Relationship Between Size of Injected Microspheres and Ultrasound Imaging of Heart by PLSR

Abstract

Purpose. To investigate the enhancement in ultrasound signal (US) in heart, for diagnostic purposes, as a function of size and number of injected air-filled microspheres by multivariate statistical methods.

Methods. The US enhancement in left ventricle was measured after injection of 31 suspensions of air-filled albumin microspheres divided on and injected in six dogs. The relationship between the size of microspheres between 1−38 μm and the US enhancement was explored by Pearson's product moment correlation analysis, ordinary least-squares regression, principal component regression (PCR) and partial least-squares regression (PLS).

Results. Relative advanced algorithms such as PCR and PLS were required to achieve accurate in vivo/in vitro correlation. The most effective microsphere sizes contributing to US enhancement in left ventricle in dogs were estimated to be in the 7−15 μm range.

Conclusions. In general, the effective in vivo sizes are dependent on the type of formulation due to the surprisingly large activein vivo sizes found for the tested concept. PCR and PLS are suitable methods forin vivo/in vitro correlation, especially for covariated and noisy data.