Developing reliable healthcare AI models requires training with representative and diverse data. In imbalanced datasets, model performance tends to plateau on the more prevalent classes while remaining low on less common cases. To overcome this limitation, we propose DiffUltra, the first generative AI technique capable of synthesizing realistic Lung Ultrasound (LUS) images with extensive lesion variability. Specifically, we condition the generative AI by the introduced Lesion-anatomy Bank, which captures the lesion's structural and positional properties from real patient data to guide the image synthesis.We demonstrate that DiffUltra improves consolidation detection by 5.6% in AP compared to the models trained solely on real patient data. More importantly, DiffUltra increases data diversity and prevalence of rare cases, leading to a 25% AP improvement in detecting rare instances such as large lung consolidations, which make up only 10% of the dataset.
@article{arxiv.2501.06356,
title = {Ultrasound Image Synthesis Using Generative AI for Lung Ultrasound Detection},
author = {Yu-Cheng Chou and Gary Y. Li and Li Chen and Mohsen Zahiri and Naveen Balaraju and Shubham Patil and Bryson Hicks and Nikolai Schnittke and David O. Kessler and Jeffrey Shupp and Maria Parker and Cristiana Baloescu and Christopher Moore and Cynthia Gregory and Kenton Gregory and Balasundar Raju and Jochen Kruecker and Alvin Chen},
journal= {arXiv preprint arXiv:2501.06356},
year = {2025}
}