WeakMitoSAM: competitive prompt aggregation for point-supervised mitochondria segmentation in electron microscopy images
Biomedical Optics Express · 01.05.2026 · 17 (5) · 2262
DOI: 10.1364/boe.592074 ↗Kurz gefasst
Fachartikel in der Zeitschrift Biomedical Optics Express, online erschienen am 09.04.2026. Verfasst von Yuanwei Li und 7 weiteren Autor:innen. Beteiligte Einrichtung: Institute of Medical Technology, Peking University Health Science Center, Peking University, Beijing, China. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 26 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: National Natural Science Foundation of China, Capital's Funds for HealthImprovement and Research, Science Foundation of Peking University Cancer Hospital.
Abstract (Originalsprache)
Mitochondrial morphology is a critical indicator of cellular metabolic status and disease pathogenesis, requiring high-resolution visualization and precise segmentation in electron microscopy (EM) images. While fully supervised deep learning models have achieved significant progress, their reliance on dense pixel-wise annotations presents a major bottleneck due to the labor-intensive labeling process and expert variability near the optical diffraction limit. Existing weakly supervised methods, primarily designed for densely packed instances, often fail to generalize to the sparse distribution of mitochondria in EM data. In this paper, we propose WeakMitoSAM, a novel weakly supervised framework for high-precision mitochondria segmentation using sparse point annotations. Our approach introduces the competitive aggregation of multiple prompts strategy, which employs a Bias-augmented Softmax mechanism to reconcile semantic ambiguities and suppress background noise, effectively converting sparse priors into high-fidelity pseudo-labels. Subsequently, segment anything model is specialized for mitochondrial ultrastructures via low-rank adaptation, ensuring parameter-efficient domain adaptation. Experimental results across four public EM datasets demonstrate that WeakMitoSAM achieves state-of-the-art performance in point-supervised scenarios and even outperforms several fully supervised benchmarks, providing an efficient and robust solution for large-scale mitochondrial morphofunctional analysis.
Autor:innen und Institutionen
- Institute of Medical Technology, Peking University Health Science Center, Peking University, Beijing, China.
- Department of Biomedical Engineering, College of Future Technology, Peking University, Beijing, China.
- National Biomedical Imaging Center, Peking University, Beijing, China.
- Beijing Key Laboratory of Intelligent Neuromodulation and Brain Disorder Treatment, Peking University Third Hospital, Beijing 100191, China.
- zhulei@stu.pku.edu.cn.
- yanye.lu@pku.edu.cn.
Förderung
- National Natural Science Foundation of China
- Capital's Funds for HealthImprovement and Research
- Science Foundation of Peking University Cancer Hospital
- National Key Research and Development Program of China
- China National Postdoctoral Program for Innovative Talents
Zitieren
APA
Li, Y., Zhang, X., He, H., Li, K., Zeng, S., Yang, C., Zhu, L., & Lu, Y. (2026). WeakMitoSAM: competitive prompt aggregation for point-supervised mitochondria segmentation in electron microscopy images. Biomedical Optics Express, 17(5), 2262. https://doi.org/10.1364/boe.592074
Vancouver
Li Y, Zhang X, He H, Li K, Zeng S, Yang C, et al. WeakMitoSAM: competitive prompt aggregation for point-supervised mitochondria segmentation in electron microscopy images. Biomedical Optics Express. 2026;17(5):2262. doi:10.1364/boe.592074
BibTeX
@article{Li2026,
author = {Li, Yuanwei and Zhang, Xinliang and He, Hangzhou and Li, Kaiwen and Zeng, Shuang and Yang, Chuqiao and Zhu, Lei and Lu, Yanye},
title = {WeakMitoSAM: competitive prompt aggregation for point-supervised mitochondria segmentation in electron microscopy images},
journal = {Biomedical Optics Express},
year = {2026},
volume = {17},
number = {5},
pages = {2262},
doi = {10.1364/boe.592074}
}
EUVIMED