Related Experiment Video
Updated: Feb 19, 2026
09:16
Supervised Machine Learning for Semi-Quantification of Extracellular DNA in Glomerulonephritis
Published on: June 18, 2020
7.4K
Development of CD3 cell quantitation algorithms for renal allograft biopsy rejection assessment utilizing open source
Andres Moon1, Geoffrey H Smith1, Jun Kong2
1Department of Pathology, Emory University, Atlanta, GA, USA.
Virchows Archiv : an International Journal of Pathology
|November 9, 2017
Summary
Automated analysis of CD3+ T-cell staining in renal allografts shows promise for reproducible assessment of rejection severity, correlating well with existing methods and potentially improving diagnostic accuracy.
Area of Science:
- Digital pathology and computational analysis in transplantation immunology.
- Quantitative immunohistochemistry for assessing renal allograft status.
Background:
- Accurate diagnosis of renal allograft rejection is crucial for patient outcomes.
- Interstitial inflammation assessment, a key rejection parameter, suffers from interobserver variability.
- Automated image analysis offers potential for reproducible quantitation.
Purpose of the Study:
- To develop and validate customized open-source image analysis methods for quantifying CD3+ T-cell density in renal allografts.
- To compare these novel methods against established commercial algorithms and pathologist visual assessment.
- To evaluate the utility of CD3+ T-cell quantitation in assessing acute cellular rejection (ACR) severity.
Main Methods:
- Whole slide imaging (WSI) of CD3 immunohistochemistry stained renal biopsy slides (n=45) with varying ACR degrees.
- Quantitation using pathologist visual assessment, commercial algorithms (Aperio nuclear, Aperio PPC), and custom ImageJ algorithms (CD3+%, CD3+ cells/mm²).
- Statistical correlation analysis between different quantitation methods and assessment of ACR grade progression.
Main Results:
- All CD3 quantitation algorithms demonstrated adequate accuracy and statistically significant correlations with each other (r=0.44 to 0.94, p<0.0001).
- Methods generally reflected a progression through ACR grades, with custom methods showing distinct results for borderline cases.
- Custom open-source algorithms showed strong correlations with established methods, indicating their potential utility.
Conclusions:
- Customizable open-source image analysis algorithms for CD3+ T-cell assessment are a promising, reproducible alternative to manual evaluation.
- These methods offer a potential 'on-slide flow cytometry' equivalent, enhancing diagnostic accuracy in renal allograft pathology.
- Further validation could integrate these tools into routine diagnostics for improved rejection assessment.
