Basal ganglia studies with 3D acquisition and 2D reconstruction on a retractable septa PET scanner.

OBJECTIVE

We have tested the feasibility of applying an approximate three-dimensional (3D) reconstruction algorithm, the single slice rebinning method followed by standard 2D filtered back-projection (SSRB), to human basal ganglia studies. Such an approximate algorithm solves the problem of long reconstruction times and large dataset size associated with mathematically correct 3D reconstruction algorithms.

MATERIALS AND METHODS

First, recovery achieved when images were reconstructed with the SSRB method was compared with that achieved when a mathematically correct 3D reconstruction algorithm (3D method) was used with a series of spherical phantom studies. The SSRB method was then applied to 14 human receptor studies and the images compared with those obtained with the 3D method. The striatum/background contrast dependence of the SSRB method performance was tested by extending the comparison of the two reconstruction methods to single frames of two dynamic scanning sequences.

RESULTS

There was no significant visual difference in the human images obtained with the two methods and a quantitative striatal/cortical activity ratio analysis showed a contrast decrease of approximately 3% for the SSRB method. The analysis of the dynamic sequence did not show any contrast dependence of the SSRB method performance.

CONCLUSION

The SSRB method was found adequate for processing data related to human striatal imaging for this particular tomograph geometry.