An improved factor analysis method for cardiac dynamic SPECT studies by using a priori information

Factor analysis of dynamic structures (FADS) can be used to extract information in dynamic studies. This technique is an alternative to commonly used region of interest measurements. The main disadvantage of FADS methods has been that the results were not unique, which made the method not quantitative. Usually, the FADS procedure first determines the low dimensional study subspace in which most of the dixels are contained. This is usually done by performing an orthogonal analysis of the data. The study subspace is spanned by a set of orthogonal basis vectors. The next step is the oblique rotation, which obtains new non-orthogonal and non-negative basis vectors (factors). The oblique rotation must also impose non-negativity on the coordinates of the dixels. The dixel coordinates are also called factor coefficients. The dimensionality of study space is equal to the number of dynamic components in the image. In order to improve quantitation of the solution of FADS, a priori physiological information can be used. In this paper, we develop a method which uses information about the physiological factors. The method is implemented for analysis of data from dynamic cardiac teboroxime-Tc-99m SPECT imaging