Virtual Colonoscopy: An Information Processing in Medical Imaging
Jerome Zhengrong Liang
Professor, Departments of Radiology and Computer Science
State University of New York
Stony Brook,
NY, U.S.A.
Abstract:
We view image-based medical diagnosis, treatment or surgical planning, and follow-up evaluation as an information processing. Images acquired from a patient carry all the subjective and objective information. This information itself may or may not be enough to make a decision. For this reason, physicians’ input as additional information would be useful. However, in more cases, presentation of the patient image information to the physician for clinical assessment is limited by available technologies and, therefore, resulting in many undetermined cases. With advent of advanced computer technologies and sophisticated image processing algorithms, computed aided diagnostic (CAD) means shall process and present the image information accurately and, therefore, decrease dramatically the number of undetermined cases.
This talk presents virtual colonoscopy as an example of information processing in terms of image formation, processing, and visualization. Image formation aims to achieve the best image quality from the measurements. Image segmentation plays a major role in image processing. Given a reconstructed image with finite voxel size, conventional segmentation algorithms label each voxel as a single class or tissue type. This kind classification limits quantitative accuracy on the contents inside the voxels and on the spatial location of tissue boundaries. Current segmentation strategy not only labels each voxel as the conventional algorithms do, but also quantifies the contents as percentages of tissues inside that voxel. Given the segmented mixels and the reconstructed image voxels, the task becomes feature analysis and extraction toward feature-based visualization and CAD. This talk will present these basic concepts and preliminary results.
.