Organizations: Vanderbilt University, Nashville TN 37235, USA · Weill Cornell Medicine, New York, NY 10021, USA · Vanderbilt University Medical Center, Nashville TN 37232, USA · UT MD Anderson Cancer Center,TX 77030, USA
Quantifying glomerular inflammation in kidney-transplant biopsies is traditionally performed on single whole-slide sections using the Banff "most-severe section" rule, despite the inherently three-dimensional nature of lesion distribution. This 2D based assessment might lead to instability when inflammation varies across slices. This paper reports a feasibility study examining the technical feasibility and clinical relevance of extending Banff scoring into three dimensions. To this end, we propose a 3D Banff lesion-scoring framework that aligns serial sections, reconstructs glomerular identities across depth, and integrates inflammatory-cell counts in 3D under three scoring strategies: Compatible, Volumetric, and Deduplicated Volumetric. In the experiments, glomerulitis (g-scores) was used as an example Banff metric and applied to multi-section renal allograft biopsies. Our findings indicate that 3D Banff g-scores are more consistent across sections and, under the semi-automatic setting, the Volumetric strategy correlates more strongly with clinical biomarkers than traditional 2D scores. These results show that 3D volumetric quantification offers promising added value, underscoring the potential benefit of 3D-aware Banff scoring for kidney transplant pathology.
Figures & tables
Figure 1: 2D Banff scoring vs. 3D Banff scoring. Upper panel: the 2D Banff pipeline analyzes each section independently and assigns the final g -score based on the most severe single section. This “worst-section” rule is sensitive to sampling variability and ignores cross-section correspondence. Lower panel: the proposed 3D Banff framework aligns serial sections and integrates information across depth for volumetric g -score computation.
Figure 2: Proposed 3D Banff scoring pipeline. Serial sections are registered and quantified in parallel. Their outputs are combined to associate glomeruli across depth, followed by Compatible, Volumetric, and Deduplicated Volumetric scoring.
Figure 3: Three 3D Banff scoring strategies enabled by cross-section glomerular alignment. Strategy 1 (Compatible): a glomerulus is labeled inflamed if any of its sectional instances exceeds the cell-count threshold—a direct 3D extension of the Banff worst-section rule. Strategy 2 (Volumetric): immune-cell counts are aggregated across all sections of a 3D glomerulus and compared against a section-normalized threshold, capturing the cumulative inflammatory burden. Strategy 3 (Deduplicated Volumetric): extends Strategy 2 by first deduplicating inflammatory cells that appear in multiple adjacent sections via cross-section cell-instance tracking, yielding a unique-cell volumetric count before thresholding. Inflamed glomeruli are shown in red and non-inflamed in blue .
Figure 4: Intra-patient WSI agreement of Banff g -scores under Semi-Automatic pipelines. For each patient, agreement is defined as the proportion of WSIs assigned the same g -score as the modal score across that patient’s sections. Five scoring strategies are evaluated: 2D worst-section scoring, 2D majority-vote aggregation (2D-Aggre.), 3D-Compatible, 3D-Volumetric, and 3D Deduplicated Volumetric (3D-Dedup. Vol.), which further incorporates cross-section cell-instance deduplication. Higher bars represent better consistency.
Figure 5: Left Panel: Qualitative comparison of glomerular inflammation scoring across three 3D Banff strategies. Upper panel (Case 1): 2D worst-section scoring detects 1/25 inflamed glomeruli (4%, g1); 3D-Compatible identifies 1/26 (3.8%, g1); 3D-Volumetric detects 0/26 (0%, g0). Lower panel (Case 2): 2D scoring finds 12/16 inflamed glomeruli (75%, g2); 3D-Compatible and 3D-Volumetric both detect 13/16 (81.3%, g3). Right Panel: Representative examples of glomerulus identity tracking across serial sections.Upper Panel: Good case. Most glomeruli are consistently matched across adjacent sections, resulting in stable identity propagation ( Green lines). Lower Panel: Failure case. While the majority of glomeruli are correctly associated, imperfect registration and alignment in individual sections lead to identity breaks( Red dashed lines and boxed regions).
Semi-Automatic
Fully-Automatic
2D Method
Δ sCr ↑
Δ BUN ↑
−Δ Ccr ↑
Δ sCr ↑
Δ BUN ↑
−Δ Ccr ↑
2D
0.308
0.077
0.154
0.227
0.136
0.045
Pooled 2D *
0.400
0.109
0.255
—
—
—
Aggre. 2D
0.462
0.231
0.308
0.227
0.136
0.045
Table 1: Kendall τb between predicted Banff g -scores and clinical outcomes ( n=8 patients with 6-month follow-up). All differences are non-significant ( p>0.10 , permutation). Bold: best within each group.
Figure 6: Alluvial comparison of section-level 2D g -scores and WSI-level 3D g -scores across 410 WSIs (50 patients) under the fully-automatic pipeline. Each WSI is assigned one 3D score; thick flows indicate frequent grade transitions. The diagram illustrates grade shifts when moving from single-section 2D scoring to: (a) Compatible scoring; and (b) Volumetric scoring.
