Computers in Biology and Medicine 36 (2006) 857 – 870 www.intl.elsevierhealth.com/journals/cobm Building medical ontologies by terminology extraction from texts: An experiment for the intensive care units Jean Charlet a, b, ∗ , Bruno Bachimont c , Marie-Christine Jaulent a a INSERM, U729, F-75006 Paris, France b Mission Recherche STIM/AP-HP, Paris, France c Institut National de l’Audiovisuel, Direction de la Recherche, 94366 Bry-sur-Marne Received 5 April 2005; accepted 5 April 2005 Abstract In many medical fields, maintenance, comparison and aggregation of unambiguous terminologies go through formal specialized clinical terminologies: ontologies. We describe a methodology to build medical ontology from textual reports using a natural language processing tool, the SYNTEX software. The methodology is illustrated in the surgical intensive care medical domain. We have tested the possibility for an expert to build a sizeable ontology in a reasonable time. The quality of the ontology has been evaluated according to its capacity to cover the ICD-10 terminology in the field. Finally, the methodology itself is discussed.  2005 Elsevier Ltd. All rights reserved. Keywords: Ontology; Natural language processing; NLP tools; Semantic commitment; Thesaurus; Top ontology 1. Introduction Surgical intensive care is a medical domain specialized in the treatment of post-operative complications and in traumatology. This field includes various pathologies and acts. They are classified in thesauri to help the physician to code his/her activity. Variability in coding is a well-known problem due in part to the ambiguity of thesauri [1]. Some work has been done on automatic coding tools to reduce ∗ Corresponding author. INSERM U729, Université Paris Descartes, Faculté de médecine, 15 rue de l’école de médecine75006 Paris, France. Tel.: +33 1 53 10 92 10; fax: +33 1 53 10 92 01. E-mail addresses: jean.charlet@spim.jussieu.fr, charlet@biomath.jussieu.fr, jc@biomath.jussieu.fr (J. Charlet). 0010-4825/$ - see front matter  2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.compbiomed.2005.04.012