利索-全球专利检索

  1. 登录
  2. 注册

搜索结果

81 条记录 (耗时 0.021 秒)

    IMAGE RENDERING METHOD FOR TOMOGRAPHIC IMAGE DATA
    2.
    发明公开
    IMAGE RENDERING METHOD FOR TOMOGRAPHIC IMAGE DATA 审中-公开

    公开(公告)号：EP3996050A1

    公开(公告)日：2022-05-11

    申请号：EP20205906.9

    申请日：2020-11-05

    申请人： Koninklijke Philips N.V.

    发明人： WIEMKER, Rafael ,  BYSTROV, Daniel ,  GOSHEN, Liran

    IPC分类号： G06T15/08 ,  G06T19/00

    摘要： A method for generating an image representation of slices through a body based on tomographic imaging data for the body. The method comprises processing reconstructed tomographic image slices to selectively embed in each slice image information from at least one 3D volume rendering of the slice plane within the 3D tomographic image dataset. This is done through a selection process wherein, based on a set of pre-defined criteria, a decision is made for each pixel in each reconstructed tomographic slice as to whether the pixel value should be replaced with a new, modified pixel value determined based on the at least one volume rendering. This may comprise simply swapping the pixel value for the value of the corresponding pixel value in the volume rendering, or it may comprise a more complex process, for instance blending the two values, or adjusting a transparency of the pixel value based on the at least one volume rendering.

    DEEP UNSUPERVISED IMAGE QUALITY ENHANCEMENT
    3.
    发明公开
    DEEP UNSUPERVISED IMAGE QUALITY ENHANCEMENT 审中-公开

    公开(公告)号：EP3965051A1

    公开(公告)日：2022-03-09

    申请号：EP20194246.3

    申请日：2020-09-03

    申请人： Koninklijke Philips N.V.

    发明人： GOSHEN, Liran

    IPC分类号： G06T5/00 ,  G06T5/20 ,  G06K9/00

    摘要： A training system (TS) for training a machine learning model for image quality enhancement in medical imagery. The system comprises an input interface (IN) for receiving a training input image ( Ĩ IN ). The system (TS) comprises artificial neural network model framework (G,D) of the generative adversarial type including a generator network (G) and a discriminator (D) network. The generative network (G) processes the training input image to produce a training output image ( Ĩ OUT ). A down-scaler (DS) of the system downscales the training input image. The discriminator attempts to discriminate between the downscaled training input image ( I ') and training output image to produce a discrimination result. A training controller (TC) adjusts parameters of the artificial neural network model framework based on the discrimination result.

    DEEP VIRTUAL CONTRAST
    4.
    发明公开
    DEEP VIRTUAL CONTRAST 审中-公开

    公开(公告)号：EP3739522A1

    公开(公告)日：2020-11-18

    申请号：EP19175068.6

    申请日：2019-05-17

    申请人： Koninklijke Philips N.V.

    发明人： GOSHEN, Liran

    IPC分类号： G06N3/04 ,  G06N3/08

    摘要： A training system for a machine learning model for image enhancement, and related methods. The system comprises an input interface (IN) for receiving at least one training input image (I) and at least two machine learning models (V, C) having parameters that define a configuration. The at least two machine learning models comprise a target machine learning model (V) to transform the training input image into an intermediate representation (CR) and to produce therefrom a training output image (I') and a first auxiliary machine learning model (C) configured to transform the intermediate representation into an estimate for the training input image. The system further comprising a model parameter adjuster (PA) configured to adjust the configuration based on i) a residue formed by a deviation between the estimate and the training input image and ii) a deviation between the training output image (I') and a training target image ( Ĩ ). The auxiliary machine learning model forms a regularization framework to foster quick training and allows achieving appreciable image contrast boost.

    DEEP ADVERSARIAL ARTIFACT REMOVAL
    5.
    发明公开
    DEEP ADVERSARIAL ARTIFACT REMOVAL 审中-公开

    公开(公告)号：EP3731144A1

    公开(公告)日：2020-10-28

    申请号：EP19170988.0

    申请日：2019-04-25

    申请人： Koninklijke Philips N.V.

    发明人： GOSHEN, Liran

    IPC分类号： G06K9/62

    摘要： A system (TS) for training a machine learning model for image processing. A training input image (Ia,If) is drawn from a set of training images. The system comprises at least three machine learning models (G, DI, DC) having parameters that define a configuration. One of the model is a target machine learning model (G) that transforms the training input image into an intermediate representation (RC) to then produce therefrom a training output image (Io). The three models further comprise a first auxiliary machine learning model (DI) for image discrimination and a second auxiliary machine learning model (DC) to facilitate enforcing a regularizing of the training of the target machine learning model. The first auxiliary machine learning model (DI) selectively processes the training output image or a sample image from the set of training images, to attempt classifying the respective image into one of two image classes, thereby producing a first intermediate result indicative of the image classification. The second auxiliary machine learning model (DC) selectively processes the intermediate representation or a sample representation, to attempt classifying the respective representation into one of two representation classes, thereby producing a second intermediate result indicative of the representation classification. A model parameter adjuster (PA) adjusts the configuration based on the first intermediate result and on the second intermediate result.

    X-RAY IMAGING DEVICE FOR AN OBJECT
    9.
    发明公开
    X-RAY IMAGING DEVICE FOR AN OBJECT 审中-公开
    用于物体的X射线成像装置

    公开(公告)号：EP3221848A1

    公开(公告)日：2017-09-27

    申请号：EP16767213.8

    申请日：2016-09-14

    申请人： Koninklijke Philips N.V.

    发明人： BRENDEL, Bernhard Johannes ,  SHECHTER, Gilad ,  GOSHEN, Liran ,  KOEHLER, Thomas

    IPC分类号： G06T11/00

    CPC分类号： G06T5/002 ,  A61B6/482 ,  A61B6/5282 ,  G06T5/50 ,  G06T11/005 ,  G06T2207/10081 ,  G06T2211/408

    摘要： The invention relates to an X-ray imaging device (10) for an object, an X-ray imaging system (100) for an object, an X-ray imaging method for an object, and a computer program element for controlling such device or system and a computer readable medium having stored such computer program element. The X-ray imaging device (10) comprises a receiving unit (11) and a processing unit (12). The receiving unit (11) is configured to receive attenuation data representing attenuation properties of the object for at least two different X-ray spectra. The processing unit (12) is configured to decompose the attenuation data into decomposed data, to reduce noise in the decomposed data to obtain de-noised data, to back-convert the de-noised data into back-converted attenuation data, to combine back-converted attenuation data and the attenuation data into combined attenuation data, and to decompose the combined attenuation data into combined decomposed data.

    摘要翻译： 用于物体的X射线成像装置（10），用于物体的X射线成像系统（100），用于物体的X射线成像方法以及用于控制这种装置的计算机程序元件技术领域本发明涉及用于物体的X射线成像装置（10），用于物体的X射线成像系统 系统和具有存储这种计算机程序单元的计算机可读介质。 X射线成像装置（10）包括接收单元（11）和处理单元（12）。 接收单元（11）被配置为接收表示对于至少两个不同的X射线频谱的物体的衰减特性的衰减数据。 处理单元（12）被配置为将衰减数据分解成分解数据，以降低分解数据中的噪声以获得降噪数据，将降噪数据反向转换为反向转换的衰减数据，将反向 将衰减数据和衰减数据转换成组合的衰减数据，并将组合的衰减数据分解为组合的分解数据。