Yet despite the universal use of laminar nomenclature in pain and spinal cord research, a surprising problem has persisted: no standardized approach for assigning cells to individual laminae has ever been broadly adopted. Laboratories typically rely on locally established workflows based on anatomical atlases, cytoarchitectonic landmarks, lamina-specific marker stainings, or manual estimation of laminar boundaries. The exact criteria used to define laminar borders are often incompletely described in published studies, making it difficult to compare datasets across laboratories and potentially contributing to an underappreciated source of inter-laboratory variability. As molecular atlases of the spinal cord expand and imaging datasets grow ever larger, the need for standardized spatial registration has become increasingly pressing.

A team at the Department of Anesthesiology of the University Medical Center Hamburg-Eppendorf, led by L. Roettger, P. Adamek, and S. G. Lechner, has now addressed this gap with Dorsal Horn Mapping, an open-source Fiji toolbox described in Frontiers in Pain Research. The framework implements a standardized workflow for laminar registration in transverse mouse spinal cord sections, assigns labeled cells to Rexed’s laminae, and transforms their positions into a common spatial reference frame. The goal is not maximal registration accuracy at any cost, but a transparent, reproducible, and accessible system that lowers the technical barrier to standardized laminar registration for the many laboratories that analyze comparatively small numbers of transverse sections using conventional immunohistochemistry, in situ hybridization, or fluorescent reporter mouse lines.

The technical design reflects this philosophy. The toolbox provides segment-specific dorsal horn reference maps covering spinal segments C1 to L5, derived from the SC_P56_Atlas published by Fiederling and colleagues, which was created from Nissl-stained sections of the Allen Brain Atlas. For each segment, the right dorsal horn was extracted from the atlas image, converted into a high-resolution vector representation, and the boundaries of the individual laminae were manually traced, vectorized, and refined to generate continuous, non-overlapping anatomical regions. Because the shape of the dorsal horn changes along the rostro-caudal axis, each segment has its own reference map with its own coordinate system; segments Th8 and Th9 as well as Th10 and Th11 share common maps because they are identical within the original atlas.

The registration principle is deliberately simple. Rather than deforming the experimental image to match an atlas, the reference map is interactively aligned to the image using user-guided rigid transformations and independent scaling in the horizontal and vertical dimensions. The resulting transformation parameters are then applied in reverse to all labeled cells, transforming their positions into normalized spatial coordinates within a common dorsal horn reference frame while leaving the original microscopy image completely unchanged. This preserves the integrity of the experimental data, keeps every transformation directly visible to the user, and minimizes the risk of introducing distortions that could arise from more complex nonlinear registration procedures. Scaling and rotation are performed around the arithmetic mean of the reference map, so when transformation parameters are re-applied to reproduce previous results, the individual steps can be applied in any order.

The workflow consists of three independent but complementary modules. The first performs image registration; the second transforms all labeled cells into normalized coordinates and assigns each cell to a Rexed lamina; the third provides visualization tools for quantitative comparison of cellular distributions across images, animals, and experimental groups. Users may employ only selected components: transformed coordinates can be exported as Excel or CSV files and imported into MATLAB or R for customized downstream analyses, while the plotting functions can be applied to coordinate data generated by alternative pipelines. For multichannel experiments, an integrated grouping tool allows cells to be assigned interactively to up to ten experimental groups, with enlarged views of each region of interest displayed in all image channels. The plugin is distributed through the official Fiji update mechanism, requires no specialized hardware beyond a recommendation of 8 GB RAM for comfortable handling of large fluorescence images, and runs on Windows, macOS, and Linux.

Validation was rigorous and multi-layered. To assess inter-user reproducibility, five independent users analyzed 48 images, 24 brightfield and 24 fluorescence images in which gray and white matter could be distinguished by intrinsic autofluorescence. The users ranged from researchers with extensive expertise in spinal cord anatomy to technical staff without previous experience in spinal cord image analysis, and after only a short introduction they aligned reference maps to sections spanning C1 to L5. Across all segments, the median overlap of all five user-aligned reference maps was 78.5 percent for brightfield-aligned maps and 74.4 percent for autofluorescence-aligned maps, with the exception of segment C2, which consistently proved more challenging. Particularly large variabilities in individual transformation parameters were usually caused by the user with no prior anatomical knowledge, indicating that the workflow remains reproducible across expertise levels.

