[Integrated diagnosis of liver angioma: comparison of Doppler color ultrasonography, computerized tomography, and magnetic resonance]

Abstract

Hemangiomas are the most frequent benign tumors of the liver which are diagnosed at clinics only when they are very big. Their differentiation from primary or secondary liver tumors is a very difficult step. To investigate the capabilities of color Doppler US, CT and MRI in characterizing liver hemangiomas, we retrospectively examined 27 patients with 35 hemangiomas of the liver, all identified with US from January, 1995, to March, 1996, plus 26 patients with color Doppler US, 23 with dynamic CT and 19 with MR findings; 16 patients had been examined with all imaging techniques. Our gold standard was red blood cell SPECT in 16 patients, clinical-diagnostic follow-up in 8 and surgical specimens in 3 patients. Morphological data, the number, size and segmental site were investigated for all lesions. Doppler US was used to study the qualitative and quantitative data of intralesional color patterns, according to a modified Tanaka's classification, and of intralesional max, systolic velocity and pulsatility index (PI). Dynamic CT was performed with the polyphasic technique, that is unenhanced scans followed by selective, dynamic sequential and delayed scans. Morphological data and dynamic time/density curves of lesion/parenchyma were studied. MRI was performed at 0.5 T with T1-weighted SE (TR/TE 280/18), T2-weighted SE (TR/TE 2000/45-90-120) and T2-weighted fast SE (TR/TE 5000/160) sequences. Dynamic T1-weighted GE sequences were performed after Gd-DTPA bolus injection, using the same times as those of dynamic CT. US showed 1.3 lesion/patient, while CT and MRI showed 1.5 and 1.7 lesion/patient, respectively. No differences were observed in lesion size with the three imaging techniques and there was 78% agreement on segmental site. As for morphological patterns, US had 46% sensitivity, considering medium-small lesions only. Intralesional color signals, with spot and branch patterns, were seen in 6/20 lesions (mean phi; 8.4 cm), with a mean intralesional systolic velocity of 0.30 m/s (1.46 KHz); mean intralesional PI value was 0.83. Color Doppler sensitivity was 23%. Combined B-mode and color Doppler US had 69% sensitivity. The typical CT pattern (ring or globular enhancement with centripetal progression and late hyperdensity) had 66% sensitivity. Hyperintensity on T2-weighted images had 96% sensitivity; Gd-DTPA MRI combined with dynamic CT had 100% sensitivity. In conclusion, the US findings of a suspected hemangioma should be diagnostically integrated with MR studies to make the correct diagnosis.