Abstract
Fabry disease (FD) is a rare lysosomal storage disorder caused by mutations in the GLA gene, resulting in globotriaosylceramide accumulation. Kidney involvement (Fabry nephropathy) significantly contributes to morbidity and mortality. Diagnosis can be difficult, especially in females or late-onset variants. Renal biopsy remains essential, but interpretation requires expert pathologists. Digital pathology and artificial intelligence (AI) offer promising solutions to support diagnosis. The study analyzed Whole-slide images from renal biopsies of Fabry nephropathy patients to develop and validate a “foamy podocytes” screening AI tool. Two computational tasks were performed: glomerular-level classification, and podocyte-level segmentation. Performance was evaluated using standard metrics. A novel ZEBRA score (fpA/tgA%) was developed to quantify disease burden, and correlations with histological scores and clinical parameters were assessed. EfficientNetB2 achieved the highest classification accuracy (79%) in identifying foamy podocytes. SegFormerB4 had the best segmentation performance (Dice = 0.46, IoU = 0.37). The ZEBRA score effectively distinguished Fabry nephropathy from controls (p < 0.001) and showed good correlation with manual scoring (rs = 0.66–0.71). The AI-assisted ZEBRA pipeline highlights high-risk Fabry nephropathy features to support nephropathologists as a screening tool.
Similar content being viewed by others
Data availability
Authors agree to make data and materials supporting the results or analyses presented in their paper available upon reasonable request. The trained models were exported in a format compatible with the WSInfer extension for QuPath, enabling user-friendly, one-click inference on annotated glomerular regions, and uploaded in the following Github repository: [https://github.com/Gizmopath/ZEBRA-ZEbra-Bodies-Recognition-by-Artificial-intelligence](https:/github.com/Gizmopath/ZEBRA-ZEbra-Bodies-Recognition-by-Artificial-intelligence).
References
Germain, D. P. Fabry disease. Orphanet J. Rare Dis. 5, 30 (2010).
Feriozzi, S. & Rozenfeld, P. Pathology and pathogenic pathways in Fabry nephropathy. Clin. Exp. Nephrol. 25, 925–934 (2021).
Pisani, A. et al. The kidney in fabry’s disease. Clin. Genet. 86, 301–309 (2014).
Branton, M. H. et al. Natural history of Fabry renal disease: influence of alpha-galactosidase A activity and genetic mutations on clinical course. Medicine 81, 122–138 (2002).
Scionti, F. et al. Genetic variants associated with Fabry disease progression despite enzyme replacement therapy. Oncotarget 8, 107558–107564 (2017).
Riccio, E., Sabbatini, M., Capuano, I. & Pisani, A. Early biomarkers of Fabry nephropathy: A review of the literature. Nephron 143, 274–281 (2019).
Tøndel, C. et al. Foot process effacement is an early marker of nephropathy in young classic Fabry patients without albuminuria. Nephron 129, 16–21 (2015).
Avarappattu, J., Gaspert, A., Spartà, G. & Rohrbach, M. Impact of kidney biopsy on deciding when to initiate enzyme replacement therapy in children with Fabry disease. Pediatr. Nephrol. 39, 131–140 (2024).
Fogo, A. B. et al. Scoring system for renal pathology in Fabry disease: report of the international study group of Fabry nephropathy (ISGFN). Nephrol. Dial Transpl. 25, 2168–2177 (2010).
Kim, I. Y., Lee, H. J. & Cheon, C. K. Fabry nephropathy before and after enzyme replacement therapy: important role of renal biopsy in patients with Fabry disease. Kidney Res. Clin. Pract. 40, 611–619 (2021).
Najafian, B. et al. One year of enzyme replacement therapy reduces globotriaosylceramide inclusions in podocytes in male adult patients with Fabry disease. PLoS One. 11, e0152812 (2016).
Tøndel, C. et al. Agalsidase benefits renal histology in young patients with Fabry disease. J. Am. Soc. Nephrol. 24, 137–148 (2013).
Rusu, E. E. et al. The Impact of Kidney Biopsy for Fabry Nephropathy Evaluation on Patients’ Management and Long-Term Outcomes: Experience of a Single Center. Biomedicines 10, (2022).
Valbuena, C. et al. Kidney biopsy findings in heterozygous Fabry disease females with early nephropathy. Virchows Arch. 453, 329–338 (2008).
Choung, H. Y. G., Jean-Gilles, J. & Goldman, B. Myeloid bodies is not an uncommon ultrastructural finding. Ultrastruct Pathol. 46, 130–138 (2022).
