Clinical CT Data Infrastructure Provides Partial Support for an Operational Ontology for Radiology.
Authors
Affiliations (6)
Affiliations (6)
- Center for Virtual Imaging Trials, Carl E. Ravin Advanced Imaging Laboratories, Department of Radiology, Duke University Medical Center, Durham, NC, USA. [email protected].
- Department of Radiology, Duke University Medical Center, Durham, NC, USA. [email protected].
- Clinical Imaging Physics Group, Duke University Health System, Durham, NC, USA. [email protected].
- Center for Virtual Imaging Trials, Carl E. Ravin Advanced Imaging Laboratories, Department of Radiology, Duke University Medical Center, Durham, NC, USA.
- Department of Radiology, Duke University Medical Center, Durham, NC, USA.
- Clinical Imaging Physics Group, Duke University Health System, Durham, NC, USA.
Abstract
Medical imaging data are increasingly used beyond patient care to support research, quality assessment, and artificial intelligence development. These applications require imaging examinations to be represented in ways that enable consistent aggregation across clinical operations to ensure reliable interpretation. In prior work, we introduced an ontology framework, the Operational Ontology for Radiology (OOR), designed to represent the radiology imaging workflow for computed tomography (CT) examinations through entities and relationships. This investigation evaluates the extent to which existing clinical imaging infrastructure can support the implementation of this framework. Using OOR, 156 data properties representing components of the imaging workflow were assessed within the clinical imaging environment at one academic medical center. Each property was classified according to its availability in institutional data systems into four categories: directly available and standardized, indirectly available and standardized, available but not standardized, or not readily available. The distribution of properties across these categories was quantified overall and across ontology classes. Among the evaluated properties, 48.1% were directly available in structured form, 7.1% were indirectly available and standardized, 33.3% required standardization, and 11.5% were not readily available in the systems. Technical acquisition parameters, device information, and patient attributes were most frequently available, while elements related to imaging tasks, contextual workflow attributes, and certain series-level descriptors required standardization or structured representation. These findings demonstrate that a portion of OOR can be supported using clinical data infrastructures, while also identifying areas where improved standardization and structured data representation are needed to enable large-scale analysis of clinical imaging data.