Interpolated Background Subtraction Method For Coronary Stenosis Quantification (original) (raw)

The determination of percent stenosis of coronary arteries is an important task in medicine. In this paper we discuss three different algorithms which can be used in conjunction with videodensitometry to measure this quantity. These algorithms may be used in subtracting the background under a vessel segment thus eliminating the need for a preinjection mask. Mathematical details of the algorithms and experimental results are presented. 1. INTRODUCTION Recent advances in the digital x-ray video imaging systems has made the visualization and measurement of small amounts of radiopaque contrast agents within the vasculature a reality. The components of a typical digital radiographic system are the x-ray generator, x-ray tube, intensifier (II), television (TV) camera, video digitizer, display and recording devices. Two general classes of measurements one can perform on the digital video imaging are: (a) geometrical8 and (b) videodensitometric quantification7,5,6. Here, we will only be concerned with the latter. Videodensitometric measurements) are based on the linear relationship between the intensity of the video signal and the thickness of the tissue transversed by x-rays causing that signal. Many physical problems, some of which are intrinsic to such imaging systems and others due to the nature of interaction of x-rays with an object, usually cause a deviation from this desired linear relationship4. Thus, it is quite essential to determine the response of the imaging chain prior to making measurements on images obtained with such systems. In this paper, we will focus on the videodensitometric measurement of coronary stenosis. The most common technique used for this purpose is the "mask subtraction" 3. In this technique, a mask image of the patient is taken prior to the injection of iodinated contrast material into the coronary vessels. Then, a pixel by pixel subtraction of the logarithmically amplified mask image from the post-injection image results in an image which "ideally" has all the noniodinated parts of the object cancelled. The net iodine signal intensity is linearly related to the thickness of the vessel containing it. Thus, a correct measurement requires, among other things, a total cancellation of the noniodinated overlaying structures. If the patient moves in between the time mask and post-injection images are acquired, then the difference images may be degraded due to motion artifacts. The measurement of coronary stenosis using such images would give rise to different degrees of error depending on the magnitude and location of motion artifacts. Since the physiological motions which occur in a patient are not simple translational motion, one can not remedy the situation by simply shifting the mask to obtain an acceptable registration with the iodinated image. Attempts to alleviate this problem by means of rubber sheet algorithms have proven to be difficult and at best, approximate. In this paper, we discuss several different algorithms to estimate the background under the iodinated vessel. Since the background is derived from the iodinated images, one does not have to use any preinjection mask images. Section 2 covers the methods. The results are presented in Section 3. The conclusions are given in Section 4. 2. METHODS 2.1 Basic Formalism The expression for a two dimensional video image Q(x,y) obtained with a TV based x-ray imaging system is given by Kruger2,