Follow us
Latest News & Events
AI-Assisted Diagnosis of Pediatric Kidney Disease with Digital Pathology
Introduction: Improving Pediatric Kidney Disease Diagnosis with AI
Pediatric kidney disease is a serious global health concern.
Many cases develop silently. Therefore, early and accurate diagnosis is essential for effective treatment.
Currently, renal biopsy remains an important method for diagnosing kidney diseases. However, traditional biopsy analysis has several challenges.
The process is often:
- Time-consuming
- Labor-intensive
- Dependent on specialist experience
In addition, differences between observers may affect diagnostic consistency.
As the number of pediatric kidney disease cases continues to grow, laboratories need more efficient diagnostic tools.
To address these challenges, KFBIO collaborated with the Children’s Hospital of Zhejiang University School of Medicine to develop an artificial intelligence (AI)-based diagnostic system.
The system, called AI-based Pediatric Kidney Diagnosis (APKD), helps pathologists analyze kidney tissue images with greater speed and consistency.
By combining AI technology with digital pathology, APKD provides a new approach for pediatric kidney disease diagnosis.
The Challenge: Diagnosing Pediatric Kidney Disease
Chronic Kidney Disease (CKD) affects millions of children worldwide.
The disease may present with symptoms such as:
- Hematuria
- Proteinuria
- Growth problems
However, accurate diagnosis often requires renal biopsy and detailed examination by experienced nephropathologists.
This process creates several challenges.
First, kidney biopsy analysis requires significant time and expertise.
Second, the global shortage of specialized pathologists increases diagnostic pressure.
Furthermore, the increasing number of pediatric CKD cases creates a growing demand for smarter diagnostic solutions.
Therefore, AI-assisted pathology has become an important direction for improving pediatric kidney disease diagnosis.
Introducing APKD: An AI-Based Pediatric Kidney Diagnosis System
To overcome these challenges, KFBIO and Zhejiang Children’s Hospital jointly developed the AI-based Pediatric Kidney Diagnosis (APKD) system.
APKD is a deep learning framework designed to analyze kidney tissue images from whole slide images (WSIs).
The system can:
- Identify kidney structures
- Classify pathological changes
- Quantify important tissue features
Unlike traditional manual analysis, APKD can process large amounts of pathology data automatically.
As a result, pathologists can obtain more consistent and efficient diagnostic support.
A Large-Scale Dataset Supports AI Training
A reliable AI system requires high-quality training data.
Therefore, researchers developed APKD using a comprehensive pediatric kidney pathology dataset.
The model was trained with:
- 2,935 pediatric patient cases
- 93,932 manually annotated histological structures
This large dataset helps the system recognize different kidney structures and pathological features.
As a result, APKD has become one of the most comprehensive AI models developed for pediatric kidney pathology.
How APKD Works: A Multi-Module AI Framework
APKD uses multiple AI modules to analyze different aspects of kidney tissue.
Each module focuses on a specific diagnostic task.
Together, these modules provide comprehensive pathology analysis.
1. Kidney Structure Segmentation
The first module identifies important kidney structures.
It can detect:
- Arteries
- Tubules
- Glomeruli
- Glomerular tufts
This helps create a detailed map of kidney tissue.
2. Nuclei Segmentation
The second module focuses on cell-level analysis.
It identifies and counts nuclei within tissue samples.
Therefore, it supports quantitative evaluation of cellular changes.
3. Glomerular Crescent Classification
The third module analyzes glomerular changes.
It distinguishes between:
- Crescentic glomeruli
- Non-crescentic glomeruli
This feature is important for evaluating progressive kidney inflammation.
4. Mesangial and Endothelium Region Segmentation
The fourth module identifies areas with increased cellular activity.
It analyzes mesangial and endothelial regions in PAS-stained images.
As a result, pathologists can better evaluate structural changes associated with kidney disease.
Clinical Validation: Improving Kidney Pathology Analysis with AI
After developing APKD, researchers evaluated its performance using pediatric kidney biopsy samples.
The results demonstrated that AI can significantly improve pathology analysis efficiency.
Compared with traditional manual evaluation, APKD provided faster and more consistent analysis.
Furthermore, the system showed strong performance across different diagnostic tasks.
Key Performance Results of APKD
Accurate Kidney Structure Detection
One of the most important steps in kidney pathology analysis is identifying glomeruli.
APKD achieved excellent performance in glomerular detection.
The system reached:
- 99% accuracy in glomerulus detection
- High consistency across different tissue samples
As a result, APKD can help pathologists quickly locate and evaluate important kidney structures.
