To compare
the information content of different MR sequences often regions of interests (ROI)
are used. A way to avoid observer-dependencies is to use ROIs from different
observer, but this leads to the question how to fuse them. We propose a new
method to combine the information obtained by multiple ROIs depending on their
similarity, making our method less sensitive to outlier. We evaluate our method
by comparing the results obtained from the traditional merging method with the
proposed algorithm. The results indicate that our method can be a valuable
extension to ROI-based, multi-observer studies.
