Zebra bodies recognition by artificial intelligence (ZEBRA): a computational tool for Fabry nephropathy

Area of Science:

• Nephrology
• Digital Pathology
• Artificial Intelligence

Background:

• Fabry disease (FD) is a rare genetic disorder causing kidney damage (Fabry nephropathy).
• Diagnosing FD, especially in certain populations, is challenging, necessitating expert renal biopsy interpretation.
• Digital pathology and AI present opportunities to enhance diagnostic accuracy and efficiency.

Purpose of the Study:

• To develop and validate an AI tool for screening "foamy podocytes" in renal biopsies from Fabry nephropathy patients.
• To create a quantitative score (ZEBRA score) for assessing disease burden.
• To evaluate the AI tool's performance in classification and segmentation tasks.

Main Methods:

• Analysis of whole-slide images from renal biopsies using AI for glomerular classification and podocyte segmentation.
• Development of a novel ZEBRA score (fpA/tgA%) to quantify disease burden.
• Performance evaluation using standard metrics and correlation analysis with histological and clinical data.

Main Results:

• The AI tool achieved 79% accuracy in identifying foamy podocytes.
• The ZEBRA score effectively differentiated Fabry nephropathy from controls (p < 0.001).
• The ZEBRA score demonstrated good correlation with manual histological scoring (rs = 0.66-0.71).

Conclusions:

• An AI-assisted ZEBRA pipeline can effectively screen for high-risk features in Fabry nephropathy.
• This tool supports nephropathologists by highlighting key features for diagnosis.
• AI in digital pathology offers a promising approach to improve Fabry nephropathy diagnosis.