Skip to main content
eScholarship
Open Access Publications from the University of California

Dermatology Online Journal

Dermatology Online Journal bannerUC Davis

ONTODerm - A domain ontology for dermatology

Main Content

ONTODerm - A domain ontology for dermatology.
Bell Raj Eapen MSc MD DNB
Dermatology Online Journal 14 (6): 16

Dip Derm (UK), Kaya Skin Clinic, Dubai, UAE. webmaster@gulfdoctor.net

Abstract

A specialty specific ontology for dermatology can be a very useful tool for conveying an accurate meaning to collaborative guideline applications and software systems. This paper describes the basics of ontology and the need for ontology in dermatology. We also propose the basic structure for one such ontology called ONTODerm. ONTODerm is grounded on DOLCE Lite foundational ontology and its purpose is to address the unique needs of dermatology as a medical specialty. It is represented in Web Ontology Language (OWL) using the Protégé OWL Plugin. ONTODerm is available for download and collaborative development under GPL license.



Introduction

Ontology is an explicit formal specification of the terms in a particular domain and describes relations among them. Ontology defines the vocabulary and the interrelationships in a particular domain that can be shared. Many medical disciplines have developed general and specialty specific ontologies that domain experts can use to share and annotate information in their fields [1].


Why develop ontology?

Health practitioners are relying more and more on software systems for tasks ranging from electronic medical records maintenance to implementation of treatment guidelines. However, various legacy systems represent and manipulate data in several ways making data exchange, updating, and reuse in a different setting next to impossible [2, 3]. Investigators in the area of software engineering emphasize the importance of making ontologies explicit when building computer programs [4]. Ontology is neither a software program nor a module. It is a definition of the concepts in a domain and the relations among them that can be shared by several software agents so that they will all be based on the same semantics.

It helps to keep the domain knowledge separate from the operational knowledge or the software system so that both can be altered without affecting the other. Any change made to the domain ontology can be easily propagated throughout the system components. It also helps in the reuse of the domain knowledge [1].

The first step in developing an ontology includes defining classes and arranging them in a hierarchy. As a result, relations between these classes are described in an unambiguous way and the possible values are filled.


Relevance of a specialty specific ontology for dermatology

Dermatology is different in many ways from the other medical specialties and has semantics of its own. The terminologies often used by dermatologists to describe lesions are unique and not very familiar to other specialists [5]. Even the medications prescribed by dermatologists and the investigations they perform are different from other specialties. Hence having a specialty specific ontology is essential to integrate dermatology with other medical software systems.


The Web Ontology Language

There are various ways of formally representing ontologies. As the World Wide Web is fast emerging as the popular data sharing platform, it is better to rely on a language that is optimized for the web for medical ontologies as well [6]. The Web Ontology Language is one such ontology language developed by the W3C Web Ontology Working Group [7]. The Web Ontology Language has more facilities for expressing meaning and semantics than other ontology languages like XML, RDF, and RDF-S. Thus, OWL goes beyond these languages in its ability to represent machine interpretable content, especially on the Web. The Web Ontology Language has 3 sublanguages: in order of decreasing expressiveness, they are OWL Full, OWL DL, and OWL Lite.


OWL Basics [8]

The basic components of OWL are classes, properties, individuals and the relationships between them. Classes are the basic building blocks of an OWL ontology arranged in a taxonomic hierarchy. For example Patients and Diseases are 2 good candidates for a top level class in a dermatology ontology. Properties can be categorized as object properties, which relate individuals to other individuals and datatype properties, which relate individuals to datatype values, such as integers, floats, and strings. A property can have a domain and range associated with it. There are various special types of properties such as functional, inverse functional, symmetric and transitive. A functional property can take only one value while two different individuals cannot have the same value in an inverse functional property. If a symmetric property links A to B, then one can infer that it links B to A. If a property links A to B and B to C, then if one can infer that it links A to C, that property is transitive. Individuals are instances of classes; properties can relate one individual to another. Various restrictions can be applied to classes and properties. Cardinality restrictions specify the number of relationships in which a class of individuals can participate. Existential restriction specifies the existence of at least one relationship along a given property to an individual that is a member of a specific class. To restrict the relationship for a given property to individuals that are members of a specific class, a universal restriction must be used. Detailed information about the OWL syntax is available from the W3C website [7].


