- Mannon R.B.
- Matas A.J.
- Grande J.
- Leduc R.
- Connett J.
- Kasiske B.
- Cecka J.M.
- Gaston R.S.
- Cosio F.
- Gourishankar S.
- Halloran P.F.
- Hunsicker L.
- Rush D.
Inflammation in areas of tubular atrophy in kidney allograft biopsies: a potent predictor of allograft failure.
- Solez K.
- Axelsen R.A.
- Benediktsson H.
- Burdick J.F.
- Cohen A.H.
- Colvin R.B.
- Croker B.P.
- Droz D.
- Dunnill M.S.
- Halloran P.F.
- Häyry P.
- Jennette J.C.
- Keown P.A.
- Marcussen N.
- Mihatsch M.J.
- Morozumi K.
- Myers B.D.
- Nast C.C.
- Olsen S.
- Racusen L.C.
- Ramos E.
- Rosen S.
- Sachs D.H.
- Salomon D.R.
- Sanfilippo F.
- Verani R.
- von Willebrand E.
- Yamaguchi Y.
- Furness P.N.
- Taub N.
- Assmann K.J.M.
- Banfi G.
- Cosyns J.-P.
- Dorman A.M.
- Hill C.M.
- Kapper S.K.
- Waldherr R.
- Laurinavicius A.
- Marcussen N.
- Martins A.P.
- Nogueira M.
- Regele H.
- Seron D.
- Carrera M.
- Sund S.
- Taskinen E.I.
- Paavonen T.
- Tihomirova T.
- Rosenthal R.
- Schinstock C.A.
- Sapir-Pichhadze R.
- Naesens M.
- Batal I.
- Bagnasco S.
- Bow L.
- Campbell P.
- Clahsen-van Groningen M.C.
- Cooper M.
- Cozzi E.
- Dadhania D.
- Diekmann F.
- Budde K.
- Lowe F.
- Orandi B.J.
- Rowshani A.T.
- Cornell L.
- Kraus E.
- Ginley B.
- Jen K.-Y.
- Han S.S.
- Rodrigues L.
- Jain S.
- Fogo A.B.
- Zuckerman J.
- Walavalkar V.
- Miecznikowski J.C.
- Wen Y.
- Yen F.
- Yun D.
- Moon K.C.
- Rosenberg A.
- Parikh C.
- Sarder P.
Materials and Methods

Patient Cohort
Tissue Samples
Characteristics | Values |
---|---|
Recipients (n = 73) | |
Age, years | 49.1 (19.8 to 70.3) |
Female sex | 30 (41.1) |
Dialysis type | |
Hemodialysis | 43 (58.9) |
Peritoneal dialysis | 18 (24.6) |
Preemptive | 12 (16.4) |
Panel-reactive antibodies ≤6 | 52 (71.2) |
Patients with retransplants | 11 (15.1) |
Graft characteristics (n = 73) | |
Donor age, years | 57.0 (31.0 to 73.0) |
Living | 35 (47.9) |
Deceased, donation after circulatory death | 16 (21.9) |
Deceased, donation after brain death | 22 (30.1) |
HLA-A mismatch | 1 (0 to 2) |
HLA-B mismatch | 1 (0 to 2) |
HLA-DR mismatch | 1 (0 to 2) |
HLA mismatch total | 3 (0 to 6) |
Cold ischemia time, hours | 11.2 (1.8 to 26.5) |
Need for dialysis <3 months after transplantation | 27 (37.0) |
Biopsy characteristics (n = 125) | |
Time between transplantation and biopsy, days | 40 (3 to 906) |
Original diagnosis | |
Rejection | 34 (27.2) |
Borderline T-cell–mediated rejection | 21 (16.8) |
Calcineurin inhibitor toxicity | 43 (34.4) |
Cytomegalovirus | 1 (0.8) |
Acute tubulus necrosis | 16 (12.8) |
Recurrence original disease | 2 (1.6) |
De novo focal segmental glomerulosclerosis | 1 (0.8) |
No diagnostic abnormalities | 7 (5.6) |
PAS-CD3 Restaining and WSI Preparation
Regions of Interest
Visual Pathologists' Assessment of the Patient Cohort Biopsies
Structure Segmentation CNN Development
