Functional renal imaging through factor analysis

Abstract

Functional images tend to be noisy, since they are formed from parameter values estimated from noisy time-activity curves. Factor analysis provides a rapid method for fitting smooth curves to these noisy curves. Noise in functional images is reduced by estimating parameter values from the smooth curves. The method is illustrated for three parameters: TMAX (time to maximum value), RISE (increase from first to maximum value), and RISMX (maximum increase between successive values). When curve-fitting through factor analysis is used to generate functional renal images from clinical studies or to estimate parameter values for simulated noisy renogram curves, noise is reduced for the TMAX and RISMX parameters and accuracy is improved for the RISE parameter.