L1-SOUL

L1-SOUL performs time delay estimation (TDE) in quasi-static ultrasound elastography using L1-norm spatial regularization to improve displacement and strain estimation for tissue mechanical-property assessment.

Key Features:

• L1-Norm Regularization: Applies L1-norm regularization to first- and second-order displacement derivatives to preserve boundaries and enhance contrast relative to L2-norm approaches.
• First- and Second-Order Derivatives: Incorporates both first- and second-order displacement derivatives to address blurring at tissue boundaries and improve visual contrast between target tissue and background.
• Handling Non-Differentiability: Smooths the sharp corners of the absolute value function to manage the non-differentiable nature of the L1-norm within optimization.
• Iterative Optimization: Uses an iterative optimization procedure to minimize a nonlinear cost function balancing data constancy and spatial continuity.
• Comparative Performance: Demonstrates improved sharpness and visual contrast compared to GLUE, OVERWIND, and the original SOUL method.
• Quantitative CNR Improvements: Achieves reported contrast-to-noise ratio (CNR) improvements of 67.8% (simulated), 46.81% (phantom), and 117.35% (in vivo) over the original SOUL method.

Scientific Applications:

• Medical imaging research: Enhances elastography-based investigations of tissue mechanical properties in research studies.
• Tissue characterization and pathology detection: Improves boundary definition and contrast to aid detection and diagnosis of pathological changes within tissues.
• Clinical and biomedical research elastography: Provides higher-fidelity displacement and strain maps for clinical assessments and biomedical studies requiring accurate quasi-static ultrasound elastography data.

Methodology:

Optimize a nonlinear energy functional that balances data constancy with spatial continuity constraints using the L1-norm for first- and second-order displacement derivatives; smooth the absolute value to handle non-differentiability and perform iterative optimization to estimate displacement and strain maps between time-series radio-frequency (RF) frames.