Digital Knowledge and Diagnostic Information

Extract

Digital representation of diagnostic imagery offers mensuration, numeric data for diagnostic clue expression, objective assessment, and the option to define standards. Quantitative measurement allows the detection and documentation of very small differences and of diagnostic information that is visually not perceived. The automated extraction of such information from microscopic imagery is beginning to yield to knowledge guided image processing and the development of image understanding systems for machine vision. In a machine vision system, knowledge guidance may be provided by an expert system that controls a top to bottom scene segmentation with constant checks on local bottom-up derived segmentation results for compliance with model specifications and final scene reconstruction. Knowledge guidance is based on a knowledge file for the fully autonomous processing of scenes from a given domain. The knowledge file includes all entities representing traditional diagnostic and histologic terms and concepts: epithelium, stroma, lumen, nucleus, secretory cell, basal cell, stroma cell, chromatin, to name a few.