Coronary arteriosclerosis is a leading cause of death in U.S., and it is to a large degree the presence or absence of severe coronary stenoses which determines the risk of death. Because many patients who suffer from this disease are asymptomatic until their first attack, a safe, inexpensive diagnostic procedure for detecting coronary artery stenoses would be beneficial. The diagnosis as currently carried out requires that a large catheter be inserted into the orifice of each artery in order to inject opaque dye to make the arteries visible with X rays. The increased contrast sensitivity of digital subtraction angiography (DSA) permits a less invasive technique involving an injection into the aortic root, a smaller catheter, and reduced dye volumes, but it has been of only limited success largely because of image degradation caused by patient motion during image acquisition.
The aim of this project is to develop and test image processing algorithms for the improvement of digital angiography of coronary arteries. The algorithms search for a geometrical transformation to produce an optimal registration of two images before subtraction. They are based on a new application of the fundamental principles of fluid dynamics, a new class of transformation functions, a new method of machine calibration, and new application of search methods developed in the field of artifical intelligence for function optimization. As a result these algorithms represent the first approach to image registration which is capable of handling true three dimensional warping. The development and evaluation of the algorithms will be carried out on both a specially designed phantom and on data derived in a new way from patient studies. The patient studies will be acquired during angioplasty procedures in which an inflated balloon and catheter will serve as phantom arteries of known dimensions. The availability of known phantoms located in the chest on a beating human heart will provide for the first time an opportunity for quantitative evaluation of the success of the image enhancement algorithms in a realistic environment. Thus, the proposed research will employ new techniques of image registration with a new technique for image evaluation to the problem of reducing motion effects in digital subtraction angiography.

Agency
National Institute of Health (NIH)
Institute
National Heart, Lung, and Blood Institute (NHLBI)
Type
Research Project (R01)
Project #
1R01HL034703-01A1
Application #
3347900
Study Section
Diagnostic Radiology Study Section (RNM)
Project Start
1986-07-01
Project End
1989-06-30
Budget Start
1986-07-01
Budget End
1987-06-30
Support Year
1
Fiscal Year
1986
Total Cost
Indirect Cost
Name
Vanderbilt University Medical Center
Department
Type
DUNS #
004413456
City
Nashville
State
TN
Country
United States
Zip Code
37203