Deep Learning Algorithms for the Prediction of Posttransplant Renal Function in Deceased-Donor Kidney Recipients: A Preliminary Study Based on Pretransplant Biopsy

Self-Supervised Learning for Feature Extraction from Glomerular Images and Disease Classification with Minimal Annotations

Background

Deep learning has great potential in digital kidney pathology. However, its effectiveness depends heavily on the availability of extensively labeled datasets, which are often limited because of the specialized knowledge and time required for their creation. This limitation hinders the widespread application of deep learning for the analysis of kidney biopsy images.

Methods

We applied self-distillation with no labels (DINO), a self-supervised learning method, to a dataset of 10,423 glomerular images obtained from 384 periodic acid–Schiff-stained kidney biopsy slides. Glomerular features extracted from the DINO-pretrained backbone were visualized using principal component analysis. We then performed classification tasks by adding either k-nearest neighbor classifiers or linear head layers to the DINO-pretrained or ImageNet-pretrained backbones. These models were trained on our labeled classification dataset. Performance was evaluated using metrics such as the area under the receiver operating characteristic curve (ROC-AUC). The classification tasks encompassed four disease categories (minimal change disease, mesangial proliferative GN, membranous nephropathy, and diabetic nephropathy) and clinical parameters such as hypertension, proteinuria, and hematuria.

Results

Principal component analysis visualization revealed distinct principal components corresponding to different glomerular structures, demonstrating the capability of the DINO-pretrained backbone to capture morphologic features. In disease classification, the DINO-pretrained transferred model (ROC-AUC=0.93) outperformed the ImageNet-pretrained fine-tuned model (ROC-AUC=0.89). When the labeled data were limited, the ImageNet-pretrained fine-tuned model's ROC-AUC dropped to 0.76 (95% confidence interval, 0.72 to 0.80), whereas the DINO-pretrained transferred model maintained superior performance (ROC-AUC, 0.88; 95% confidence interval, 0.86 to 0.90). The DINO-pretrained transferred model also exhibited higher AUCs for the classification of several clinical parameters. External validation using two independent datasets confirmed DINO pretraining's superiority, particularly when labeled data were limited.

Conclusions

The application of DINO to unlabeled periodic acid–Schiff-stained glomerular images facilitated the extraction of histologic features that could be effectively used for disease classification.

A digital pathology model for predicting radioiodine-avid metastases on initial post-therapeutic 131I scan in patients with papillary thyroid cancer

Accurate postoperative assessment is critical for optimizing 131I therapy in patients with papillary thyroid cancer (PTC). This study aimed to develop a pathology model utilizing postoperative digital pathology slides to predict lymph node and/or distant metastases on post-therapeutic 131I scan after initial 131I treatment in PTC patients. A retrospective analysis was conducted on 229 PTC patients who underwent total or near-total thyroidectomy and subsequent 131I treatment after levothyroxine (LT4) withdrawal between January 2022 and August 2023. The pathology model was developed through two stages: patch-level prediction and WSI-level prediction. The clinical model was constructed using statistically significant variables identified from univariate and multivariate logistic regression analysis. Of the 229 patients, 19.6% (45/229) exhibited 131I-avid metastatic foci in post-therapeutic 131I scan. Multifactorial analysis identified stimulated thyroglobulin (sTg) as the sole independent risk factor. The AUC of the pathology model in the training and test cohorts were 0.976 (95% CI 0.948-1.000) and 0.805 (95% CI 0.660-0.951), respectively, which were significantly higher than the clinical model (AUC 0.652 and 0.548, Pall < 0.05). This model has the potential to serve as a valuable tool for clinicians in tailoring treatment strategies, thereby optimizing therapeutic outcomes for PTC patients.

Advanced Prediction of Glomerular Filtration Rate After Kidney Transplantation Using Gradient Boosting Techniques

OBJECTIVES

Clinicians often face uncertainty when interpreting whether a decline in estimated glomerular filtration rate is within the patient's expected range of fluctuation or if the decline signals a substantial deviation. Thus, accurate predictions of glomerular filtration rate can be an early warning system, prompting timely interventions, such as biopsies to preclude early graft rejection and adjustments in immunosuppression. Traditional models, encompassing linear and conventional methods, typically struggle with variabilities and complexities in posttransplant data.

