Problems in Signal-to-Noise Ratio for Attenuation Correction in High Resolution PET

Abstract

In high resolution PET, the number of events required for a high signal-to-noise ratio in the measured attenuation correction is an order of magnitude higlher (50-100 million) than that required for the emission image. This is due to the large number of elements in the attenuation correction sinogram (e.g. 50,000) and the large attenuation factors through the thicker sections of the body (e.g. 50). Another major contribution to noise in high resolution PET is contamination of true coincidences by accidentals and their subtraction from the prompt coincidences. In this work three methods which reduce noise in the measured attenuation are evaluated. The methods are: (1) optimization of energy threshold, (2) processing of random sinogram before subtraction and (3) spatial averaging of the tranismission sinogram before performing attenuation correction.