Improving µCT image segmentation through architectural enhancements in the U-Net model

Articles in Press

Document Type : Research Paper

Authors

1 Simulation and Data Processing Laboratory, School of Mining Engineering, College of Engineering, University of Tehran, Tehran, Iran.

2 Civil and Environmental Engineering Dept, School of Mining & Petroleum Engineering, Faculty of Engineering, University of Alberta, Edmonton, Alberta, Canada.

3 School of Mining Engineering, College of Engineering, University of Tehran, Tehran, Iran.

4 Civil and Environmental Engineering Dept, School of Mining & Petroleum Engineering, Faculty of Engineering, University of Alberta, Edmonton, Alberta, Canada.

10.22059/ijmge.2025.399223.595283

Abstract

Micro-Computed Tomography (µCT) is an indispensable non-destructive technique for characterizing the internal microstructures of materials like porous media, but any quantitative analysis hinges on accurate image segmentation—a task complicated by image noise, poor contrast, and intricate features. While deep learning, and the U-Net architecture in particular, has shown considerable promise in automating this process, this study presents a comparative analysis between the standard U-Net and a bespoke, modified architecture for segmenting multi-phase µCT images of Bentheimer sandstone. Our modified U-Net introduces several architectural enhancements: optimized convolutional blocks for superior feature extraction, spatial dropout for more effective regularization, and L2 weight regularization to mitigate overfitting. We trained both models on 2D slices from two core samples and subsequently evaluated them against an independent blind test set of 1872 slices from a third core, which contained four distinct phases: porosity, quartz, clay, and feldspar. The quantitative results reveal that the modified U-Net decidedly outperforms its standard counterpart, achieving a macro-averaged Dice Similarity Coefficient (DSC) of 0.92 versus 0.88, and a macro-averaged Intersection over Union (IoU) of 0.85 versus 0.80. Most notably, our model demonstrated substantial gains in segmenting the challenging minority phases; the DSC for clay surged from 0.71 to 0.85, and for feldspar, it rose from 0.85 to 0.87, all while maintaining stable performance on the majority phases of porosity and quartz. These statistical improvements are corroborated by qualitative visual assessments, which confirm superior boundary delineation and a reduction in misclassifications. Ultimately, our findings indicate that the proposed architectural refinements yield a more accurate and robust segmentation model for µCT imagery, providing a more reliable foundation for downstream Digital Rock Physics (DRP) applications critical to the mining and geo-engineering sectors, such as geomechanical stability assessment and mineral liberation analysis.

