Automated Glomerulosclerosis Analysis in Mouse Kidney Sections
The GSI Analyzer – Mice module enables fully automated analysis of PAS-stained kidney sections from mouse models. AI-based image analysis detects glomeruli in whole-slide images and automatically evaluates the degree of glomerulosclerosis.
The results are stored in a structured format and can be analyzed at the level of individual glomeruli, individual sections, or entire mouse cohorts.
Automatic Glomerulus Detection and Sclerosis Classification
The GSI Analyzer automatically detects glomeruli in complete kidney sections and analyzes their morphological properties using trained AI models. Each detected glomerulus is segmented and assigned to a sclerosis class.
The analysis results are visualized directly on the tissue section, allowing detected structures and classifications to be verified immediately. In addition, individual glomeruli are extracted and displayed separately, enabling detailed inspection of each structure.
This visual feedback enables transparent and traceable analysis of large histological datasets.
Structured Quantification of Analysis Results
All detected glomeruli are automatically summarized in structured result tables. For each glomerulus, relevant parameters such as position within the section, classification, and additional quantitative features are stored.
The results can be aggregated at different levels:
• per glomerulus
• per kidney section
• per mouse
• across complete study cohorts
The structured data organization allows results to be efficiently filtered, compared, and exported for further statistical analysis.
Visualization of Sclerosis Distribution
In addition to tabular results, the module automatically generates clear visualizations of the analysis results. Charts show, for example, the distribution of sclerosis classes within a section or across multiple animals.
These visualizations provide a quick overview of experimental results and facilitate comparison between different experimental groups. Changes in the distribution of sclerosis can therefore be identified immediately and interpreted in the context of disease progression or therapeutic interventions.
Automated analysis compared with manual scoring
I scored the GSI myself once more using the microscope. The results correlate very well, which is encouraging. I do not think one could expect more: a similar result would probably have emerged if I had performed the scoring manually twice. This means that, at least in mice, we can have strong confidence in the automated analysis.
