A three-object model for the similarity searches of chest CT images.

摘要: Summary We propose in this paper a three-object model specifically for the archiving and retrieval of chest CT images. To calculate parameters model, each image needs to be processed segment three main objects then features extracted describe objects' properties relationships. In segmentation part, we applied knowledge modality on images modified traditional watershed algorithm including four-step merging After segmentation, mediastinum two lung lobes are identified. The object is mainly described by shape-related while texture features. A was exploited spatial relationship among objects. test capability similarity searches images, developed CBIR system which distinct query modes were provided. They ‘searching ARGs’, shape mediastinum’, lobes’. experimental results show that demonstrates impressive power searching Among modes, mediastinum’ lobes’ provide user choices search with high similarities specific rather than whole precision rate either mode high, an average around 80% out first 30 result justified as similar, fully automatic using content Nevertheless, concentrate slightly better similar ARGs’ mode.