Evidential Calibrated Uncertainty-Guided Interactive Segmentation Paradigm for Ultrasound Images
IEEE Transactions on Instrumentation and Measurement · 2026 · 75 · 1-16
DOI: 10.1109/tim.2026.3687293 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Instrumentation and Measurement, 2026. Verfasst von Jiang Shang und 3 weiteren Autor:innen. Die Arbeit nennt 53 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Natural Science Foundation of Shanghai Municipality, Jie Bang Gua Shuai Project from the Digital Medical Research Institute, School of Medicine, Shanghai University.
Abstract (Originalsprache)
Accurate and robust ultrasound image segmentation is critical for computer-aided diagnostic systems. Nevertheless, the inherent challenges of ultrasound imaging, such as blurry boundaries and speckle noise, often cause traditional segmentation methods to struggle with performance. Despite recent advancements in universal image segmentation, such as the segment anything model (SAM), existing interactive segmentation methods still suffer from inefficiency and a lack of specialization. These methods rely heavily on extensive, accurate manual or random sampling prompts for interaction, necessitating numerous prompts and iterations to reach satisfactory performance. In response to this challenge, we propose evidential uncertainty-guided interactive segmentation (EUGIS), a tiered two-stage interactive segmentation framework for ultrasound image segmentation based on evidential uncertainty estimation. Specifically, EUGIS harnesses evidence-based uncertainty estimation, grounded in Dempster–Shafer theory (DST) and subjective logic (SL), to gauge the level of uncertainty in the predictions of the model for different regions. By prioritizing sampling the high-uncertainty region, our method can effectively simulate the interactive behavior of well-trained radiologists, enhancing the targeted sampling while reducing the number of prompts and iterations required. Additionally, we propose a trainable calibration mechanism for uncertainty estimation, which can further optimize the boundary between certainty and uncertainty, thereby enhancing the confidence of uncertainty estimation. Extensive experiments on three ultrasound datasets show that EUGIS achieves strong and consistent performance relative to the compared noninteractive and interactive segmentation methods within the ultrasound domain. These results suggest that evidential uncertainty-guided interactive segmentation is a promising direction for ultrasound image segmentation. Code and data are available at https://github.com/JiangVentinal/EUGIS
Förderung
- National Natural Science Foundation of China
- Natural Science Foundation of Shanghai Municipality
- Jie Bang Gua Shuai Project from the Digital Medical Research Institute, School of Medicine, Shanghai University
- Shanghai Technical Service Center of Science and Engineering Computing, Shanghai University
Zitieren
APA
Shang, J., Wu, Y., Han, X., & Zhang, Q. (2026). Evidential Calibrated Uncertainty-Guided Interactive Segmentation Paradigm for Ultrasound Images. IEEE Transactions on Instrumentation and Measurement, 75, 1-16. https://doi.org/10.1109/tim.2026.3687293
Vancouver
Shang J, Wu Y, Han X, Zhang Q. Evidential Calibrated Uncertainty-Guided Interactive Segmentation Paradigm for Ultrasound Images. IEEE Transactions on Instrumentation and Measurement. 2026;75:1-16. doi:10.1109/tim.2026.3687293
BibTeX
@article{Shang2026,
author = {Shang, Jiang and Wu, Yuanmeng and Han, Xiaoxiang and Zhang, Qi},
title = {Evidential Calibrated Uncertainty-Guided Interactive Segmentation Paradigm for Ultrasound Images},
journal = {IEEE Transactions on Instrumentation and Measurement},
year = {2026},
volume = {75},
pages = {1--16},
doi = {10.1109/tim.2026.3687293}
}
EUVIMED