In vivo quantification of arterial active mechanics using deep learning-assisted pressure-area analysis

Area of Science:

• Cardiovascular Physiology
• Biomedical Engineering
• Medical Imaging

Background:

• Active arterial mechanics, driven by vascular smooth muscle, are vital for cardiovascular health and disease.
• Current methods struggle to isolate smooth muscle's contribution to arterial stiffness.
• Quantitative assessment of active arterial properties is challenging.

Purpose of the Study:

• To develop and validate a novel framework for quantifying active arterial mechanics.
• To investigate the recovery of active arterial mechanics post-exercise.
• To explore the relationship between blood pressure and smooth muscle tone during recovery.

Main Methods:

• Developed a pressure-area analysis framework integrating ultrasound, blood pressure, and neural networks.
• Utilized neural network-based segmentation for arterial cross-sectional area.
• Employed biomechanical model-driven inversion to infer active mechanical properties.
• Recruited 233 volunteers for network training and 10 for exercise testing.

Main Results:

• The neural network demonstrated robust performance in segmenting arterial areas.
• Active arterial mechanics remained elevated for ~15 minutes post-exercise, longer than systolic pressure (~5 minutes).
• A dissociation between blood pressure and smooth muscle recovery was observed.

Conclusions:

• The proposed framework effectively quantifies active arterial mechanics.
• Post-exercise recovery shows distinct temporal patterns for blood pressure and vascular smooth muscle activity.
• Findings provide new insights into vascular smooth muscle regulation under physiological stress.