A machine learning based algorithm accurately stages liver disease by quantification of arteries.

Area of Science:

• Hepatology
• Medical Imaging
• Digital Pathology

Background:

• Cirrhosis involves liver architecture loss and vascular changes.
• Quantifying these changes is crucial for staging liver disease.
• Existing methods may not fully capture vascular alterations.

Purpose of the Study:

• To develop and validate qVessel, a tool for quantifying arterial density (AD) in liver biopsies.
• To assess the relationship between AD, collagen proportionate area (CPA), and liver fibrosis stage across different etiologies.
• To explore the potential of AD as a novel biomarker for liver disease staging.

Main Methods:

• Utilized qVessel algorithm on second harmonic generation/two-photon images from unstained liver biopsy slides.
• Quantified arterial density (AD) and collagen proportionate area (CPA).
• Analyzed samples from patients with chronic hepatitis B (CHB), primary biliary cholangitis (PBC), and metabolic dysfunction-associated steatotic liver disease (MASLD) staged using METAVIR.

Main Results:

• Both AD and CPA significantly increased with fibrosis progression (F1-F4) across all three etiologies (P < 0.05).
• No significant differences in AD or CPA were observed between CHB/PBC and MASLD at any fibrosis stage (P > 0.05).
• AD and CPA demonstrated comparable diagnostic efficacy for liver cirrhosis and correlated significantly with METAVIR stages.

Conclusions:

• qVessel effectively quantifies AD in liver biopsies.
• AD and CPA are reliable indicators of fibrosis progression and staging across diverse liver diseases.
• Arterial density shows promise as a novel component for liver disease staging systems.