An Unsupervised Learning Tool for Plaque Tissue Characterization in Histopathological Images

Area of Science:

• Digital Pathology
• Computational Biology
• Cardiovascular Research

Background:

• Stroke is a leading global cause of death and disability.
• Carotid artery atherosclerotic plaques are a primary cause of severe cerebrovascular events.
• Accurate histopathological analysis of carotid plaques is crucial for patient stratification and complication prevention.

Purpose of the Study:

• To apply an unsupervised learning approach for analyzing whole-slide images (WSIs) of atherosclerotic carotid plaques.
• To enable simple and fast examination of relevant features within complex carotid plaque WSIs.
• To provide qualitative and quantitative tools to assist pathologists in plaque analysis.

Main Methods:

• Unsupervised machine learning applied to whole-slide images (WSIs) of carotid atherosclerotic plaques.
• Textural-based feature analysis to identify relevant regions within the plaques.
• Development and availability of all analysis code.

Main Results:

• The proposed unsupervised method facilitates a more effective examination of complex WSIs.
• The approach offers both qualitative and quantitative insights into plaque features.
• The study provides a foundation for enhanced diagnostic tools in cardiovascular pathology.

Conclusions:

• The unsupervised learning approach offers a valuable tool for pathologists analyzing carotid atherosclerotic plaques.
• The method aids in the efficient examination of complex whole-slide images.
• Future research should validate these findings using supervised methods and expert annotations.