Keywords

Main Subjects

References

[1] Zou, S., & Sun, C. (2020). X-ray microcomputed imaging of wettability characterization for multiphase flow in porous media: A review. Capillarity, 3(3), 36–44. doi: https://doi.org/10.46690/capi.2020.03.01
[2] Li, X., Li, B., Liu, F., Li, T., & Nie, X. (2023). Advances in the application of deep learning methods to digital rock technology. Advances in Geo-Energy Research, 8(1), 5–18. doi: https://doi.org/10.46690/ager.2023.04.02
[3] Guntoro, P. I., Ghorbani, Y., Koch, P.-H., & Rosenkranz, J. (2019). X-ray Microcomputed Tomography (µCT) for Mineral Characterization: A Review of Data Analysis Methods. Minerals, 9(3), 183. https://doi.org/10.3390/min9030183
[4] Tung, Patrick, Halim, Amalia, Wang, Helen, Rich, Anne, Chen, Xiao, Regenauer-Lieb, Klaus, & Marjo, Christopher. (2024). Deep learning 3D-mineral liberation analysis with micro-X-ray fluorescence, micro-computed tomography, and deep learning segmentation. BIO Web Conf., 129, 27004. https://doi.org/10.1051/bioconf/202412927004
[5] Feali, M., Pinczewski, W. V., Cinar, Y., Arns, C., Arns, J.-Y., Turner, M., Senden, T., Francois, N., & Knackstedt, M. A. (2012). Qualitative and quantitative analyses of the three-phase distribution of oil, water, and gas in Bentheimer sandstone by use of micro-CT imaging. SPE Reservoir Evaluation & Engineering. doi: https://openresearch-repository.anu.edu.au/items/099f39fc-2db0-4964-913b-62ab8c499492
[6] Liu, M., & Mukerji, T. (2022). Digital transformation in rock physics: Deep learning and data fusion. The Leading Edge, 41(9), 591–598. doi: https://doi.org/10.1190/tle41090591.1
[7] Center of Innovation for Flow Through Porous Media (COIFPM). "Micro-scale Tomography." COIFPM, coifpm.org/index.php/research/micro-scale-tomography. Accessed 31 May 2025. https://www.coifpm.org/index.php/research/micro-scale-tomography
[8] Wetzel, M., Kempka, T., & Kühn, M. (2020). Hydraulic and mechanical impacts of pore space alterations within a sandstone quantified by a flow velocity-dependent precipitation approach. Materials, 13(14), 3100. doi: https://doi.org/10.3390/ma13143100
[9] Shah, S. M., Crawshaw, J. P., & Boek, E. S. (2014). Bentheimer Sandstone [Dataset]. figshare. doi: https://doi.org/10.6084/m9.figshare.1200096.v1
[10] Brondolo, F., & Beaussant, S. (2024). DINOv2 rocks geological image analysis: Classification, segmentation, and interpretability [Preprint]. arXiv. https://arxiv.org/abs/2407.18100
[11] Nickerson, S., Shu, Y., Zhong, D., Könke, C., & Tandia, A. (2019). Permeability of porous ceramics by X-ray CT image analysis. Acta Materialia, 172, 121–130. doi: https://doi.org/10.1016/j.actamat.2019.04.053
[12] Wildenschild, D., & Sheppard, A. P. (2013). X-ray imaging and analysis techniques for quantifying pore-scale structure and processes in subsurface porous medium systems. Advances in Water Resources, 51, 217–246. doi: https://doi.org/10.1016/j.advwatres.2012.07.018
[13] TESCAN. (n.d.). Micro-CT solutions. Retrieved June 4, 2025, from https://info.tescan.com/micro-ct
[14] Cobos, S. F., Norley, C. J., Pollmann, S. I., & Holdsworth, D. W. (2022). Cost-effective micro-CT system for non-destructive testing of titanium 3D printed medical components. PLOS ONE, 17(10), e0275732. doi: https://doi.org/10.1371/journal.pone.0275732
[15] Cengiz, I. F., Oliveira, J. M., & Reis, R. L. (2018). Micro-CT – A digital 3D microstructural voyage into scaffolds: A systematic review of the reported methods and results. Biomaterials Research, 22, 26. doi: https://doi.org/10.1186/s40824-018-0136-8
[16] Karimpouli, S., & Tahmasebi, P. (2019). Segmentation of digital rock images using deep convolutional autoencoder networks. Computers & Geosciences, 126, 142–150. doi: https://doi.org/10.1016/j.cageo.2019.02.003
