Atherosclerosis Quantitation by Computer Image Analysis

Existing methods to quantify atherosclerosis in living persons directly from arterial images have virtually all been developed for use with conventional angiograms for the obvious reason that practical alternative methods for arterial visualization have not been available. As will be described later in this session, a number of important new noninvasive imaging techniques have emerged recently, which will provide additional possibilities for lesion quantitation. Intravenous digital radiography, for example, can be used to visualize most of the major arteries except the coronaries, and ultrasound systems are available for routine carotid artery imaging. Experimental ultrasound systems are under development that may be capable of imaging deeper arteries including the coronaries. Imaging systems using nuclear magnetic resonance have successfully visualized arteries in small animals and phantoms that simulate arteries in the millimeter range (1). While these noninvasive methods appear very likely to be used for quantitative lesion assessment in the near future, relatively little quantitative experience has been gained to date. Conversely, computer methods to quantify atherosclerosis from conventional selective angiograms have reached a stage of relative maturity, in which their use in clinical studies is practical and on-going. The intent of this paper is to outline lesion quantitation methods using the angiographic techniques for illustration and to discuss some of the physical factors, performance measurements, and image analysis problems that have common applicability to both invasive and noninvasive imaging.