Experimental measurement of impulse response and noise for an emission computed tomography system.

Experimental studies have been made of the impulse response and noise characteristics of a tomographic system using a gamma camera. Fourier transform, deconvolution and iterative methods have been used with a CDC 6600 computer to reconstruct images from data obtained for various experimental arrangements of sources in a cylindrical phantom. It is shown that with an appropriate attenuation correction the impulse response in the reconstruction is substantially constant, independent of the position of the source in the phantom and that the reconstruction technique used is of secondary importance. The resolution obtained for the impulse response and the relative noise level throughout the 'non image' part of the reconstructions is shown for different experimental situations. The measured variance in the reconstruction of an extended uniform activity source was found to be somewhat below the theoretical value except at high count densities (above 1,000 counts per image element) where the limit of accuracy of the reconstruction is shown to be imposed by the variation in the camera sensitivity over the field of view.