The reconstruction of 2-D emissivity profiles from soft. X ray tomogra
phy measurements constitutes a highly underdetermined and ill-posed in
version problem, because of the restricted viewing access, the number
of chords and the increased noise level in most plasma devices. An unb
iased and consistent probabilistic approach within the framework of Ba
yesian inference is provided by the maximum entropy method, which is i
ndependent of model assumptions, but allows any prior knowledge availa
ble to be incorporated. The formalism is applied to the reconstruction
of emissivity profiles in an NBI heated plasma discharge to determine
the dependence of the Shafranov shift on beta, the reduction of which
was a particular objective in designing the advanced W7-AS stellarato
r.