Kwon, H., Kim, D. (2026). Stealthy watermarking for medical AI models via spatial patch trigger and two-stage training. Scientific Reports. https://doi.org/10.1038/s41598-026-59042-4