Workshop Organizers:
Carol A. Bean, NIH (cabean@mail.nih.gov)
Ram D. Sriram, NIST (sriram@nist.gov)

Advances in molecular and genomic imaging technologies in combination with anatomical and functional medical imaging technologies are poised to have a tremendous impact on health care. Hence, it is important to integrate effectively various types of images generated by these technologies into clinical decision-making. Standards have enormous potential to facilitate and coordinate such integration and interoperability of biomedical imaging applications and practices for clinical decision-making. However, the development and implementation of standards are extremely resource-intensive and can take years to accomplish; further, standards evolve so they must be tracked and maintained through ongoing efforts.

Numerous standards exist in the health care arena, most of which focus on a specific biomedical or clinical subject domain at some level of granularity or a specific type of health care encounter. Within each domain, there are standards that may address the meanings of terms and the relationships between them, the overall conceptual model for the domain or enterprise, or the ways in which information must be formatted in order to be transmitted electronically and used by machines.

The DICOM family was designed primarily as a data interchange standard to deal with the acquisition and transportation of digital diagnostic images and associated information over networks. Today, virtually all imaging modalities in radiology (e.g., CT, MRI, ultrasound, RF, PET, etc.) support the DICOM standard. An interest in linking all images associated with a single patient into some form of integrated master patient record is spurring other specialty domains (e.g., pathology, ophthalmology, dermatology, gastroenterology, etc.) to develop and implement extensions to the DICOM standard. Similarly, the HL7 series comprises a set of data interchange formats for use in other applications of the health care enterprise.

Additional dimensions of standardization beyond data interchange formats are important for using interpretations of these images in the context of clinical decision-making. These semantic dimensions correspond to the meaning of terms used to express concepts, the relationships that obtain between and among concepts and the knowledge structures used to represent them, and various means of modifying the attributes of both concepts and relationships.

The primary purpose of this workshop is to explore the potential of semantic standards for enhancing meaningful integration of biomedical imaging into the clinical decision-making process.

The following themes are offered to guide the discussions:

1. Clinical decision-making relies increasingly on information gathered from a variety of sources. Because of differences in perspective and tradition, health care providers in different specialty domains that use images (such as radiologists, surgeons, internists, etc.) tend to use different terms for describing the same anatomical features and landmarks as well as for clinical observations and interpretations. How can standards be used to integrate and coordinate the annotation of medical images with clinical observations from different specialty domains?

2. The same image object corresponding to a set of measurable observations may be interpreted in a different manner, yielding variation or conflict in treatment. Further, these interpretations are likely to be highly contextual, depending on the patient data, anthropological gene pool, clinical objectives, and so on. Often these criteria address degree or severity; for example, the same tumor image object may be interpreted as mild by an oncologist and as severe by a radiologist. To what extent do such criteria that influence decision-making need to be standardized? How can ontologies be used for standardization of decision-making criteria? Are existing standards sufficient for these tasks?

While discussing the above, we would like to address the following questions: What are the deficiencies and what enhancements and/or extensions are needed to existing standards? What obstacles (both technical and organizational) inhibit sharing, coordinating, and integrating biomedical image data across modalities and across sites for support of clinical decision-making? What near-term actions can be taken to overcome these obstacles for timely application?