ASNC: Quality Myocardial Perfusion Images in Half the Acquisition Time

MONTREAL, Sept. 12 -- Speeding up imaging time without sacrificing image quality is at the top of the wish list for nuclear cardiologists.

Manufacturers have responded with new technology promising top quality images in half the acquisition time. But do they really work? Data presented here at the annual meeting of the American Society of Nuclear Cardiology (ASNC) suggests that they do.

"A number of vendors have been introducing methods to improve image quality and at the same time decrease acquisition time, which is exactly what we want in nuclear cardiology because it will give us better studies, better diagnostic accuracy, increased efficiency, and increased patient throughput," said E. Gordon De Puey, M.D., of St. Luke's Roosevelt Hospital Center in New York. "So, it's really our goal, but before we can embrace that, we really have to make sure that they give us adequate diagnostic accuracy, and they don't introduce artifacts."

In a late breaking abstract session, Dr. De Puey presented his results comparing single photon emission computed tomography (SPECT)-gated myocardial perfusion images (MPIs) obtained in 50 patients.

They used standard, full acquisition time filtered-back projection (FBP) with two novel reconstruction technologies touted to offer at least as high quality images with half-time acquisition: wide beam reconstruction (WBR, UltraSpect Ltd.), and ordered set expectation maximization with optimized resolution recovery (OSEM RR). WBR is currently available while OSEM RR remains investigational. All acquisitions were made using a GE Healthcare camera, and GE Healthcare provided support for the study.

The investigators followed the standard MPI imaging protocol, followed by acquisitions at half-time using the new technologies. They performed both rest and stress images at full-time and half-time. Readers of the images were blinded to method used for obtaining the image and clinical information about patients.

"The patients were regular clinical patients," said Dr. De Puey "There were no particular selection criteria. They ranged from 120 pounds up to 265 pounds, and it also did high count density stress studies and low count density resting studies. So, really it addressed a very broad spectrum of types of images we get with myocardial perfusion scans."

Based on the ratings of the blinded evaluators on a five-point scale, where 0=uninterpretable and 4=excellent, image quality of summed stress score (SSS) with FBP received an average score of 2.9, while summed rest score (SRS) received a 2.6. For WBR, image quality scores were 3.3 and 2.9, respectively, and for OSEM RR, these scores were 3.1 and 2.6. FBP/ WBR correlation was 0.85 and FBP/OESM RR correlation was 0.86. Diagnostic certainties were approximately 0.8 for all three methods.

"We find that the image quality is equivalent or actually better with wide beam reconstruction [and OSEM RR] half-time images than the standard filtered back projection whole acquisitions," said Dr. De Puey.

While these findings are encouraging, Dr. De Puey warned that they cannot be extrapolated to other systems, particularly those of other manufacturers. Each system must undergo its own validation process.

At a symposium sponsored by Philips, Gary Heller, M.D., and Gavin Noble, M.D., both of Hartford (Conn.) Hospital, demonstrated similarly encouraging data on use of the Philips Astonish hardware and software solution for half-time acquistions for SPECT. The solution employs iterative reconstruction, depth dependent resolution recovery, scatter correction, noise reduction, and detector distance optimization to produce superior image quality in half the acquisition time and can be coupled with attenuation correction to improve images even more.

In an early pilot study in which both Dr. Heller and Noble took part, 57 patients underwent SPECT MPI using a standard protocol. Of these patients, 37 had recent catheterization data, and the other 20 were considered to be low risk. The data was then analyzed using standard FBP, Astonish, and Astonish with every other frame removed to simulate half-time acquisition.

Three blinded readers read the images by consensus and compared them by looking at all three on one screen (randomized as to whether these were rest or stress studies) as well as by comparing only rest studies with rest studies and stress studies with stress studies. Randomization ensured that the three images acquired for each patient were separated both in time and numerical assignment.

No difference was found among the three type of images with respect to sensitivity and specificity for detecting >50% or >70% lesion severity, when looked at on a per patient or per vessel basis. The only statistically significant finding among the three methods was a slight overall decrease in specificity for the right coronary artery (RCA) with half-time Astonish. It is unclear what this might mean clinically, but the investigators say it should be taken into account when designing larger trials.

Both Astonish and half-time Astonish provided better image quality than FBP, based on subjective assessment of rest images, and a similar, nonsignificant trend in this direction was also seen for stress images.

Interestingly, Astonish and half-time Astonish yielded, somewhat higher SS scores than FBP, although not enough to be clinically relevant, said Dr. Noble. This may reflect the learning curve necessary to properly read this technology, he suggested.

"So many laboratories, particularly private outpatient laboratories are trying to push through 12 patients per camera for day or sometimes even more," said Dr. De Puey. "The technologists are very rushed, and they can only accommodate so many people, and there's a backlog of patients. By halving the acquisition time, they can increase their efficiency and do more patients. That will ultimately serve more patients and be more profitable for the labs."