Temporal image fusion

Presents a method to make dynamic image analysis by image fusion. This method is unsupervised and does not require previous knowledge of the number of relevant parameters as in statistical methods. It uses set operations to segment recursively a given region if at any following time this region appears not to be homogeneous. This algorithm has been tested with success on two sets of positron emission tomography images (a set of 10 images of a phantom and a set of 22 images of a human brain). It allows an automatic region of interest detection and it displays a kinetic image that could help image interpretation.