@article{Zhang2026, 
author = {Rui Zhang and Xianzhi Song and Linqi Zhu and Branko Bijeljic and Gensheng Li and Martin J. Blunt},
title = {SAMamba3D: Adapting segment anything for generalizable three-dimensional segmentation of multiphase pore-scale images},
year = {2026},
journal = {Advances in Geo-Energy Research},
volume = {21},
number = {2},
pages = {109-124},
keywords = {X-ray micro-computed tomography, multiphase pore-scale imaging, three-dimensional image segmentation, foundational model adaptation, cross-domain generalization},
url = {https://www.sciopen.com/article/10.46690/ager.2026.08.03},
doi = {10.46690/ager.2026.08.03},
abstract = {Reliable segmentation of multiphase pore-scale X-ray images of rocks is necessary to quantify fluid saturation, connectivity, and interfacial geometry. However, current three-dimensional segmentation methods are typically dataset-specific, requiring retraining or extensive fine-tuning whenever rock type, fluid pattern, scanner, or acquisition conditions change. Foundational models such as the Segment Anything Model provide strong two-dimensional boundary priors, but they are not directly applicable to three-dimensional data. This study presents SAMamba3D, a parameter-efficient framework that adapts a largely frozen segment anything model encoder to generalizable three-dimensional pore-scale segmentation by coupling it with Mamba-based volumetric context modeling and progressive cross-scale feature interaction. For sandstone and carbonate datasets, with different fluids, wettability, and scanning conditions, SAMamba3D matches or outperforms current three-dimensional baselines while reducing the need for case-specific retraining. The resulting segmented images preserve physically meaningful descriptors, including fluid saturation, connectivity, and interface morphology, enabling more reliable and rapid analysis of large three-dimensional multiphase images.}
}