Computation of semantic similarity within an ontology of breast pathology to assist inter-observer consensus.

Computer-assisted consensus in medical imaging involves automatic comparison of morphological abnormalities observed by physicians in images. We built an ontology of morphological abnormalities in breast pathology to assist inter-observer consensus. Concepts of morphological abnormalities extracted from existing terminologies, published grading systems and medical reports were organized in an taxonomic hierarchy and furthermore linked by the relation "is a diagnostic criterion of" according to diagnostic meaning. We implemented position-based, content-based and mixed semantic similarity measures between concepts in this ontology and compared the results with experts' judgment. The position-based similarity measure using both taxonomic and non-taxonomic relations performed as well as the other measures and was used for automatic comparison of morphological abnormalities within the IDEM computer-assisted consensus platform.

Integration of multiple ontologies in breast cancer pathology.

The diagnostic variability in pathology, widely reported in the literature, is partly due to the use of different classification systems by pathologists. The descriptions of morphological characteristics on the same image within different classification systems can be considered as different points of view of pathologists. Our aim is to represent the points of view of the experts in pathology during image interpretation and to propose a method ological and technical solution in order to implement interoperability between these points of view. According to the hybrid ontology approach, we developed a system in three stages consisting in 1) the representation of the various points of view in local ontologies 2) the realization of a shared vocabulary and the development of a mapping tool used to allow the matching of local ontologies and shared vocabulary 3) the development of a transcoding algorithm for the translation of a case description from one point of view to another. A first evaluation of the transcoding algorithm was conducted for 33 cases of breast pathology. Our results show that the pathologists generally produce descriptions of the cases which do not follow rigorously the interpretation rules corresponding to the point of view they assert to adopt. While most of the concepts of local ontologies can be transcoded from a local ontology to another one (varying from 62.5 % to 100% according to the local ontology), the transcoding of a description which is valid according to a certain point of view, often results in a description which is not rigorously in accordance with the new point of view. These results underline the differences of interpretation rules existing in the different points of view.