The use of ontologies to facilitate semantic annotation of medical images has been a widely used approach. A particular limitation of this approach is the lack of ontologies with a high level of completeness, mainly because the problem that ontology evolution represent. This article proposes an approach that facilitates the evolution of ontologies from the contributions made by domain experts through social tagging of medical images. The method guides the collaborative process during the discovery of change. Additionally, we present a tool build on Web Protegeto support the proposed method.