Siemens Healthcare Gmbh

Siemens Medical Solutions Usa, Inc.

New York University

Siemens Corporation

Siemens Healthineers Ag

Jin-hyeong Park

Shaohua Kevin Zhou

Dorin Comaniciu

Zhoubing Xu

Bogdan Georgescu

Atilla Peter Kiraly

Sasa Grbic

Ankur Kapoor

Daguang Xu

Constantine Simopoulos

Mingqing Chen

Puneet Sharma

Ali Kamen

Shanhui Sun

Yefeng Zheng

Mehmet Akif Gulsun

John I Jackson

Kaloian Petkov

Xiaoguang Lu

Thomas Pheiffer

Joanne Otsuki

Tae Soo Kim

Hui Ding

Tommaso Mansi

Michal Sofka

Rui Liao

Wei Zhang

Matthias Fenchel

Esther Raithel

Young-Ho Kim

Dong Yang

Haofu Liao

Sebastien Piat

Abdoul Aziz Amadou

Shaolei Feng

Dana Lin

Tao Xiong

Feng Guo

Sri Venkata Anirudh Nanduri

Yuankai Huo

Bo Liu

Jarrod Collins

Gustavo Henrique Monteiro De Barros Carneiro

Xianjun S Zheng

Systems and methods for automatically navigating a catheter in a patient are provided. An image of a current view of a catheter in a patient is received. A set of actions of a robotic navigation system for navigating the catheter from the current view towards a target view is determined using a machine learning based network. The catheter is automatically navigated in the patient from the current view towards the target view using the robotic navigation system based on the set of actions.

Smart image navigation for intracardiac echocardiography

Methods and systems for artificial intelligence based medical image segmentation are disclosed. In a method for autonomous artificial intelligence based medical image segmentation, a medical image of a patient is received. A current segmentation context is automatically determined based on the medical image and at least one segmentation algorithm is automatically selected from a plurality of segmentation algorithms based on the current segmentation context. A target anatomical structure is segmented in the medical image using the selected at least one segmentation algorithm.

Method and system for artificial intelligence based medical image segmentation

Medical image data is received at a data processing system, which is an artificial intelligence-based system. An identification process is performed at the data processing system to identify a subset of the medical image data representing a region of interest including one or more target tendons. A determination process is performed at the data processing system to determine one or more characteristics relating to one or more abnormalities of the one or more target tendons. Abnormality data is output, the abnormality data relating to the one or more abnormalities and being based on the one or more characteristics.

Medical image data

An anatomical structure is detected () in a volume of ultrasound data by identifying () the anatomical structure in another volume of ultrasound data and generating () an image of the anatomical structure and an anatomical landmark. A group of images are generated () of the original volume and compared () to the image of the other volume. An image of the group of images is selected () as including the anatomical structure based on the comparison.

Intervolume lesion detection and image preparation

Methods and systems are provided for registration of preoperative images to ultrasound images. The preoperative images are segmented using a shape model. An ultrasound procedure is performed to acquire the ultrasound images. The path of an ultrasound transducer used in the ultrasound procedure is tracked. The path is used to deform the segmented preoperative images, providing an alignment. The ultrasound images are registered to the preoperative images using the alignment.

Deformable registration of preoperative volumes and intraoperative ultrasound images from a tracked transducer

Medical image data may be applied to a machine-learned network learned on training image data and associated image segmentations, landmarks, and view classifications to classify a view of the medical image data, detect a location of one or more landmarks in the medical image data, and segment a region in the medical image data based on the application of the medical image data to the machine-learned network. The classified view, the segmented region, or the location of the one or more landmarks may be output.

Segmentation, landmark detection and view classification using multi-task learning

A method and apparatus for automated reconstruction of a 3D computed tomography (CT) volume from a small number of X-ray images is disclosed. A sparse 3D volume is generated from a small number of x-ray images using a tomographic reconstruction algorithm. A final reconstructed 3D CT volume is generated from the sparse 3D volume using a trained deep neural network. A 3D segmentation mask can also be generated from the sparse 3D volume using the trained deep neural network.

