Clinical Evaluation of Respiratory Gated Magnetic Resonance-Based Attenuation Correction of PET/MR

Background: Quiescent period gating (QPG) of positron emission tomography (PET) studies has been shown to reduce motion artifacts. In addition, QPG of the attenuation correction (AC) measurement data has potential to improve the registration of QPG PET data to the attenuation maps, increasing the accuracy of AC and quantification of standardized uptake values (SUV).

Purpose: The purpose of this study was to evaluate QPG magnetic resonance (MR)-based attenuation correction (MRAC) on PET/MR and determine whether it creates PET images with improved repeatability of SUV. Methods and Materials: In this IRB approved study, thirty patients (17 males, 13 females) underwent a single 10 min QPG PET scan on the GE Healthcare PET/MR scanner with MP26 software features after their clinically indicated PET/CT or PET/MR. Patients were enrolled from 11/8/2016 to 06/22/2107. Simultaneous to the PET acquisition, five MRAC sequences followed by five QPG MRAC sequences (Q.MRAC) were sequentially acquired. The QPG PET data was reconstructed using 3D iterative reconstruction and each of the MRAC and Q.MRAC attenuation maps. The Q.MRAC PET images were evaluated for diagnostic acceptability by a radiologist and a nuclear medicine physician based on a Blinded Image Evaluation (BIE). The diagnostic acceptance was determined based upon a consensus review between the two readers. For 20 of these patients, the Q.MRAC PET images were rated on the Likert scale (1 to 5, with 5 indicating excellent image quality) by two additional nuclear radiologists. The reconstructed PET images were sent to a workstation for analysis. The SUV_mean in the lung and liver were measured using regions of interest (ROIs) drawn two slices superior and inferior to dome of the liver, respectively. The mean and standard deviation (SD) of liver and lung SUV_mean were then computed for the MRAC and Q.MRAC PET images for each patient, with the SD representing the repeatability.

Results: The reviewers rated all Q.MRAC PET images as diagnostically acceptable. In addition, the Q.MRAC PET images were rated as acceptable to good (Likert rating 3.9 ± 0.8) as some of the images showed mild artifact that did not impact diagnostic confidence. No Q.MRAC PET images were rated below 3. In 26 patients the mean liver SUV was approximately 10% greater for the PET Q.MRAC than the PET MRAC images. Also, in 20 out of 30 patients, the repeatability of liver SUV improved when Q.MRAC was employed. There was negligible difference in the mean lung SUV, and in 17 patients the repeatability of the lung SUV improved when Q.MRAC was employed. Conclusion: The use of QPG MRAC for a PET/MR study produces diagnostically acceptable images with improved repeatability of SUV over non-gated MRAC.