Diagnostic value of signal-enhanced color Doppler sonography for cutis-involving lesions.

Purpose: Prospective evaluation of the diagnostic outcome of plain and sign al-enhanced color Doppler sonography for the differentiation between benign and malignant cutis-involving lesions. Materials and Methods: 39 patients with 40 suspected malignant skin lesions received an preoperative ultrasoun d examination before and after application of 4 g Levovist(R) i.v. Spectral Doppler, conventional color and power Doppler were performed. The images w ere analyzed semiquantitatively using a specially developed software. The d iagnoses were confirmed by histological analysis (37 cases) or by follow up controls (3 cases). Results: Whereas the B-mode criteria such as echogenoi ty, borderline and homogenecity, and the spectral Doppler analysis were not useful for differentiation between malignant and benign lesions, the analy sis of the intratumorously visible number of vessels and of the ratio of th e vascularized area using a special software, especially after injection of the signal enhancing agent, provided valuable information. In 10 of 18 ben ign but only in 1 of 22 malignant lesions, no intratumorous vessels were vi sible after application of Levovist (plain: 10/18 benign and 6/22 malignant lesions). The number of visible vessels increased strongly after signal en hancement. Using a "ratio of the vascularized area (percentage vessel area) > 5 %" as a criterion of malignancy, 2 false positive and 2 false negative results were observed after application of Levovist(R) (without/with Levov ist(R) sensitivity 73 %/91 %, specificity 89 %/89 %). Conclusions: The sign al-enhanced color Doppler sonography provides useful additional information for the differentiation between benign and malignant cutis-involving lesio ns, especially the semiquantitative vascularization analysis. The latter is superior to B-mode ultrasound and to spectral Doppler analysis. The signal enhancer increases the sensitivity, whereas the specificity remains unchan ged.