 
Summary: A measure of the information content of EIT data
Andy Adler1
, Richard Youmaran2
, William R.B. Lionheart3
1
Systems and Computer Engineering, Carleton University, Ottawa, Canada
2
School of Information Technology and Engineering, University of Ottawa, Canada
3
School of Mathematics, University of Manchester, U.K.
Email: adler@sce.carleton.ca
Abstract. We ask: how many bits of information (in the Shannon sense) do we
get from a set of EIT measurements? Here, the term information in measurements
(IM) is defined as: the decrease in uncertainty about the contents of a medium, due
to a set of measurements. This decrease in uncertainly is quantified by the change
from the the interclass model, q, defined by the prior information, to the intraclass
model, p, given by the measured data (corrupted by noise). IM is measured by the
expected relative entropy (KullbackLeibler divergence) between distributions q and
p, and corresponds to the channel capacity in an analogous communications system.
Based on a Gaussian model of the measurement noise, n, and a prior model of the
