Improved PET Detection of Focal Brain Activity Using Subset-dependent Relaxation `Dynamic´ Row-action Maximum Likelihood Algorithm (DRAMA)

The SHR-7700 scanner (Hamamatsu Photonics, K.K.) is a high-resolution 20 cm bore research scanner that is used to acquiring images from both animals and human brains. The scanner has a reconstructed resolution of 2.5 mm and can image small amounts of tracer. To improve the recovery of low signals the DRAMA reconstruction method was implemented for the Hamamatsu scanner. To improve the signal-to-noise ratios the algorithm of subset-dependent relaxation ´dynamic´ row-action maximum likelihood algorithm (DRAMA)[Tanaka, et al., -] was evaluated using a data set of three groups of subjects (n = 9, 9, and 8) received 5 mCi of [F-18] fluorodeoxyglucose at rest and during delayed recall task on 2 separate days. The optimum filter will be determined by varying the filter size. One filter has been evaluated and the resulting data was evaluated with a pixel wise one-sample t statistics were performed between task-resting conditions following pixel normalization and 3D Gaussian smoothing (9mm FWHM). Evoked brain activities were recorded with Z scores and spatial coordinates for comparisons. Significant task-evoked brain activities were detected by FBP and DRAMA (chi-square p < 0.001) in all 3 groups, but DRAMA produced 6.9 times more pixels than expected from random chance in comparison to 5.3 times by FBP. This method will be applied to the rest of the filter choices. This method will be also applied to a small animal study where the injected dose was only 15 to 25 muCi of activity to see if there is any improvement in the results. These results indicate a potential for DRAMA to improve the statistical power of brain-activation test.