A computational pipeline for clustering left atrial appendage morphology via elastic shape analysis

Area of Science:

• Cardiology
• Medical Imaging
• Computational Biology

Background:

• Left atrial appendage (LAA) morphology varies among patients with atrial fibrillation (AF).
• These morphological variations are linked to differing ischemic stroke risks.
• Current methods for LAA morphology description are qualitative and inconsistent, hindering research.

Purpose of the Study:

• To develop a quantitative pipeline for categorizing LAA morphology in AF patients.
• To overcome the limitations of current qualitative LAA descriptions.
• To improve understanding of stroke pathogenesis in AF.

Main Methods:

• Utilized elastic shape analysis to quantify LAA geometry.
• Employed unsupervised learning for clustering and categorization of LAA shapes.
• Applied the pipeline to LAA morphology data from AF patients.

Main Results:

• The quantitative pipeline reliably clustered LAA morphologies based on geometric features.
• Demonstrated consistent categorization of LAA shapes.
• Provided a robust method for objective LAA morphology assessment.

Conclusions:

• The developed quantitative pipeline offers an objective approach to LAA morphology categorization in AF.
• This method addresses the inconsistencies of qualitative descriptions.
• Facilitates advancements in stroke risk stratification for AF patients.