Ground Truth
Network Design
Post-Processing
Structure Segmentation Performance
Indirect Segmentation Method for Interstitial Fibrosis and IFTA
Lymphocyte Detection CNN
Automated Assessment of the Patient Cohort Based on CNN Results
Image Registration
Automatically Quantified Tissue Features
Correlation between Automated Feature Quantification and Visual Banff Lesion Scoring
Correlation between Automated and Visual Scoring of Chronic Lesions and the Course of Kidney Function
Results
Visual Banff Lesion Scoring of the Patient Cohort by Pathologists
Structure Segmentation Performance of the CNN
Tissue class | Precision | Recall | Dice |
---|---|---|---|
Glomeruli | 0.96 | 0.94 | 0.95 |
Sclerotic glomeruli | 0.78 | 0.90 | 0.84 |
Empty Bowman capsule | 0.38 | 0.58 | 0.45 |
Proximal tubuli | 0.96 | 0.88 | 0.92 |
Distal tubuli | 0.85 | 0.86 | 0.86 |
Atrophic tubuli | 0.44 | 0.63 | 0.52 |
Arteries/arterioles | 0.60 | 0.93 | 0.73 |
Interstitium | 0.91 | 0.89 | 0.90 |
Capsule | 0.53 | 0.90 | 0.66 |
Weighted average | — | — | 0.88 |
Validation of the Indirect Interstitial Fibrosis and IFTA Segmentation Method with Visually Estimated Percentages

Segmentations and Cell Detections of the Patient Cohort by the CNNs



Agreement between Automated Feature Quantification and Visual Banff Lesion Scoring
Group | ICC |
---|---|
Mean pathologists | 0.977 |
Mean pathologists: CNN (NSG) | 0.972 |
Mean pathologists: CNN (NSG + GSG) | 0.941 |
Computed feature | Banff lesion | Spearman ρ | P value |
---|---|---|---|
(Interstitial fibrosis pixels/total cortical area pixels) × 100% | Interstitial fibrosis (ci) | 0.785 | <0.001 |
(Atrophic tubuli objects/total tubuli objects) × 100% | Tubular atrophy (ct) | 0.773 | <0.001 |
CD3+ cell density in total cortical area, cells/mm2 | Total inflammation (ti) | 0.838 | <0.001 |
CD3+ cell density in cortical area–interstitial fibrosis, cells/mm2 | Inflammation (i) | 0.806 | <0.001 |
Highest CD3+ cell count inside nonatrophic tubuli object segmentations | Tubulitis (t) | 0.551 | <0.001 |
CD3+ cell density in interstitial fibrosis area, cells/mm2 | Inflammation in regions with interstitial fibrosis and tubular atrophy (i-IFTA) | 0.706 | <0.001 |
Highest CD3+ cell count inside atrophic tubuli object segmentations | Tubulitis in regions with interstitial fibrosis and tubular atrophy (t-IFTA) | 0.632 | <0.001 |

Correlation between Chronic Tissue Scores and the Course of Kidney Function
Discussion
- Bouteldja N.
- Klinkhammer B.M.
- Bülow R.D.
- Droste P.
- Otten S.W.
- Freifrau von Stillfried S.
- Moellmann J.
- Sheehan S.M.
- Korstanje R.
- Menzel S.
- Bankhead P.
- Mietsch M.
- Drummer C.
- Lehrke M.
- Kramann R.
- Floege J.
- Boor P.
- Merhof D.
- Jayapandian C.P.
- Chen Y.
- Janowczyk A.R.
- Palmer M.B.
- Cassol C.A.
- Sekulic M.
- Hodgkin J.B.
- Zee J.
- Hewitt S.H.
