Simultaneously acquired MR/PET images compared with sequential MR/PET and PET/CT: alignment quality.

Abstract

To compare anatomic alignment between morphologic and positron emission tomography (PET) images acquired in the abdomen and pelvis by using simultaneous magnetic resonance (MR)/PET, PET/computed tomography (CT), and retrospective MR/PET fusion and to compare alignment between MR and PET in simultaneous and sequential thoracic MR/PET by using different breathing, registration, and gating protocols. Informed consent and institutional review board approval were obtained. The misalignment in 28 abdominal patient datasets was analyzed in simultaneous MR/PET, PET/CT, and retrospective MR/PET fusion. In seven thoracic MR/PET datasets, the effect of different breathing protocols, simultaneous, sequential, and MR-gated data acquisition was evaluated, and the effect of rigid registration versus nonregistered images was evaluated. Analysis of variance and subsequent Tukey test were used for statistical analysis. The misalignment in all abdominal organs was reduced in simultaneous MR/PET compared with retrospectively fused MR and PET images (means, 5.8 mm vs 11.9 mm; P < .001), and in the urinary bladder compared with PET/CT (means, 5.9 mm vs 11.0 mm; P < .007). Thoracic MR/PET with inspiratory breath-hold MR showed the largest misalignment (mean, 24.5 mm; P < .001). None of the other thoracic protocols that were composed of different acquisition, registration, or gating procedures differed significantly from one another. The alignment of hybrid datasets acquired in simultaneous whole-body MR/PET was more accurate than retrospective fusion in all investigated abdominal organs, and more accurate than PET/CT in the urinary bladder; the alignment of thoracic MR/PET with expiratory breath hold or free-breathing MR was more exact than with inspiratory MR. However in this preliminary work the clinical importance of these degrees of misalignment was not assessed.