Diagnostic Aid for the Early Detection of Breast Cancers

Mammography's role in the detection of breast cancer is well known. Although more accurate than other existing techniques, mammography's general inability to detect small tumors and other features motivates our study of the Mammograph Display System [MDS]. The MDS is based on the application of wavelets, artificial neural networks, and other advanced signal processing techniques. The MDS will enhance radiologists' perception of mammographic features and improve the accuracy of diagnosis. The Phase I study will lead to significantly enhanced mammogram images which complement existing modalities and allow radiologists to interactively examine diagnostic features. The Phase I study will define a Phase II prototype MDS which will be used in a pre-commercialization clinical testing program. During the Phase I period, we shall demonstrate that wavelets can significantly enhance the visualization of breast carcinomas, improving diagnostic accuracy and the probability of early detection. In addition, we shall demonstrate that this enhanced imaging capability can be achieved in an affordable hardware package. We will also provide the MDS with a prototype radiologist assistant system [RAS] which identifies possible subtle and difficult to detect pathologies and directs the radiologist to their location. An MDS development and experimental system will also be developed.