DEVO: an ontology to assist with dermoscopic feature standardization

Area of Science:

• Medical Informatics
• Dermatology
• Ontology Engineering

Background:

• Dermoscopic analysis is crucial for skin disease diagnosis by physicians and AI.
• The rapid, uncontrolled expansion of dermoscopy vocabulary hinders standardization.
• A domain-specific ontology is needed to formally represent dermoscopic feature knowledge.

Purpose of the Study:

• To develop a domain-specific ontology for formal knowledge representation of dermoscopic features.
• To standardize the complex and evolving vocabulary of dermoscopy.

Main Methods:

• Developed a two-phase ontology: a fundamental-level ontology for visualization elements (shapes, colors) and a domain ontology for dermoscopic metaphorical terms.
• Formalized definitions of dermoscopic terms using the fundamental-level ontology.

Main Results:

• The Dermoscopy Elements of Visuals Ontology (DEVO) comprises 1047 classes and 63 properties.
• DEVO achieved a superior semiotic score compared to existing ontologies in the same domain.
• Human annotators validated DEVO's consistency, complexity, and applicability.

Conclusions:

• The DEVO ontology successfully defines metaphoric dermoscopic terms by breaking them into visual components.
• Future applications include educational tools for trainees and diagnostic support for dermatologists.
• The ontology aims to facilitate query responses and integrate features for improved skin disease diagnosis.