Three-dimensional throat segmentation from core CT images is a key step in digital core analysis, significantly impacting the accuracy and reliability of pore structure characterization and flow simulation.In this paper, we propose a novel segmentation algorithm based on an enhanced 3D-nnUNet framework, employing a multi-branch attention fusion mechanism and multi-scale feature enhancement.In this algorithm, a Channel-spatial Enhancement Aattention (CEA) module is developed by parallel integration of Convolutional Block Attention Module (CBAM) and Efficient Spatial Pyramid Attention (EPSA), combined with multi-scale dilated convolution operations, the ability to distinguish multi-scale throat features is thereby significantly improved.To further augment contextual representation, a Spatial Pyramid Pooling-Fast (SPPF) module is incorporated into the network bottleneck, employing a progressive pooling with fixed-size kernels to enhance multi-scale feature learning while mininizing computational overhead.The entire network is implemented on the nnUNetv2 platform with automatic architecture adaptation and trained and evaluated on binarized pore-throat CT datasets.Experimental results achieve a Dice coefficient of 91.20%, an IoU of 88.42%, and a correct recognition rate () of 89.20%, outperforming other mainstream traditional and learning-based approaches.The proposed approach provides an effective and automated tool for three-dimensional throat segmentation in digital core analysis and reservoir flow simulation.
Al-KharusiA S, BluntM J.Network extraction from sandstone and carbonate pore space images [J].J Petrol Sci Eng, 2007, 56(4): 219-231.
[2]
DongH.Micro-CT imaging and pore network extraction [D].London: Department of Earth Science and Engineering, Imperial College London, 2008.
[3]
DongH, BluntM J. Pore-network extraction from micro-computerized-tomography images[J].Phys Rev E, 2009, 80(3): 036307.
[4]
LiuY C, TengQ Z, HeX H, et al.Optimization algorithm for throat segmentation in 3D core images [J].Journal of Sichuan University (Engineering Science Edition), 2012, 44(S1): 171-176.
JennaneR, AufortG, BenhamouC L, et al.A new method for 3D thinning of hybrid shaped porous media using artificial intelligence.application to trabecular bone [J].J Med Syst, 2012, 36(2): 497-510.
[7]
BertrandG, CouprieM.Powerful parallel and symmetric 3D thinning schemes based on critical kernels [J].J Math Imag Vis, 2014, 48(1): 134-148.
[8]
LiuJ T.A new method of extracting objects' curve-skeleton: A new method of extracting objects' curve-skeleton [J].Acta Autom Sin, 2008, 34(6): 617-622.
[9]
CorneaN D, SilverD, YuanX, et al.Computing hierarchical curve-skeletons of 3D objects [J].Vis Comput, 2005, 21(11): 945-955.
[10]
YuanH.Research on fracture extraction and pore-throat segmentation algorithm in 3D core images [D].Chengdu: Sichuan University, 2015.
[11]
袁浩.岩心三维图像裂缝提取及孔喉划分算法研究[D].成都: 四川大学, 2015.
[12]
GongX M, TengQ Z, WangZ Y, et al.Pore-throat segmentation algorithm based on medial axis in 3D core images [J].Journal of Sichuan University (Engineering Science Edition), 2016, 48(S2): 100-106.
SilinD, PatzekT.Pore space morphology analysis using maximal inscribed spheres [J].Phys A Stat Mech Appl, 2006, 371(2): 336-360.
[15]
YamaguchiT.Skeleton extraction by distance and geodesic methods in porous media segmentation [J].Journal of Microscopy, 2013, 250(3): 238-247.
[16]
ChenZ Z.Core CT image analysis based on deep learning [D].Chengdu:Sichuan University,2024.
[17]
陈忠照.基于深度学习的岩心CT图像分析[D].成都: 四川大学,2024.
[18]
IsenseeF, JaegerP F, KohlS A A, et al.nnU-Net: A self-configuring method for deep learning-based biomedical image segmentation [J].Nat Methods, 2021, 18(2): 203-211.
[19]
JiangP, ZhangJ, ChenJ.Enhanced rain removal network with convolutional block attention module (CBAM):A novel approach to image de-raining [J].EURASIP Journal on Advances in Signal Processing, 2025(1): 9.
[20]
YangZ, LiY, XuT, et al.EPSA-Net: Efficient spatial pyramid attention network for High-Resolution remote sensing image scene classification [J].IEEE T Geosci Remote, 2022, 60: 1-17.
[21]
HeK, ZhangX, RenS, et al.Spatial pyramid pooling in deep convolutional networks for visual recognition[J].IEEE T Pattern Anal, 2015, 37(9): 1904-1916.
[22]
UlyanovD, VedaldiA, LempitskyV.Instance normalization: The missing ingredient for fast stylization [C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2017: 872-881.
[23]
XuB, WangN, ChenT, et al.Empirical evaluation of rectified activations in convolutional networks [C]//Proceedings of the 32nd International Conference on Machine Learning (ICML), 2015: 95-99.
[24]
MaX, HuangH, WangY, et al.Normalized loss functions for deep learning with noisy labels [C]//Proceedings of the 37th International Conference on Machine Learning, 2020: 6543-6553.
[25]
HeK, ZhangX, RenS,et al.Spatial pyramid pooling in deep convolutional networks for visual recognition [C]//European Conference on Computer Vision, 2014: 346-361.
[26]
DoofealyA, FawziO.Convergence analysis of Nesterov acceleration for stochastic gradient methods [C]//International Conference on Machine Learning, 2020: 2597-2607.
[27]
LoshchilovI, HutterF.SGDR:Stochastic gradient descent with warm restarts [C]//International Conference on Learning Representations(ICLR), 2017.
[28]
SchroederW, MartinK, LorensenB.The visualization toolkit:An object-oriented approach to 3D graphics [M].4th ed.Clifton Park, NY, USA: Kitware Inc., 2006.
[29]
HouQ, ZhouD, FengJ.Coordinate attention for efficient mobile network design [C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021: 13708-13717.
[30]
VaswaniA, ShazeerN, ParmarN, et al.Attention is all you need [C]//Proceedings of the 31st International Conference on Neural Information Processing Systems (NeurIPS), 2017: 5998-6008.
[31]
HuJ, ShenL, SunG.Squeeze-and-excitation networks [C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018: 7132-7141.
[32]
WangQ, WuB, ZhuP, et al.ECA-net: Efficient channel attention for deep convolutional neural networks [C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020: 11531-11539.
[33]
PeiW, MaY, ZhangQ, et al.PSA: A novel efficient attention module for image classification [C]//Proceedings of the 29th ACM International Conference on Multimedia (MM '21), 2021: 1090-1098.
[34]
YuF, KoltunV.Multi-scale context aggregation by dilated convolutions [PP/OL].V3.arXiv (2016-04-30)[2025-03-25].
[35]
ZhaoH, ShiJ, QiX, et al.Pyramid scene parsing network [C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).Honolulu HI, USA: IEEE, 2017: 6230-6239.
[36]
TianZ, ShenC, ChenH, et al.Self-calibrated convolutions for semantic segmentation [C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020: 10711-10720.