Publications

Manuel Möller, Patrick Ernst, Daniel Sonntag and Andreas Dengel: “Automatic Spatial Plausibility Checks for Medical Object Recognition Results Using a Spatio-Anatomical Ontology”, Proc. of the International Conference on Knowledge Discovery and Information Retrieval (KDIR 2010), 25 – 28 October 2010, Valencia, Spain [BibTex]

Abstract:We present an approach to use medical expert knowledge represented in formal ontologies to check the results of automatic medical object recognition algorithms for spatial plausibility. Our system is based on the comprehensive Foundation Model of Anatomy ontology which we extend with spatial relations between a number of anatomical entities. These relations are learned inductively from an annotated corpus of 3D volume data sets. The induction process is split into two parts: First, we generate a quantitative anatomical atlas using fuzzy sets to represent inherent imprecision. From this atlas we abstract onto a purely symbolic level to generate a generic qualitative model of the spatial relations in human anatomy. In our evaluation we describe how this model can be used to check the results of a state-of-the-art medical object recognition system for 3D CT volume data sets for spatial plausibility. Our results show that the combination of medical domain knowledge in formal ontologies and sub-symbolic object recognition yields improved overall recognition precision.

Manuel Möller, Patrick Ernst, Andreas Dengel: “Spatial Reasoning for Plausibility Checks of Medical Object Recognition Results Using the Foundational Model of Anatomy“, to appear in Proceedings of the 2nd Malaysian Joint Conference on Artificial Intelligence 2010 (MJCAI 2010), Kuching, Sarawak, Malaysia, 26th – 30th July 2010 [BibTex]

Abstract: We present a rule-based system using medical expert knowledge represented in a formal ontology to check the results of automatic medical object recognition algorithms for anatomical plausibility. Our system is based on the comprehensive Foundation Model of Anatomy ontology and uses a set of forward rules executed by a Prolog engine. In our evaluation we describe how this approach can be used to check the results of a state-of-the-art medical object recognition system for 3D CT volume data sets for anatomical plausibility. Our results show that the combination of sub-symbolic object recognition, medical domain knowledge represented in formal ontologies and yields an improved overall recognition precision.

Manuel Möller, Patrick Ernst, Michael Sintek, Sascha Seifert, Gunnar Grimnes, Alexander Cavallaro, Andreas Dengel: “Combining Patient Metadata Extraction and Automatic Image Parsing for the Generation of an Anatomic Atlas”, in Proc. of the 14th International Conference on Knowledge-Based and Intelligent Information & Engineering Systems (KES2010), Cardiff, UK, 8-10 September 2010 [BibTex]

Abstract: We present a system that integrates ontology-based metadata extraction from medical images with a state-of-the-art object recognition algorithm for 3D volume data sets generated by Computed Tomography scanners. Extracted metadata and automatically generated medical image annotations are stored as instances of OWL classes. This system is applied to a corpus of over 750 GB of clinical image data. A spatial database is used to store and retrieve 3D representations of the generated medical image annotations. Our integrated data representation allows to easily analyze our corpus and to estimate the quality of image metadata. A rule-based system is used to check the plausibility of the output of the automatic object recognition technique against the Foundational Model of Anatomy ontology. All combined, these methods are used to determine an appropriate set of metadata and image features for the automatic generation of a spatial atlas of human anatomy.

Manuel Möller, Christian Folz, Michael Sintek, Sascha Seifert, Pinar Wennerberg: “Extending the Foundational Model of Anatomy with Automatically Acquired Spatial Relations”, to appear in Proceedings of the International Conference on Biomedical Ontologies (ICBO), Buffalo, NY, USA, July 24-26, 2009, [PDF on Nature Precedings] [BibTex]

Abstract: Formal ontologies have gained a lot of impact in bioscience over the last ten years. Among them, the Foundational Model of Anatomy Ontology (FMA) is the most comprehensive model for the spatio-structural representation of human anatomy. In the research project THESEUS MEDICO we use the FMA as our main source of background knowledge about human anatomy. Our ultimate goals are to use spatial knowledge about anatomy the FMA to (1) improve automatic parsing algorithms for 3D volume data sets generated by Computed Tomography and Magnetic Resonance Imaging and (2) to generate semantic annotations using the concepts from the FMA to allow semantic search on medical image repositories. We argue that in this context more spatial relation instances are needed than currently available in the FMA.  We present a technique for the automatic inductive learning of missing spatial relation instances by generalizing from expert-annotated volume datasets. The result is stored using the formalism of the FMA and subsequently available for spatial reasoning.