TUESDAY, Jan. 12 (HealthDay News) -- A classification procedure using metabolic brain imaging with positron emission tomography (PET) can identify different forms of Parkinsonism early and with high accuracy, preventing misdiagnosis and ensuring correct treatment, according to a report published online Jan. 11 in The Lancet Neurology.
Chris C. Tang, M.D., of the Feinstein Institute for Medical Research in Manhasset, N.Y., and colleagues used an automated image-based procedure to classify 167 patients who presented from 1998 through 2006 with Parkinson's disease features but uncertain clinical diagnosis. The patients had fluorine-18-labelled-fluorodeoxyglucose-PET brain scans, and patterns consistent with Parkinson's disease, multiple system atrophy, or progressive supranuclear palsy were converted to mathematical scores expressing the likelihood of having each of the three diseases. The disease classification then was made on the basis of probability. The patients subsequently were assessed by movement disorders specialists (blinded to the PET results) during a mean of 2.6 years follow-up to reach a final clinical diagnosis, which was compared to the PET classification.
The researchers found that the PET-based classification for Parkinson's disease had a sensitivity of 84 percent, a specificity of 97 percent, a positive predictive value of 98 percent, and a negative predictive value of 82 percent. The PET classifications were comparably accurate for multiple system atrophy and progressive supranuclear palsy.
"Automated image-based classification has high specificity in distinguishing between parkinsonian disorders and could help in selecting treatment for early-stage patients and identifying participants for clinical trials," the authors write.