A machine learning approach for quantifying age-related histological changes in the mouse kidney

Area of Science:

• Histopathology
• Computational Biology
• Gerontology

Background:

• Quantifying age-related histological changes is vital for assessing interventions targeting health span.
• Existing methods for evaluating aging in tissues can be subjective and lack reproducibility.
• Developing objective, high-throughput methods is essential for advancing aging research.

Approach:

• Developed and trained a machine learning architecture to detect and quantify aging in mouse kidney histology.
• Validated the model using held-out data, comparing its output to pathologist scores from the Geropathology Research Network aging grading scheme.
• Provided trained classifiers for H&E-stained slides and tutorials for creating custom classifiers for other stains and tissues.

Key Points:

• The machine learning model provides reproducible and quantifiable age scores for histological samples.
• The quantification offers insights into image appearance changes independent of specific aging lesions.
• The architecture enables high-throughput quantification for mouse aging studies, particularly for kidney tissues.

Conclusions:

• Machine learning offers a powerful tool for objective quantification of aging in histological samples.
• This approach enhances the evaluation of interventions aimed at improving health span.
• The developed resources facilitate scalable and reproducible analysis of aging in preclinical research.