公开(公告)号：US20240144441A1

公开(公告)日：2024-05-02

申请号：US17975899

申请日：2022-10-28

Applicant: GE Precision Healthcare LLC

Inventor： Michel Souheil Tohme ,  German Guillermo Vera Gonzalez ,  Ludovic Boilevin Kayl ,  Vincent Bismuth ,  Tao Tan

IPC: G06T5/00 ,  G06T5/20 ,  G06T7/00

CPC classification number: G06T5/002 ,  G06T5/20 ,  G06T7/0014 ,  A61B6/5258 ,  G06T2200/24 ,  G06T2207/10116 ,  G06T2207/20081 ,  G06T2207/30061

Abstract: Various methods and systems are provided for training a denoising system for a digital imaging system. The denoising system can be a deep learning denoising system formed as a blind or non-blind denoising system in which the training dataset provided to the denoising system includes a noisy image formed with simulated noise added to a clean digital image, and a reference image formed of the clean image having residual noise added thereto, where the residual noise is a fraction of the simulated noise used to form the noisy image. The use of the residual noise within the reference image of the training dataset teaches the DL network in the training process to remove less than all the noise during subsequent inferencing of digital images from the digital imaging system. By leaving selected amounts of noise in the digital images, the denoiser can be tuned to improve image attributes and texture.

公开(公告)号：US20230394296A1

公开(公告)日：2023-12-07

申请号：US17805375

申请日：2022-06-03

Applicant: GE Precision Healthcare LLC

Inventor： Tao Tan ,  Gopal B. Avinash ,  Ludovic Boilevin Kayl ,  Vincent Bismuth ,  Michel S. Tohme ,  German Guillermo Vera Gonzalez

IPC: G06N3/08

CPC classification number: G06N3/08

Abstract: Systems/techniques that facilitate improved neural network inferencing efficiency with fewer parameters are provided. In various embodiments, a system can access a medical image on which an artificial intelligence task is to be performed. In various aspects, the system can facilitate the artificial intelligence task by executing a neural network pipeline on the medical image, thereby yielding an artificial intelligence task output that corresponds to the medical image. In various instances, the neural network pipeline can include respective skip connections from the medical image, prior to any convolutions, to each convolutional layer in the neural network pipeline.

公开(公告)号：US20230252614A1

公开(公告)日：2023-08-10

申请号：US18304947

申请日：2023-04-21

Applicant: GE Precision Healthcare LLC

Inventor： Tao Tan ,  Pál Tegzes ,  Levente Imre Török ,  Lehel Ferenczi ,  Gopal B. Avinash ,  László Ruskó ,  Gireesha Chinthamani Rao ,  Khaled Younis ,  Soumya Ghose

IPC: G06T5/50 ,  G06V10/772 ,  G06V10/774 ,  G06V10/762 ,  G06V10/74 ,  G06V10/776 ,  G06T7/00 ,  G06V10/82 ,  G06F18/21 ,  G06F18/22 ,  G06F18/23 ,  G06F18/28 ,  G06F18/214

CPC classification number: G06T5/50 ,  G06V10/772 ,  G06V10/774 ,  G06V10/762 ,  G06V10/761 ,  G06V10/776 ,  G06T7/00 ,  G06V10/82 ,  G06F18/217 ,  G06F18/22 ,  G06F18/23 ,  G06F18/28 ,  G06F18/214

Abstract: Techniques are described for optimizing deep learning model performance using image harmonization as a pre-processing step. According to an embodiment, a method comprises decomposing, by a system operatively coupled to a processor, an input image into sub-images. The method further comprises harmonizing the sub-images with corresponding reference sub-images of at least one reference image based on two or more different statistical values respectively calculated for the sub-images and the corresponding reference-sub images, resulting in transformation of the sub-images into modified sub-images images. In some implementations, the modified sub-images can be combined into a harmonized image having a more similar appearance to the at least one reference image relative to the input image. In other implementations, harmonized images and/or modified sub-images generated using these techniques can be used as ground-truth training samples for training one or more deep learning model to transform input images with appearance variations into harmonized images.

公开(公告)号：US20230018833A1

公开(公告)日：2023-01-19

申请号：US17379003

申请日：2021-07-19

Applicant: GE Precision Healthcare LLC

Inventor： Bipul Das ,  Rakesh Mullick ,  Utkarsh Agrawal ,  KS Shriram ,  Sohan Ranjan ,  Tao Tan

IPC: G16H50/20 ,  G06N5/04

Abstract: Techniques are described for generating multimodal training data cohorts tailored to specific clinical machine learning (ML) model inferencing tasks. In an embodiment, a method comprises accessing, by a system comprising a processor, multimodal clinical data for a plurality of subjects included in one or more clinical data sources. The method further comprises selecting, by the system, datasets from the multimodal clinical data based on the datasets respectively comprising subsets of the multimodal clinical data that satisfy criteria determined to be relevant to a clinical processing task. The method further comprises generating, by the system, a training data cohort comprising the datasets for training a clinical inferencing model to perform the clinical processing task.

公开(公告)号：US20210334598A1

公开(公告)日：2021-10-28

申请号：US16858862

申请日：2020-04-27

Applicant: GE Precision Healthcare LLC

Inventor： Tao Tan ,  Pál Tegzes ,  Levente Imre Török ,  Lehel Ferenczi ,  Gopal B. Avinash ,  László Ruskó ,  Gireesha Chinthamani Rao ,  Khaled Younis ,  Soumya Ghose

IPC: G06K9/62

Abstract: Techniques are described for optimizing deep learning model performance using image harmonization as a pre-processing step. According to an embodiment, a method comprises decomposing, by a system operatively coupled to a processor, an input image into sub-images. The method further comprises harmonizing the sub-images with corresponding reference sub-images of at least one reference image based on two or more different statistical values respectively calculated for the sub-images and the corresponding reference-sub images, resulting in transformation of the sub-images into modified sub-images images. In some implementations, the modified sub-images can be combined into a harmonized image having a more similar appearance to the at least one reference image relative to the input image. In other implementations, harmonized images and/or modified sub-images generated using these techniques can be used as ground-truth training samples for training one or more deep learning model to transform input images with appearance variations into harmonized images.

公开(公告)号：US20210232909A1

公开(公告)日：2021-07-29

申请号：US16773156

申请日：2020-01-27

Applicant: GE Precision Healthcare LLC

Inventor： Tao Tan ,  Min Zhang ,  Gopal Biligeri Avinash ,  Lehel Ferenczi ,  Levente Imre Török ,  Pál Tegzes

IPC: G06N3/08 ,  G06N20/00

Abstract: Systems and techniques that facilitate freeze-out as a regularizer in training neural networks are presented. A system can include a memory and a processor that executes computer executable components. The computer executable components can include: an assessment component that identifies units of a neural network, a selection component that selects a subset of units of the neural network, and a freeze-out component that freezes the selected subset of units of the neural network so that weights of output connections from the frozen subset of units will not be updated for a training run.