- O'Toole J.
- Toro P.
- Sedor J.R.
- Barisoni L.
- Madabushi A.
- Jayapandian C.P.
- Chen Y.
- Janowczyk A.R.
- Palmer M.B.
- Cassol C.A.
- Sekulic M.
- Hodgkin J.B.
- Zee J.
- Hewitt S.H.
- O'Toole J.
- Toro P.
- Sedor J.R.
- Barisoni L.
- Madabushi A.
- Bouteldja N.
- Klinkhammer B.M.
- Bülow R.D.
- Droste P.
- Otten S.W.
- Freifrau von Stillfried S.
- Moellmann J.
- Sheehan S.M.
- Korstanje R.
- Menzel S.
- Bankhead P.
- Mietsch M.
- Drummer C.
- Lehrke M.
- Kramann R.
- Floege J.
- Boor P.
- Merhof D.
- Ginley B.
- Jen K.-Y.
- Han S.S.
- Rodrigues L.
- Jain S.
- Fogo A.B.
- Zuckerman J.
- Walavalkar V.
- Miecznikowski J.C.
- Wen Y.
- Yen F.
- Yun D.
- Moon K.C.
- Rosenberg A.
- Parikh C.
- Sarder P.
- Mannon R.B.
- Matas A.J.
- Grande J.
- Leduc R.
- Connett J.
- Kasiske B.
- Cecka J.M.
- Gaston R.S.
- Cosio F.
- Gourishankar S.
- Halloran P.F.
- Hunsicker L.
- Rush D.
Inflammation in areas of tubular atrophy in kidney allograft biopsies: a potent predictor of allograft failure.
- Yi Z.
- Salem F.
- Menon M.C.
- Keung K.
- Xi C.
- Hultin S.
- Al Rasheed M.R.H.
- Li L.
- Su F.
- Sun Z.
- Wei C.
- Huang W.
- Fredericks S.
- Lin Q.
- Banu K.
- Wong G.
- Rogers N.M.
- Farouk S.
- Cravedi P.
- Shingde M.
- Smith R.N.
- Rosales I.A.
- O'Connell P.J.
- Colvin R.B.
- Murphy B.
- Zhang W.
- Yi Z.
- Salem F.
- Menon M.C.
- Keung K.
- Xi C.
- Hultin S.
- Al Rasheed M.R.H.
- Li L.
- Su F.
- Sun Z.
- Wei C.
- Huang W.
- Fredericks S.
- Lin Q.
- Banu K.
- Wong G.
- Rogers N.M.
- Farouk S.
- Cravedi P.
- Shingde M.
- Smith R.N.
- Rosales I.A.
- O'Connell P.J.
- Colvin R.B.
- Murphy B.
- Zhang W.
- Borovec J.
- Kybic J.
- Arganda-Carreras I.
- Sorokin D.V.
- Bueno G.
- Khvostikov A.V.
- Bakas S.
- Chang E.I.C.
- Heldmann S.
- Kartasalo K.
- Latonen L.
- Lots J.
- Noga M.
- Pati S.
- Punithakumar K.
- Ruusuvuori P.
- Skalski A.
- Tahmasebi N.
- Valkonen M.
- Venet L.
- Wang Y.
- Weiss N.
- Wodzinski M.
- Xiang Y.
- Xu Y.
- Yan Y.
- Yushkevich P.
- Zhao S.
- Munoz-Barrutia A.
