Structural Informatics involves the systematic classification of 3D reconstructions which can range from ultrasound images, protein structures to anatomical data. This methodology is applied to organise medical data but can be expanded to information areas beyond the field of medicine. The term was coined in 1990 by Jim Brinkley who had been recruited from Stanford to join the Digital Anatomist Program a few years earlier. In 1997, the research group consisting of computer scientists, engineers, and biologists finally was renamed to the UW Structural Informatics Group. The team's major objective constitutes the design of methods for representing both spatial and symbolic information about the physical organisation of the human body for their use in computer programmes, as to solve problems in clinical medicine, research and education.
Within the Digital Anatomist project, these general principles are concretised in the implementation of an anatomy information system whose components are organised along two axes of representation that are spatial/symbolic and data/knowledge. This results in four types of information resources that are accessible via the Internet in a client-server setting:
- The Spatial Database contains spatial information in a coordinate system of one or more dimensions, on individual structural objects, like 2D images, 3D volume data sets, etc.
- The Symbolic Database comprises "text" data, like anatomic terminology, definitions, glossaries, and semantic relationships, that define the corresponding spatial data.
- The Spatial Knowledge Base includes classes of spatial models describing the shape and range of variation within structural objects, like for instance all normal kidneys.
- The Symbolic Knowledge Base encompasses non-spatial information on classes of structural objects, lying at the basis of expert systems or decision models.
The spatial database in the Digital Anatomist's latest version forms an image repository built out of annotated images and animations being packaged into interactive atlases which represent different body regions. The 3D models are manually segmented, reconstructed and annotated by processing the visual input derived from the Visible Human Project or from clinical image volumes. In a later stage, the research team will apply knowledge-based segmentation programmes to automatically generate patient-specific 3D models. There are two ways to retrieve the atlases, whether via CD-ROM or through the Web by a set of Common Gateway Interface (CGI) programmes, referred to as the DA-CGI package.
The Digital Anatomist anatomy information system is already being provided as a learning tool to faculty students at the UW Health Sciences Center, and to universities participating in UW's regional medical education programme, in the four state area, formed by Washington, Alaska, Montana, and Idaho. The ongoing work will particularly concentrate on an optimised integration of spatial and symbolic information. As a result, the authoring tools will gain a higher level of user-friendliness which will stimulate more experts to develop image-based atlases for consultation over the Web. More technical details on this Structural Informatics project are available at the home page of the Digital Anatomist.