Siemens Medical Solutions Usa, Inc.

Siemens Healthcare Gmbh

Siemens Aktiengesellschaft

Siemens Healthineers Ag

Yiqiang Zhan

Xiang Sean Zhou

Gerardo Hermosillo Valadez

Zhigang Peng

Shu Liao

Fitsum Aklilu Reda

Maneesh Dewan

Arun Krishnan

Yoshihisa Shinagawa

Matthias Fenchel

Marcos Salganicoff

Matthias Wolf

Parmeet Singh Bhatia

Berthold Kiefer

Luca Bogoni

Liang Zhao

Zhennan Yan

Gerardo Hermosillo-Valadez

Zhao Yi

Gerhard Kleinszig

Martin Harder

Heiko Meyer

Vikas Raykar

Esther Raithel

Senthil Periaswamy

Le Lu

Jun Ma

Yu Zhao

Halid Ziya Yerebakan

Yimo Gao

Mingzhong Li

Yimo Tao

Mert Dikmen

A method to automatically align magnetic resonance (MR) scans for diagnostic scan planning includes acquiring a three-dimensional (3D) localizer image of an anatomical object. One or more initial landmarks are identified in the 3D localizer image using a landmarking engine. One or more main axes associated with the anatomical object are identified based on the one or more initial landmarks. The 3D localizer image is registered to a canonical space based on the main axes associated to yield a registered 3D localizer image. The landmarking engine is applied to the registered 3D localizer image to yield one or more updated landmarks. A plurality of reference points for performing a MR scan are computed based on the one or more updated landmarks.

Combined steering engine and landmarking engine for elbow auto align

A visualization framework based on document representation learning is described herein. The framework may first convert a free text document into word vectors using learning word embeddings. Document representations may then be determined in a fixed-dimensional semantic representation space by passing the word vectors through a trained machine learning model, wherein more related documents lie closer than less related documents in the representation space. A clustering algorithm may be applied to the document representations for a given patient to generate clusters. The framework then generates a visualization based on these clusters.

Visualization framework based on document representation learning

A framework for patient management based on anatomic measurements is described herein. In accordance with one aspect, patient records are clustered into a set of sub-populations based on first anatomic measurements and characteristics extracted from first patient data associated with a population of patients. A representative sub-population similar to a patient may be determined from the set of sub-populations based on the patient data of the patient. A report that presents the second anatomic measurements associated with the patient in relation to corresponding first anatomic measurements associated with the representative sub-population may then be generated.

Patient management based on anatomic measurements

In a method for alpha angle measurements of a bone joint based on magnetic resonance imaging (MRI) data of the bone joint, a three-dimensional (3D) bone surface image of the bone joint is generated based on two-dimensional (2D) segmentations on radial plane slices of the bone joint. Based on the 3D bone surface image, a head center of a head of the bone joint and a neck axis of a neck of the bone joint are estimated. An alpha angle model is constructed based on the estimated head center and neck axis. Further, the alpha angle measurement is determined for each of the radial plane slices of the bone joint based on the constructed alpha angle model.

System and method for angle measurements using magnetic resonance images

A framework for cross-modality image synthesis. A first and second model may be trained using respective first and second pairs of complementary images and corresponding first and second ground truth images that represent first and second views of a region of interest. The first and second pairs of complementary images may be acquired by a first modality and the first and second ground truth images may be acquired by a second modality. A combinational network may further be trained to combine features from the first and second models. At least one synthetic second modality image may then be generated by passing current images through the trained first or second model and the combinational network, wherein the current images are acquired by the first modality and represent the first or second view of the region of interest.

Cross-modality image synthesis

A framework for automated measurement. In accordance with one aspect, the framework detects a centerline point of a structure of interest in an image. A centerline of the structure of interest may be traced based on the detected centerline point. A trained segmentation learning structure may be used to generate one or more contours of the structure of interest along the centerline. One or more measurements may then be extracted from the one or more contours.

