The goals of the proposed research are to develop computer-aided diagnostic schemes for detection of microcalcifications in digital mammograms, detection of lung nodules and quantitative analysis of lung textures in digital chest images, and also to develop multiple slit beam imaging techniques with an image intensifier (I.I)-TV digital system for producing high-quality digital chest images. We beleive that these computer-aided diagnostic schemes, which provide the radiologist with the location and/or quantitative measures of suspicious lesions, have the potential to improve diagnostic accuracy in the detection of cancer, by reducing human errors associated with radiologic diagnosis. Specifically, we plan to (1) investigate characteristic features of breast microcalcifications, lung nodules and lung textures; (2) establish a data base for these patterns and their characteristic features; (3) investigate efficient filters for enhancement or suppression of microcalcifications (and lung nodules) for removal of structured background by use of a difference image technique; (4) develop feature-extraction techniques for clusters of microcalcifications (and lung nodules); (5) investigate quantitative measures of lung textures to detect and/or characterize interstitial diseases based on the power spectrum of the underlying texture fluctuations by taking into account the overall background trend and the visual system response of human observers; (6) develop an efficient method for computerized localization of inter-rib spaces for lung texture analysis; and (7) evaluate the effectiveness of computer-aided diagnostic methods compared with radiologists' performance using ROC analysis. For development of multiple slit beam imaging techniques with the I.I.-TV digital system, we plan to (8) (implement and evaluate a continuous beam scan technique to improve image acquisition time and x-ray beam utilization efficiency; (9) investigate high- resolution techniques (2048 x 2048 matrix format) for improvement of spatial resolution properties of reconstructed images; and (10) implement and evaluate a prototype chest unit with a large format (57 cm) I.I.-TV digital system.
Showing the most recent 10 out of 116 publications
