Ja. Thie et al., LINEAR LEAST-SQUARES COMPARTMENTAL-MODEL-INDEPENDENT PARAMETER-IDENTIFICATION IN PET, IEEE transactions on medical imaging, 16(1), 1997, pp. 11-16
Citations number
19
Categorie Soggetti
Engineering, Biomedical","Radiology,Nuclear Medicine & Medical Imaging
A simplified approach involving linear-regression straight-line parame
ter fitting of dynamic scan data is developed for both specific and no
nspecific models, Where compartmental-model topologies apply, the meas
ured activity may be expressed in terms of: its integrals, plasma acti
vity and plasma integrals-all in a linear expression with macroparamet
ers as coefficients, Multiple linear regression, as in spreadsheet sof
tware, determines parameters for best data fits, Positron emission tom
ography (PET)-acquired gray-matter images in a dynamic scan are analyz
ed: both by this method and by traditional iterative nonlinear least s
quares, Both patient and simulated data were used, Regression and trad
itional methods are in expected agreement, Monte-Carlo simulations eva
luate parameter standard deviations, due to data noise, and much small
er noise-induced biases, Unique straight-line graphical displays permi
t visualizing data influences on various macroparameters as changes in
slopes, Advantages of regression fitting are: simplicity, speed, ease
of implementation in spreadsheet software, avoiding risks of converge
nce failures or false solutions in iterative least squares, and provid
ing various visualizations of the uptake process by straight line grap
hical displays, Multiparameter model-independent analyses on lesser un
derstood systems is also made possible.