The arrival of high speed digital electronics and communication technologies to radiological science is changing the methods of acquiring, storing, viewing, and transmitting diagnostic images. Currently, about 20% of radiologic images are acquired digitally. These include computed tomography (CT), digital fluorography (DF), ultrasound (US), nuclear medicine (NM), and magnetic resonance (MR) images. There is also a trend that the remaining 80%, consisting of conventional projectional x-ray procedures, will gradually be replaced by computed radiography (CR). As a result, a very large volume of digital diagnostic images will be generated. It is therefore essential to investigate methods for compressing these images into a more compact form before storage and transmittal. For this reason, a two-phase research plan to study radiologic image compression was planned four years ago. The first phase, which is the current funding period, was to complete four specific aims. They are: (1) refinement of irreversible image compression methods, (2) generation of a database for image compression study, (3) design and implementation of a compression hardware module, and (4) conduct a visual perception study on the quality of reconstruction images resulting from compression of images, selected form the database. The completion of the first phase demonstrates that radiologic image compression is technologically possible. The second phase, which is the current application, has two primary goals. The first is to establish that the quality of images reconstructed from compressed data is acceptable for diagnosis for all imaging modalities. The second is to demonstrate that the hardware compression module can be integrated into a radiologic image management system to be used in clinical environment. To achieve these goals the experimental design and methods is comprised of four steps. These are: (1) extended development of compression module so that it can compress images from all radiologic imaging modalities; (2) demonstration of compression in clinical environment; (3) extension of database to include images from all modalities for compression study, and (4) image quality evaluation study of images from all modalities. By the end of the second phase study it is expected that the following can be firmly established: (1) Irreversible image compression with compression ratio of 10:1 or higher for a 2,048 image is achievable. The image reconstructed from the compressed data will not have any degradation in diagnostic quality. (2) The compression times will be four seconds for 2,048 radiographic images, one second for 1,024 images, and less than a half-second for 512 CT or 256 MR images. (3) Image compression can be implemented in clinical environment.
Showing the most recent 10 out of 22 publications
