Characterization of sampling schemes for whole body PET imaging

Abstract

Whole body PET images suffers by relatively high noise levels due to inherent poor counting statistics in the emission data. It is therefore important to optimize the acquisition parameters to minimize any additional noise contamination. It has previously been shown that using a continuous or redundant acquisition scheme, improvements in noise characteristics and image quality are seen. In this work the continuous sampling is further characterized and compared to the conventional step and shoot acquisition. The main sources of noise contamination using conventional sampling is due to the normalization procedure which is applied to the emission data. By using continuous sampling and using a common normalization for all planes, the statistical noise in the normalization is improved by factor close to the number of planes in the scanner. The continuous sampling showed to be less sensitive to small patient movements (<5 mm) compared to conventional scanning. Previously reported problems of data handling have been resolved by rewriting the acquisition firmware to allow on-line addition of the redundant sampled data in hardware.<>