The analysis of image sequences in nuclear medicine: 2 - factor analysis methods

This article is a review of the Factor Analysis methods used in Nuclear Medicine. These methods consist of processing temporal or energetic dynamic scintigraphic series with Descriptive Multivariate Analysis methods. The principle for using these methods is the following: the dynamic series is regarded as a table in which each row is a dixel and each column is an image. The three methods used in Nuclear Medicine are Principal Components Analysis and Factor Correspondence Analysis which are Orthogonal Analysis methods and Factor Analysis which is an Oblique Analysis method. The results of Orthogonal Analysis consist of a limited number of non correlated factor curves and images which can be used for generating parametric images, for nosologic classification, for smoothing or for data compression. The software of Oblique Analysis used in Nuclear Medicine is called factor Analysis of Dynamic Structures (FADS). It generates a limited number of correlated oblique factor images and curves which are estimates of Dynamic Structures and of kinetics in these structures