Evaluation of Maximum-Likelihood based Attenuation Correction in Positron Emission Tomography

In positron emission tomography, transmission scans are used to correct the acquired data for the effect of photon attenuation. The noise present in the transmission measurement often has a significant effect on the signal to noise ratio of the final attenuation corrected image of the tracer distribution. This study evaluates the effect of different attenuation correction strategies on the performance of human observers in a tumor detection task. The four strategies considered are: no attenuation correction, multiplication with the count ratio of blank and transmission sinograms, reprojection of a maximum-likelihood reconstruction, and reprojection of a maximum-a-posteriori reconstruction of the transmission sinogram. Performance in tumor detection was quantified as the contrast at which the number of errors increased beyond 20%. No statistically significant difference was found between classical attenuation correction and maximum-likelihood based correction. With maximum-a-posteriori based attenuation correction performance was significantly better than with the other methods.