Digital Pathology Approach to Mapping Collagen VI Tumor Spatial Arrangement in COVID-19 Tissue Slides Using ImageJ

Background:
Collagen VI is a structural extracellular matrix protein involved in cell adhesion, migration, and signaling. Altered colagen architecture is a feature of diseases including SARS-CoV-2 and Hodgkin lymphoma. Traditional histopathology and pixel-intensity metrics inadequately capture the spatial organization and topological complexity of collagen networks. This study developed and validated a reproducible digital pathology workflow using ImageJ, an image analysis platform allowing image segmentation, skeletonization, and quantitative morphometric analysis, to quantify collagen VI cellular spatial interactions in diseased tissue.
Aim: To develop and validate a digital pathology workflow using ImageJ for quantitative assessment of collagen VI-cellular spatial arrangements in diseased tissue.
Methods:
A Hodgkin lymphoma tissue image was used to establish workflow, which was subsequently applied to COVID-19 lung tissue cohorts. Structural elements were binarized and skeletonized in ImageJ; allowing extraction of branching-related topological metrics including Branching Complexity Index (BCI), brach density, and junction density. Collagen fractional area was quantified using color thresholding. Associations between structural metrics and collagen deposition were evaluated using Spearman’s rank correlation. Reproducibility between analyzers was assessed using Bland-Altman analysis.
Results:
BCI demonstrated no significant association with collagen fractional area in normal lung tissue (Spearmen ρ ≈ −0.07), but positive associations in COVID-19 cohorts (UK cohort ρ ≈ 0.28; Italian cohort ρ ≈ −0.76). Branch density correlated moderately with collagen fractional area in normal tissue (ρ ≈ 0.50), the UK COVID cohort (ρ ≈ 0.64), and the Italian cohort (ρ ≈ 0.40). Junction density demonstrated positive associations across normal lung (ρ ≈ 0.63), UK COVID tissue (ρ ≈ 0.65), and Italian COVID tissue (ρ ≈ 0.46). Bland-Altman analysis demonstrated minimal systematic bias for collagen area quantification (mean difference -0.36) and BCI measurements (mean difference -0.03).

Conclusion:
ImageJ-based skeleton analysis provides a reproducible framework for quantifying ECM remodeling beyond conventional area-based measures, supporting a multi-metric assessment of pathological tissue architecture.