MATERIALS AND METHODS

We evaluated the efficacy of a gradient boosting model in predicting posttransplant glomerular filtration rate, to potentially enhance accuracy over traditional prediction approaches. Our patient dataset included 68 pediatric patients aged 1 to 18 years who underwent kidney transplant between 2017 and 2023 at Baskent University Hospital (Ankara, Turkey). The dataset comprised 2285 glomerular filtration rate measurements, along with patient demographics and transplant-related data. For our model, we included "days to transplant" (glomerular filtration rate values pretransplant), "days from transplant" (glomerular filtration rate values up to 7 days posttransplant), patient age, sex, and donor types. We divided the dataset into a training set (70%) and a test set (30%). To evaluate model performance, we used mean absolute error and root mean squared error, with a focus on the accuracy of glomerular filtration rate predictions at various posttransplant stages.

RESULTS

In the training set, the gradient boosting model demonstrated a significant improvement in prediction accuracy, achieving an mean absolute error of ~5.64 mL/min/1.73 m².

CONCLUSIONS

Our model underscored the promise of advanced machine learning techniques in refining prediction of glomerular filtration rate after kidney transplant. With its augmented precision, the model can support clinicians in making informed decisions regarding early biopsies and interventions, thus highlighting the vital role of sophisticated analytical methods in medical prognosis and the monitoring of pediatric patient care.

[Artificial intelligence in kidney transplant pathology]

BACKGROUND

Artificial intelligence (AI) systems have showed promising results in digital pathology, including digital nephropathology and specifically also kidney transplant pathology.

AIM

Summarize the current state of research and limitations in the field of AI in kidney transplant pathology diagnostics and provide a future outlook.

MATERIALS AND METHODS

Literature search in PubMed and Web of Science using the search terms "deep learning", "transplant", and "kidney". Based on these results and studies cited in the identified literature, a selection was made of studies that have a histopathological focus and use AI to improve kidney transplant diagnostics.

RESULTS AND CONCLUSION

Many studies have already made important contributions, particularly to the automation of the quantification of some histopathological lesions in nephropathology. This likely can be extended to automatically quantify all relevant lesions for a kidney transplant, such as Banff lesions. Important limitations and challenges exist in the collection of representative data sets and the updates of Banff classification, making large-scale studies challenging. The already positive study results make future AI support in kidney transplant pathology appear likely.

Deep learning applications for kidney histology analysis

PURPOSE OF REVIEW

Nephropathology is increasingly incorporating computational methods to enhance research and diagnostic accuracy. The widespread adoption of digital pathology, coupled with advancements in deep learning, will likely transform our pathology practices. Here, we discuss basic concepts of deep learning, recent applications in nephropathology, current challenges in implementation and future perspectives.

RECENT FINDINGS

Deep learning models have been developed and tested in various areas of nephropathology, for example, predicting kidney disease progression or diagnosing diseases based on imaging and clinical data. Despite their promising potential, challenges remain that hinder a wider adoption, for example, the lack of prospective evidence and testing in real-world scenarios.

SUMMARY

Deep learning offers great opportunities to improve quantitative and qualitative kidney histology analysis for research and clinical nephropathology diagnostics. Although exciting approaches already exist, the potential of deep learning in nephropathology is only at its beginning and we can expect much more to come.

European Society for Organ Transplantation (ESOT)-TLJ 3.0 Consensus on Histopathological Analysis of Pre-Implantation Donor Kidney Biopsy: Redefining the Role in the Process of Graft Assessment

The ESOT TLJ 3.0. consensus conference brought together leading experts in transplantation to develop evidence-based guidance on the standardization and clinical utility of pre-implantation kidney biopsy in the assessment of grafts from Expanded Criteria Donors (ECD). Seven themes were selected and underwent in-depth analysis after formulation of PICO (patient/population, intervention, comparison, outcomes) questions. After literature search, the statements for each key question were produced, rated according the GRADE approach [Quality of evidence: High (A), Moderate (B), Low (C); Strength of Recommendation: Strong (1), Weak (2)]. The statements were subsequently presented in-person at the Prague kick-off meeting, discussed and voted. After two rounds of discussion and voting, all 7 statements reached an overall agreement of 100% on the following issues: needle core/wedge/punch technique representatively [B,1], frozen/paraffin embedded section reliability [B,2], experienced/non-experienced on-call renal pathologist reproducibility/accuracy of the histological report [A,1], glomerulosclerosis/other parameters reproducibility [C,2], digital pathology/light microscopy in the measurement of histological variables [A,1], special stainings/Haematoxylin and Eosin alone comparison [A,1], glomerulosclerosis reliability versus other histological parameters to predict the graft survival, graft function, primary non-function [B,1]. This methodology has allowed to reach a full consensus among European experts on important technical topics regarding pre-implantation biopsy in the ECD graft assessment.

