Late-Phase Pulse-Inversion Sonography Using the Contrast Agent Levovist: Differentiation Between Benign and Malignant Focal Lesions of the Liver

Transcription

1 Downloaded from by on 01/28/18 from IP address Copyright RRS. For personal use only; all rights reserved lexandra von Herbay 1 Christoph Vogt Dieter Häussinger Received September 13, 2001; accepted after revision pril 26, Supported by the Freunde und Förderer der Heinrich- Heine-Universität Düsseldorf and by the Deutsche Krebshilfe. 1 ll authors: Department of Medicine and Department of Hepatology, Gastroenterology and Infectious Diseases, University of Düsseldorf, Moorenstr. 5, D Düsseldorf, Germany. ddress correspondence to. von Herbay. JR 2002;179: X/02/ merican Roentgen Ray Society Late-Phase Pulse-Inversion Sonography Using the Contrast gent Levovist: Differentiation etween enign and Malignant Focal Lesions of the Liver OJECTIVE. We assessed the ability of contrast-enhanced sonography to reveal differences between benign and malignant focal hepatic lesions. SUJECTS ND METHODS. We examined 67 patients with focal hepatic lesions in a prospective study. The causes of the lesions were confirmed by histology, CT, MR imaging, or scintigraphy. The liver was screened for focal lesions using sonography. Thereafter, 2 g of Levovist (300 mg/ml; 1 ml/sec) was injected IV as a bolus. fter a delay of at least 2.5 min without scanning, the liver was examined via three different scans using pulse-inversion sonography. RESULTS. For the discrimination of malignant versus benign liver lesions, contrast-enhanced sonography improved sensitivity from 85% to 100% and specificity from 30% to 63%, as compared with baseline sonography. Receiver operating characteristic analysis revealed a significant improvement in this discrimination ( z = ± at baseline sonography, z = ± with contrast-enhanced sonography, p < 0.001). Furthermore, a lower interobserver variability was found for contrast-enhanced sonography (weighted κ = 0.947), as compared with baseline sonography (weighted κ = 0.469). ll lesions that had homogeneous enhancement in the late phase of Levovist enhancement were benign. In distinction, 90% of lesions without contrast enhancement in the late phase were malignant. ll lesions were malignant that were isoechoic (invisible) on baseline sonography but visible because of lack of enhancement after injection. CONCLUSION. Contrast-enhanced sonography has greater specificity and sensitivity than baseline sonography for the differentiation of benign and malignant liver lesions. S onography is usually the first line of investigation in the assessment of liver disease because it is noninvasive and widely available. The detection of focal hepatic lesions is relatively straightforward with sonography, but the differentiation between benign and malignant lesions is often difficult. Therefore, a technique that could improve the ability of sonography to reveal the differences between benign and malignant lesions would be a major clinical benefit for both patients and investigators. Sonographic contrast agents have been studied to improve imaging of the liver. One of these agents, Levovist (galactose 99.9%, palmitic acid 0.1%; Schering, erlin, Germany), has a specific late phase with selective enhancement of the parenchyma of the liver and spleen [1]. This postvascular phase becomes dominant 2 3 min after a bolus injection [1 2]. The exact site of Levovist accumulation in the liver parenchyma is still unknown, but it may accumulate in the reticuloendothelial cells [2]. The accumulation of Levovist in the parenchyma of the liver and spleen can be made visible on sonography using the pulse-inversion mode [3]. With pulse-inversion imaging, two separated pulses, 180 out of phase, are transmitted into the tissue, and the resulting echoes are added to form the final sonographic signal [4 8]. The linear echoes reflected from the tissue are canceled, whereas the nonlinear echoes reflected from the contrast medium produce a detectable signal. Other studies in the late phase of Levovist enhancement have shown that metastases have no contrast enhancement, whereas the surrounding liver parenchyma shows a bright and homogeneous Levovist accumulation [3, 9, 10]. The aim of our study was to demonstrate the ability of contrast-enhanced sonography to discriminate between benign and malignant focal lesions of the liver. JR:179, November

