Experimental Validation of Tomographic Gamma Scanning for Small Quantities of Special Nuclear Material*

We have observed that conventional tomographic reconstruction algorithms, such as those used in medical imaging, will give biased results when assaying small quantities of nuclear material. Bias is affected by our treatment of the statistical structure of gamma-ray spectroscopy data. For safeguards applications the bias introduced by some algorithms can be significant when the amount of material is near the accountability limit. Consequently, we are developing alternative statistical methods to analyze tomographic gamma scanning data for small quantities or highly shielded nuclear materials that have reduced bias. The underlying principles are discussed, and results of experiments to characterize the bias for small quantities of 235U and 239Pu are presented.