BanffNET, a Deep Learning System for Comprehensive Histological Lesion Quantification in Kidney Transplant Biopsies
Abstract
Accurate, reproducible interpretation of kidney allograft biopsies is critical for the diagnosis of graft injury and for informing prognosis and clinical management. The international Banff classification is a consensus diagnostic system based on semiquantitative histological lesion scoring according to either lesion extent or severity in kidney transplant biopsies. However, pathologist scoring is limited by interobserver variability, constrained scalability, and the inherent nature of the scoring system itself. Here we present BanffNET, a weakly supervised, probabilistic deep learning framework that combines self-supervised feature extraction with a novel Bayesian multiple-instance learning framework to predict (continuously) the full spectrum of Banff lesion scores directly from whole-slide images (WSIs). Using lesion-specific aggregation functions tailored to localized (modeling severity) and diffuse histological lesions (modeling extent), BanffNET generates interpretable, patch-level probability maps and calibrated slide-level scores. BanffNETs performance was assessed relative to consensus, biological correlates of rejection and clinical outcome, demonstrating superior consistency, transportability and generalization. Trained on 7,533 WSIs from three cohorts, BanffNET demonstrates consistent performance on 12,687 WSIs across five external validation cohorts, matching or surpassing individual expert pathologists across lesion assessments. BanffNET scores align more closely than pathologist Banff scores with molecular profiles of rejection, offering an objective, transparent, biologically grounded framework for computational pathology with relevance beyond kidney transplantation.
Related
- Validation of the Incremental Prognostic Value of Deceased Donation Pre-implantation Kidney Transplant Biopsies through Comprehensive Lesion Quantification with Deep Learning
- Continuous Diagnostic Indices from BanffNET Automated Lesion Scores for Kidney Transplant Pathology: Development and Evaluation of a Diagnostic Prediction Model
- Automated segmentation and quantification of histological liver features for MASH/MASLD scoring
- Calcineurin Inhibitors and Male Fertility in Renal Transplant Recipients: A Comprehensive Review of Molecular Pathomechanisms and Reproductive Toxicity of Cyclosporine A and Tacrolimus.
- Automatic segmentation of histological zones in rat distraction osteogenesis using a self-configuring deep learning pipeline
BibTeX
@misc{buzzanca2026banffnet,
title = {{BanffNET, a Deep Learning System for Comprehensive Histological Lesion Quantification in Kidney Transplant Biopsies}},
author = {G. Buzzanca and C. Pala and J. He and R. Hofstraat-Boersma and A. Tammaro and D. van Midden and R. Buelow and D. L. Hoelscher and A. S. Muehlfeld and m. Koeller and N. Kozakowski and G. Boehmig and P. F. Halloran and D. van der Helm and S. Meziyerh and J.-H. Venhuizen and S. Haitjema and J. Dijkstra and L. B. Hilbrands and E. J. Steenbergen and A. D. van Zuilen and A. S. Nurmohamed and F. J. Bemelman and I. B. Bruns and G. Callegaro and B. van de Water and T. T. Pieters and G. E. Breimer and G. M. Rossi and E. Fiaccadori and U. Maggiore and J. J. T. H. Roelofs and F. Testa and F. Fontana and A. A. Abiola and M. Delsante and G. L. Corthals and H. Peters-Sengers and Ngu},
howpublished = {medRxiv},
year = {2026},
month = aug,
doi = {10.64898/2026.08.28.26360029},
url = {https://www.medrxiv.org/content/10.64898/2026.08.28.26360029v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - BanffNET, a Deep Learning System for Comprehensive Histological Lesion Quantification in Kidney Transplant Biopsies
AU - G. Buzzanca
AU - C. Pala
AU - J. He
AU - R. Hofstraat-Boersma
AU - A. Tammaro
AU - D. van Midden
AU - R. Buelow
AU - D. L. Hoelscher
AU - A. S. Muehlfeld
AU - m. Koeller
AU - N. Kozakowski
AU - G. Boehmig
AU - P. F. Halloran
AU - D. van der Helm
AU - S. Meziyerh
AU - J.-H. Venhuizen
AU - S. Haitjema
AU - J. Dijkstra
AU - L. B. Hilbrands
AU - E. J. Steenbergen
AU - A. D. van Zuilen
AU - A. S. Nurmohamed
AU - F. J. Bemelman
AU - I. B. Bruns
AU - G. Callegaro
AU - B. van de Water
AU - T. T. Pieters
AU - G. E. Breimer
AU - G. M. Rossi
AU - E. Fiaccadori
AU - U. Maggiore
AU - J. J. T. H. Roelofs
AU - F. Testa
AU - F. Fontana
AU - A. A. Abiola
AU - M. Delsante
AU - G. L. Corthals
AU - H. Peters-Sengers
AU - Ngu
PY - 2026
DA - 2026/08/28
PB - medRxiv
DO - 10.64898/2026.08.28.26360029
UR - https://www.medrxiv.org/content/10.64898/2026.08.28.26360029v1
AB - Accurate, reproducible interpretation of kidney allograft biopsies is critical for the diagnosis of graft injury and for informing prognosis and clinical management. The international Banff classification is a consensus diagnostic system based on semiquantitative histological lesion scoring according to either lesion extent or severity in kidney transplant biopsies. However, pathologist scoring is limited by interobserver variability, constrained scalability, and the inherent nature of the scoring system itself. Here we present BanffNET, a weakly supervised, probabilistic deep learning framework that combines self-supervised feature extraction with a novel Bayesian multiple-instance learning framework to predict (continuously) the full spectrum of Banff lesion scores directly from whole-slide images (WSIs). Using lesion-specific aggregation functions tailored to localized (modeling severity) and diffuse histological lesions (modeling extent), BanffNET generates interpretable, patch-level probability maps and calibrated slide-level scores. BanffNETs performance was assessed relative to consensus, biological correlates of rejection and clinical outcome, demonstrating superior consistency, transportability and generalization. Trained on 7,533 WSIs from three cohorts, BanffNET demonstrates consistent performance on 12,687 WSIs across five external validation cohorts, matching or surpassing individual expert pathologists across lesion assessments. BanffNET scores align more closely than pathologist Banff scores with molecular profiles of rejection, offering an objective, transparent, biologically grounded framework for computational pathology with relevance beyond kidney transplantation.
ER -