Comparison of automated chemical-guided segmentation and human annotation of soil organic matter in X-ray microcomputed tomography imaging in contrasted soil types

Authors: HE, JIA HAO - University Of Illinois; MARGENOT, ANDREW - University Of Illinois; NAKAYAMA, YUHEI - University Of Illinois; NIKITIN, VIKTOR - Argonne National Laboratory; FINFROK, Y. ZOU - Argonne National Laboratory; QAFOKU, ODETA - Pacific Northwest National Laboratory; VARGA, TAMAS - Pacific Northwest National Laboratory; TAS, NESLIHAN - Lawrence Berkeley National Laboratory; GILLISPIE, ELIZABETH - Washington State University; Poret-Peterson, Amisha; McElrone, Andrew; Rippner, Devin

Submitted to: Geoderma
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/7/2026
Publication Date: 3/31/2026
Citation: He, J., Margenot, A.J., Nakayama, Y., Nikitin, V., Finfrok, Y., Qafoku, O., Varga, T., Tas, N., Gillispie, E.C., Poret Peterson, A.T., McElrone, A.J., Rippner, D.A. 2026. Comparison of automated chemical-guided segmentation and human annotation of soil organic matter in X-ray microcomputed tomography imaging in contrasted soil types. Geoderma. 469:117769. https://doi.org/10.1016/j.geoderma.2026.117769.
DOI: https://doi.org/10.1016/j.geoderma.2026.117769

Interpretive Summary: Soil organic matter (OM) is an important part of soil. It is formed from formerly living organism like plants and is important for improving soil structure, water holding capacity, and as a source of nutrients for plants and microorganisms. Our understanding of OM is limited by the techniques we have to study it which mostly involves sieving soil and burning it to measure soil organic carbon content in different soil particle size fractions. We used X-ray microcomputed tomography (µCT) with iodine staining and computer image processing to automate the visual identification of OM in µCT scans of soils. This work will improve our understanding of OM formation and persistence in soils. It will also enable the rapid generation of training data for modern AI algorithms to identify OM in µCT scans without human bias.

Technical Abstract: Soil organic matter (OM) formation and persistence is strongly influenced by the spatial distribution of organic substrates and microscale soil heterogeneity by dictating OM accessibility to microorganisms. However, traditional size and/or density fractionation techniques disrupt aggregate architecture, eliminating spatial information needed to fully understand intra-aggregate OM distribution. To quantify three-dimensional OM spatial distribution and automate segmentation in X-ray micro-computed tomography (µCT) imaging without human annotation bias, we developed an iodine (I2) based staining workflow that eliminates labor-intensive manual annotation while maintaining segmentation accuracy, using aggregates from four taxonomically diverse soils (Xerofluvent, Haploxeroll Sphagnofibrist, Palehumult). Human annotation of 10 µCT slices introduced bias up to 3% variations in Dice similarity coefficient (DSC), with additional manually annotated slices expected to compound segmentation inconsistencies. Dual-energy µCT imaging at 33.1 keV (below the I2 K-edge) and 33.2 keV (above the I2 K-edge) was used to capture aggregate microstructure following I2 staining. The automated image subtraction pipeline identified OM regions by the I2 K-edge induced brightness increases, achieving DSCs ranging from 0.62 to 0.94. The pipeline without GPU acceleration achieved 9.6- to 43.2-fold greater processing efficiency than manual annotation. Using GPU-accelerated image post-processing and affine transformation matrices, the pipeline successfully generated ground truth for large-scale datasets (3232×3232 pixel, 2048 slices) within ~4500 s from raw file acquisition to segmented output. The high-throughput approach can be applied to quantify OM spatial distribution across heterogeneous soil systems.