US 12,141,900 B2
Machine learning generation of low-noise and high structural conspicuity images
Rajesh Veera Venkata Lakshmi Langoju, Bengaluru (IN); Utkarsh Agrawal, Bangalore (IN); Bipul Das, Chennai (IN); Risa Shigemasa, Tokyo (JP); Yasuhiro Imai, Tokyo (JP); and Jiang Hsieh, Brookfield, WI (US)
Assigned to GE Precision Healthcare LLC, Waukesha, WI (US)
Filed by GE Precision Healthcare LLC, Milwaukee, WI (US)
Filed on Dec. 6, 2021, as Appl. No. 17/543,234.
Prior Publication US 2023/0177747 A1, Jun. 8, 2023
Int. Cl. G06T 11/00 (2006.01); G06N 20/20 (2019.01); G06T 5/50 (2006.01); G06T 5/70 (2024.01)
CPC G06T 11/008 (2013.01) [G06N 20/20 (2019.01); G06T 5/50 (2013.01); G06T 5/70 (2024.01); G06T 2207/20084 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A system, comprising:
a processor that executes computer-executable components stored in a computer-readable memory, the computer-executable components comprising:
a receiver component that accesses an image;
a preprocessing component that applies at least one of image denoising or image resolution enhancement to the image, thereby yielding a first intermediary image;
a kernel synthesis component that generates, via execution of a plurality of machine learning models trained using different respective frequency kernels, a plurality of second intermediary images based on the first intermediary image, wherein each machine learning model in the plurality of machine learning models receives as input the first intermediary image and produces as output a respective second intermediary image of the plurality of second intermediary images, and wherein the respective second intermediary image represents a frequency kernel-transformed version of the first intermediary image according to a respective frequency kernel at which the machine learning model was trained; and
a blending component that generates a blended image based on the plurality of second intermediary images.