OntoNeuroBase is an application ontology that is being developed within the NeuroBase project, which seeks to create a federated system for the management of distributed and heterogeneous information sources in neuroimaging. Having adopted a specific, multi-layered, modular approach to ontology design, we used DOLCE as a foundational ontology together with three core ontologies: I&DA for modelling documents (texts and images), COPS for modelling programs and software and OntoKADS for modelling problem solving activities. Here, we report on how we built OntoNeuroBase by refining the concepts present in the existing modules. Neuroimaging is a very active and rapidly changing field. It is essential to ensure that a newly developed ontology is compatible with other available ontologies and to enable extension of the new ontology to a variety of neuroscience applications. The work reported here is in line with these ambitious objectives.