elements and the boundary of the object in the image based on the determined subvolumes. The surface model is then adapted in accordance with the determined distances, in order to segment the object. This segmentation, which is based on the subvolumes of the image and the neural network, is improved in comparison to known techniques which rely, () for instance, on a sampling of candidate points along lines being perpendicular to the respective surface element and on a determination of likelihoods for the candidate points that they indicate a boundary of the object.

Segmentation system for segmenting an object in an image

There is provided a computer-implemented method and system () for determining regions of hyperdense lung parenchyma in an image of a lung. The system () comprises a memory () comprising instruction data representing a set of instructions and a processor () configured to communicate with the memory and to execute the set of instructions. The set of instructions, when executed by the processor (), cause the processor () to locate a vessel in the image, determine a density of lung parenchyma in a region of the image that neighbours the located vessel, and determine whether the region of the image comprises hyperdense lung parenchyma based on the determined density, hyperdense lung parenchyma having a density greater than &#x2212;800 HU.

Determining regions of hyperdense lung tissue in an image of a lung

The invention relates to a magnetic resonance imaging data processing system () for processing motion artifacts in magnetic resonance imaging data sets using a deep learning network () trained for the processing of motion artifacts in magnetic resonance imaging data sets. The magnetic resonance imaging data processing system () comprises a memory () storing machine executable instructions () and the trained deep learning network (). Furthermore, the magnetic resonance imaging data processing system () comprises a processor () for controlling the magnetic resonance imaging data processing system. Execution of the machine executable instructions () causes the processor () to control the magnetic resonance imaging data processing system () to: receive a magnetic resonance imaging data set (), apply the received magnetic resonance imaging data set () as an input to the trained deep learning network (), process one or more motion artifacts present in the received magnetic resonance imaging data set () using the trained deep learning network ().

Growing community of inventors

Hamburg, Germany

Tom Brosch