Ronald H Huesman, Gregory J. Klein, and Bryan W. Reutter
Center for Functional Imaging
Lawrence Berkeley National Laboratory
Positron emission tomography (PET) provides the ability to measure the moment to moment concentration of a broad variety of biologically active tracers in the organs of the body. One of its principal applications has been in the heart, where this capability provides a unique avenue to assess myocardial perfusion and metabolism.
Over the past 25 years, the spatial resolution attainable by PET systems has improved dramatically -- from a resolution of 17 mm in 1974 to approximately 4-5 mm for today's commercial cardiac scanners. With this improved resolution, there is the potential to obtain detailed maps of myocardial perfusion and metabolism. However, this potential remains largely unfulfilled since current data acquisition and analysis strategies do not account for the contractile and respiratory motion of the heart, which has a combined amplitude more than twice the resolution of contemporary scanners. The effects of gross patient motion compound the problem. While the resulting blurred images are reasonable qualitative estimates of the left ventricular myocardial activity, they fall far short of the high-resolution quantitative images that are potentially attainable with modern PET systems. In addition, heart motion prevents a clean separation of myocardium from blood in the ventricular cavities, thus biasing estimates of the blood pool activity and affecting the reliability of kinetic analyses used to quantify myocardial perfusion and metabolism.
Early work in gated single photon emission computed tomography (SPECT) using Tl-201 demonstrated feasibility as well as statistical limitations of retrospectively separating events according to the phase of the cardiac cycle. Hardware for cardiac gating in PET and SPECT scanners was subsequently developed. In these devices, data are separated according to time after detection of the R-wave signal from an EKG unit, while ungated datasets can still be synthesized by summing the gated data. Dividing the data in this manner effectively stops the contractile motion of the heart; however, lack of sufficient numbers of events often requires that the data be summed to improve image quality for visualization. Unfortunately, quantitative accuracy is generally not a great deal better for the gated datasets than it is for the corresponding ungated datasets. We believe that this is due to the lack of compensation for respiratory motion of the heart in addition to the loss of events outside of a single chosen gating interval.
The effects of respiratory motion of the heart have been virtually ignored in cardiac emission tomography, although the problem has been recognized and was described in 1982. The movement of the diaphragm and heart due to respiration in a human in the supine position during tidal breathing is approximately 15 mm. In addition to heart motion due to tidal breathing, another type of respiratory motion known as "cardiac creep" has been observed in Tl-201 stress-redistribution SPECT studies and in Rb-82 rest/stress myocardial perfusion PET studies. Cardiac creep is probably related to physiological changes in the volume and distribution of air in the lungs between rest and stress, which result in changes in the mean position of the diaphragm and thus the heart. An increase in mean lung volume during stress causes a downward cardiac creep, followed by a post-stress upward creep as the mean lung volume returns to normal.
Current technological development is expected to lead to better than 3 mm resolution in cardiac PET. In order to take advantage of existing and future technology in quantitative cardiac PET, methods for the compensation of heart motion must be developed. The thrust of current research in this area is demonstration of the feasibility of detecting motion of the heart and providing efficacious means of compensating for motion in order to make possible high-resolution, low-noise images for greater clinical utility. The principal methods entail superposition of spatially registered volume image data from multiple respiratory phases to account for respiratory motion of the heart and superposition of deformable motion models to account for contractile cardiac motion.