Knowledge-based approach to image interpretation

摘要

This paper presents an automatic interpretation system to support diagnostic biomedical image analysis. The system is organized in three modules: image segmentation; low level object interpretation; and high level biomedical object interpretation, in terms of clinical finding using a semantic model. The domain-specific knowledge used concerns the nailfold capillaries of the human finger, and is organized in a semantic network. User-defined semantic networks may be supplied to extend the system to other problem domains.