News Release

New AI Tool Learns to Read Medical Images With Far Less Data

A recent work from the research group of Pengtao Xie, an associate professor in the Department of Electrical and Computer Engineering, is featured in UC San Diego Today.

You can read the story here: https://today.ucsd.edu/story/new-ai-tool-learns-to-read-medical-images-with-far-less-data

The work was published in Nature Communications: https://www.nature.com/articles/s41467-025-61754-6

We proposed an end-to-end, downstream-task-guided framework for generating synthetic training data for medical image segmentation. It significantly reduces the need for manual annotations — by a factor of 8 to 20 — while maintaining high segmentation accuracy.

The method demonstrates strong generalization across 11 medical image segmentation tasks and 19 datasets, covering various diseases, organs, and modalities. It improves performance by 10–20% (absolute) in both same- and out-of-domain settings.

Media Contacts

Liezel Labios
Jacobs School of Engineering
858-246-1124
llabios@ucsd.edu