Ge Precision Healthcare LLC

General Electric Company

Rahul Venkataramani

Hariharan Ravishankar

Vikram Reddy Melapudi

Prasad Sudhakar

Vivek Prabhakar Vaidya

Krishna Seetharam Shriram

Sheshadri Thiruvenkadam

Deepa Anand

Pavan Annangi

Sai Hareesh Anamandra

Prasad Sudhakara Murthy

Aditi Garg

Dattesh Dayanand Shanbhag

Rakesh Mullick

Chandan Kumar Mallappa Aladahalli

Sandeep Suryanarayana Kaushik

Sohan Rashmi Ranjan

Chandan Kumar Aladahalli

Eigil Samset

Harihan Ravishankar

Bharath Ram Sundar

Annangi P Pavan Kumar

Systems and methods for self-supervised representation learning as a means to generate context-specific pretrained models include selecting data from a set of available data sets; selecting a pretext task from domain specific pretext tasks; selecting a target problem specific network architecture based on a user selection from available choices or any customized model as per user preference; and generating a pretrained model for the selected network architecture using the selected data obtained from the set of available data sets and a pretext task as obtained from domain specific pretext tasks.

Self-supervised learning framework to generate context specific pretrained models

Systems/techniques that facilitate machine learning image analysis based on explicit equipment parameters are provided. In various embodiments, a system can access a medical image generated by a medical imaging device. In various instances, the system can perform, via execution of a machine learning model, an inferencing task on the medical image. In various cases, the machine learning model can receive as input the medical image and a set of equipment parameters. In various aspects, the set of equipment parameters can indicate how the medical imaging device was configured to generate the medical image.

Machine learning image analysis based on explicit equipment parameters

Systems/techniques that facilitate variable confidence machine learning are provided. In various embodiments, a system can access a medical image. In various aspects, the system can perform, via execution of a machine learning model, a regression task on the medical image, wherein the machine learning model can receive as input both the medical image and a user-specified confidence indicator.

Variable confidence machine learning

Methods and systems are provided for generating ultrasound probe motion recommendations. In one example, a method includes obtaining an ultrasound image of a source scan plane, the ultrasound image acquired with an ultrasound probe at a first location relative to a patient, entering the ultrasound image as input to a probe recommendation model trained to output a set of recommendations to move the ultrasound probe from the first location to a plurality of additional locations at which a plurality of target scan planes can be imaged, and displaying the set of recommendations on a display device.

Medical imaging method

Systems/techniques that facilitate generation of model suitability coefficients are provided. In various embodiments, a system can access a model trained on a training dataset, and the system can compute a coefficient indicating whether the model is suitable for deployment on a target dataset, based on analyzing activation maps associated with the model. In some cases, the system can: train a generative adversarial network (GAN) to learn a distribution of training activation maps produced by the model; generate a set of target activation maps of the model, by feeding samples from the target dataset to the model; cause a generator of the GAN to generate synthetic training activation maps from the learned distribution of training activation maps; iteratively perturb inputs of the generator until distances between the synthetic training activation maps and the target activation maps are minimized; and aggregate the minimized distances to form the coefficient.

Model suitability coefficients based on generative adversarial networks and activation maps

Systems, machine-readable media, and methods for ultrasound imaging can include acquiring three-dimensional data for one or more patient data sets and generating a three-dimensional environment based on one or more transition areas identified between a plurality of volumes of the three-dimensional data. A method can also include generating a set of probe guidance instructions based at least in part on the one or more transition areas and the plurality of volumes of the three-dimensional data, and acquiring, using an ultrasound probe, a first frame of two-dimensional data for a patient. The method can also include executing the set of probe guidance instructions to provide probe feedback for acquiring at least a second frame of two-dimensional data.

Systems and methods for generating ultrasound probe guidance instructions

Systems, methods and computer program products are provided to collect ultrasound (US) data. A processor is configured to acquire the US data along one or more acquisition scan planes. The US data defines a plurality of image frames that have a first image quality. The processor is further configured to apply a generative model to at least one of the US data or plurality of image frames to generate a synthetic scan plane image along a synthetic scan plane. The generative model is defined based on one or more training ultrasound data sets. The synthetic scan plane image has an image quality that is common with the first image quality of the plurality of image frames. The system further comprises a display configured to display the synthetic scan plane image.

