Directions and magnitudes of misregistration of CT attenuation-corrected myocardial perfusion studies: incidence, impact on image quality, and guidance for reregistration.

UNLABELLED CT-based attenuation-corrected (AC) myocardial perfusion imaging (MPI) studies may show significant artifacts caused by misregistration between SPECT and CT data. The present study aimed at identifying the directions and magnitudes of misregistration with greatest impact on AC myocardial perfusion image quality. METHODS The incidence, magnitude, and direction of misregistration were assessed in 248 consecutive stress-rest MPI studies in 124 patients. In addition, cardiac SPECT/CT registration was artificially modified in 40 studies, shifting CT data by +/-1, +/-2, and +/-3 pixels along the cephalad/caudal, dorsal/ventral, and left/right axes. Percentage of change in 5-wall AC-MPI polar map scoring was calculated for each region, and the direction of the shift along each axis was analyzed statistically (Student t test, P < 0.05) and compared to determine the region most significantly affected by each shift (Newman-Keuls test, P < 0.05). Changes in the normal and abnormal summed stress score (SSS) due to artificial misregistration were assessed (kappa-statistics, McNemar differences). RESULTS SPECT/CT misregistration of more than 1 pixel was found in 73% (181/248) of studies and more than 2 pixels in 23% of studies (57/248). A 3-pixel ventral shift most significantly affected polar map scoring (-15.4% +/- 6.1% change in lateral wall; -7.5% +/- 5.5% change in inferior wall). A 3-pixel dorsal shift resulted in a -9.5% +/- 5.3% apical and -8.8% +/- 5.8% septal change. Polar map scoring was least affected by the cephalad/caudal shift (<5% average change in all regions except for the anterior wall; -9.9% +/- 7.4% change for 3-pixel caudal shift). The most significant changes occurred in the lateral and anterior walls when the myocardium on SPECT overlapped lung tissue on CT, encountered in 16% of studies (40/248). Clinically significant changes (in SSS) occurred for the 3-pixel caudal, dorsal, and right shifts. CONCLUSION A misregistration of significant magnitude occurred in 23% of studies and in the direction of the most severe artifacts in 16% of studies. Severe misregistration along the dorsal/ventral axis most significantly affected AC-MPI. Quality control of SPECT and CT registration and manual realignment should be performed routinely, with the highest priority in AC studies showing an overlay of the myocardium on SPECT with lung tissue on the CT component of AC-MPI.