Reconstruction of elasticity: a stochastic model-based approach in ultrasound elastography

Abstract

BackgroundThe convectional strain-based algorithm has been widely utilized in clinical practice. It can only provide the information of relative information of tissue stiffness. However, the exact information of tissue stiffness should be valuable for clinical diagnosis and treatment.MethodsIn this study we propose a reconstruction strategy to recover the mechanical properties of the tissue. After the discrepancies between the biomechanical model and data are modeled as the process noise, and the biomechanical model constraint is transformed into a state space representation the reconstruction of elasticity can be accomplished through one filtering identification process, which is to recursively estimate the material properties and kinematic functions from ultrasound data according to the minimum mean square error (MMSE) criteria. In the implementation of this model-based algorithm, the linear isotropic elasticity is adopted as the biomechanical constraint. The estimation of kinematic functions (i.e., the full displacement and velocity field), and the distribution of Young’s modulus are computed simultaneously through an extended Kalman filter (EKF).ResultsIn the following experiments the accuracy and robustness of this filtering framework is first evaluated on synthetic data in controlled conditions, and the performance of this framework is then evaluated in the real data collected from elastography phantom and patients using the ultrasound system. Quantitative analysis verifies that strain fields estimated by our filtering strategy are more closer to the ground truth. The distribution of Young’s modulus is also well estimated. Further, the effects of measurement noise and process noise have been investigated as well.ConclusionsThe advantage of this model-based algorithm over the conventional strain-based algorithm is its potential of providing the distribution of elasticity under a proper biomechanical model constraint. We address the model-data discrepancy and measurement noise by introducing process noise and measurement noise in our framework, and then the absolute values of Young’s modulus are estimated through the EFK in the MMSE sense. However, the initial conditions, and the mesh strategy will affect the performance, i.e., the convergence rate, and computational cost, etc.

Cross-talk between shoulder and neck pain

“…[ 21 ] In addition to traditional B-mode and Doppler US imaging, the development of elastography further allows the physicians to evaluate elastiticy of tendons and muscles on shoulder girdles. [ 22 – 24 ] A recent US study using shear wave elastography demonstrated tendon softening in rotator cuff disorders. [ 25 ] In the present study, we targeted a population with rotator cuff tendon tears, which are common and more easily defined than other shoulder pathologies on the basis of high resolution US imaging.…”

Systematic feasibility analysis of a quantitative elasticity estimation for breast anatomy using supine/prone patient postures

“…The direct approach reconstructs an elastographic image by converting the strain value at each voxel to a relative Young's modulus using a simple model constraint, such as Hooke's law. 21 This approach is limited, however, and has been shown to be inferior when compared to the inverse approach. 22 The inverse approach, or model-based method, is an iterative approach that allows for more reasonable and realistic model constraints.…”

A survey of ultrasound elastography approaches to percutaneous ablation monitoring

“…93 Accurate and reproducible determination of the modulus distribution of thermal lesions would be preferred because it would provide quantitative modulus information as opposed to the qualitative information offered by displacement vector and strain tensor images. 93 Lu et al 65 proposed a stochastic model-based modulus reconstruction method for modulus imaging and applied it to visualize the modulus distribution of the thermal lesion produced by RFA therapy of liver cancer in vivo. It was shown that the thermal lesion reflected by the modulus image is more in line with the position of the liver tumor in the CT image than reflected by the strain image.…”