Year: 2006 | ![Adobe Acrobat Reader Adobe Acrobat Reader](https://webarchive.library.unt.edu/eot2008/20090114063945im_/http://www.lhncbc.nlm.nih.gov/lhc/images/lhc/misc/getacro.gif)
Download Free Adobe Acrobat Reader |
2006-004 |
Biomedical Multimedia Database Projects |
Rodney Long |
2006-01-09 |
The area of Biomedical Multimedia Databases is related to Content-Based Image Retrieval in the medical image R&D work in the Communications Engineering Branch. Our Biomedical Multimedia Database work focuses on state-of-the-art technical approaches to digitize, store, disseminate, and visually enhance biomedical images for users who are primarily in the fields of research, clinical practice, and education, but may be also include users in the general public with other interests. Content-Based Image Retrieval focuses on advanced techniques to query image databases by the direct use of image content. In both areas, we currently work with two main data sources: (1) data from the National Health and Nutrition Surveys (NHANES) II and III, including 17,000 digitized x-ray images from NHANES II; and (2) data from the National Cancer Institute's Guanacaste Project, which includes longitudinal data collected on about 10,000 women in Guanacaste, Costa Rica. This is a region with high incidence of cervical cancer, and the collected data includes 60,000 cervicography images, along with several thousand associated histology and PAP smear images. The NCI Guanacaste data is the content source for new applications being developed, including the Boundary Marking Tool, which allows the manual identification of important regions in uterine cervix cervicography images, the Virtual Microscope, which enables the viewing and evaluation of cervix histology slides over the Internet, the Multimedia Database Tool, which will allow query and retrieval of any of the Guanacaste images collected by NCI, along with the associated clinical text data, and the NCI Teaching Tool, which is a Web-based teaching and evaluation tool being created to train clinical workers in consistent techniques for the visual evaluation of cervicography and colposcopy images for cervical cancer screening. The work that we are doing in the area of Biomedical Multimedia Databases is complemented by work being done in parallel to develop efficient techniques to allow information retrieval by direct query of image contents. This Content-Based Image Retrieval capability is expected to be become an additional and powerful capability for biomedical information retrieval in the future. With this new capability, a user of the WebMIRS of the future would be able to search directly for images with particular visual characteristics-for example, images with vertebrae having bone spurs (osteophytes) or dislocation (subluxation). Our research in the area of Content-Based Image Retrieval is described in a separate presentation on this topic. |
|