[17] Madonna, C., Almqvist, B. S. G., & Saenger, E. H. (2012). Digital rock physics: Numerical prediction of pressure-dependent ultrasonic velocities using micro-CT imaging. Geophysical Journal International, 189(3), 1475–1482. doi: https://doi.org/10.1111/j.1365-246X.2012.05437.x
[18] Al-Marzouqi, H. (2018). Digital rock physics: Using CT scans to compute rock properties. IEEE Signal Processing Magazine, 35(2), 121–131. doi: https://doi.org/10.1109/MSP.2017.2784459
[19] Balcewicz, M., Siegert, M., Gurris, M., Ruf, M., Krach, D., Steeb, H., & Saenger, E. H. (2021). Digital rock physics: A geological driven workflow for the segmentation of anisotropic Ruhr sandstone. Frontiers in Earth Science, 9, 673753. https://doi.org/10.3389/feart.2021.673753
[20] Wang, H., Guo, R., Dalton, L. E., Crandall, D., Hosseini, S. A., Fan, M., & Chen, C. (2024). Comparative assessment of U-Net-based deep learning models for segmenting microfractures and pore spaces in digital rocks. SPE Journal, 29, 5779–5791. doi: https://doi.org/10.2118/215117-PA
[21] Roslin, A., Marsh, M., Provencher, B., Mitchell, T. R., Onederra, I. A., & Leonardi, C. R. (2023). Processing of micro-CT images of granodiorite rock samples using convolutional neural networks (CNN), Part II: Semantic segmentation using a 2.5D CNN. Minerals Engineering, 195, 108027. doi: https://doi.org/10.1016/j.mineng.2023.108027
[22] Palacio-Mancheno, P. E., Larriera, A. I., Doty, S. B., Cardoso, L., & Fritton, S. P. (2014). 3D assessment of cortical bone porosity and tissue mineral density using high-resolution µCT: Effects of resolution and threshold method. Journal of Bone and Mineral Research, 29(1), 142–150. doi: https://doi.org/10.1002/jbmr.2012
[23] Li, Q., Ma, C., Zhang, C., Guo, Y., & Zhou, T. (2024). Study on the Microstructure and Permeability Characteristics of Tailings Based on CT Scanning Technology. Applied Sciences, 14(24), 12032. doi: https://doi.org/10.3390/app142412032
[24] Chawshin, K., Berg, C. F., Varagnolo, D., & et al. (2022). Automated porosity estimation using CT-scans of extracted core data. Computational Geosciences, 26, 595–612. doi: https://doi.org/10.1007/s10596-022-10143-9
[25] Kornilov, A., Safonov, I., & Yakimchuk, I. (2022). A Review of Watershed Implementations for Segmentation of Volumetric Images. Journal of Imaging, 8(5), 127. doi: https://doi.org/10.3390/jimaging8050127
[26] Zhao, L., Zhang, H., Sun, X., Ouyang, Z., Xu, C., & Qin, X. (2024). Application of ResUNet-CBAM in Thin-Section Image Segmentation of Rocks. Information, 15(12), 788. doi: https://doi.org/10.3390/info15120788
[27] Wang, H., Yang, X., Zhou, C., Yan, J., Yu, J., & Xie, K. (2025). Constructions of multi-scale 3D digital rocks by associated image segmentation method. Frontiers in Earth Science, 12. doi: https://doi.org/10.3389/feart.2024.1518561
[28] Kornilov, A. S., & Safonov, I. V. (2018). An Overview of Watershed Algorithm Implementations in Open Source Libraries. Journal of Imaging, 4(10), 123. doi: https://doi.org/10.3390/jimaging4100123
[29] Yu, Q., Wang, G., Cheng, H., Guo, W., & Liu, Y. (2024). The segmentation and intelligent recognition of structural surfaces in borehole images based on the U2-Net network. PLOS ONE, 19(3), e0299471. doi: https://doi.org/10.1371/journal.pone.0299471
[30] Minaee, S., Boykov, Y., Porikli, F., Plaza, A., Kehtarnavaz, N., & Terzopoulos, D. (2022). Image segmentation using deep learning: A survey. IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(7), 3523–3542. doi: https://doi.org/10.1109/TPAMI.2021.3059968
[31] Long, J., Shelhamer, E., & Darrell, T. (2015). Fully convolutional networks for semantic segmentation. arXiv. https://arxiv.org/abs/1411.4038
[32] Liu, X., Deng, Z., & Yang, Y. (2019). Recent progress in semantic image segmentation. Artificial Intelligence Review, 52, 1089–1106. doi: https://doi.org/10.1007/s10462-018-9641-3
