US 12,307,560 B2
Systems and methods for processing medical images with invertible neural networks
Zhang Chen, Cambridge, MA (US); Siyuan Dong, New Haven, CT (US); Shanhui Sun, Cambridge, MA (US); Xiao Chen, Cambridge, MA (US); Yikang Liu, Cambridge, MA (US); and Terrence Chen, Lexington, MA (US)
Assigned to Shanghai United Imaging Intelligence Co., Ltd., Shanghai (CN)
Filed by Shanghai United Imaging Intelligence Co., Ltd., Shanghai (CN)
Filed on Aug. 19, 2022, as Appl. No. 17/891,668.
Prior Publication US 2024/0062438 A1, Feb. 22, 2024
Int. Cl. G06T 5/70 (2024.01); G06T 5/10 (2006.01); G06T 5/20 (2006.01); G06T 5/73 (2024.01); G06T 7/00 (2017.01); G06T 11/00 (2006.01)
CPC G06T 11/008 (2013.01) [G06T 5/10 (2013.01); G06T 5/20 (2013.01); G06T 5/70 (2024.01); G06T 5/73 (2024.01); G06T 7/0014 (2013.01); G06T 2207/10088 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30008 (2013.01)] 17 Claims
OG exemplary drawing
 
1. An apparatus, comprising:
one or more processors configured to:
obtain a reconstructed medical image of an anatomical structure based on under-sampled medical data collected from a medical data space;
process the reconstructed medical image of the anatomical structure through an invertible neural network (INN) to derive a refined medical image of the anatomical structure, wherein the refined medical image includes an improvement to the reconstructed medical image with respect to at least one of a sharpness of the refined medical image, a resolution of the refined medical image, or an amount of noise in the refined medical image, wherein the INN is pre-trained to learn a mapping from the refined medical image to the reconstructed medical image, and wherein, during the derivation of the refined medical image, the one or more processors are configured to:
generate a preliminary medical image based on the reconstructed medical image and an inverse of the mapping learned by the INN;
convert the preliminary medical image into medical data in the medical data space;
update at least a portion of the medical data based on the under-sampled medical data; and
generate the refined medical image based on the updated medical data; and
store or transmit the refined medical image.