Acknowledgments
Supplemental Data




- Supplemental Table S1
- Supplemental Table S2
- Supplemental Table S3
References
- Scoring total inflammation is superior to the current Banff inflammation score in predicting outcome and the degree of molecular disturbance in renal allografts.Am J Transpl. 2009; 9: 1859-1867
- T cell-mediated rejection is a major determinant of inflammation in scarred areas in kidney allografts.Am J Transpl. 2018; 18: 377-390
- The causes, significance and consequences of inflammatory fibrosis in kidney transplantation: the Banff i-IFTA lesion.Am J Transpl. 2018; 18: 364-376
- Inflammation in areas of tubular atrophy in kidney allograft biopsies: a potent predictor of allograft failure.Am J Transpl. 2010; 10: 2066-2073
- A 2018 reference guide to the Banff classification of renal allograft pathology.Transplantation. 2018; 102: 1795-1814
- The Banff 2015 kidney meeting report: current challenges in rejection classification and prospects for adopting molecular pathology.Am J Transpl. 2017; 17: 28-41
- International standardization of criteria for the histologic diagnosis of renal allograft rejection: the Banff working classification of kidney transplant pathology.Kidney Int. 1993; 44: 411-422
- Reproducibility of the Banff classification in subclinical kidney transplant rejection.Clin Transpl. 2005; 19: 518-521
- International variation in the interpretation of renal transplant biopsies: report of the CERTPAP Project.Kidney Int. 2001; 60: 1998-2012
- Reproducibility of the Banff classification of renal allograft pathology: inter- and intraobserver variation.Transplantation. 1995; 60: 1083-1089
- International variation in histologic grading is large, and persistent feedback does not improve reproducibility.Am J Surg Pathol. 2003; 27: 805-810
- Banff survey on antibody-mediated rejection clinical practices in kidney transplantation: diagnostic misinterpretation has potential therapeutic implications.Am J Transpl. 2019; 1: 123-131
- Computer-assisted topological analysis of renal allograft inflammation adds to risk evaluation at diagnosis of humoral rejection.Kidney Int. 2017; 92: 214-226
- Interstitial fibrosis evolution on early sequential screening renal allograft biopsies using quantitative image analysis.Am J Transpl. 2011; 11: 1456-1463
- Renal graft fibrosis and inflammation quantification by an automated Fourier-transform infrared imaging technique.J Am Soc Nephrol. 2016; 27: 2382-2391
- From detection of individual metastases to classification of lymph node status at the patient level: the CAMELYON17 challenge.IEEE Trans Med Imaging. 2018; 38: 550-560
- Automated deep-learning system for Gleason grading of prostate cancer using biopsies: a diagnostic study.Lancet Oncol. 2020; 21: 233-241
- Diagnostic assessment of deep learning algorithms for detection of lymph node metastases in women with breast cancer.JAMA. 2017; 318: 2199-2210
- A survey on deep learning in medical image analysis.Med Image Anal. 2017; 42: 60-88
- Region-based convolutional neural nets for localization of glomeruli in trichrome-stained whole kidney sections.J Am Soc Nephrol. 2018; 29: 2081-2088
- Segmentation of glomeruli within trichrome images using deep learning.Kidney Int Rep. 2019; 4: 955-962
- Computational segmentation and classification of diabetic glomerulosclerosis.J Am Soc Nephrol. 2019; 30: 1953-1967
- CNN cascades for segmenting sparse objects in gigapixel whole slide images.Comput Med Imaging Graph. 2019; 71: 40-48
- Automated computational detection of interstitial fibrosis, tubular atrophy, and glomerulosclerosis.J Am Soc Nephrol. 2021; 32: 837-850
- Deep learning-based histopathologic assessment of kidney tissue.J Am Soc Nephrol. 2019; 30: 1968-1979
- Learning to detect lymphocytes in immunohistochemistry with deep learning.Med Image Anal. 2019; 58: 101547
Ronneberger O, Fischer P, Brox T: U-Net: Convolutional networks for biomedical image segmentation. Edited by Navab N, Hornegger J, Wells W, Frangi A. Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015. Lecture Notes in Computer Science, Vol 9351, p. 234–241.
