Come on, baby, let's do the twist: Detecting and correcting cardiac torsion effects in myocardial perfusion SPECT

Abstract

In this issue of the Journal of Nuclear Cardiology, Nichols et al describe a technique to detect cardiac torsion and use the information generated on a subjectby-subject basis to correct for any artifacts that cardiac torsion produces. Cardiac torsion is defined as the dynamic twisting component of cardiac motion about the apex-to-base axis as the heart contracts. This particular component of cardiac motion is not accounted for in routine myocardial perfusion single photon emission computed tomography (SPECT), even when such studies are acquired and processed in a gated fashion. Nichols et al hypothesized that maps of perfusion and wall thickening are motion-blurred by torsion and sought to “deblur” the maps based on the detected torsion. Imaging studies of cardiac motion, including torsion, in experimental animals and human beings have used a variety of modalities, including radiography, echocardiography, and magnetic resonance imaging (MRI). In general, these studies rely on “tagging” the myocardium with a series of reference points (eg, surgically attached steel markers for radiography, right ventricular insertion points or papillary muscles for echocardiography) or reference lines (eg, modulated magnetization grids for MRI) to directly assess movement (ie, change in location) of each point on or in the myocardium through the cardiac cycle. These studies have shown that most twisting occurs at the apex in normal persons, with different abnormal rotation patterns in the presence of cardiac disease. The study of cardiac torsion in the context of myocardial perfusion SPECT is potentially interesting in 2 ways: (1) cardiac torsion may generate artifacts that reduce diagnostic or quantitative accuracy, and (2) patterns of cardiac torsion may be (independently) diagnostic for cardiac disease. Nichols et al explicitly identified 3 goals for their work: (1) to document the frequency and degree of confidence with which it is feasible to detect torsion motion patterns, (2) to determine whether corrections for torsion improve myocardial perfusion and wall thickening quantitation, and (3) to discover whether abnormal twisting patterns seen by MRI are also detected by SPECT. In this regard, their work can be seen as the first step in addressing the two potential implications or uses of cardiac torsion detection in myocardial perfusion SPECT. Their first and third goals represent necessary prerequisites to accept their detection method as accurate; their second goal is the first step in determining whether cardiac torsion correction is important in myocardial perfusion SPECT. Nichols et al studied their approach in 52 subjects who underwent both technetium 99m sestamibi gated myocardial perfusion SPECT studies and contrast angiography. They divided the 52 subjects into 3 groups: (1) 12 subjects with normal angiography and gated myocardial perfusion SPECT results, (2) 12 subjects with abnormal angiography results and abnormal perfusion by SPECT, and (3) 28 subjects after angioplasty, representing a mixture of normal and abnormal perfusion, prior infarction, wall motion, and coronary artery bypass grafting. The authors developed a fully automated algorithm to detect and quantify torsion from the 8-frame gated SPECT data. The algorithm is based on determining the location of right ventricular insertion points (anterior and especially inferior wall insertion points) in the polar plot of each frame; the change in angular location of these points through the cardiac cycle represents torsion. In essence, their algorithm outputs a rotational (angular) shift at a given phase of the cardiac cycle; this shift is proportional to the magnitude of torsion, and torsion correction can be accomplished simply by rotating each polar map in the opposite direction by the corresponding shift. If this is performed before adding the 8 maps together (to simulate a “static” or ungated map), then the resultant static map does not contain the angular blurring that torsion would otherwise produce. Nichols et al found that torsion magnitude and patterns were consistent with those previously reported from MRI studies. They also found that torsion correcFrom Johns Hopkins University, Baltimore, Md. Reprint requests: Jonathan M. Links, PhD, Radiology Health Sciences/ Environmental Health, Johns Hopkins Medical Institute, 615 N Wolfe St, Baltimore, MD 21205-2179. J Nucl Cardiol 2002;9:561-2. Copyright © 2002 by the American Society of Nuclear Cardiology. 1071-3581/2002/$35.00 0 43/39/125917 doi:10.1067/mnc.2002.125917