Automated quantification of renal interstitial fibrosis for computer-aided diagnosis: A comprehensive tissue...

Area of Science:

• Nephrology
• Digital Pathology
• Medical Image Analysis

Background:

• Interstitial fibrosis in renal biopsies indicates chronic kidney disease extent.
• Current visual quantification by pathologists has reproducibility issues.
• Automated systems can provide consistent fibrosis measurements.

Purpose of the Study:

• Develop an automated system for interstitial fibrosis quantification in renal biopsies.
• Enhance diagnostic consistency and reduce inter-observer variability.
• Provide a reliable tool for assessing chronic kidney disease progression.

Main Methods:

• Image analysis system using color and structural features for tissue segmentation.
• Interstitial fibrosis regions identified by eliminating non-fibrotic structures.
• Validation against a pathologist-annotated ground truth dataset.
• Comparison with manual quantification by multiple pathologists.

Main Results:

• Automated system achieved quantification accuracy comparable to pathologists.
• Error rate of the automated system matched intra-observer variability.
• Demonstrated good correlation between automated and visual evaluations.
• Improved correlation among pathologists' estimations when using the system.

Conclusions:

• The automated quantification system is accurate and reliable for renal biopsy analysis.
• It effectively reduces variability in interstitial fibrosis assessment.
• The system serves as a valuable diagnostic aid for pathologists in managing chronic kidney disease.