|
|
I-Browse - An Image Retrieval and Annotation System Based on Semantic Content |
INVESTIGATORS
Dr. Lilian H. Y. Tang, Prof. Horace H. S. Ip, Dr. R. Hanka, Dr. Ringo W. K. Lam, Mr. Kent K. T. Cheung,
BRIEF DESCRIPTION
There is a need to access efficiently large collections of digital images by their content. Medical images pose additional challenges to content-based image retrieval and browsing. I-Browse project aims to develop techniques which enable a physician to search image archives through a combination of semantic and iconic contents.
FUNDING AGENCY
The I-Browse project is supported by the Hong Kong Jockey Club Charities Trust
PUBLICATIONS
Lilian H Y Tang, Rudolf Hanka, Horace H S Ip, "A System Architecture for Integrating Semantic and Iconic Content for Intelligent Browsing of Medical Images", Medical Imaging 1998: PACS Design and Evaluation: Engineering and Clinical Issues, Steven C. Horii, M.D., G. James Blaine, Editors, Proceedings of SPIE 3339,pp. 572-580 San Diego, California, USA, 21-27 February 1998.
Lilian H Y Tang, Rudolf Hanka, H H S Ip, R Lam, "Extraction of Semantic Features of Histological Images for Content- Based Retrieval of Images", Proceedings of SPIE Medical Imaging ?9, Vol. 3662, pp. 360-368, San Diego, 20-26 February 1999.
Lilian H Y Tang, Rudolf Hanka, Horace H S Ip, "A Review of Intelligent Content-Based Indexing and Browsing of Medical Images", Health Informatics Journal, Vol. 5 No.1, pp. 40-49, March 1999.
Lilian H. Y. Tang, R. Hanka, Ringo W. K. Lam, Kent K T Cheung, Horace H. S. Ip, "Automatic Semantic Analysis for Histological Images", Medical Imaging Understanding and Analysis 99 (MIUA'99), pp. 53-56, Oxford, UK, 19-20 July, 1999.
Kent K. T. Cheung, Horace H. S. Ip, Ringo W. K. Lam, R. Hanka, Lilian H. Y. Tang and G. Fuller, "An Object-oriented Framework for Content-based Image Retrieval Using a 5-tier Architecture", Proc. Asia Pacific Software Engineering Conference 99, 7-10 Dec. 1999, Takamatsu, Japan, pp 174-177.
Lilian H Y Tang, Horace H S Ip, Rudolf Hanka, Kent K T Cheung, Ringo Lam, "Semantic Query Processing and Annotation Generation for Content-based retrieval of Histological Images", Proc of SPIE Medical Imaging, 2000, San Diego, CA, USA, 20-26 February 2000. (Cum Laude Best Poster Award)
Tang, L. H. Y., Hanka, R., Ip, H.H.S., Cheung, K.K.T. and Lam, R., "Integration of intelligent engines for a large scale medical image database", Proc. 13th IEEE Symposium on Computer-Based Medical Systems 2000, 22-24 Jun., 2000, Houston, USA, pp 235-240.
Ringo W. K. Lam, Horace H. S. Ip, Kent K. T. Cheung, Lilian H. Y. Tang and R. Hanka, "A Multi-Window Approach To Classify Histological Features", Proc. 15th Intl. Conf. on Pattern Recognition, 3-7 Sep., 2000, Barcelona, Spain, pp 259-262.
Ringo W. K. Lam, Horace H. S. Ip, Kent K. T. Cheung, Lilian H. Y. Tang and R. Hanka, "Similarity Measures for Histological Image Retrieval", Proc. 15th Intl. Conf. on Pattern Recognition, 3-7 Sep., 2000, Barcelona, Spain, pp 295-295.
Kent K. T. Cheung, Ringo W. K. Lam, Horace H. S. Ip, Lilian H. Y. Tang, R. Hanka, "A Software Framework for Combining Iconic and Semantic Content for Retrieval of Histological Images", Proc. 4th Intl. Conf. Visual 2000, 2-4, Nov., 2000, Lyon, France, pp. 488-499.
Lilian H Y Tang, Horace H S Ip, Rudolf Hanka, Ringo Lam, Kent K T Cheung, "An Image Retrieval and Annotation System Based on Semantic Content", Proceedings of 5th World Multiconference on Systemics, Cybernetics and Informatics (SCI 2001), Vol. 1, pp. 311-315, Orlando, Florida, USA, July 22-25, 2001.
Application Domain: Histology of GI Tract
Histological images pose additional challenges
GI tract has a general structure common to all segments
GI tract exhibits regional variations in structure and in magnifications
Various admixtures of different appearances
Five main areas: oesophagus, stomach, small intestine, large intestine (including appendix), and anus
I-Browse also acts like a postman, put the unknown image into a correct pigeon hole
Image Examples
Semantic Features
Coarse histological features
characterised by color and texture
detected by color histogram and Gabor filters
15 coarse semantic labels (5 coarse regions and their junctions)
Fine histological features
characterised by local textures
detected by Gabor filters and other fine feature detectors
76 fine feature labels at the present stage, more frequently only 63 features used
I-Browse Architecture
Automatic Annotation