[33] Ronneberger, O., Fischer, P., & Brox, T. (2015). U-Net: Convolutional networks for biomedical image segmentation. In N. Navab, J. Hornegger, W. M. Wells, & A. F. Frangi (Eds.), Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015 (Vol. 9351, pp. 234–241). Springer. doi: https://doi.org/10.1007/978-3-319-24574-4_28
[34] Azad, R., Khodapanah Aghdam, E., Rauland, A., Jia, Y., Haddadi Avval, A., Bozorgpour, A., Karimijafarbigloo, S., Cohen, J. P., Adeli, E., & Merhof, D. (2024). Medical image segmentation review: The success of U-Net. IEEE Transactions on Pattern Analysis and Machine Intelligence, 46(12), 10076–10095. doi: https://doi.org/10.1109/TPAMI.2024.3435571
[35] Aldi, F., Yuhandri, & Tajuddin, M. (2024). Enhanced U-Net architecture for glottis segmentation with VGG-16. Journal of Informatics and Visualization, 8(4), 2173–2180. doi: https://doi.org/10.62527/joiv.8.4.3088
[36] Safarov, F., Khojamuratova, U., Komoliddin, M., Kurbanov, Z., Tamara, A., Nizamjon, I., Muksimova, S., & Cho, Y. I. (2025). Lightweight evolving U-Net for next-generation biomedical imaging. Diagnostics, 15(9), 1120. doi: https://doi.org/10.3390/diagnostics15091120
[37] Alshawi, R., Hoque, M. T., & Flanagin, M. C. (2023). A Depth-Wise Separable U-Net Architecture with Multiscale Filters to Detect Sinkholes. Remote Sensing, 15(5), 1384. doi: https://doi.org/10.3390/rs15051384
[38] Chandra, N., Sawant, S., & Vaidya, H. (2023). An efficient U-Net model for improved landslide detection from satellite images. PFG – Journal of Photogrammetry, Remote Sensing and Geoinformation Science, 91, 13–28. doi: https://doi.org/10.1007/s41064-023-00232-4
[39] Mahmoudi, S., Asghari, O., & Boisvert, J. (2025). Addressing class imbalance in micro-CT image segmentation: A modified U-Net model with pixel-level class weighting. Computers & Geosciences, 196, 105853. doi: https://doi.org/10.1016/j.cageo.2025.105853
[40] Sarsembayeva, T., Mansurova, M., Abdildayeva, A., & Serebryakov, S. (2025). Enhancing U-Net Segmentation Accuracy Through Comprehensive Data Preprocessing. Journal of Imaging, 11(2), 50. doi: https://doi.org/10.3390/jimaging11020050
[41] Jin, Z., Li, X., Yang, H., Wu, B., & Zhu, X. (2023). Depthwise separable convolution U-Net for 3D seismic data interpolation. Frontiers in Earth Science, 10, 1005505. doi: https://doi.org/10.3389/feart.2022.1005505
[42] Cheng, H. (2024). Lightweight depthwise separable U-Net for laptop-based abdominal multi-organ segmentation. OpenReview. https://openreview.net/forum?id=R9MagNE8ub
[43] Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., & Salakhutdinov, R. (2014). Dropout: A simple way to prevent neural networks from overfitting. Journal of Machine Learning Research, 15(56), 1929–1958. http://jmlr.org/papers/v15/srivastava14a.html
[44] Tompson, J., Goroshin, R., Jain, A., LeCun, Y., & Bregler, C. (2015). Efficient object localization using convolutional networks. arXiv. https://arxiv.org/abs/1411.4280
[45] Lee, S., & Lee, C. (2020). Revisiting spatial dropout for regularizing convolutional neural networks. Multimedia Tools and Applications, 79, 34195–34207. doi: https://doi.org/10.1007/s11042-020-09054-7
[46] Simplilearn. (2025, March 26). The best guide to regularization in machine learning. https://www.simplilearn.com/tutorials/machine-learning-tutorial/regularization-in-machine-learning
[47] Liang, J., Sun, Y., Lebedev, M., Gurevich, B., Nzikou, M., Vialle, S., & Glubokovskikh, S. (2022). Multi-mineral segmentation of micro-tomographic images using a convolutional neural network. Computers & Geosciences, 168, 105217. https://doi.org/10.1016/j.cageo.2022.105217
[48] Kingma, D. P., & Ba, J. (2017). Adam: A method for stochastic optimization [Preprint]. arXiv. https://arxiv.org/abs/1412.6980
[49] Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press. https://www.deeplearningbook.org/
International Journal of Mining and Geo-Engineering

Articles in Press, Accepted Manuscript
Available Online from 01 November 2025

Files

Share

How to cite

Statistics