Improving liver lesion characterisation using retrospective fusion of FDG PET/CT and MRI

AimTo compare retrospectively fused FDG PET/CT and MRI (PET/MRI) to FDG PET/CT and MRI for characterisation of indeterminate focal liver lesions as malignant or benign in patients with a known primary malignancy.Materials and methodA retrospective review of 70 patients (30 females, 40 males; mean age 56 ± 14 years) with 150 indeterminate lesions after FDG PET/CT and MRI (mean scan time interval 21 ± 11 days). HERMES® software was used to fuse PET/CT and MRI scans which were reviewed by 2 readers using the Likert score (scale 1–5) to characterise lesions as benign (1–3) or malignant (4–5). Final diagnosis was determined by histopathology or follow up imaging. Results for fused PET/MRI were compared to PET/CT and MRI alone.ResultsFor detection, MRI and fused PET/MRI detected all the lesions while PET/CT detected 89.4%. Characterisation of liver lesions as malignant on PET/CT alone yielded sensitivity, specificity, accuracy, PPV and NPV of 55.6%, 83.3%, 66.7%, 83.3%, 55.6% respectively and 67.6%, 92.1%, 80%, 89.3%, 74.5% for MRI, respectively. The sensitivity, specificity, accuracy, PPV and NPV for characterising lesions as malignant increased to 91.9%, 97.4%, 94.7%, 97.1%, 92.5% with PET/MRI fusion. The sensitivity, specificity, accuracy, PPV and NPV of fused PET/MRI for characterising lesions as malignant remained superior to PET/CT and MRI.ConclusionRetrospective fusion of PET with MRI has improved characterisation of indeterminate focal liver lesions compared to MRI or FDG PET/CT alone.