2 von Herbay et al. Downloaded from by on 01/28/18 from IP address Copyright RRS. For personal use only; all rights reserved Subjects and Methods In a prospective study, 67 patients with focal hepatic lesions (33 men and 34 women; mean age, 57 ± 14 years) were examined. The protocol was approved by the institutional review board at our hospital. efore the sonographic examination was started, the purpose of the study was explained to each patient, and oral informed consent was obtained according to the ethical guidelines from Helsinki [11]. Patients Our study population was selected from patients who presented for abdominal sonography. large number of patients undergo sonography in our institution; for practical reasons, we could not include every patient with focal hepatic lesions. We selected patients whose final diagnosis was confirmed by histology, CT, MR imaging, or scintigraphy and in whom all lesions had a similar appearance (so as to minimize the possibility of different causes of lesions in the examined liver). Only patients who agreed to provide oral informed consent were included. The final diagnosis for each subject was confirmed by liver biopsy, CT, MR imaging, or scintigraphy (Table 1), performed as part of the clinical workup of the patients. In 40 of the 67 patients, a malignant liver lesion was diagnosed as hepatocellular carcinoma (n = 18); cholangiocellular carcinoma (n = 1); metastatic liver disease (n = 19), or liver infiltration by non-hodgkin s lymphoma (n = 2). Patients with metastatic liver disease had the following diagnoses: adenocarcinoma of the colon (n = 6), adenocarcinoma of the pancreas (n = 2), adenocarcinoma of the cardia or stomach (n = 2), adenocarcinoma of unknown primary origin (n = 3), small cell carcinoma of the lung (n = 1), carcinoid (n = 2), melanoma (n = 2), and primitive neuroendocrine tumor (n = 1). We found no evidence of malignancy of the lesions in the remaining 27 patients. Their diagnoses included focal nodular hyperplasia (n = 8), hepatic adenoma (n = 1), focal hyposteatosis or hypersteatosis (n = 6), hemangioma (n = 8), and regenerative cirrhotic nodule (n = 4). The interval between the examination used as the confirmation standard and performance of the sonographic examination was 1 8 days for patients who underwent liver biopsy, 1 26 days for patients who underwent CT or MR imaging, and from 5 days to 1 year for patients whose focal nodular hyperplasia was diagnosed by scintigraphy. In all of the patients with focal nodular hyperplasia, the size of the lesion was stable during the interval between scintigraphy and sonography. Contrast gent Levovist is a sonographic contrast agent consisting of galactose microparticles (99.9%) and palmitic acid (0.1%) with a mean microbubble diameter of 2 5 µm (97% < 7 µm) [2, 12]. Originally, Levovist was developed to enhance the signal of color and power Doppler images during vascular sonography [13 16]. fter IV injection, Levovist circulates in the general blood pool for several minutes. Study Design aseline examinations included gray-scale, color, and power Doppler sonography. ll examinations were performed with a Sonoline Elegra sonography unit (Siemens, Erlangen, Germany) using a 3.5C40-multifrequency transducer (Siemens). We screened the liver for focal lesions in both native - and tissue harmonic imaging modes and documented the number and location of focal lesions. To ensure that identical areas of the liver were compared, we performed identical scans in each patient before and after Levovist injection. TLE 1 Final Diagnosis in 67 Patients Using Histology, CT, MR Imaging, or Scintigraphy Modality Final Diagnosis Patients Histology CT MR Imaging Scintigraphy Malignant focal hepatic lesions Hepatocellular carcinoma Metastasis Cholangiocellular carcinoma 1 1 Non-Hodgkin s lymphoma 2 2 Total enign focal hepatic lesions Focal nodular hyperplasia denoma 1 1 Hemangioma Regenerative cirrhotic nodule 4 4 Focal hyposteatosis or hypersteatosis 6 6 Total Pulse-inversion sonography was performed using the Ensemble Contrast Harmonic Imaging software (Siemens, Erlangen, Germany). Levovist was administered via bolus injection using an IV catheter having at least a 21-gauge diameter. bolus of 2 g of Levovist (300 mg/ml; 1 ml/sec) was injected IV followed by 10 ml of a 0.9%-saline bolus. fter a scanning delay of 2.5 min, the liver was examined via three distinct