The Protégé Approach [9]

Ontology development is a difficult task because it has to be represented in a restricted, formal language. Hence a visual development tool with a user friendly view is needed for creating ontology and for its maintenance. Protégé is one such generic and flexible, visual ontology development environment. In the Protégé approach the ontology is represented in a machine-readable format. This makes it possible for software systems to interact directly with the ontology. Protégé's powerful editing interface consists of several tabs that display different aspects of the ontology (Fig. 1). The details about a selected object are displayed by forms and widgets [10]. The functionalities of Protégé can be enhanced using plug-ins.


Figure 1
Figure 1. Screenshot of Protégé 3.1

The Protégé OWL Plugin [11]

The OWL Plug-in is one such plug-in used to load and save ontology in OWL formats and to edit OWL ontologies using owl specific widgets. The Web Ontology Language plug-in provides the functionality to edit OWL classes, properties, forms, individuals, and ontology metadata within Protégé framework in place of the default frame model.


Basics of ontology creation [12]

Dermatology as a medical specialty is a very complex domain for modeling and representing intended meaning. Hence building domain ontology from scratch is a complicated task and it requires a general, foundational ontology to interoperate and support soft modularization. The foundational ontology implements the most appropriate set of principles and speed up the ontology building process as "reinventing the wheel" is avoided [13].

During the specification stage of an ontology it is important to define the purpose of ontology. To make the ontology clear and unambiguous, knowledge units, not directly useful to the domain need to be avoided. It is also important to ensure that the ontology is coherent with consistent definitions, extensible with the addition of new concepts and has minimal encoding bias.

The next step is to formalize the concept and to implement it using the appropriate tools. However, an ontology needs to be updated with new concepts and maintained. A good ontology will mature over time with the contribution from various domain experts.


ONTODerm Methodologies and Proposals

ONTODerm is grounded on the Dolce-Lite-Plus (DOLCE+), a foundation ontology that belongs to the "WonderWeb Foundational Ontology Library" [13]. It is expressed in the OWL DL language using the Protégé owl plug-in. As mentioned before, an ontology should represent only the intended models for its domain of interest for the tasks the ontology is meant to accomplish.

ONTODerm is intended to serve as a specialty specific ontology that will complement the existing core Medical ontologies and the ones that will be developed in the future. Hence, ONTODerm concentrates on 6 important concepts for which dermatology has a unique view of the world different from the other medical specialties. They are:

1. Description of lesions with terms like macule or papule

2. Description of patients with terms like skin type or atopy

3. Investigations (histopathological examination cytodiagnostic techniques, MED determination)

4. Interventions (cosmetic procedures, various surgical treatments)

5. Drugs in the form of lotions, creams or oral medications

6. Grades or stages in the form of PASI, SLE DAI, and others

These six concepts form the top level concepts for ONTODerm and are mapped to the DOLCE+ (Descriptive Ontology for Linguistic and Cognitive Engineering) foundation ontology as per Figure 2. This grounding of top level concepts is a difficult task because the foundation ontology has a broad philosophical view that applies to everything [13]. Moreover, the present grounding may not be the most appropriate one and it can evolve with the help of ontology experts.


Figure 2
Figure 2. The ONTODerm schema with the top level classes

Only a few of the classes and properties are presently defined in ONTODerm. It is a project open to all domain and ontology experts licensed under GPL to contribute and enhance. The links to the project page [14] and ONTODerm wiki [15] are available from the project home page [16].


