Medicine & Health28 April 2026
AI-Driven Liver Ultrasound Image Enhancement Gives Old Scanners New Vision
Source PublicationScientific Publication
Primary AuthorsHuh J, Choi JH, Lee ES, Ye JC, Lee JE, Park HJ, Choi BI.

These results were observed under controlled laboratory conditions, so real-world performance may differ.
The Power of Liver Ultrasound Image Enhancement
Researchers trained a neural network, known as a Switchable CycleGAN, to act as a digital polisher. They fed the AI grainy scans from a 12-year-old machine and high-definition images from a 4-year-old device. The AI learned to bridge the gap using unsupervised learning, essentially teaching itself how to recognise and reconstruct a high-quality scan. The study measured significant gains in:- Overall image brightness and visibility.
- Visual contrast between healthy and diseased tissue.
- Inter-reader agreement among medical experts.
Cite this Article (Harvard Style)
Huh J, Choi JH, Lee ES, Ye JC, Lee JE, Park HJ, Choi BI. (2026). 'Image quality improvement of liver ultrasound using unsupervised deep learning. '. Scientific Publication. Available at: https://doi.org/10.1371/journal.pone.0348137