Multiview Clustering to Identify Novel Kidney Donor Phenotypes for Assessing Graft Survival in Older Transplant Recipients

Key Points

An unsupervised machine learning clustering algorithm identified distinct deceased kidney donor phenotypes among older recipients. Recipients of certain donor phenotypes were at a relatively higher risk of all-cause graft loss even after accounting for recipient factors. The use of unsupervised clustering to support kidney allocation systems may be an important area for future study.

Background

Older transplant recipients are at a relatively increased risk of graft failure after transplantation, and some of this risk may relate to donor characteristics. Unsupervised clustering using machine learning may be a novel approach to identify donor phenotypes that may then be used to evaluate outcomes for older recipients. Using a cohort of older recipients, the purpose of this study was to (1 ) use unsupervised clustering to identify donor phenotypes and (2 ) determine the risk of death/graft failure for recipients of each donor phenotype.

Methods

We analyzed a nationally representative cohort of kidney transplant recipients aged 65 years or older captured using the Scientific Registry of Transplant Recipients between 2000 and 2017. Unsupervised clustering was used to generate phenotypes using donor characteristics inclusive of variables in the kidney donor risk index (KDRI). Cluster assignment was internally validated. Outcomes included all-cause graft failure (including mortality) and delayed graft function. Differences in the distribution of KDRI scores were also compared across the clusters. All-cause graft failure was compared for recipients of donor kidneys from each cluster using a multivariable Cox survival analysis.

Results

Overall, 23,558 donors were separated into five clusters. The area under the curve for internal validation of cluster assignment was 0.89. Recipients of donor kidneys from two clusters were found to be at high risk of all-cause graft failure relative to the lowest risk cluster (adjusted hazards ratio, 1.86; 95% confidence interval, 1.69 to 2.05 and 1.73; 95% confidence interval, 1.61 to 1.87). Only one of these high-risk clusters had high proportions of donors with established risk factors (i.e. , hypertension, diabetes). KDRI scores were similar for the highest and lowest risk clusters (1.40 [1.18–1.67] and 1.37 [1.15–1.65], respectively).

Conclusions

Unsupervised clustering can identify novel donor phenotypes comprising established donor characteristics that, in turn, may be associated with different risks of graft loss for older transplant recipients.

Identify Hard-to-Place Kidneys for Early Engagement in Accelerated Placement With a Deep Learning Optimization Approach

Recommended practices that follow match-run sequences for hard-to-place kidneys succumb to many declines, accruing cold ischemic time and exacerbating kidney quality that may lead to unnecessary kidney discard. Hard-to-place deceased donor kidneys accepted and transplanted later in the match-run sequence may threaten higher graft failure rates. Accelerated placement is a practice for organ procurement organizations (OPOs) to allocate high-risk kidneys out of sequence and reach patients at aggressive transplant centers. The current practice of assessing hard-to-place kidneys and engaging in accelerated kidney placements relies heavily on the kidney donor profile index (KDPI) and the number of declines. Although this practice is reasonable, it also accrues cold ischemic time and increases the risk for kidney discard. We use a deep learning optimization approach to quickly identify kidneys at risk for discard. This approach uses Organ Procurement and Transplantation Network data to model kidney disposition. We filter discards and develop a model to predict transplant and discard of recovered and not transplanted kidneys. Kidneys with a higher probability of discard are deemed hard-to-place kidneys, which require early engagement for accelerated placement. Our approach will aid in identifying hard-to-place kidneys before or after procurement and support OPOs to deviate from the match-run for accelerated placement. Compared with the KDPI-only prediction of the kidney disposition, our approach demonstrates a 10% increase in correctly predicting kidneys at risk for discard. Future work will include developing models to identify candidates with an increased benefit from using hard-to-place kidneys.