Automated measurement based on deep learning

A framework for coarse orientation detection in image data. In accordance with one aspect, the framework trains a learning structure to recognize a coarse orientation of the anatomical structure of interest based on training images. The framework may then pass one or more current images through the trained learning structure to generate a coarse orientation of the anatomical structure of interest. The framework then outputs the generated coarse orientation of the anatomical structure of interest.

Coarse orientation detection in image data

A framework for facilitating visualization, including localizing at least one anatomical structure of interest in image data. The structure of interest is then highlighted by reformatting the image data by mapping landmarks associated with the structure of interest to corresponding points along a contour of a geometric shape and warping the image data based on the mapped landmarks. The resulting reformatted image data is rendered for display to a user.

Visualization with anatomical intelligence

A framework for pedicle screw positioning is described herein. In accordance with one aspect, the framework segments at least one vertebra of interest in image data. The framework then automatically determines a pedicle region within the segmented vertebra of interest, and a safe region within the segmented vertebra of interest. An optimal insertion path passing through the pedicle region may then be generated within the safe region.

Image-based pedicle screw positioning

A framework for automatic retrieval of medical images. In accordance with one aspect, the framework detects patches in a query image volume that contain at least a portion of an anatomical region of interest by using a first trained classifier. The framework determines disease probabilities by applying a second trained classifier to the detected patches, and selects, from the patches, a sub-set of informative patches with disease probabilities above a pre-determined threshold value. For a given patch from the sub-set of informative patches, the framework retrieves, from a database, patches that are most similar to the given image. Image volumes associated with the retrieved patches are then retrieved from the database. A report based on the retrieved image volumes may then be generated and presented.

Automatic medical image retrieval

In a magnetic resonance (MR) apparatus and segmentation method, a region in an MR image, acquired from a scan of a patient with an MR scanner of the apparatus, is provided to a computer for segmentation of the region from the overall image dataset. The segmentation takes place based on a classification of image elements of the image dataset, and the classification is iteratively re-trained in a weakly supervised learning algorithm based on examination-specific information provided to the computer.

Method and apparatus for atlas/model-based segmentation of magnetic resonance images with weakly supervised examination-dependent learning

Disclosed herein is a framework for facilitating image processing. In accordance with one aspect, the framework receives first image data acquired by a first modality and one or more articulated models. The one or more articulated models may include at least one section image acquired by the first modality and aligned with a local image acquired by a second modality. The framework may align an anatomical region of the first image data with the section image and non-rigidly register a first region of interest extracted from the section image with a second region of interest extracted from the aligned anatomical region. To generate a segmentation mask of the anatomical region, the registered first region of interest may be inversely mapped to a subject space of the first image data.

Anatomy aware articulated registration for image segmentation

A framework for anatomy orientation detection is described herein. In accordance with one aspect, a pre-trained regressor is applied to appearance features of the image volume to predict a colatitude of the structure of interest. An optimal longitude corresponding to the predicted colatitude is then determined. In response to the colatitude being more than a pre-determined threshold, the image volume is re-oriented based on the predicted colatitude and the optimal longitude, and the predicted colatitude and optimal longitude determination is repeated for the re-oriented image volume.

Detection of anatomy orientation using learning-based regression

A framework for visualization is described herein. In accordance with one implementation, one or more structures of interest are localized in a three-dimensional image. A position of an anatomical label may be determined using a positioning technique that is selected according to a view type of a visualization plane through the image, wherein the position of the anatomical label is outside the one or more structures of interest. The anatomical label may then be displayed at the determined position in the visualization plane.

Visualization of anatomical labels

Disclosed herein is a framework for facilitating automatic planning for medical imaging. In accordance with one aspect, the framework receives first image data of a subject. One or more imaging parameters may then be derived using a geometric model and at least one reference anatomical primitive detected in the first image data. The geometric model defines a geometric relationship between the detected reference anatomical primitive and the one or more imaging parameters. The one or more imaging parameters may be presented, via a user interface, for use in acquisition, reconstruction or processing of second image data of the subject.

Growing community of inventors

Berwyn, PA, United States of America

Yiqiang Zhan