Chronic kidney disease is characterized by structural alterations in renal tissue and progressive loss of kidney function. While glomerular subtype ratios provide a compact and clinically meaningful summary of glomerular composition, representing pathomic information using whole-slide image (WSI)-level mean values diminishes glomerulus-level heterogeneity and the contribution of less prevalent patterns. In this proof-of-concept study, we investigated whether modeling individual glomeruli using multiple-instance learning (MIL) could provide complementary information to conventional subtype ratios for estimated glomerular filtration rate (eGFR) classification. We analyzed 162 kidney-biopsy WSIs from 162 patients, categorizing renal function as eGFR < 60 or >= 60 mL/min/1.73 m^2. Quantitative size, shape, intensity, and texture features were extracted from segmented glomeruli, representing each patient as a bag of glomerular instances. An attention-based MIL model was trained to aggregate these heterogeneous pathomic profiles. We evaluated subtype ratios, pathomics, and their combination using ABMIL, with a mean-pooled multilayer perceptron baseline. In five-fold cross-validation, the subtype-ratio-only model achieved an AUC of 0.7262 +/- 0.0596. MIL using pathomics alone achieved an AUC of 0.7362 +/- 0.0788, whereas integrating pathomics with subtype ratios produced the highest mean AUC of 0.7630 +/- 0.0665. SHapley Additive exPlanations (SHAP) were used to examine individual feature contributions. These results demonstrate that MIL-based pathomics can complement established glomerular subtype ratios by capturing additional glomerulus-level information for quantitative kidney-biopsy analysis.
Confocal microscopy of optically cleared and swelled tissue resolves complex biological structures in 3D, but such acquisitions are highly anisotropic: along the under-sampled axial direction the structure can appear discontinuous, hampering reconstruction and automated quantitative analysis. The usual remedy upsamples the axial dimension to an isotropic volume before training a segmentation model, which requires dense annotations in the upsampled space, a prohibitive labeling burden. We present an end-to-end, GPU-accelerated framework that overcomes this without additional annotations. The model is trained on the native acquisition volume; random rotation of training patches leverages the well-resolved lateral plane to supply the missing axial information, and a z-axis continuity loss keeps neighboring slices consistent. We adapt both a convolutional (3D U-Net) and a transformer (SwinUNETR) backbone, aggregate overlapping patches by Gaussian consensus, and compute point-spread-function-corrected membrane thickness by ray-surface intersection on the GPU. We apply the method to the glomerular basement membrane (GBM), a thin, highly convoluted part of the kidney's filtration barrier that grows more irregular in disease. Segmentation accuracy matches inter-expert agreement. Continuity-aware training improves reconstruction smoothness and suppresses a periodic terracing artifact at minimal accuracy cost. We quantify GBM thickness across the reconstructed 3D surface and capture disease-related thickening, enabling fully automated anisotropic 3D morphometry of biological structures without dense volumetric labels or image restoration.
Arash Fatehi, Robin Ebbestad, Linus Butt +7
Institute for Biomedical Informatics, Faculty of Medicine and University Hospital Cologne, University of Cologne, Cologne, Germany · Center for Molecular Medicine Cologne (CMMC), University of Cologne, Faculty of Medicine and University Hospital Cologne, Cologne, Germany · Science for Life Laboratory, Department of Women’s and Children’s Health, Karolinska Institutet, Solna, Sweden +6
Fine-grained evaluation of glomerular pathology must distinguish normal glomeruli from abnormalities such as global and segmental glomerulosclerosis, obsolescent, ischemic, solidified, disappearing, and atubular glomeruli. Supervised classification requires labeled examples of every category, which is impractical when subtypes are rare or absent from the training cohort. One-class anomaly detection offers an alternative by modeling normal data and scoring deviations, allowing previously unseen abnormalities to be detected. We use the frozen residual U-Net backbone of Omni-Seg, pretrained to segment structurally normal renal primitives without abnormal-subtype labels. We propose NoRDeC (Normal-Reference Detection and Characterization), a framework combining Mahalanobis normal-reference scoring with layer-wise representation analysis to determine whether and where glomerular pathology is encoded, how spatial aggregation affects detection, and whether abnormalities alter inter-layer relationships differently. Using glomerular images from two institutions, we evaluate backbone layers and aggregation strategies, compare NoRDeC with PaDiM and PatchCore, and analyze representations using centered kernel alignment (CKA). Layer 4 with Center-70 aggregation achieved a pooled AUROC of 0.926±0.013. NoRDeC achieved the highest AUROC in six of seven abnormality categories and in the pooled analysis, while CKA suggested subtype-dependent changes in inter-layer relationships not captured by anomaly scores alone. The normal-reference model is fitted using only normal glomeruli; abnormality labels are used for configuration selection, evaluation, and grouping in the representation analysis. These results show that a frozen renal feature extractor can support both detection and representation-level characterization of glomerular abnormalities without using abnormal examples to fit the detector.
Greta Hasko, Rachit Saluja, Tianyu Shi +11
Cornell University, Ithaca, NY 14853, USA · Cornell Tech, New York, NY 10044, USA · Sichuan University, Chengdu, 610207, CN +6