A combined Bayesian approach to classifying venous flow during contrast-agent injection using doppler ultrasound

Abstract

The administration of intravenous contrast media during CT examinations is routine, but carries with it a risk of extravasation. With a new Doppler ultrasound monitoring technique, we propose a method for automatic classification of injection flow states. The method combines a Bayesian network and a sparse kernel classifier. The network captures the dependencies between latent variables, observations and previous system states. The sparse kernel classifier is a Relevance Vector Machine that is well suited for spectral analysis and which provides a probabilistic estimate. We present preliminary results showing a challenging input signal variance and how the method applies to empirical data. © 2008 Springer-Verlag.