SPECT (Single Photon Emission Computed Tomography) techniques have bee
n applied to a wide range of medical studies, The stability of a SPECT
model depends strongly upon the data collected. We show that a SPECT
model is full rank and well-conditioned (stable) if the projection dat
a are large enough. Condition number estimates for a linear model are
given. Numerical results for a class of linear models confirm our theo
retical analysis.