Trajectory-based deformation correction in ultrasound images

Abstract

Tissue deformation in ultrasound imaging poses a challenge to the development of many image registration techniques, including multimodal image fusion, multi-angle compound image and freehand three-dimensional ultrasound. Although deformation correction methods are desired to provide images of uncompressed tissue structure, they have not been well-studied. A novel trajectory-based method to correct a wide range of tissue deformation in ultrasound imaging was developed. In order to characterize tissue deformation under different contact forces, a force sensor provides contact force measurement. Template based image-flow techniques were applied to RF A-lines under different contact forces. A two-dimensional displacement trajectory field was constructed, where pixel coordinates in each scan were plotted against the corresponding contact force. Nonlinear extrapolation algorithms are applied to each trajectory to relocate the corresponding pixel to where it would have been had there been no contact, thereby correcting tissue deformation in the images. This method was validated by using a combination of FEM deformation and ultrasound simulation. It was shown that deformation of the simulated pathological tissue could be corrected. Furthermore, nonlinear polynomial regression was found to give better estimates, than linear regression, when large deformation was present. Estimation accuracy was not improved significantly for a polynomial regression larger than second order.

An Iterative Method for Estimating Nonlinear Elastic Constants of Tumor in Soft Tissue from Approximate Displacement Measurements

“…When a tissue is inspected with the US imaging system, the tissue is scanned with respect to the probe; in other words, when the tissue is compressed with the probe, the presented features of the tissue in the image seemingly move upward, although in point of fact they would move downward. In furtherance of simulating the postcompression RF signals and B-mode images in the probe coordinate system, instead of the phantom's surface in contact with the probe, the surface in front of the probe was assigned to be moving [40–42].…”

Quantitative elastography and its application to blood pressure estimation: Theoretical and experimental results

“…The tissue is compressed by approximately 5 percent of the axial length of the tissue and a multi-scale block-matching method, as described by Sun et al, estimates the displacement between each pair of ultrasound B-Mode images [8]. The gradient of the displacement distribution forms the strain image, which indicates an approximate qualitative elasticity map.…”

Deformation-Aware Robotic 3D Ultrasound

“…The tissues with high stiffness can resist deformations, while the tissues with low stiffness will suffer larger deformations during scans. However, most model-based deformation correction methods directly compute the pixel displacements with respect to the contact force (F c ) rather than to the stiffness (k) [2], [3], [7]. Since F c is applied externally, it cannot reflect tissue's deformationresistant capacity.…”

“…To address this drawback, Dahmani et al applied a linear elastic biomechanical model to estimate the personalized mechanical parameters of the tissues along the deformation field [6]. In addition, Sun et al proposed a method to achieve a zeropressure image using an empirical regressive model of US image deformation with respect to the applied force [7]. Nevertheless, the method was developed to correct the deformation in 2D slice.…”