Method and system for managing image quality utilizing a generative model

Systems and methods for tissue specific time gain compensation of an ultrasound image are provided. The method comprises acquiring an ultrasound image of a subject and displaying the ultrasound image over a console. The method further comprises selecting by a user a region within the ultrasound image that requires time gain compensation. The method further comprises carrying out time gain compensation of the user selected region of the ultrasound image. The method further comprises identifying a region having a similar texture to the user selected region and carrying out time gain compensation of the user selected region by an artificial intelligence (AI) based deep learning module.

Tissue specific time gain compensation methods and systems

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

Various methods and systems are provided for bicuspid valve detection with ultrasound imaging. In one embodiment, a method comprises acquiring ultrasound video of a heart over at least one cardiac cycle, identifying frames in the ultrasound video corresponding to at least one cardiac phase, and classifying a cardiac structure in the identified frames as a bicuspid valve or a tricuspid valve. A generative model such as a variational autoencoder trained on ultrasound image frames at the at least one cardiac phase may be used to classify the cardiac structure. In this way, relatively rare occurrences of bicuspid aortic valves may be automatically detected during regular cardiac ultrasound screenings.

Methods and systems for bicuspid valve detection with generative modeling

A system is presented. The system includes an acquisition subsystem configured to obtain images corresponding to a target domain. Moreover, the system includes a processing subsystem in operative association with the acquisition subsystem and including a memory augmented domain adaptation platform configured to compute one or more features of an input image corresponding to a target domain, identify a set of support images based on the features of the input image, where the set of support images corresponds to the target domain, augment an input to a machine-learnt model with a set of features, a set of masks, or both corresponding to the set of support images to adapt the machine-learnt model to the target domain, and generate an output based at least on the set of features, the set of masks, or both. Additionally, the system includes an interface unit configured to present the output for analysis.

System and method for memory augmented domain adaptation

The subject matter discussed herein relates to systems and methods for generating a clinical outcome based on creating a task-specific model associated with processing raw image(s). In one such example, input raw data is acquired using an imaging system, a selection input corresponding to a clinical task is received, and a task-specific model corresponding to the clinical task is retrieved. Using the task-specific model, the raw data is mapped onto an application specific manifold. Based on the mapping of the raw data onto the application specific manifold the clinical outcome is generated, and subsequently providing the clinical outcome for review.

Systems and methods for predicting outcomes using raw data

Methods and systems are provided for automatically determining a phase shift and noise insensitive similarity metric for electrocardiogram (ECG) beats in a Holter monitor recording. In one embodiment, a deep neural network may be trained to map an ECG beat to a phase shift insensitive and noise insensitive feature space embedding using a training data triad, wherein the training data triad may be produced by a method comprising: selecting a first beat and a second beat recorded via one or more Holter monitors, determining a dynamic time warping (DTW) distance between the first beat and the second beat, setting a similarity label for the first beat and the second beat based on the DTW distance, and storing the first beat, the second beat, and the similarity label, in a location of non-transitory memory as an ECG training data triad.

System and methods for electrocardiogram beat similarity analysis using deep neural networks

Methods and systems are provided for assessing image quality of ultrasound images. In one example, a method includes determining a probe position quality parameter of an ultrasound image, the probe position quality parameter representative of a level of quality of the ultrasound image with respect to a position of an ultrasound probe used to acquire the ultrasound image, determining one or more acquisition settings quality parameters of the ultrasound image, each acquisition settings quality parameter representative of a respective level of quality of the ultrasound image with respect to a respective acquisition setting used to acquire the ultrasound image, and providing feedback to a user of the ultrasound system based on the probe position quality parameter and/or the one or more acquisition settings quality parameters, the probe position quality parameter and each acquisition settings quality parameter determined based on output from separate image quality assessment models.

Systems and methods for ultrasound image quality determination

The present approach relates to a system capable of life-long learning in a deep learning context. The system includes a deep learning network configured to process an input dataset and perform one or more tasks from among a first set of tasks. As an example, the deep learning network may be part of an imaging system, such as a medical imaging system, or may be used in industrial applications. The system further includes a learning unit communicatively coupled to the deep learning networkand configured to modify the deep learning network so as to enable it to perform one or more tasks in a second task list without losing the ability to perform the tasks from the first list.

Growing community of inventors

Bengaluru, India

Rahul Venkataramani