scans. Scanning was not performed during the 2.5-min delay to avoid the destruction of the microbubbles before the late phase. The delayed contrast-enhanced pulse-inversion scanning was terminated within 10 min. The equipment settings for late-phase imaging were set to CHI-mode, frequency of MHz, parallel processing turned on, maximal frame rate (>10 frames per second), persistence turned off, and high (>0.8) mechanical index (MI). Scanning was performed according to the procedure described by lbrecht et al. [5] (first sweep: right liver lobe, high single-focus zone [4 cm], transmit power 40% [MI > 0.8]; second sweep: right liver lobe, deep single-focus zone [10 cm], transmit power 100% [MI > 1.3]; third sweep: left liver lobe, high single-focus zone [4 cm], transmit power 100% [MI > 1.3]). The patient was asked not to breathe during the scanning procedure. The scan was started at the upper edge of the liver and then a fast sweep lasting approximately 3 4 sec was performed through the entire lobe of the liver. ll examinations were stored on magnetic optical disks and on S-VHS videotapes. Image nalysis nalysis of the baseline sonograms and of the contrast-enhanced studies was performed by two independent observers who were unaware of the final diagnosis of the lesions. oth observers evaluated all of the videotapes and the digitally stored images. The videotapes were coded by consecutive numbers so that the observers were unaware of the patient s name and diagnosis at the time of the retrospective analysis. We used a 5-point grading system for baseline and for contrast-enhanced images (1 = definitely benign, 2 = probably benign, 3 = indeterminate, 4 = probably malignant, 5 = definitely malignant). nalysis of the grading was performed for both baseline -mode sonography and contrastenhanced sonography. subjective evaluation of the lesion was used to rate the lesions at baseline sonography. This evaluation included consideration of the echogenicity, shape, and location of the lesions relative to neighboring vessels. Hyperechoic lesions next to veins were suspected to be hemangiomas, and lesions with a central starlike sonographic pattern were suspected to be focal nodular hyperplasia. Lesions with a hypoechoic rim or with polygonal edges were suspected to be malignant. For imaging performed after contrast agent administration, the lesion s enhancement characteristics were evaluated on a 5-point scale, with 1 being definitely benign and 5 being definitely ma JR:179, November 2002

3 Contrast-Enhanced Sonography of Liver Lesions Downloaded from by on 01/28/18 from IP address Copyright RRS. For personal use only; all rights reserved lignant. On the basis of prior experience, we rated lesions with homogeneous bright enhancement as benign; lesions without enhancement or with inhomogeneous enhancement were given higher ratings for malignancy. Lesions that could not be classified as either benign or malignant were given a score of 3. The sonographer was one of the two physicians experienced in sonography who performed the review of these examinations. In addition to the 5-point scale, contrast-enhanced images were classified according to the following features: homogeneous bright enhancement, dark nodules without enhancement surrounded by bright enhancement of the liver, and inhomogeneous enhancement. The two observers performed this classification working in consensus after the independent blinded ratings on the 5- point scale were made. Statistical nalysis nalysis of our data was made on a by patient basis. In patients with multiple lesions in whom one lesion was diagnosed at biopsy of the liver (n = 44), that one lesion was chosen for analysis. In the remaining patients, the observer chose a lesion that was seen without difficulty on baseline sonography to ensure that the contrast enhancement of the lesion could be evaluated. Only six of the patients who did not undergo liver biopsy had multiple lesions. In these patients, the appearance of all of the various lesions revealed on sonography, CT, or MR imaging was similar in any given patient. (Patients with multiple lesions of varied appearance had been excluded from the study.) mean rating score was computed for each lesion as the average of the ratings provided by the two independent observers. To calculate sensitivity and specificity, we counted a score of 3 or higher as a positive test result. Our rationale