Method and system for 3D reconstruction of X-ray CT volume and segmentation mask from a few X-ray radiographs

A 2D medical image is colorized. In one approach, a deep-learnt classifier is trained to colorize from color 2D medical images. The color 2D medical images for training are cinematically rendered from slabs to add color. In another approach, a deep machine-learnt generator creates slices as if adjacent to the 2D medical image. The slices and 2D medical image form a slab, which is cinematically rendered to add color. The result is a colorized 2D medical image.

Two-dimensional cinematic medical imaging in color based on deep learning

A computer-implemented method for performing deformable registration between Magnetic Resonance (MR) and Ultrasound (US) images include receiving an MR volume depicting an organ and segmenting the organ from the MR volume to yield a first 3D point representation of the organ in MR coordinates. Additionally, a US volume depicting an organ is received and the organ is segmented from the US volume to yield a second 3D point representation of the organ in US coordinates. Next, a plurality of point correspondences between the first 3D point representation and the second 3D point representation are determined. Then, a biomechanical model is applied to register the MR volume to the US volume. The plurality of point correspondences are used as displacement boundary conditions for the biomechanical model.

Deformable registration of magnetic resonance and ultrasound images using biomechanical models

A method and apparatus for automated vertebra localization and identification in a 3D computed tomography (CT) volumes is disclosed. Initial vertebra locations in a 3D CT volume of a patient are predicted for a plurality of vertebrae corresponding to a plurality of vertebra labels using a trained deep image-to-image network (DI2IN). The initial vertebra locations for the plurality of vertebrae predicted using the DI2IN are refined using a trained recurrent neural network, resulting in an updated set of vertebra locations for the plurality of vertebrae corresponding to the plurality of vertebrae labels. Final vertebra locations in the 3D CT volume for the plurality of vertebrae corresponding to the plurality of vertebra labels are determined by refining the updated set of vertebra locations using a trained shape-basis deep neural network.

Deep image-to-image recurrent network with shape basis for automatic vertebra labeling in large-scale 3D CT volumes

Automated detection of structure is provided in ultrasound M-mode imaging. A coarse and fine search for structure is used. For example, a less noise susceptible initial position or range of positions for a given structure is determined. This position is then refined. The coarse positioning and/or the refined position may use machine-trained classifiers. The positions of other structure may be used in either coarse or fine positioning, such as using a Markov Random Field. The structure or structures may be identified in the M-mode image without user input of a location in the M-mode image or along the line.

Detection of structure in ultrasound M-mode imaging

A method and system for segmentation of mitral valve inflow (MI) patterns in Doppler echocardiogram images is disclosed. Trained root detectors are used to detect left root candidates, right root candidates, and peak candidates in an input Doppler echocardiogram image. Two global structure detectors, a single triangle detector for non-overlapping E-waves and A-waves and a double triangle detector for overlapping E-waves and A-waves, are used to detect single triangle candidates and double triangle candidates based on the left root, right root, and peak candidates. A shape profile is used to determine a shape probability for each of the single triangle candidates and each of the double triangle candidates. The best single triangle candidate and the best double triangle candidate are selected based on shape probability and detection probability. One of the best single triangle candidate and the best double triangle candidate is selected as the final segmentation result based on a shape probability comparison.

Method and system for automatic detection and measurement of mitral valve inflow patterns in doppler echocardiography

A method for view classification includes providing a frame of an object of interest, detecting a region of interest within the object of interest for each of a plurality of detectors (e.g., binary classifiers), wherein each binary classifier corresponds to a different view, performing a global view classification using a multiview classifier for each view, outputting a classification for each view, fusing outputs of the multiview classifiers, and determining and outputting a classification of the frame based on a fused output of the multiview classifiers.

Automatic cardiac view classification of echocardiography

Growing community of inventors

Princeton, NJ, United States of America

Jin-hyeong Park