Related Experiment Video
Updated: Jun 1, 2026
09:16
Supervised Machine Learning for Semi-Quantification of Extracellular DNA in Glomerulonephritis
Published on: June 18, 2020
Two-Stage Deep Learning Networks for Diagnosing and Staging Membranous Glomerulonephritis From Electron Microscopy
Mihai Gabriel Constantin1, George Terinte-Balcan2, Ioana Maria Lambrescu2
1AI Multimedia Lab, National University of Science and Technology Politehnica Bucharest, Bucharest, Romania.
Summary
A new AI model using transmission electron microscopy images can accurately detect and stage membranous glomerulonephritis, a kidney disease. This deep learning approach shows promise for improving diagnosis in clinical settings.
Area of Science:
- Nephrology
- Medical Imaging
- Artificial Intelligence
Background:
- Membranous glomerulonephritis is a leading cause of nephrotic syndrome in adults.
- Accurate staging of membranous glomerulonephritis is crucial for effective treatment and prognosis.
- Current diagnostic methods may benefit from advanced computational tools.
Purpose of the Study:
- To develop a deep-learning-based artificial intelligence model for detecting and staging membranous glomerulonephritis.
- To utilize transmission electron microscopy (TEM) images as the primary data source.
- To enhance diagnostic accuracy and efficiency for this kidney condition.
Main Methods:
- A two-stage deep learning model utilizing Vision Transformer networks was developed.
- The first stage identified membrane regions, while the second stage classified disease stage.
- Image-level classification was determined by aggregating region-level predictions.
Main Results:
- The model achieved high performance metrics, including an accuracy of 0.9231, sensitivity of 0.9240, and specificity of 0.9167.
- Strong generalization was observed on an independent external dataset, with an accuracy of 0.8905.
- The model demonstrated robustness when evaluated by different practitioners.
Conclusions:
- The proposed two-stage model offers enhanced accuracy, robustness, and clinical interpretability for staging membranous glomerulonephritis.
- This AI tool shows significant potential for integration into routine diagnostic workflows for kidney diseases.
- The study highlights the utility of deep learning in analyzing complex medical images like TEM micrographs.