- Adam: a method for stochastic optimization.ArXiv. 2014; (1412.6980)
- Resolution-agnostic tissue segmentation in whole-slide histopathology images with convolutional neural networks.PeerJ. 2019; 7: e8242
- Robust, fast and accurate: a 3-step method for automatic histological image registration.ArXiv. 2019; (1903.12063)
- Artificial intelligence and machine learning in nephropathology.Kidney Int. 2020; 98: 65-75
- Digital pathology and computational image analysis in nephropathology.Nat Rev Nephrol. 2020; 16: 669-685
- Artificial intelligence driven next-generation renal histomorphometry.Curr Opin Nephrol Hypertens. 2020; 29: 265-272
- Artificial intelligence in nephrology: core concepts, clinical applications, and perspectives.Am J Kidney Dis. 2019; 74: 803-810
- Glomerulus classification with convolutional neural networks.in: Medical Image Understanding and Analysis. MIUA 2017. Communications in Computer and Information Science. Vol 723. 2017: 839-849
- Deep learning-based segmentation and quantification in experimental kidney histopathology.J Am Soc Nephrol. 2021; 32: 52-68
- Glomerulosclerosis identification in whole slide images using semantic segmentation.Comput Methods Programs Biomed. 2020; 184: 105273
- Nephrotic Syndrome Study Network (NEPTUNE): development and evaluation of deep learning-based segmentation of histologic structures in the kidney cortex with multiple histologic stains.Kidney Int. 2021; 99: 86-101
- Computerized image analysis of Sirius Red-stained renal allograft biopsies as a surrogate marker to predict long-term allograft function.J Am Soc Nephrol. 2003; 14: 1662-1668
- What is the best way to measure renal fibrosis? a pathologist's perspective.Kidney Int Sup. 2014; 24: 9-15
- Renal interstitial fibrosis: mechanisms and evaluation.Curr Opin Nephrol Hypertens. 2012; 21: 289-300
- Morphometric and visual evaluation of fibrosis in renal biopsies.J Am Soc Nephrol. 2011; 22: 176-186
- Deep learning-driven quantification of interstitial fibrosis in digitized kidney biopsies.Am J Pathol. 2021; 8: 1442-1453
- Association of pathological fibrosis with renal survival using deep neural networks.Kidney Int Rep. 2018; 3: 464-475
- Delta analysis of posttransplantation tubulointerstitial damage.Transplantation. 2004; 78: 434-441
- Inflammation lesions in kidney transplant biopsies: association with survival is due to the underlying diseases.Am J Transpl. 2011; 11: 489-499
- Infiltrates in protocol biopsies from renal allografts.Am J Transpl. 2007; 7: 356-365
- Deep learning identified pathological abnormalities predictive of graft loss in kidney transplant biopsies.Kidney Int. 2021; 101: 288-298
- ANHIR: automatic non-rigid histological image registration challenge.IEEE Trans Med Imaging. 2020; 39: 3042-3052
Article Info
Publication History
Publication stage
In Press Journal Pre-ProofFootnotes
Supported by the ERACoSysMed initiative (project SysMIFTA) as part of the European Union's Horizon 2020 Framework Program offered by ZonMw grant 9003035004 and the Dutch Kidney Foundation project DEEPGRAFT grant 17OKG23 and project DIAGGRAFT grant 21OK+012 . The development of the lymphocyte detection network and the automated analysis pipeline was supported by the Dutch Cancer Society Fund Alpe dHuZes' AQUILA project grant KUN 2014-7032 .
Disclosures: J.A.W.M.v.d.L. is a member of the advisory boards of Philips, the Netherlands, and ContextVision, Sweden, and received research funding from Philips, the Netherlands, ContextVision, Sweden, and Sectra, Sweden, in the last 5 years. He is chief scientific officer and shareholder of Aiosyn BV, the Netherlands. F.C. is advisory board member and shareholder of Aiosyn BV, the Netherlands, and received consulting fees from TRIBVN Healthcare, France.
Identification
Copyright
User License
Creative Commons Attribution (CC BY 4.0) |
Permitted
- Read, print & download
- Redistribute or republish the final article
- Text & data mine
- Translate the article
- Reuse portions or extracts from the article in other works
- Sell or re-use for commercial purposes
Elsevier's open access license policy