US 12,141,990 B2
Image registration
Shanhui Sun, Lexington, MA (US); Zhang Chen, Brookline, MA (US); Xiao Chen, Lexington, MA (US); Terrence Chen, Lexington, MA (US); and Junshen Xu, Cambridge, MA (US)
Assigned to Shanghai United Imaging Intelligence Co., Ltd., Shanghai (CN)
Filed by Shanghai United Imaging Intelligence Co., Ltd., Shanghai (CN)
Filed on Sep. 15, 2021, as Appl. No. 17/475,534.
Prior Publication US 2023/0079164 A1, Mar. 16, 2023
Int. Cl. G06T 7/30 (2017.01); G06N 3/02 (2006.01)
CPC G06T 7/30 (2017.01) [G06N 3/02 (2013.01); G06T 2207/10081 (2013.01); G06T 2207/10088 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30004 (2013.01)] 18 Claims
OG exemplary drawing
 
1. A method of image registration, the method comprising:
obtaining a first image of an anatomical structure;
obtaining a second image of the anatomical structure;
determining a plurality of transformation parameters for registering the first image with the second image, wherein the plurality of transformation parameters is determined by at least:
obtaining initial values of the plurality of transformation parameters;
determining, through one or more iterations, respective updates for the plurality of transformation parameters using an artificial neural network (ANN), wherein each of the updates is determined based on a respective present state of the plurality of transformation parameters associated with each of the one or more iterations; and
obtaining, utilizing an ordinary differential equation (ODE) solver, final values of the plurality of transformation parameters based on the respective updates determined by the ANN; and
registering the first image with the second image using the final values of the plurality of transformation parameters, wherein:
the ANN comprises a first sub-network and a second sub-network;
the first sub-network comprises a first neural ODE solver characterized by a first step size or a first error tolerance level, the first sub-network configured to determine a first set of transformation parameters for registering the first image with the second image based on respective versions of the first image and the second image that are characterized by a first resolution; and
the second sub-network comprises a second ODE solver characterized by a second step size that is different than the first step size or by a second error tolerance level that is different than the first error tolerance level, the second sub-network configured to determine a second set of transformation parameters for registering the first image with the second image based on the first set of transformation parameters and respective versions of the first image and the second image that are characterized by a second resolution that is different than the first resolution.