ESTIMATION OF TISSUE PROPORTIONS IN X-RAY CT IMAGES USING A NEW MIXED PIXEL DISTRIBUTION

Abstract

A new probability distribution is obtained, termed the mixed pixel distribution, appropriate for the mixing proportion in digital images when the point spread function has a bivariate normal density. It is used to derive the distribution of pixel values in X-ray CT images where pixels may be a mixture of two tissue types. In a simulation study it is shown that, by fitting this distribution to histograms of pixel values, tissue proportions are estimated more accurately than using threshola-based methods.