Organizations: Vanderbilt University, Tennessee, USA · Vanderbilt University Medical Center, Tennessee, USA · Division of Nephrology, Department of Medicine, Taipei Veterans General Hospital, Taipei, Taiwan · School of Medicine, College of Medicine, National Yang Ming Chiao Tung University, Taipei, Taiwan
Quantifying disease-associated renal tubular remodeling is challenging because tubular appearances are heterogeneous. We introduce TuMAP (Tubular Morphotype-Aware Pathomics), a pipeline combining (1) dual-view foundation-model representations and provisional morphotype annotation, (2) compositional and within-morphotype pathomics, and (3) patient-level statistical inference. From 456,158 tubules in 358 PAS-stained kidney biopsy slides, TuMAP identified eight image-based morphotypes, seven of which showed stable subsampling assignments. Across 219 patients in six diagnostic groups, diabetic nephropathy (DN) and tubulointerstitial nephritis/acute tubular injury (TIN/ATI) each differed from minimal change disease in seven morphotype proportions, while IgA nephropathy differed in six. DN showed reduced nuclear solidity and increased eccentricity across all eight morphotypes. Area-normalized nuclear density exposed additional within-morphotype differences. TuMAP provides a framework for separating changes in tubular morphotype composition from phenotypic differences within morphotypes.
Figures & tables
Figure 1: Overview of the TuMAP pipeline. (1) Dual-view tubular representation: Texture and shape views encoded with frozen UNI2-h are harmonized and clustered into provisional M1–M8 morphotypes. (2) Dual-level morphotype-aware pathomics: Composition and within-morphotype phenotypes are quantified. (3) Patient-level statistical inference: Measurements are aggregated per patient and morphotype (red circle: M1 median solidity in one DN biopsy; note different y-axis scales), followed by regression with HC3 robust standard errors and FDR control.
Diagnosis
Patients
Tubules
Tubules/patient
GS (%)
MCD
23
33,486
1446 [1156-1722]
4.3 [0.0–13.4]
MN
37
53,603
1426 [1154-1714]
4.6 [0.0–15.6]
LN
28
43,628
1418 [1197-1980]
9.4 [2.9–23.6]
IgAN
39
43,228
1049 [879-1234]
22.2 [10.0–50.0]
TIN/ATI
28
30,931
1070 [805-1402]
20.0 [0.0–34.4]
DN
64
78,236
1164 [1008-1388]
52.7 [33.3–64.6]
Table 1: Comparison cohort (219 patients). Tubules: total on analyzed sections; tubules/patient: median [IQR]. GS: globally sclerosed glomeruli (%), median [IQR] (190 patients with available GS%).
Figure 2: Eight provisional morphotypes (456,158 tubules; 358 slides). Left: UMAP (200,000 points). Right: M1–M8 counts, median area/circularity, and representative tubules (20-µm bars). Segment-like names are not immunohistochemically confirmed.
Figure 3: Patient-level comparisons vs. MCD ( n=219 ). (a) Mean morphotype shares (%) with adjusted logit-share effects (SD). (b,c) Tubular/nuclear effects (SD), labeled by M number; dots mark FDR q<0.05 . (d–f) M1 tubular solidity, nuclear minor axis, and assigned nuclei per 10,000 µm 2 of contour area (including lumen). Points represent patients; asterisks mark FDR q<0.05 .
Institute of Biopharmaceutical and Health Engineering, Tsinghua Shenzhen International Graduate School, Tsinghua University, Shenzhen, China · Interdisciplinary Institute for Medical Engineering, Fuzhou University, Fuzhou, China · Medical Optical Technology R&D Center, Research Institute of Tsinghua, Pearl River Delta, Guangzhou, China +6