OBJECTIVE. The objective of our study was to assess the diagnostic utility of the "salt-and-pepper noise" sign on fat-fraction maps by chemical-shift-encoded MRI (CSE-MRI) compared with the halo sign on fat-suppressed T2-weighted imaging and mean attenuation on CT for differentiating bone islands from osteoblastic metastases. MATERIALS AND METHODS. Twenty-nine patients with 43 sclerotic vertebral bone marrow lesions (26 osteoblastic metastases, 17 bone islands) were included retrospectively. All patients underwent CT and MRI, including a CSE-MRI sequence on a 1.5-T MRI system, from November 2016 to January 2019. The salt-and-pepper noise sign was defined as the speckled appearance of white and black pixels that is similar to the appearance of background air on a fat-fraction map. ROC curves were analyzed to compare the diagnostic performance of the salt-and-pepper noise sign, halo sign, and mean CT attenuation between the two groups. RESULTS. The salt-and-pepper noise sign was significantly associated with bone islands (p < 0.001). The sensitivity, specificity, and accuracy for discriminating bone islands from osteoblastic metastases were 92.31-96.15%, 100%, and 95.35-97.67% for the salt-and-pepper noise sign; 88.46-92.31%, 88.24-94.12%, and 90.70% for the halo sign; and 96.15%, 94.12-100%, and 95.35-97.67% for mean CT attenuation, respectively. There was no statistically significant difference of diagnostic performances among the imaging characteristics for differentiating between bone islands and osteoblastic metastases (p > 0.05). Interobserver agreement for the salt-and-pepper noise sign, halo sign, and mean CT attenuation was almost perfect (κ ≥ 0.953, κ = 0.905, and ICC = 0.966, respectively). CONCLUSION. The salt-and-pepper noise sign is present in bone islands on fat-fraction maps by CSE-MRI and can aid in differentiating bone islands from osteoblastic metastases.
Keywords: MRI; bone islands; chemical-shift–encoded MRI; osteoblastic metastases; proton density fat fraction.