Feature wise representation for both still and motion 3D medical images

With the arrival of the state of the art medical imaging equipment, a plethora of images are acquired not only in higher dimensions (3D+) but also with various presenting forms of either still or motion, complicating data management systems even further. This paper offers, from an application point of view, representations of content features from both still 3D MR brain images and 3D ultrasound cardiac video sequences by demonstrating a developed online content-based image retrieval system, MIRAGE. The approaches of 3D SIFT coupled with sparse code have been appointed to facilitate the representation of image features, whereas widely applied four texture based approaches are also implemented to allow users' benefit of different retrieving intentions.