Comparing the effects of standard and segmented attenuation correction

We have evaluated a segmentation algorithm developed by General Electric for the Advance positron emission scanner (PET). Phantom studies were performed to measure the accuracy in emission scans reconstructed with segmented, attenuation data as a function of transmission scan time. The results indicated errors of less than 2% will be made in emission scan data reconstructed with transmission scan times of 3 minutes. Based on the phantom results, 185 patient data sets were acquired with both long (15 min.) non-segmented and short (3 min.) segmented attenuation scans. Comparisons of scan data in foci of abnormal uptake yielded a correlation coefficient between long and short scan SUV maximum values of 0.99 and a mean absolute difference of 4.6%. The average SUV values in lung between long and short has a correlation coefficient of 0.99 and a mean absolute difference of 3.1%. The corresponding values from the liver had a correlation coefficient of 0.96 and mean absolute difference of 7.4%. Visual review by physicians noted minor differences, but when grading the images on a scale of 1 to 5, 91% of the time there was no difference. In all cases comparing the long and short attenuation and no abnormal sites were missed