Fig 1.
Study flow diagram.
Table 1.
Acquisition parameters for the applied MR-sequences (A-F).
Fig 2.
Both PET/CT (A,B) as well as PET/MRI (C; D; VIBE, portal venous phase) show a lesion with elevated FDG-uptake and ill-defined lesion borders as well as central contrast enhancement as signs of malignancy. Based on these findings the lesion was correctly identified as metastasis in both modalities.
Fig 3.
cePET/CT (A+B) shows a hypodense liver lesion of 10 mm in diameter without elevated tracer uptake which is therefore rated as indeterminate. In PET/MRI (C-F) the lesion is hyperintense in T2w TSE (D) and shows signs of restricted diffusion (E: b1000; F: ADC map). Therefore, based on PET/MRI the lesion is rated as metastasis. Based on T2w imaging additional 4 metastases are visualized in PET/MRI in the contralateral lobe.
Fig 4.
Patient without liver metastases.
PET/CT with a false positive result showing a hypodense pseudolesion with a diameter 9 mm in the CT dataset without correlate in PET. In the later acquired PET/MRI (C-E) no correlate in PET nor in the morphological datasets (D: T1w VIBE portal-venous phase; E: T2w TSE fs).
Table 2.
Further characterization of benign liver lesions.
Table 3.
Liver metastases not visible in PET/CT: detection rate in different MR-sequences.
Table 4.
Sensitivity, Specificity, Accuracy (area under the curve, AUC), PPV, NPV with 95% CI for each reader as well as in combination.
Significant differences between PET/CT and PET/MRI are indicated (*: p<0.05; **: p<0.01, ***:p<0.001).
Fig 5.
Receiver operating characteristic (ROC) curve for PET/CT and PET/MRI for reader 1 (left), reader 2 (middle) and both readers (right).
Significantly higher accuracy (area under the curve, AUC) for PET/MRI (p<0.01 for reader 1 and 2; p<0.001 overall).
Table 5.
Lesion conspicuity and diagnostic confidence in PET/CT and PET/MRI.
Values are reported as mean±SD [median; range].