Probable uses of ONTODerm

ONTODerm is not a software system with a well defined set of uses. It is a system with which other software systems can interact. Hence, it is difficult to predict its possible uses. Potential uses include teaching, decision support for clinical practice, and semantic assistance for data processing tools. An example for the last mentioned use would be a context-sensitive search function for research articles even if they do not contain the correct keyword. Existing context-sensitive image search [17] and medical resource search systems [18] can be made more efficient with ONTODerm.

ONTODerm could have an important application in decision support systems for clinical practice and clinical guideline systems. There are various computer-interpretable guideline models like Asbru, EON, GLIF, GUIDE, PRODIGY and PROforma [19]. Some of them are based on ontologies. The domain ontology in EON has made medical-specialty ontology an explicit component of its core guideline ontology [20, 21]. ONTODerm aims to be such a medical-specialty ontology defining the particular needs of dermatology.

Though the field of medicine has benefited from the recent advances in molecular biology and bioinformatics, a still significant information gap exists between these specialties [22, 23, 24, 25]. ONTODerm is intended to bridge the gap between data at the molecular level with clinico-pathological data.


Other medical ontologies

The National center for Biomedical Ontology is developing algorithms and tools for accessing, visualizing, and analyzing biomedical knowledge and data. They have a virtual library of biomedical ontologies in the OBO format [26].

The ontology project at the Lister Hill National Center for Biomedical Communications aims to verify the formal properties of medical concepts and relationships for consistency and accuracy by using UMLS, SNOMED-RT, GALEN, and Pubmed as the primary knowledge base [27]. The Foundational Model of Anatomy (FMA) is a domain ontology that represents a coherent body of explicit declarative knowledge about human anatomy maintained by the Structural Informatics Group at the University of Washington. Because the classes represented in the Foundational Model of Anatomy generalize to essentially all biomedical domains it can also be considered a foundational ontology [28]. Ontologies have also been used for various other medical tasks [12, 29, 30, 31].


Conclusion

There is no canonical model for how ontologies for any application area can be defined. Ontology development fundamentally is a creative modeling activity of several domain experts that will define the basic domain rules for communication between software systems. ONTODerm is one such step toward a semantic dermatology.

References

1. Yu AC. Methods in biomedical ontology. J Biomed Inform 2006; 39: 252-66. PubMed

2. Coyle JF, Mori AR, Huff SM. Standards for detailed clinical models as the basis for medical data exchange and decision support. Int J Med Inform 2003; 69: 157-74. PubMed

3. Muller ML, Uckert F, Burkle T et al. Cross-institutional data exchange using the clinical document architecture (CDA). Int J Med Inform 2005; 74: 245-56. PubMed

4. Shegogue D, Zheng WJ. Integration of the Gene Ontology into an object-oriented architecture. BMC Bioinformatics 2005; 6: 113. PubMed

5. Boer A. A dictionary of dermatology should serve to clarify terminology but not be yet another textbook of skin diseases! J Am Acad Dermatol 2005; 52: 735. PubMed

6. Grutter R, Eikemeier C, Fierz W et al. Conception of an XML-based ontology for a Web-based medical information service. Stud Health Technol Inform 2000; 77: 832-6. PubMed

7. Deborah L. McGuinness FvH. OWL Web Ontology Language Overview. 2004. http://www.w3.org/TR/owl-features/ [Accessed on 20/06/2006]

8. Smith MK WC, Deborah L. OWL Web Ontology Language Guide. 2004. http://www.w3.org/TR/2004/REC-owl-guide-20040210/ [Accessed on 19/06/2006]

9. Noy NF, Crubezy M, Fergerson RW et al. Protege-2000: an open-source ontology-development and knowledge-acquisition environment. AMIA Annu Symp Proc 2003: 953. PubMed

10. Matthew Horridge HK, Alan Rector, Robert Stevens, Chris Wroe. A Practical Guide To Buiding OWL Ontologies Using The Protege-OWL Plugin and CO-ODE Tools Edition 1.0. 2004. http://www.co-ode.org/resources/tutorials/ProtegeOWLTutorial.pdf [Accessed on 20/05/2006]

