[Quantification of repositioning errors in PET studies through superimposition of emission and transmission scans....

The purpose of this study was to quantify positioning discrepancies between emission (E) and transmission (T) scan using image fusion. A direct comparison of one-step and two-step acquisitions was performed where all studies were analyzed in respect to artifacts caused by inaccurate positioning. In addition, phantom measurements were conducted to estimate the consequences of repositioning errors on standardized uptake value calculations (SUV).

40 patients were examined by two-step whole-body scans using PET and 15 patients were subject to one-step examinations in the head/neck area. Repositioning between the scans was achieved by a laser matrix positioning system in combination with external body markings. After reconstruction and image fusion of the scans, the positioning discrepancies were measured as the distances between the outer boundaries of E and T in four different body regions. Additional evaluations of the SUV by increasing E-T dislocation were performed using a Jaszczak phantom containing hollow spheres.

For the two-step acquisitions, the mean spatial deviations along the three orthogonal axes x, y, and z were between 8.9 mm and 13.8 mm, whereas for the one-step examinations mean values between 3.5 mm and 4.3 mm were determined (level of significance in each direction p < 0.0001). Artifacts were found in 47.5% of the whole body scans, but in none of the head/neck studies. The development of image artifacts was simulated by phantom studies. In contrast, the deviations of the computed SUV caused by increasing positioning discrepancies were minimal because of the minimal differences between the attenuation coefficients of the media involved.

The presented data show that an artifactfree reconstruction of attenuation-corrected studies requires a precise positioning of the patient. One-step examination protocols without repositioning are advantageous due to the significantly lower positioning discrepancies. The additional reconstruction of nonattenuation-corrected studies has proven to be useful in discovering image artifacts and is therefore recommended.