VECTOR WIENER FILTER FOR DUAL ISOTOPE SPECT

点击翻译按钮获取中文摘要

DESCRIPTION (Adapted from Applicant's Abstract): Interest in dual-isotope imaging in nuclear medicine is growing, both for studies employing two tracers to simultaneously examine two different processes in a single organ, and for simultaneous transmission/emission imaging in SPECT. A major problem with all dual-isotope imaging is the "crosstalk" between the two isotopes (i.e., the contribution of one isotopes's counts into the other isotope s energy window). The long term goal of the proposed research is to improve the quantitative accuracy and precision of SPECT imaging in the setting of mild disease, where quantification is of most value. Toward that end, the applicants previously described the use of Fourier-based restoration filtering in (single-isotope) SPECT and PET to improve quantitative accuracy. They hypothesized that a significant extension of such filtering can be used in dual-isotope imaging to simultaneously (1) improve contrast in each isotope's "direct" image, (2) reduce image noise, and (3) reduce the crosstalk contribution from the other isotope. Based on the derivation described below, the applicants called such a filter a "vector Wiener filter." The applicants proposed a series of computer simulations and phantom experiments which mimic clinically mild brain and heart disease; specifically: Aim 1: To implement a two-dimensional shift-invariant vector Wiener filter for projection data, and a three-dimensional shift-invariant vector Wiener filter for reconstructed images, and to validate their quantitative accuracy in computer simulations, including realistic brain and cardiac imaging conditions; Aim 2: To assess the effects of vector Wiener filtering on accuracy and precision, in a series of dual-isotope phantom studies, including both dual-isotope brain and cardiac phantom studies, and simultaneous transmission/emission phantom studies; and Aim 3: To study implementation issues, including estimation of power spectral density, to derive a parametric form of the vector Wiener filter, and to derive and initially evaluate a shift-variant form of the filter.