? In this project we hypothesize that a comprehensive clinical case library with intelligent agents can sort and render clinically similar cases and present clinically significant features, to assist the radiologist in interpreting mammograms. This computer library system is designed to integrate, search, and analyze clinical data that are analogous to the study case. In normal clinical practice, experienced radiologists often refer to their personal mental images of previous proven cases in making diagnoses and patient management decisions. Providing a computerized library can refresh a radiologist's mental memory with a broad array of proven cases and concrete visualizations. The system can also assist less experienced radiologists in making diagnoses by referring them not only to histologically proven cases but also to the statistical distribution of features. Freed from bias induced by their recent experience, radiologists can use such a library system to improve the diagnostic accuracy at large. This approach differs greatly from conventional computer-aided detection (CAD) or computer-aided diagnosis (CADx) methods. One can also expect that by adding the radiologist's guidance - for example BI-RADS descriptors - into the proposed mammographic library system will lead to increase the searching accuracy and achieve a greater diagnostic outcome. Technically speaking, this project differs from previous mammographic CADx and image-based retrieval methods in its emphasis on using local vector features and sector fuzzy features of the mammographic masses supported by our newly-invented maximum likelihood fuzzy shadow techniques and multiple circular path convolution neural network. In addition, the recent technical advances in image retrieval using the fuzzy feature matching method and hyper-space analysis techniques have made the success of the proposed approach highly possible. This computer-based library system can be extended to include other imaging modalities for breast care. It is conceivable that this comprehensive breast care database can further link local breast care centers through the coming high-speed network system. In addition, this system could also serve as a learning center for clinicians and scientists. This project represents a significant trend in the development of a comprehensive image data system for breast care in the next several decades. ? ? ?

Agency
National Institute of Health (NIH)
Institute
National Cancer Institute (NCI)
Type
Exploratory/Developmental Grants (R21)
Project #
1R21CA102960-01A2
Application #
6969741
Study Section
Special Emphasis Panel (ZRG1-SBIB-J (90))
Program Officer
Baker, Houston
Project Start
2005-09-01
Project End
2007-08-31
Budget Start
2005-09-01
Budget End
2006-08-31
Support Year
1
Fiscal Year
2005
Total Cost
$179,881
Indirect Cost
Name
Georgetown University
Department
Radiation-Diagnostic/Oncology
Type
Schools of Medicine
DUNS #
049515844
City
Washington
State
DC
Country
United States
Zip Code
20057
Wei, Jun; Chan, Heang-Ping; Sahiner, Berkman et al. (2009) Computer-aided detection of breast masses on mammograms: dual system approach with two-view analysis. Med Phys 36:4451-60