3D software boosts diagnostic accuracy for Alzheimer's By Wayne Forrest AuntMinnie.com staff writer December 1, 2009
CHICAGO - 3D stereotactic surface projection (3D-SSP) software enhances the diagnostic accuracy and confidence of both beginner and expert radiologists when interpreting FDG-PET scans of patients with mild cognitive impairment or Alzheimer's disease, according to a study from the Mayo Clinic in Rochester, MN.
The study, presented Monday at the RSNA annual meeting, received the RSNA Research Trainee Award.
Alzheimer's disease currently affects between 4 and 5 million Americans. "Early, accurate diagnosis of Alzheimer's disease is essential for optimum treatment and could help with implementation of potential disease-modifying drugs," said lead author and presenter Dr. Vance Lehman.
"Relatively recently, 3D stereotactic surface projection -- or 3D-SSP -- has been used in some centers and is thought to improve the accuracy of experts," he added. "3D-SSP offers fully automated PET quantification and can regularly be implemented into clinical practice."
The study retrospectively analyzed 54 patients (23 normal and 31 cognitively impaired individuals) who had been enrolled at the Mayo Alzheimer's Disease Research Center. Each patient received a baseline FDG-PET scan and quantification by 3D-SSP.
Two readers with less than one year of experience and two nuclear medicine experts with at least 10 years of experience viewed and rated all the PET scans. The readers ranked the scans from 1 to 5, with 1 as a normal scan and 5 as severe dementia. They also evaluated the scans for diagnostic confidence, with 1 as uncertain diagnosis and 5 as complete diagnostic certainty.
During the 3 to 3.5 years of patient follow-up, no one in the normal group developed cognitive impairment, which confirmed their initial diagnosis, Lehman noted.
"Twelve of the 18 patients with mild cognitive impairment progressed to dementia, with 11 of those 12 to Alzheimer's dementia," he added. "All 13 people with Alzheimer's disease demonstrated an increase in disease severity."
Without 3D-SSP, Lehman said that the experienced readers performed "moderately better" than the beginners in terms of accuracy, with the experts at 70%, compared with the beginners' accuracy of 61%.
Specificity was relatively low (26%) for the beginners without the use of 3D-SSP, as the beginners tended to call normal scans as abnormal due to false positives. Specificity for the experienced readers was 57%.
With the addition of 3D-SSP, the accuracy of the beginners "improved substantially" from 61% to 76%, Lehman said. He credited the improvement to a marked increase in specificity (from 26% without 3D-SSP to 63% with 3D-SSP), with no loss in sensitivity (86% without 3D-SSP to 86% with 3D-SSP).
The addition of 3D-SSP also helped the experienced readers improve their accuracy from 70% without 3D-SSP to 82% with 3D-SSP. Again the gain was due to greater specificity, from 57% without 3D-SSP to 87% with 3D-SSP, and little difference in sensitivity (81% to 79%, respectively).
Overall, diagnostic confidence increased by 20% for both the beginners and the expert readers with 3D-SSP, with the experienced readers' diagnostic confidence greater than that of the beginners by 20%.
Given the results, Lehman and colleagues wrote that accurate diagnosis of Alzheimer's disease with quantitative FDG-PET "would help direct appropriate treatment early in the disease course and facilitate future testing of potential disease modifying drugs."
At the end of the presentation, Lehman received the RSNA Research Trainee Award from Dr. Milton Guiberteau, chairman of the RSNA nuclear medicine subcommittee.
Guiberteau serves as academic chief in the department of medical imaging/radiology and division chief of nuclear medicine at St. Joseph Medical Center in Houston. He also is a professor of clinical radiology at the University of Texas Medical School in Houston.