Anatomical accuracy was tested against established laminar markers. Isolectin B4, which labels non-peptidergic primary afferent terminals in inner lamina II, and protein kinase Cγ, which labels a characteristic band of interneurons around the transition between inner lamina II and lamina III, were superimposed onto the registered maps. Across all spinal segments, a median of 81 percent of IB4-positive pixels mapped to lamina II for brightfield-aligned maps, with a 95 percent confidence interval of 68.4 to 86.2 percent, and 75.5 percent for autofluorescence-aligned maps. The majority of PKCγ-positive pixels were distributed across inner lamina II and lamina III, with medians of 69.9 and 73.7 percent in lamina III for the two alignment modalities. The authors note that laminar borders are fluid anatomical gradients rather than abrupt changes, so neither marker exclusively nor completely labels its expected lamina; nevertheless, the agreement demonstrates that registration based solely on the outline of the dorsal horn reliably reproduces the position of established landmarks without consuming dedicated fluorescence channels.

Downstream cell assignment proved equally robust. Spinal cord sections were immunolabeled for PAX2, a marker expressed by the majority of inhibitory interneurons distributed throughout the dorsal horn, and the five participants independently quantified laminar cell counts after both alignment modalities. Of 144 paired comparisons between brightfield and autofluorescence alignment, 16 yielded nominal P-values below 0.05, but none remained significant after Holm or Benjamini–Hochberg correction. A linear mixed-effects model with alignment modality and lamina as fixed effects estimated an average difference of 0.08 cells across all users, segments, and laminae, with a 95 percent confidence interval of −0.68 to 0.83 and P = 0.843. All lamina-specific confidence intervals included zero, and the modality-by-lamina interaction was not significant. Because dorsal horn neurons typically have cell body diameters of approximately 20 micrometers, small differences in template alignment have only limited influence on laminar assignment except for neurons immediately adjacent to laminar boundaries.

A representative application illustrates the practical power of the workflow. The team analyzed the spatial distribution of calretinin- and reelin-expressing neurons in transverse sections of the L4 dorsal horn, assigning manually identified cells to three groups: calretinin-positive/reelin-negative, calretinin-negative/reelin-positive, and double-positive. After registration, the plotting module generated composite reference maps combining all neurons from all analyzed L4 sections into a single standardized anatomical framework, alongside stacked bar graphs quantifying each group within the individual laminae. In contrast to the traditional presentation of one or a few representative micrographs, composite maps display the position of every analyzed cell, allowing readers and reviewers to assess the complete spatial distribution of the dataset. The approach is equally applicable to reporter mouse lines, in situ hybridization, activity-dependent labeling, viral tracing, and emerging spatial transcriptomic methods.

The authors acknowledge limitations. The reference maps derive from a standardized atlas of P56 C57BL/6 mice and cannot capture all biological variability between animals, developmental stages, strains, species, or disease models. The workflow is designed for two-dimensional transverse sections and does not replace comprehensive three-dimensional atlas-registration platforms such as SpinalJ, which remains indispensable for large-scale anatomical studies. Manual registration retains a user-dependent component, although the validation experiments demonstrate that this has only a minor influence on downstream laminar assignment. Future developments may include automated landmark detection, machine-learning-assisted registration, probabilistic assignment of cells near laminar boundaries, and additional templates covering other developmental stages or pathological conditions.

The broader significance may lie in reporting standards. By introducing normalized spatial coordinates within a common dorsal horn reference frame and routine publication of coordinate datasets alongside microscopy images, the toolbox establishes a common anatomical reference system that facilitates direct comparison of cellular localization across experiments and laboratories. The source code, example datasets, and documentation are openly available through the project repository, and the plugin installs directly from within Fiji without manual compilation. In a field where molecularly defined neuronal populations number in the dozens and datasets continue to expand, a shared spatial language for the dorsal horn could prove as consequential as the laminae themselves.

Subject of Research: Dorsal horn mapping: a Fiji toolbox for standardized laminar registration and transparent reporting of spinal cord imaging data from adult C57Bl/6 mice

Article Title: Dorsal horn mapping: a Fiji toolbox for standardized laminar registration and transparent reporting of spinal cord imaging data from adult C57Bl/6 mice

Article References: Roettger, L., Adamek, P., & Lechner, S. G. (2026). Dorsal horn mapping: a Fiji toolbox for standardized laminar registration and transparent reporting of spinal cord imaging data from adult C57Bl/6 mice. Frontiers in Pain Research, 7, Article 1961590. https://doi.org/10.3389/fpain.2026.1961590

Image Credits: AI Generated

DOI: 10.3389/fpain.2026.1961590

Keywords: Dorsal, horn, mapping, Fiji, toolbox, standardized, laminar, registration, transparent, reporting, spinal, cord