Towards a “Surgical GPS”: Combining Surgical Ontologies with Physician-Designated Anatomical Landmarks

Abstract

In Computer Assisted Surgery (CAS) the application of generic surgical planning models on individual patient cases is often limited by the number of cases incorporated into the models, as well as the variation that exists between surgical departments in individual hospitals due to local practices. This study seeks to provide a software foundation for the integration of physician-designated anatomical landmarks with existing surgical ontologies for surgical process modeling, in conjunction with Resource Description Framework (RDF) models. Various ontologies were combined with physician-verified landmarks to form surgical process model components, encoded as RDF triples, using the software Karma. These generic, but approach-specific surgical procedure models have been augmented by the physician-submitted landmark information, which is unique for both the patient of interest and the workflows adopted by the physicians themselves. The landmark information is collected through a simple graphical user interface (GUI) using a basic form submission plugin from a website. The data are then integrated into existing ontologies as distinct values through Karma. These newly constructed RDFs are validated using the Apache Jena Fuseki software tool for interoperability, based on their compatibility with SPARQL (SPARQL Protocol and RDF Query Language). Anatomical landmark datasets were easily submitted through the Author’s public website GUI, which also provides a baseline set of generic landmark information and notifies the user of changes made to this default dataset. These datasets and associated ontologies were then successfully joined in Karma to produce viable Terse RDF Triple Language (Turtle) entries. These Turtles specify surgeon-submitted anatomical landmarks, which are now represented by the object, the third part of the semantic triple. Apache Jena Fuseki accepted these models and successfully implemented SPARQL queries that allow users to obtain critical information based on content filtering. The information obtained through SPARQL may be utilized in conjunction with medical imaging-based multi-surface anatomical models to enhance patient care and descriptively guide surgeons through procedures by highlighting the anatomical landmarks along a surgical corridor. Augmented surgical ontologies may enhance the outcome of surgical interventions through their use within the field of computer-assisted surgery (CAS) and can provide a representation of the methods present in pre-operative, intra-operative and post-operative procedures. A combination of physician-determined, patient-specific, landmarks and verified ontologies, ultimately produce a Turtle that can be incorporated into any combination of surgical workflows. Upon development of the “Surgical GPS”, which will utilize these patient-specific anatomical landmarks and personalized multi-surface anatomical models, surgeons will navigate a patient’s surgical corridor through RDF-based highlighting of patient anatomy.