An H(∞) approach for elasticity properties reconstruction

Area of Science:

• Biomechanics
• Medical Imaging
• System Identification

Background:

• Quantifying object elasticity is crucial for technical applications and medical disease diagnosis.
• Accurate elasticity parameter recovery from noisy kinematic data is challenging due to high dimensionality.
• Existing methods like least-square (LS) and extended Kalman filter (EKF) have limitations with noise and initialization.

Purpose of the Study:

• To develop a robust system identification paradigm for quantitative analysis of object elasticity.
• To improve the accuracy of elasticity parameter recovery in the presence of unknown noise types and levels.
• To overcome the limitations of traditional LS and EKF methods in elasticity quantification.

Main Methods:

• The proposed method is derived from H(∞) filtering principles, offering robustness against unknown disturbances.
• System identification is employed for quantitative analysis of object elasticity properties.
• The approach is designed to handle large parameter dimensionality inherent in elasticity quantification.

Main Results:

• The novel H(∞) filtering-based method demonstrates superior performance compared to LS and EKF algorithms.
• The proposed method reliably identifies object elastic modulus distributions even with various noise types (Gaussian, Poisson) and levels.
• Sensitivity analyses using synthetic data confirm the robustness and effectiveness of the new approach.

Conclusions:

• The developed framework provides a powerful tool for accurate elasticity quantification in complex scenarios.
• Validation with phase contrast imaging of canine hearts and human MRI data showcases the practical applicability.
• The method enhances the potential for improved medical disease diagnosis through precise elasticity measurements.