Study claims cardiac PET/CT demonstrates 40% false positive rate

July 5, 2007

Current PET/CT scanners wrongly indicate coronary disease in up to 40% of patients, according to a study published this month in the Journal of Nuclear Medicine.

Current PET/CT scanners wrongly indicate coronary disease in up to 40% of patients, according to a study published this month in the Journal of Nuclear Medicine.

Misregistration of PET and CT data sets causes these false positives, according to lead author Dr. K. Lance Gould, a professor of medicine/cardiology at the University of Texas Medical School and executive director of the Weatherhead PET Center for Preventing and Reversing Atherosclerosis at the Memorial Hermann-Texas Medical Center, both in Houston.

"This medical product is out there on the market, and physicians and their patients trust it, yet there is a big problem with it," Gould said. "Since PET/CT proved ideal for cancer, body, and brain imaging, everyone assumed its application to the heart would work with the same protocols and software - a big mistake."

Vendors contend, however, that the potential for misregistration to cause misdiagnosis has been known for a long time and has been resolved. GE Healthcare, provider of the 16-slice PET/CT installed at the Weatherhead PET Center, claims that software available from the company for more than a year solves the misregistration problem.

"Commercial software does not account for movement in the lungs and heart while the patient is breathing during the test, and as a result, the PET and CT images don't always match or coregister," Gould said.

This makes these scanners prone to false-positive results when doing cardiac assessments, according to Gould. He asserted that his group at the University of Texas "found errors that are inherent in the technology as now used." (Gould was not available to DI SCAN to elaborate on these comments.)

"I don't know which device he (Gould) is using now, but for sure he does not have our latest," Hadi Moufarrej, global general manager for molecular imaging, nuclear medicine and PET at GE, told DI SCAN.

GE installed a PET/CT at the Weatherhead PET Center several years ago, according to Moufarrej.

"Probably, this is what he is using without any new software from us," Moufarrej saud.

Dr. Partha Ghosh, clinical marketing manager for Siemens molecular imaging, said the makers of PET/CT scanners have understood the risks of misregistration for some time and have taken steps to fix it. At Siemens, the approach has been to adapt the clinical workflow for cardiac PET/CTs so that the technologist can examine the alignment of images done automatically with software and make manual adjustments when necessary.

"This is a workflow issue that takes into account the inaccuracies that can happen," Ghosh said. "We have tested this and have found that the misregistration is very, very small, once you take into consideration this manual realignment."

The workflow process, including software for aligning PET and CT data sets, has been available on the Biograph 64 since it became commercially available 18 months ago, he said.

Tests on 259 patients imaged at the Weatherhead PET Center produced false positives in 40% of the results, according to the JNM study. False abnormalities suggesting severe heart disease were found in 23% of patients. These abnormalities would have resulted in unwarranted, emergency heart procedures if additional tests had not been run to confirm Gould's suspicions of inaccurate results.

The patients participating in the study underwent diagnostic rest-dipyridamole myocardial perfusion PET/CT using rubidium-82 on a 16-slice scanner. Spiral CT attenuation correction was performed with breathing and also at end-expiratory breath-hold, according to the JNMarticle. Cine CT data during breathing were averaged.

Misregistration on superimposed PET/CT fusion images was measured in millimeters and correlated with associated quantitative size and severity of PET defects. Misregistration artifacts were defined as PET defects with corresponding misregistration on spiral PET/CT fusion images that resolved after correct coregistration using a repeat CT scan, cine CT averaged attenuation during normal breathing, or shifted cine CT data that coregistered with PET data.

Since documenting the misregistration problem, Gould and Tinsu Pan, Ph.D., an associate professor in the department of imaging physics at the University of Texas M.D. Anderson Cancer Center, wrote software for their PET/CT to account for respiratory motion. It properly aligns the PET and CT images, according to Gould. This software, however, is not yet commercially available to the general imaging community.

When data are properly matched, PET/CT is a valuable tool for detecting and managing coronary artery disease, Gould said. He hopes other cardiologists and their patients will insist that scanner manufacturers resolve problems with the software to produce the most accurate results.