Abstract
Information system designers face many challenges with regards to selecting appropriate semantic technologies and deciding on a modeling approach for their system. However, there is no clear methodology yet to evaluate “semantically enriched” information systems. In this paper we present a case study on different modeling approaches for annotating medical images and introduce a conceptual framework that can be used to analyze the fitness of information systems and help designers to spot the strengths and weaknesses of various modeling approaches as well as managing trade-os between modeling eort and their potential benefits.
Highlights
Information systems can have very different shapes and variants and designing such systems involves taking important modelling decisions to optimise information retrieval
We will outline five modelling approaches for medical image annotations that range from a simple text-based information system to fullfledged ontology-based information systems based on an established medical ontology, namely SNOMED CT.1
SNOMED CT is strictly speaking an ontology that can be expressed in OWL and conforms to the OWL 2 profile OWL EL [7], we merely use it in the sense of a thesaurus for this modelling approach, i.e. each image is “tagged” with a list of SNOMED CT classes and we only make use of the synonyms and the taxonomical information that are contained in the ontology and do not capture any relational information
Summary
Information systems can have very different shapes and variants and designing such systems involves taking important modelling decisions to optimise information retrieval. The more expressive the schema, the more “meaning” can be potentially modelled in the image annotations This might result in better retrieval performance and in more modelling effort while creating the annotations, e.g. depending on whether or not the information extraction can be done automatically or has to be done manually. There is a trade-off between how much effort is involved in the design of the system and the creation of meaningful image annotations and how this leads to better queriability and better quality of the retrieval results. In order to understand this trade-off and to analyse and compare different approaches for modelling medical image annotations we will use a conceptual framework [9] that has been designed for the evaluation of information systems with a particular focus on queriability. We will outline five modelling approaches for medical image annotations that range from a simple text-based information system to fullfledged ontology-based information systems based on an established medical ontology, namely SNOMED CT. Applying the framework to these modelling approaches will highlight their strengths and weaknesses and the framework’s measurements will allow us to compare the modelling approaches with each other
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.