Deconvolution of shift-variant broadening for Compton scatter imaging

A technique is presented for deconvolving shift-variant Doppler broadening of singly Compton scattered gamma rays from their recorded energy distribut ion. Doppler broadening is important in Compton scatter imaging techniques employing gamma rays with energies below roughly 100 keV. The deconvolution unfolds an approximation to the angular distribution of scattered photons from their recorded energy distribution in the presence of statistical nois e and background counts. Two unfolding methods are presented, one based on a least-squares algorithm and one based on a maximum likelihood algorithm. Angular distributions unfolded from measurements made on small scattering t argets show less evidence of Compton broadening. This deconvolution is show n to improve the quality of filtered backprojection images in multiplexed C ompton scatter tomography. Improved sharpness and contrast are evident in t he images constructed from unfolded signals. (C) 1999 Published by Elsevier Science B.V. All rights reserved.