Abstract
1486
Objectives Misalignment of images in cardiac PET-CT imaging may lead to erroneous attenuation correction (AC) and misdiagnosis. Such misalignment is usually corrected by hand prior to clinical assessment. The aim of the study is to assess whether a single automatically aligned rest CT can be used for stress PET AC.
Methods 26 patients underwent PET-CT, consisting of 82Rb PET at stress, and 3 free breathing CTs at each of rest and stress. The best aligned stress CT is selected for clinical use.An automatic mutual information registration was used to align each rest CT to the stress PET, first using translation, and then deformable. Registrations were assessed visually, and translations compared to manual alignments. We performed 17-segment polar plot analysis using 4DM-SPECT on each registered image: for each patient, consistency of scores across plots was measured by counting segments with different scores.
Results The automatic methods aligning rest CTs to stress PET gave visually acceptable results for 98% of cases. Comparison of the automatic and manual translations showed a mean difference of 8mm +/- 4mm. Consistency across the 3 AC images for each patient showed that there were differences in 13% of segments with manual registration, 8% with automatic translation and 7% with automatic deformable.Comparing to the clinical stress CT chosen for AC, 9% of segments differed with manual, 5% with automatic translation and 5% with deformable.
Conclusions Our analysis has shown that automatic registration is a viable option for aligning rest CT to stress PET for AC, with results comparable to that of a manual alignment in comparison to the clinical chosen CT.
- © 2009 by Society of Nuclear Medicine