Texture based region annotation in HRCT lung images
Pramod Kumar Singh, Paul Compton · 2006
Content-based image retrieval (CBIR) in medicine is critically important in patient diagnosis, case-based patient management and medical teaching. Annotation plays a crucial role for the indexing and performance of a CBIR system. Prevalentretrievaltechniquesinvolvehuman-supplied text annotations of the content of an image regions. We are developing a system which annotates and indexes the regions of an image as soon as it is registered into the system. The proposed approach has two main components. At first a submitted image is partitioned into homogeneous regions by a segmentation technique. Then the segmented regions are annotated using a knowledge base system (KBS). The KBS contains pattern signatures created by machine learning and heuristic rules acquired from the domain experts.