Characterization of focal liver lesions with SonoVue®-enhanced sonography: International multicenter-study in comparison to CT and MRI

2009 
AIM: To evaluate in a multicenter study whether the sonographic characterization of focal liver lesions can be improved using SonoVue®-enhancement; and to compare this method with computed tomography (CT) and magnetic resonance imaging (MRI). METHODS: One hundred and thirty four patients with one focal liver lesion detected in baseline ultrasound (US) were examined with conventional US, contrast-enhanced US (n = 134), contrast-enhanced CT (n = 115) and/or dynamic contrast-enhanced MRI (n = 70). The lesions were classified as malignant, benign or indeterminate and the type of lesion was determined. The final diagnosis based on the combined information of all imaging examinations, clinical information and histology (n = 32) was used. Comparisons were made to see whether the addition of contrast-enhanced US led to the improvement of the characterization of doubtful focal liver lesions. RESULTS: In comparison with unenhanced US, SonoVue® markedly improves sensitivity and specificity for the characterization (malignant/benign) of focal liver lesions. In comparison with CT and/or dynamic MRI, SonoVue®-enhanced sonography applied for characterization of focal liver lesions was 30.2% more sensitive in the recognition of malignancy and 16.1% more specific in the exclusion of malignancy and overall 22.9% more accurate. In the subgroup with confirmative histology available (n = 30), sensitivity was 95.5% (CEUS), 72.2% (CT) and 81.8% (MRI), and specificity was 75.0% (CEUS), 37.5% (CT) and 42.9% (MRI). The sensitivity and specificity of CEUS for the identification of focal nodular hyperplasia (FNH) and hemangiomas was 100% and 87%, resulting in an accuracy of 94.5%. CONCLUSION: SonoVue®-enhanced sonography emerges as the most sensitive, most specific and thus most accurate imaging modality for the characterization of focal liver lesions.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    31
    References
    162
    Citations
    NaN
    KQI
    []