Faster Pathological Feature Analysis
Traditional kidney biopsy analysis requires careful manual examination.
However, reviewing large numbers of tissue structures can be time-consuming.
With AI assistance, APKD can automatically analyze important pathological features.
The system achieved approximately:
5.5 times faster analysis efficiency compared with manual assessment.
Therefore, AI can help laboratories reduce repetitive work and improve diagnostic workflow.
Supporting More Consistent Diagnosis
Pathology diagnosis often requires expert interpretation.
However, differences in experience may lead to variations in evaluation.
APKD provides standardized image analysis based on trained AI models.
As a result, it can support more consistent diagnostic decisions.
Moreover, AI assistance allows pathologists to focus more on complex cases that require expert judgment.
The Role of Digital Pathology in AI-Assisted Kidney Diagnosis
AI pathology requires high-quality digital images.
Therefore, whole slide imaging (WSI) plays an essential role in AI-based diagnosis.
In this project, researchers used the KFBIO KF-PRO-400 Digital Slide Scanner to digitize kidney biopsy samples.
The scanner converts traditional glass slides into high-resolution digital images.
These images provide the foundation for:
- AI model training
- Quantitative tissue analysis
- Automated pathology evaluation
- Digital collaboration
Furthermore, high-quality scanning improves the reliability of AI analysis.
Without accurate digital images, AI systems cannot fully recognize complex tissue structures.
KF-PRO-400: Supporting Advanced Digital Pathology Applications
The KF-PRO-400 Digital Slide Scanner is designed for high-performance digital pathology workflows.
It provides:
High-Resolution Whole Slide Imaging
The scanner captures detailed tissue information from kidney biopsy samples.
Therefore, researchers can analyze both overall tissue structure and cellular-level features.
Efficient Slide Digitization
Large-scale AI projects require efficient image acquisition.
The KF-PRO-400 supports rapid slide scanning and reliable image quality.
As a result, laboratories can create high-quality digital pathology datasets more efficiently.
AI-Ready Imaging Workflow
AI models require standardized image input.
The KF-PRO-400 provides consistent digital images that support AI analysis.
Therefore, it helps connect laboratory scanning workflows with advanced AI applications.
Transforming Pediatric Kidney Disease Diagnosis
The collaboration between KFBIO and Zhejiang Children’s Hospital demonstrates the potential of AI in pediatric pathology.
By combining:
- Artificial intelligence
- Digital pathology
- Whole slide imaging
- Expert pathology knowledge
APKD provides a smarter approach for kidney disease diagnosis.
This technology can help laboratories:
- Improve diagnostic efficiency
- Reduce repetitive manual tasks
- Standardize pathology analysis
Support clinical decision-making
Future of AI in Pediatric Pathology
AI technology continues to expand in healthcare.
In the future, AI pathology systems may support more applications, including:
- Automated disease classification
- Quantitative biomarker analysis
- Treatment response prediction
- Large-scale pathology screening
Furthermore, combining AI with digital pathology will create new opportunities for precision medicine.
Looking ahead, KFBIO will continue developing intelligent pathology solutions to support laboratories worldwide.
Conclusion: Building Smarter Kidney Pathology Workflows with AI
Pediatric kidney disease diagnosis requires accurate, efficient and consistent analysis.
However, traditional workflows face challenges such as time consumption and limited specialist resources.
The collaboration between KFBIO and Zhejiang Children’s Hospital demonstrates how AI can address these challenges.
Through APKD, AI-assisted pathology can improve kidney tissue analysis and support pathologists with faster and more standardized results.
Combined with KFBIO digital pathology technologies, including the KF-PRO-400 Digital Slide Scanner, AI provides a powerful foundation for next-generation pathology workflows.
Ultimately, AI and digital pathology will continue to improve disease diagnosis and support better patient care.
Access the published dataset and learn more:
https://github.com/ChunyueFeng/Kidney-DataSet
Feng, C., Ong, K., Young, D. M., Chen, B., Li, L., Huo, X., Lu, H., Gu, W., Liu, F., Tang, H., Zhao, M., Yang, M., Zhu, K., Huang, L., Wang, Q., Marini, G. P. L., Gui, K., Han, H., Sanders, S. J., Li, L., Yu, W., & Mao, J. (2024). Artificial intelligence-assisted quantification and assessment of whole slide images for pediatric kidney disease diagnosis. Bioinformatics, btad740. https://doi.org/10.1093/bioinformatics/btad740

