Rationale and objectives: To compare B-mode ultrasonography (US) alone with the combination of B-mode and contrast-enhanced (Sonazoid) late-phase pulse-inversion US for the detection of hepatic metastases by use of jackknife free-response receiver-operating characteristic (JAFROC) analysis.
Materials and methods: Twenty-seven patients with 57 hepatic metastases and 6 patients without hepatic metastases underwent B-mode and contrast-enhanced US. We used the diagnoses established by contrast-enhanced computed tomography and contrast-enhanced US as the standard of reference. All ultrasonographic scanning was performed by an experienced radiologist with a routine clinical procedure. All scanning data were archived with digital cine clips. A review system, which can display pairs of cine clips for B-mode and contrast-enhanced US side by side, was developed for off-site observer study. Seven radiologists interpreted each case individually first by B-mode US only, and then by the combination with contrast-enhanced US by identifying locations of possible candidates for hepatic metastasis with their confidence ratings. The figure-of-merit (FOM) values, sensitivity, and false-positives per case were estimated for B-mode US alone, and for the combination of B-mode and contrast-enhanced US.
Results: The sensitivities of the combined ultrasonographic imaging (mean, 72.2%) were clearly improved from that of B-mode US alone (mean, 41.6%) while reducing the average number of false positives from 1.1 to 0.5 per case. In the jackknife analysis, there was a statistically significant difference between mean FOM values for the combined imaging (0.76) and for B-mode US alone (0.44, P < .00001).
Conclusion: Evaluating cine clips of contrast-enhanced liver US together with B-mode US could improve physicians' accuracy for detection of hepatic metastases.