General Electric Company

Vivek Prabhakar Vaidya

Rakesh Mullick

Rahul Venkataramani

Prasad Sudhakar

Hariharan Ravishankar

Sheshadri Thiruvenkadam

Shekhar Dwivedi

Vidya Pundalik Kamath

Fei Zhao

Xiaoxing Li

Soma Biswas

Christopher Judson Hardy

Robert David Darrow

Thomas K F Foo

Krishna Seetharam Shriram

Jiayu Chen

Thomas K Foo

Chandan Kumar Mallappa Aladahalli

Arathi Sreekumari

Xiaodong Tao

Sohan Rashmi Ranjan

Daniel J Blezek

Pavan Kumar Veerabhadra Annangi

Maggie M Fung

Kunlin Cao

Umesha Perdoor Srinivas Adiga

Hima Mukesh Patel

Bharath Ram Sundar

Hidenori Shikata

Harihan Ravishankar

A method for image segmentation includes receiving an input image. The method further includes obtaining a deep learning model having a triad of predictors. Furthermore, the method includes processing the input image by a shape model in the triad of predictors to generate a segmented shape image. Moreover, the method includes presenting the segmented shape image via a display unit.

System and method for image segmentation using a joint deep learning model

The present approach relates to feature ranking within deep neural networks in a multi-task and/or multi-label setting. Approaches are described to identify features that are task-specific as well as features that are shared across multiple tasks. In addition to facilitating interpretability, the selected subset of features can be used to make efficient models leading to better stability & regularization along with reduced compute and memory.

Multi-task feature selection neural networks

A method for assisted reading of automated ultrasound image volumes includes receiving a plurality of scan images generated from an imaging device, wherein the plurality of scan images comprises a chest wall region. The method further includes determining a chest wall model representative of the chest wall region based on the plurality of scan images. The method also includes determining a plurality of segmented scan images segmented along the chest wall region based on the chest wall model. In addition, the method includes determining lesion information using an automated lesion detection technique applied to the plurality of segmented scan images. The method also includes displaying the plurality of scan images along with at least one of the lesion information and the chest wall model.

System and method for visualization of ultrasound volumes

A method for determining optimized deep learning architecture includes receiving a plurality of training images and a plurality of real time images corresponding to a subject. The method further includes receiving, by a medical practitioner, a plurality of learning parameters comprising a plurality of filter classes and a plurality of architecture parameters. The method also includes determining a deep learning model based on the plurality of learning parameters and the plurality of training images, wherein the deep learning model comprises a plurality of reusable filters. The method further includes determining a health condition of the subject based on the plurality of real time images and the deep learning model. The method also includes providing the health condition of the subject to the medical practitioner.

System and method for optimization of deep learning architecture

A method for image segmentation includes receiving an input image (). The method further includes obtaining a deep learning model () having a triad of predictors (). Furthermore, the method includes processing the input image by a shape model in the triad of predictors () to generate a segmented shape image (). Moreover, the method includes presenting the segmented shape image via a display unit ().

A method and system for communicating medical data is presented. Patient data, scan parameters, and/or reference information may be received from a patient unit communicatively coupled to a remote unit over a communication network. The patient data may include at least an image corresponding to a patient. Further, one or more anatomical regions in the image may be identified. Additionally, ranks corresponding to the one or more anatomical regions may be computed based on the patient data, the scan parameters, and/or the reference information. Further, one or more portions of the image corresponding to the one or more anatomical regions may be iteratively transmitted from the patient unit to the remote unit based on the computed ranks.

System and method for efficient transmission of patient data

A method is provided for detecting lesions in ultrasound images. The method includes acquiring ultrasound information, determining discriminative descriptors that describe the texture of a candidate lesion region, and classifying each of the discriminative descriptors as one of a top boundary pixel, a lesion interior pixel, a lower boundary pixel, or a normal tissue pixel. The method also includes determining a pattern of transitions between the classified discriminative descriptors, and classifying the candidate lesion region as a lesion or normal tissue based on the pattern of transitions between the classified discriminative descriptors.

Method and system for lesion detection in ultrasound images

A method is provided for detecting lesions in ultrasound images. The method includes acquiring an ultrasound image, generating a Fisher-tippett (FT) distribution-based edge feature map from the acquired ultrasound image, generating gradient concentration (GC) scores for pixels of the acquired ultrasound image using the FT distribution-based edge feature map, and identifying a candidate lesion region within the acquired ultrasound image based on the GC scores.

Systems and methods for landmark correction in Magnetic Resonance Imaging (MRI) are provided. One method includes acquiring at least one calibration image or at least one localizer image of an object, identifying in the calibration or localizer images a region of the object as a reference point, wherein the reference point defines a landmark position. The method further includes determining an offset between an initial landmark position and the identified landmark position. The method also includes using the determined offset for MRI.

Systems and methods for landmark correction in magnetic resonance imaging

A method is provided for generating a metavolume for reconstructing an image. The method provides for acquiring a plurality of low-resolution images of an imaged volume, the plurality of low-resolution images obtained from one or more imaging planes and merging the plurality of low-resolution images using a statistical operation to generate a metavolume. Systems and computer programs that afford functionality of the type defined by this method may be provided by the present technique.

System and method for generating high resolution images

An apparatus, system and method to determine a coordinate system of a heart includes an imager and a computer. The computer is programmed to acquire a first set of initialization imaging data from an anatomical region of a free-breathing subject. A portion of the first set of initialization imaging data includes organ data, which includes cardiac data. The computer is further programmed to determine a location of a central region of a left ventricle of a heart, where the location is based on the organ data and a priori information. The computer is also programmed to determine a short axis of the left ventricle based on the determined location, acquire a first set of post-initialization imaging data from the free-breathing subject from an imaging plane orientation based on the determination of the short axis, and reconstruct at least one image from the first set of post-initialization imaging data.

Growing community of inventors

Banaglore, India

Vivek Prabhakar Vaidya