Non-negative constraint for image-based breathing gating in ultrasound hepatic perfusion data

In‐vitro evaluation of accuracy of dynamic contrast‐enhanced ultrasound (DCEUS)‐based parametric perfusion imaging with respiratory motion‐compensation

“…The dynamic kinetic model assumed that the activity of each element V K × T in the DCEUS sequence I H × W × T was a linear combination of kinetic factors P r × T with the factor coefficients Q K × r, as per the following equation:where H , W , and T are the height, width, and frame number of DCEUS sequence, respectively; K = H × W is the pixel number in V ; the number of factor r is 2 in this study. The non‐negative kinetic factor P was treated as a NMF problem and solved by the following optimization equation:where is an approximation algorithm to minimize the distance from QP to V , and the Euclidian distance D ( Q , P ) is the approximation standard.…”

“…Respiratory motion compensation (rMoCo) strategies, including breath‐holding, image registration, and online or offline breathing gating have been extensively investigated to minimize the disturbances of respiratory motion on abdominal DCEUS studies. However, the former three strategies are limited by the long acquisition time, inefficiency for out‐of‐plane motion, unavailable landmarks, extra tracking devices, and missing perfusion details .…”

“…According to respiratory kinetics, displacements are first estimated at each respiratory phase and then TICs are corrected using a compensation algorithm . However, ICA and PCA are limited by an uncertain independent assumption between respiratory and contrast‐enhanced kinetics and by a negative orthonormal component in some cases . With these limitations, correct rMoCo for PPI cannot be guaranteed …”

“…Respiratory kinetics and hemodynamics can be considered as a non‐negative matrix factorization (NMF) problem to address these drawbacks . The performance of NMF on DCEUS has been preliminarily verified through in‐vitro respiratory kinetic estimation, and its feasibility on PPI has been only illustrated by in‐vivo perfusion experiments . The performance of non‐negative rMoCo in in‐vivo PPIs is qualitatively compared against results without rMoCo .…”

“…However, the ground truths of respiratory kinetics and static PPIs without respiratory motion disturbances cannot be determined because of the limitations of in‐vivo perfusion experiments. Thus, it is impossible to evaluate the accuracy of respiratory kinetic estimation and PPIs with rMoCo in in‐vivo studies . This study aimed to modify the DCEUS‐based PPI with non‐negative rMoCo and further evaluate its accuracy through in‐vitro controllable perfusion experiments.…”