BLUE RIBBON POSTER RADIOPHARMACEUTICALS: Improving Myocardial Quantitation Using Data-Driven Gated PET/CT
Abstract
Purpose
Accurate PET/CT registration is essential for quantitative cardiac PET. Common approaches include manual rigid registration of CT to PET (MRR) and average CT derived from cine CT (ACT). We evaluated whether data-driven gated PET combined with end-expiration CT for attenuation correction (DDG) provides superior registration and quantitative accuracy compared with MRR and ACT.
Methods
In whole-body FDG PET/CT, cine CT was acquired over regions prone to PET/CT misregistration, often caused by excessive inhalation during CT. From cine CT, both ACT and end-expiration CT (EECT) were generated, with EECT obtained by selecting images with the highest lung density. Twenty of such studies with significant myocardial FDG uptake were identified. Two sets of attenuation-corrected PET images were reconstructed using MRR, and ACT. The DDG set was obtained by data-driven gated PET between 30-80% of the phase range from end expiration and R-value threshold of 9, followed by attenuation correction using EECT. Polar maps were generated using Cedars QPS for all methods and for misregistered PET images, which served as baseline. Quantitative analysis used 17-segment SUV percent difference (SUV%Diff) relative to baseline.
Results
Without manual registration, DDG achieved average SUV%Diff value ~8% higher than ACT and 18% higher than MRR. Statistically significant differences were observed between DDG and ACT, and between ACT and MRR, particularly in inferior wall. Liver displacement was approximately twice that of the heart; thus, heart-focused MRR often left the liver misregistered, leading to reduced attenuation correction and SUV underestimation in the inferior wall. ACT was sensitive to irregular respiration, whereas DDG was not. The lateral and superior walls showed the greatest SUV variability due to interpatient differences in breathing depth.
Conclusion
DDG provides more accurate PET/CT registration and quantification than ACT or MRR in cardiac PET, without requiring manual registration, and is more robust to irregular respiration.