Systems and methods for motion estimation
AssigneeShanghai United Imaging Intelligence Co., Ltd.
InventorsShanhui Sun, Hanchao Yu, Xiao Chen, Terrence Chen
Life of this patent
Estimating the term…
Abstract
Described herein are neural network-based systems, methods and instrumentalities associated with estimating the motion of an anatomical structure. The motion estimation may be performed using a feature pyramid and/or a motion pyramid that correspond to multiple image scales. The motion estimation may be performed using neural networks and parameters that are learned via a training process involving a student network and a teacher network pre-pretrained with abilities to apply progressive motion compensation.
Claims20
An apparatus for motion tracking, comprising one or more processors, wherein the one or more processors are configured to:
- derive a source image of an anatomical structure and a reference image of the anatomical structure;
- at each of multiple image scales, generate a first representation of features from the source image, a second representation of features from the reference image, and a motion field based on the first representation of features and the second representation of features that indicates a change of the anatomical structure from the source image to the reference image, wherein the respective first representations of features and the respective second representations of features associated with the multiple image scales form a feature pyramid and the respective motion fields associated with the multiple image scales form a motion pyramid;
- determine a preliminary motion field based on the feature pyramid;
- and
- refine the preliminary motion field based on the motion pyramid to obtain a refined motion field.
A method for tracking a motion of an anatomical structure, the method comprising:
- receiving a source image of an anatomical structure and a reference image of the anatomical structure;
- at each of multiple image scales, generating a first representation of features from the source image, a second representation of features from the reference image, and a motion field based on the first representation of features and the second representation of features that indicates a change of the anatomical structure from the source image to the reference image, wherein the respective first representations of features and the respective second representations of features associated with the multiple image scales form a feature pyramid and the respective motion fields associated with the multiple image scales form a motion pyramid;
- determining a preliminary motion field based on the feature pyramid;
- and
- refining the preliminary motion field based on the motion pyramid to obtain a refined motion field.
A method for training one or more artificial neural networks to prediction a motion of an anatomical structure, the method comprising:
- training a teacher neural network to predict a motion field based on a source training image of the anatomical structure and a reference training image of the anatomical structure, wherein, during the training, the teacher neural network is configured to:
- predict a first motion field based on the source training image and the reference training image;
- obtain a warped image based on the source image and the first motion field;
- and
- predict a second motion field based on the source training image and the warped image;
- predicting, using the one or more artificial neural networks, a motion field based on the source training images of the anatomical structure and the reference training image of the anatomical structure;
- and
- adjusting parameters of the one or more artificial neural networks based on a difference between the motion field predicted by the artificial neural network and the motion field predicted by the teacher neural network.
Dependencies are inferred from the claim text.