NvidiaPatent applicationAI & softwareInterest 6/10

Nvidia patents AI method to align medical images using simulated data

Patent drawing
Patent drawing from the official filing.

Nvidia describes a system that trains neural networks to align two images by finding matching features between them. Instead of relying solely on real-world examples, the system uses a simulator to create artificial pairs of images and their correct alignments. The network learns to predict how to shift one image to match the other by minimizing the error between its guess and the simulated truth.

Why it mattersThis technique could improve the accuracy of medical imaging tools that need to overlay scans from different times or angles, such as MRI or CT data.

In the filing’s own words

Apparatuses, systems, and techniques to perform registration among images. In at least one embodiment, one or more neural networks are trained to indicate registration of features in common among at least two images by generating a first correspondence by simulating a registration process of registering an image and applying the at least two images and the first correspondence to a neural network to derive a second correspondence of the features in common among the at least two images.

What exactly they claim (claim 1)

1. A system comprising: one or more processors to calculate parameters corresponding to one or more neural networks for performing image registration indicating registration of features in common among at least two images, at least in part, by simulating a process of image registration indicating registration of features in common between a training image and a correspondence image, applying the training image and the correspondence image to the neural network, and providing a loss function for the neural network to optimize wherein the loss function is a first function of differences between a simulated displacement field and an output displacement field that the one or more neural network outputs based on attempted registration of the training image and the correspondence image; and one or more memories to store the parameters, wherein the parameters are a second function optimizing for the loss function.

Facts

Company
Nvidia (NVIDIA Corporation)
Official title
NEURAL NETWORK FOR IMAGE REGISTRATION AND IMAGE SEGMENTATION TRAINED USING A REGISTRATION SIMULATOR
Publication number
US20260359984A1
Application number
19739915
Filed
Jul 13, 2026
Published
Nov 12, 2026
Inventors
4
Claims
15

Classified under

G06T 7/0012 Physics & computingG06F 30/20 Physics & computingG06N 3/08 Physics & computingG06T 7/11 Physics & computingG06T 7/344 Physics & computingG06T 2207/20081 Physics & computing

More from Nvidia in AI & software

Summaries are simplified from the filing’s own abstract and first claim. A patent filing is not a product announcement.
Patent data and drawings: U.S. Patent and Trademark Office (USPTO). Nebula is not affiliated with or endorsed by the USPTO.