Whole Body PET Imaging Using Variable Acquisition Times

Abstract

Whole Body PET scans are typically performed as a series of image sets acquired at discrete axial positions to cover most or all of the body. The acquisition time at each axial position is typically kept fixed for all positions although the total imaging time is typically adjusted according to the patient's weight. Because of the varying amount of attenuation for different sections of the body, it is expected that the image Signal-to-Noise (S/N) will vary accordingly. The aim of this work is to investigate the possibility of varying the acquisition time at different sections of the body such that the image S/N is kept relatively constant for all slices. To estimate the acquisition times for the different sections of the body we propose to use the attenuation correction (AC) sinogram generated from the CT scan that is acquired prior to the PET scan. Both simulations and phantom measurements of different diameter cylinders with activity distributions were performed. The image noise was estimated in every pixel from multiple replicate image sets. The image noise was compared to the average AC factors through the center of each body slice. A simple polynomial function was found for both the simulations and the phantom measurement images to accurately describe the image noise as a function of AC factors. These results indicate that the noise properties of whole body images can be made more uniform axially by adjusting the acquisition time according to the amount of attenuation. Instead of using a fixed scan time per bed position, the acquisition time can be reduced in areas of lower attenuation and increased in more absorbing sections of the body. Since there is a strong correlation of the image noise and the AC factors, the relative acquisition times can be quickly calculated using a simple functional relationship.