Multiple Source Data Fusion in Blood Vessel Imaging

We have developed a processing module which can accept data from a number of different sources about a particular vessel segment and can reconcile and combine (fuse) this data. It can be used to test the hypothesis that two or more vessel segments from different sources are the same and combine the data if they are compatible. Each source provides a description in terms of attributes distributed along the segment. Each attribute is expressed in terms of three probability density functions, one for the attribute position, one for its longitudinal extent, and one for the type of property being measured (for example cross-sectional area or width). The reconciliation process depends on reasoning about evidence against certain property values being correct, as well as evidence in favour.