MRI-based two-way fluid-structure interaction simulation for discriminating symptomatic carotid atherosclerosis.

Area of Science:

• Biomedical Engineering
• Cardiovascular Imaging
• Computational Fluid Dynamics

Background:

• Carotid artery disease poses a significant risk for stroke.
• Accurate discrimination of symptomatic carotid disease is crucial for timely intervention.
• Current methods may not fully capture the biomechanical factors contributing to plaque vulnerability.

Purpose of the Study:

• To evaluate the added value of two-way fluid-structure interaction (twFSI)-derived biomechanical metrics.
• To enhance artery-level discrimination of symptomatic carotid disease.
• To assess the utility of an attention-based multiple-instance learning framework for integrating imaging and biomechanical data.

Main Methods:

• Retrospective analysis of 97 patients (125 carotid arteries) with high-resolution vessel wall imaging (HR-VWI).
• 3D reconstruction of lumen, vessel wall, and plaque followed by twFSI simulations to derive hemodynamic and structural indices.
• Attention-based multiple-instance learning framework aggregated slice-wise features to artery level for classification.

Main Results:

• A clinical-morphology model achieved an AUC of 0.716.
• Incorporating twFSI-derived biomechanical features (CM-BM) improved discrimination to an AUC of 0.821.
• The CM-BM model identified key descriptors including plaque type, deformation, and wall shear stress.

Conclusions:

• twFSI-derived biomechanical metrics offer incremental discriminative value for symptomatic carotid disease.
• Attention-based MIL aids interpretability by providing slice-level weightings.
• Combined structural-dynamic phenotyping shows potential for symptom-oriented carotid assessment, requiring external validation.