The effect of gadolinium enhancement on cardiac PET / MR imaging based on MR-Dixon attenuation correction sequence

Purpose: To investigate the influence of gadolinium contrast agent on attenuation correction method based on MR-Dixon sequence in cardiac PET / MR imaging, and analyze its effect on PET semi- quantitative results.

Methods: Seventeen patients with glucose loading underwent 18F-fluorodeoxyglucose (FDG) cardiac PET/MR scan. Cardiac PET was acquired with ECG gated list mode during free breathing for ten minutes, and cardiac MR images were collected during expiratory phase with breath holding. Two attenuation maps (umap_Gd- and umap_Gd+) were calculated from the respiratory-gating Dixon MRI sequence acquired before and after gadolinium contrast medium was injected. Two sets of PET images (PET_Gd- and PET_Gd+) were reconstructed using umap_Gd- and umap_Gd+ combined with OSEM point spread function correction methods, respectively. The two umaps were fused by using TrueD software of Siemens MMWP workstation and their accuracy was compared visually. Mean SUV values (SUVmean) of PET_Gd- and PET_Gd+ in cardiac AHA 1-20 segment of each patient were calculated using Cedars QPS cardiac function. The difference of SUVmean between PET_Gd- and PET_Gd+ was compared by using paired t-test and the correlation was analyzed by Pearson correlation.

Results: All patients completed the examinations. The umap_Gd- and umap_Gd+ of 10 patients were segmented into soft tissue, the SUVmean of PET_Gd- and PET_{Gd +} had a good correlation ((5.83 ± 2.61 vs 6.21 ± 2.85, r = 0.986). There were significant statistical differences (P <0.0001); the umap_Gd- and umap_Gd+ of 3 patients were segmented into adipose tissue, there was a good correlation between SUVmean of PET_Gd- and PET_Gd+ (2.65±1.55 vs 2.58±1.48, r = 0.996), and there was a significant difference (P < 0.0001). The umap_Gd- was segmented into soft tissue while umap_Gd+ was segmented into adipose tissue in 4 patients, there was a good correlation between PET_Gd- (SUVmean = 8.14±3.87) and PET_Gd+ (SUVmean = 6.70±3.86) (r = 0.982), and they showed significant difference (P < 0.0001).

Conclusions: In cardiac PET/MR imaging, gadolinium contrast agent influences the attenuation correction method based on MR-Dixon sequence and leads to inaccurate SUV results. Therefore, MR-Dixon attenuation correction sequence and PET imaging should be performed before the injection of contrast agents. In addition, it is recommended that umap should be combined with PET / MR image to avoid PET interpretation errors in case that the segmentation fails.