SPECT MPI: The Lowdown on Lowering Dose

Advancements in hardware, software and protocols now allow nuclear medicine groups to lower both dose and acquisition time when performing SPECT MPI studies. For some labs, though, folding new technologies into routine practice may require some sacrifice.

Balancing cost & benefit

In 2010, the American Society of Nuclear Cardiology (ASNC) set a goal for the nuclear imaging community to lower patient radiation exposure in SPECT myocardial profusion imaging (MPI) to less than 9 mSv in half of patients by 2014 (J Nucl Cardiol 2010;17:709-718). ASNC encouraged nuclear cardiology specialists to consider technologies and techniques that allowed them to reduce the injected radiopharmaceutical dose and consequently reduce the radiation dose absorbed by the patient—all while maintaining, or even improving, image quality. To meet ASNC’s challenge, nuclear cardiologists, radiologists and technologists may need to modernize.

According to a survey by ASNC and the consulting company MedAxiom, only 3 percent of cardiology practices that provided the age of their cameras reported having equipment that was less than a year old. About four in five cameras were six years old or older, and one in five were 11 years or older. The survey was conducted in 2011 and included 111 cardiology practices in the U.S.  

Purchasing a state-of-the-art dedicated cardiac camera solves the sometimes competing issues of time and dose. Typically, longer imaging time allows for lower dose and shorter time requires higher dose to ensure images with diagnostic accuracy to detect ischemia in patients with coronary artery disease. Some of today’s instruments can accomplish both.

But dose typically trumps time, says E. Gordon DePuey, MD, director of nuclear medicine at St. Luke’s-Roosevelt Hospital in New York City. Guidelines such as Image Wisely, Choosing Wisely and ASNC’s patient-centered initiative emphasize minimizing patient exposure and tailoring imaging plans to best serve each patient. Time becomes a factor in scenarios when the patient cannot or will not remain still for the needed duration or where efficiency and throughput are a priority.

“The issue of reducing time is attractive but not overwhelmingly attractive,” DePuey says. “The issue of dose, however, is very important. We really want to reduce dose.”

And money may trump all. While the price of a new, state-of-the-art camera varies, it still falls at a minimum in the six-digit range. The supply contracting company Novation placed the cost of a single-head gamma camera at $200,000 and a SPECT/CT system at $600,000 in its 2012 Diagnostic Imaging Watch report. Other estimates for a new camera suggest $750,000 or higher. Labs may not be able to justify the price, despite the equipment’s technical wizardry.

“Money is tight now,” DePuey says. “Not all labs have enough volume to have a dedicated cardiac camera. They instead might want to use a general nuclear medicine camera for that purpose and other purposes as well. In that case, they can reduce the injected dose and, thus, patient radiation exposure by incorporating new software.”   

Ernest V. Garcia, PhD, director of nuclear cardiology research and development at Emory University School of Medicine in Atlanta, includes value in the equation. The dedicated cardiac SPECT hardware in his lab has allowed a fivefold increase in counts, he says, at no loss in quality. Increased count sensitivity in solid-state cameras means scans can be completed in less time, or physicians can sacrifice some count sensitivity and lower the injected dose.

“Solid-state [cameras] in general do cost more than the conventional cameras but the question is how do you measure cost?” Garcia says. “You pay more but it is more efficient and at a lower dose.” With the ability to perform scans on more patients, the program at Emory also was able consolidate patient flow to one camera and one site.   

His group found that new software let them goose efficiencies by an additional factor of two, for a tenfold improvement. “The other beauty is not only an increase in efficiency, or a reduction in dose, but in fact the images have higher quality,” he says. Their goal is to further reduce dose by another factor of three.  

Alternative routes

Nuclear cardiologists and radiologists also have less expensive options. New developments in the design of components raise the possibility of adding a cardio-focused collimator, for instance, which places the heart in the sweet spot of highest magnification. Or purchasers can explore software solutions such as reconstruction algorithms designed to shorten acquisition time or lower the dose.

“Many people would say the better way to do it would be to go with the new camera, which also incorporates the new software,” DePuey says. “However, if you don’t have half a million dollars lying around, then the software solution alone is easier.”

Most new software is available only on new computer platforms, DePuey says, but he and Garcia add there are exceptions, with software that can be retrofitted to older cameras and integrated into conventional systems. The approximate $20,000 cost may be more palatable for practices.  

Given the latest in software, how low can you go? Piotr Slomka, a medical biophysicist and research scientist in the Artificial Intelligence in Medicine Program at Cedars-Sinai Medical Center in Los Angeles, and colleagues designed a retrospective study using a dedicated cardiac camera to answer that question (J Nucl Med online Jan. 15, 2013). They hypothesized that today’s high-sensitivity dedicated camera systems could be leveraged to identify the lower limits for dosing that would still achieve accurate images.

Slomka and his colleagues used existing scans of 79 consecutive patients who underwent rest-stress MPI with a dedicated cardiac camera. The 79 studies also had been evaluated with a fully automated quantitative analysis algorithm designed by Slomka, who specializes in automated software for quantifying myocardial perfusion.  

To determine the lower count limits, they reframed each patient’s list-mode raw data to create a truncated view to simulate reduced-dose data. The views were then reconstructed in the same manner as full-count scans. They applied software to quantify total perfusion deficit and ejection fraction. “That is a good way to test how low you can go with the dose because you are not risking a poor quality scan,” Slomka explains.

Left ventricular region counts as low as 1 million retained excellent agreement in quantitative perfusion and function parameters seen at high-count images, even in the presence of perfusion defects. In a 14-minute acquisition at a 1 million count, injected dose could be reduced by a factor of eight for an average effective radiation dose to the patient of 0.71 mSv, or 1 mSv for a 10-minute stress scan. The study informs clinicians of the technology’s potential, but the findings need to be further tested in a larger, multicenter prospective trial, according to the researchers.

But achieving lower dose also may require that practitioners and researchers embrace newer software. The scans in the study used software that fully automated the quantitative analysis, an approach that Slomka and his colleagues have shown reduces variability and bias with diagnostic and prognostic performance equivalent to an expert reader. With lower dose, it may be more difficult for physicians to visually assess images, he says, making quantitative analysis critical.

“The bottom line is that it [the automated algorithm] allows you to really lower the dose but not subjectively,” Slomka says. “When we apply this quantitative technique, we can precisely and scientifically say it is OK to lower the dose.”

Costs and caution may have contributed to reluctance in the medical community to incorporate hardware and software advances, but that may change as the ASNC’s deadline approaches. “As with everything in medicine, it takes a while to convince people this is the way to go,” DePuey says. “But we also are under the gun to decrease radiation exposure.”