Cardiac motion recovery: continuous dynamics, discrete measurements, and optimal estimation

Area of Science:

• Biomedical Engineering
• Medical Imaging Analysis
• Control Systems Theory

Background:

• Cardiac motion is a continuous physiological process.
• Medical imaging provides discrete, sampled measurements.
• Reconciling continuous dynamics with discrete data is challenging.

Purpose of the Study:

• To develop a sampled-data filtering framework for cardiac motion recovery.
• To couple continuous cardiac dynamics with discrete imaging measurements.
• To address parameter uncertainty and noise in imaging data.

Main Methods:

• Stochastic multi-frame filtering frameworks were constructed.
• Continuous-time state equations predicted estimates between observations.
• Discrete measurements updated state estimates.
• Continuous-discrete Kalman filter and sampled-data H-infinity filter were applied.

Main Results:

• The framework yields physically meaningful and accurate cardiac motion estimation.
• Sampled-data H-infinity filtering offers robust results without prior noise statistics.
• Validation on synthetic data and canine MR images demonstrated advantages and clinical relevance.

Conclusions:

• The presented sampled-data filtering framework effectively recovers cardiac motion from medical image sequences.
• The approach integrates continuous physiological dynamics with discrete imaging data for improved accuracy.
• The method shows significant clinical relevance for cardiac motion analysis.