Development of a PET protocol to equalize noise for PET/CT leg acquisitions


Journal Article

The goal of this work is to use phantom and patient data to determine if a single PET leg protocol can be developed that will reduce or redistribute scan time and equalize noise throughout the resulting images by varying scan times by bed position (BP). Most PET protocols use a constant time per BP regardless of patient size. The extreme but consistent tapering (across patients) of legs from pelvis to foot is in contrast to the wide distribution of patient sizes in the upper body. It may be possible to define an acquisition protocol that redistributes scan time per BP such that the image noise is relatively constant through the length of the leg, for all patients. Phantom data were taken from four pairs of bottles, positioned as legs, with diameters between 4.5 to 19.5 cm that were filled with equal radioactive concentrations. ROI measurements were made to relate diameter to noise which could then be used to calculate scan times, for a range of diameters, that would equalize noise throughout the image slices (assuming an inverse quadratic relationship between noise and scan time and assuming random, scatter, and attenuation, would be comparable for the bottle pairs as for legs). 55 PET/CT patient leg studies were used. Geometric measurements were taken from the CT to calculate each patient's effective leg diameter. 95% of all patients had effective leg diameters that tapered at a rate that was within 30% of the mean. The leg measurements were translated into scan times per BP, resulting in a scan sequence that would require 75% less scan time for some BPs. The calculated scan times suggest that redistributing time from the lower-legs and feet to the pelvis and upper-legs will more closely equalize noise throughout the image without increasing the total scan time. It appears that a single protocol may be suitable for all patients and will be an improvement over the current practice of a constant time for each BP, if equalized noise is truly the correct optimization. © 2007 IEEE.

Full Text

Duke Authors

Cited Authors

  • Wilson, JM; Turkington, TG

Published Date

  • December 1, 2007

Published In

Volume / Issue

  • 5 /

Start / End Page

  • 3246 - 3249

International Standard Serial Number (ISSN)

  • 1095-7863

Digital Object Identifier (DOI)

  • 10.1109/NSSMIC.2007.4436831

Citation Source

  • Scopus