Output details

Discriminative segmentation-based evaluation through shape dissimilarity.

<23> This article introduces the concept of using a shape dissimilarity metric for the assessment of segmentation-based evaluation techniques. It is the first work that illustrates that by adding shape dissimilarity to the evaluation criterion it is possible to discriminate image registration results quantitatively which wasn't possible before. The work adds an important component to overcome the lack of sensitivity of criticized but popular segmentation-based evaluation of registration. The implementation of the shape dissimilarity metric has been made publicly available.