Related Experiment Video
Updated: May 17, 2026
An Efficient and Fast Method for Labeling and Analyzing Mouse Glomeruli
Published on: February 9, 2024
Glo-MMF: A modular multi-model framework for automated morphometry of glomerular ultrastructural features.
Zhentai Zhang1, Danyi Weng1, Guibin Zhang1
1School of Biomedical Engineering, Southern Medical University, Guangzhou, 510515, China; Guangdong Provincial Key Laboratory of Medical Image Processing, Southern Medical University, Guangzhou, 510515, China; Guangdong Province Engineering Laboratory for Medical Imaging and Diagnostic Technology, Southern Medical University, Guangzhou, 510515, China.
We developed Glo-MMF, a modular framework for automated glomerular ultrastructural analysis. This system simultaneously quantifies key features, improving renal pathology diagnostics and research efficiency.
Area of Science:
- Renal pathology and diagnostic imaging analysis.
- Application of deep learning in medical image quantification.
Background:
- Automated analysis of glomerular ultrastructures aids diagnosis by reducing pathologist workload and enhancing accuracy.
- Existing single-model approaches struggle with the heterogeneity of multiple quantification objectives (structure measurement, state assessment, lesion localization).
Purpose of the Study:
- To develop Glo-MMF, a modular framework for integrated automated analysis of glomerular ultrastructures.
- To jointly quantify key ultrastructural features for renal pathology research and diagnostic assistance.
Main Methods:
- Glo-MMF utilizes three deep learning models: ultrastructural segmentation, glomerular filtration barrier (GFB) region classification, and electron-dense deposits (EDD) detection.
- A post-processing workflow with computer vision modules integrates model outputs for feature measurement, including adaptive GFB cropping and measurement location screening.
- This modular approach enhances measurement reliability and provides interpretable quantitative results.
Main Results:
- The Glo-MMF framework, trained on 372 images, simultaneously quantifies glomerular basement membrane (GBM) thickness, foot process effacement (FPE) degree, and EDD location.
- Automated quantification showed strong agreement with pathological reports across 115 test cases and 9 renal pathological types.
- Analysis per case (GBM thickness, FPE degree, EDD location) averaged 4.23 ± 0.48 seconds on a CPU.
Conclusions:
- The modular Glo-MMF framework supports flexible extensibility for joint quantification of multiple glomerular ultrastructural features.
- It demonstrates robust performance and clinical applicability across diverse renal pathologies, offering an efficient auxiliary role in glomerular pathological analysis.