11. Vassanyi I, Barcza J, Tobak T. Making two-level standards work in legacy communication frameworks. Stud Health Technol Inform 2005; 116: 689-94. PubMed

12. Dieng-Kuntz R, Minier D, Ruzicka M et al. Building and using a medical ontology for knowledge management and cooperative work in a health care network. Comput Biol Med 2006; 36: 871-92. PubMed

13. MasoloC BS, Gangemi A, Guarino N, Oltramari A Schneider L. The WonderWeb Library of Foundational Ontologies. 2003. http://wonderweb.semanticweb.org [Accessed on 19/06/2006]

14. Eapen BR. ONTODerm - Ontology for Dermatology. 2006. http://www.sourceforge.net/projects/ontoderm/ [Accessed on 19/06/2006]

15. Eapen BR. ONTODerm Wiki. 2006. http://www.gulfdoctor.net/wiki/index.php?title=ONTODerm&lang=en [Accessed on 19/06/2006]

16. Eapen BR. ONTODerm - A Step towards semantic dermatology. 2006. http://www.gulfdoctor.net/ontoderm/ [Accessed on 16/06/2006]

17. Kamel Boulos M. Map of dermatology: Web image browser for differential diagnosis in dermatology. Indian J Dermatol Venereol Leprol 2006; 72: 72-4. PubMed

18. Kamel Boulos M, Roudsari A, Carso N E. HealthCyberMap: a semantic visual browser of medical Internet resources based on clinical codes and the human body metaphor. Health Info Libr J 2002; 19: 189-200. PubMed

19. Peleg M, Tu S, Bury J et al. Comparing computer-interpretable guideline models: a case-study approach. J Am Med Inform Assoc 2003; 10: 52-68. PubMed

20. Tu SW, Musen MA. Modeling data and knowledge in the EON guideline architecture. Medinfo 2001; 10: 280-4. PubMed

21. Musen MA. Domain ontologies in software engineering: use of Protege with the EON architecture. Methods Inf Med 1998; 37: 540-50. PubMed

22. Smith J, Protti D. Bioinformatics meets clinical informatics. Stud Health Technol Inform 2005; 116: 27-32. PubMed

23. Knaup P, Ammenwerth E, Brandner R et al. Towards clinical bioinformatics: advancing genomic medicine with informatics methods and tools. Methods Inf Med 2004; 43: 302-7. PubMed

24. Kohane IS. Bioinformatics and clinical informatics: the imperative to collaborate. J Am Med Inform Assoc 2000; 7: 512-6. PubMed

25. Altman RB. The interactions between clinical informatics and bioinformatics: a case study. J Am Med Inform Assoc 2000; 7: 439-43. PubMed

26. OBO - open biomedical ontologies. 2006. http://obo.sourceforge.net/ [Accessed on 20/06/2006]

27. Chen H FS, Jersh WR, Friedman C. In: Medical Informatics: Advances in Knowledge Management and Data Mining in Biomedicine.: Springer-Verlag, 2005: 211-36.

28. Mejino JL, Jr., Rosse C. Conceptualization of anatomical spatial entities in the Digital Anatomist Foundational Model. Proc AMIA Symp 1999: 112-6. PubMed

29. Steichen O, Bozec CD, Thieu M et al. Computation of semantic similarity within an ontology of breast pathology to assist inter-observer consensus. Comput Biol Med 2006; 36: 768-88. PubMed

30. Henegar C, Bousquet C, Lillo-Le Louet A et al. Building an ontology of adverse drug reactions for automated signal generation in pharmacovigilance. Comput Biol Med 2006; 36: 748-67. PubMed

31. Foss L. Psychooncology: a view from medical ontology. Adv Mind Body Med 1999; 15: 259-63; discussion 75-81. PubMed

© 2008 Dermatology Online Journal