The concept of digital radiology is ascending from a laboratory exploration into common practice. A completely digital radiology department must integrate all the imaging modalities: computed tomography (CT), magnetic resonance (MR), digital fluorography (DF), ultrasound (US), and computed radiography (CR) into one network interlinked picture archiving and communication system (PACS). Image compression methods for archival and transmission are essential for implementing a clinically viable PACS. During our two previous funded phases of research on full-frame cosine transform based compression, we have achieved excellent results in terms of image quality, compression ratio, compression speed, and clinical implementation, however, the emphasis has been on static square-shaped images only. During the third phase, we will extend the realm of our compression method to general areas of applications involving arbitrary image formats as well as dynamic images. Both types of images are common in radiological practice. Our approach to dynamic image compression will be styled after but not complaint with the JPEG and MPEG standards. Specifically, we will make full-frame customizations to eliminate block artifacts associated with these two standards. Quality issues for dynamics images will require innovative methodologies to be developed alongside the compression technique, 3D CT and MR images will be treated separately as volumetric sets and not as dynamic sequences. This category as well as images with odd formats will be handled with creative implementation of cosine transforms. Finally, we will conclude our previous quality studies on static images with a massive and organized ROC evaluation involving multiple international sites.
The specific aims to achieve these goals will be: (1) Extension of the algorithm to 3D and 4D image sets, (2) Extension of the algorithm to non-square images, (3) Development of theory and hardware module for real time compression, (4) Exploring filtering techniques during image compression and decompression, (5) Develop protocols for quality evaluations of dynamic images, (6)Collaborative image quality studies on compressed static images. By the end of the third phase, we believe that we will have accomplished the following: (1) developed the theory and hardware implementation for compressing every type of radiological images including static images of all sizes and shapes, and dynamic images with both open-ended and finite sequences, (2) developed techniques for image filtering in frequency domain to be applied during image decompression, (3) confirmed the diagnostic accuracy of compressed static images based on international consensus, (4) developed protocols for the quality evaluation of compressed dynamic images.
Showing the most recent 10 out of 22 publications