Colpart, P. & Félix, S. Fabry nephropathy. Arch. Pathol. Lab. Med. 141, 1127–1131 (2017).
Capuano, I., Buonanno, P., Riccio, E., Crocetto, F. & Pisani, A. Parapelvic cysts: an imaging marker of kidney disease potentially leading to the diagnosis of treatable rare genetic disorders? A narrative review of the literature. J. Nephrol. 35, 2035–2046 (2022).
L’Imperio, V. et al. Improvements in digital pathology equipment for renal biopsies: updating the standard model. J. Nephrol. 37, 221–229 (2024).
L’Imperio, V. et al. Digital pathology for the routine diagnosis of renal diseases: a standard model. J. Nephrol. 34, 681–688 (2021).
Cazzaniga, G. et al. Time for a full digital approach in nephropathology: a systematic review of current artificial intelligence applications and future directions. J. Nephrol. 37, 65–76 (2024).
Xiong, Z. et al. Advances in kidney biopsy lesion assessment through dense instance segmentation. Artif. Intell. Med. 164, 103111 (2025).
Weis, C. A. et al. Assessment of glomerular morphological patterns by deep learning algorithms. J. Nephrol. 35, 417–427 (2022).
Chen, R. J. et al. Towards a general-purpose foundation model for computational pathology. Nat. Med. 30, 850–862 (2024).
Kaczmarzyk, J. R. et al. Open and reusable deep learning for pathology with WSInfer and QuPath. NPJ Precis Oncol. 8, 9 (2024).
Hermsen, M. et al. Deep Learning-Based histopathologic assessment of kidney tissue. J. Am. Soc. Nephrol. 30, 1968–1979 (2019).
Ma, P. et al. AI-based system for analysis of electron microscope images in glomerular disease. JAMA Netw. Open. 8, e2534985 (2025).
Smerkous, D. et al. Development of an automated Estimation of foot process width using deep learning in kidney biopsies from patients with Fabry, minimal change, and diabetic kidney diseases. Kidney Int. 105, 165–176 (2024).
Bankhead, P. et al. QuPath: open source software for digital pathology image analysis. Sci. Rep. 7, 16878 (2017).
Asadi-Aghbolaghi, M. et al. Learning generalizable AI models for multi-center histopathology image classification. NPJ Precis Oncol. 8, 151 (2024).
Acknowledgements
We would like to acknowledge Google for supporting this computational pathology research. We would also like to acknowledge the Italian Ministry of the University MUR Dipartimenti di Eccellenza 2023-2027 (IMPACT MEDICINE, l. 232/2016, art. 1, commi 314-337).
Funding
The work has been funded by the European Union - Next Generation EU - NRRP M6C2 - Investment 2.1 Enhancement and strengthening of biomedical research in the NHS (DIPLOMAT - PNRR-MR1-2022-12375735).
Author information
Authors and Affiliations
Contributions
GioCa and VL defined the study design; MC provided the first annotations for the algorithm development; RB performed statistical analysis; GaCa helped in the scanning procedure; GioCa provided the statistical and computational analysis of the data; LN, GP, AE, VL, FaPa contributed with specialized renal pathology perspective; IC, GV, RM, FePi, LC, ED, MA, AP, SV, GG contributed with enrollment of cases; AE, GG and FP provided the funding acquisition and administrative support. All authors were involved in writing the paper and had final approval of the submitted and published versions.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Ethical consideration
This study, conducted as part of the project ‘DIPLOMAT: DIgital PLatform for OMics and Artificial intelligence in Transplant and native rare renal diseases’ (Grant no. PNRR-MR1-2022-12375735), complies with the principles of the Declaration of Helsinki and all relevant ethical regulations. All participant data were fully anonymized. Ethical approval was granted by the main coordinating center (Integrated University Hospital of Verona, protocol no. 16681, March 16, 2023), with integrated amendments approved by the Ethics Committee 2 of Catania (protocol no. 15/CEL) and the Ethics Committee of Monza/University of Milano-Bicocca (UNIMIB, file no. 4269).
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Below is the link to the electronic supplementary material.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Cazzaniga, G., Carbone, M., Barretta, R. et al. Zebra bodies recognition by artificial intelligence (ZEBRA): a computational tool for Fabry nephropathy. Sci Rep (2026). https://doi.org/10.1038/s41598-026-35466-w
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-026-35466-w