for treating a score of 3 (i.e., indeterminate) as malignant was to avoid missing possibly malignant lesions. In addition, calculations for unweighted and weighted kappa statistics were based on a 5 5 table of observer ratings for the comparison between the independently performed reviews of baseline sonography and contrast-enhanced sonography [17, 18]. The kappa analysis was performed using software [8.2/FREQ; SS Institute, Cary, NC]. The quadratic-weighted kappa was weighted according to the interval between the rating scores (5-point system) of both reviewers. Receiver operating characteristic curves were obtained for baseline and contrast-enhanced sonography according to the 5-point system as described. Comparison of receiver operating characteristic curves was performed with the Dorfman-erbaum-Metz laboratory algorithm to compare multiple reviewers and multiple cases [19, 20]. This program uses jackknifing and analysis of variance techniques to test the statistical significance of the differences between cases and reviewers. p value of less than 0.05 was considered significant. TLE 2 Fig. 1. Receiver operating characteristic curve. Graph shows differentiation of malignant from benign lesions using contrast-enhanced sonography in late phase of contrast enhancement. We set cutoff value at 3. roken line represents baseline sonography; solid line represents contrast-enhanced sonography. Sensitivity and specificity of contrast-enhanced sonography for differentiation between benign and malignant lesions were significantly higher than those of -mode sonography. Sensitivity (True-Positive) Results oth baseline and contrast-enhanced sonography was performed in all patients without technical problems. Contrast enhancement was visible for a sufficient depth to reach at least the upper edge of the lesion in question in all patients. The deepest enhancement was seen at 17 cm. In 15 patients, enhancement was limited to a depth of cm and did not reach the deepest part of the liver. The mean rating scores of the observers for the 40 malignant lesions and the 27 benign lesions are summarized in Table 2. aseline -mode sonography detected 85% (34/40) of malignant lesions on the basis of a mean score of 3 or higher, with a specificity of 30% (8/27 lesions). Contrast-enhanced sonography detected 100% (40/40) of malignant lesions on the basis of a mean score of 3 or higher, with a specificity of 63% (17/27 lesions). Lesions in 23 patients were rated as indeterminate (mean score, 3) on baseline imaging because classification as benign or malignant was not possible. On contrast-enhanced imaging, lesions in only nine patients were rated as indeterminate. Interobserver agreement during baseline imaging was poor, with a weighted kappa value of Interobserver agreement during contrastenhanced imaging improved markedly, with a weighted kappa of Receiver operating characteristic curves are presented in Figure 1. The area under the fitted curve for contrast-enhanced sonography ( z = 0.947) is significantly higher (p < 0.001) than the area under the curve for baseline sonography ( z = 0.692). Rating of Focal Hepatic Lesions by Two Observers in 67 Patients Mean Rating Scores by Technique a enign Lesions (n = 27) Malignant Lesions (n = 40) aseline -mode sonography Scores of 1 and Scores of 2 and Score of Scores of 3.5 and Scores of 4.5 and Contrast-enhanced sonography b Scores of 1 and Scores of 2 and Score of Scores of 3.5 and Scores of 4.5 and a ased on 5-point grading system: 1 = definitely benign, 2 = probably benign, 3 = indeterminate, 4 = probably malignant, 5 = definitely malignant with cutoff value of 3.0. b Levovist; Schering, erlin, Germany Specificity (False-Positive) JR:179, November

4 von Herbay et al. TLE 3 -Mode Pulse-Inversion Sonography in Late Phase of Levovist dministration in 67 Patients Diagnosis of Focal Hepatic Lesion Patients Lesion Enhancement in Late Phase (n) None a Inhomogeneous Homogeneous Detection of Occult Lesions b Downloaded from by on 01/28/18 from IP address Copyright RRS. For personal use only; all rights reserved Malignant focal hepatic lesions Hepatocellular carcinoma c 0 5 Metastasis Cholangiocellular carcinoma Non-Hodgkin s lymphoma Total d 3 0 d 13 d enign focal hepatic lesions Focal nodular hyperplasia denoma Hemangioma e 0 Regenerative cirrhotic nodule 4 1 f 1 c 2 Focal hyposteatosis or hypersteatosis Total 27 4 d 6 17 d 0 d a Clear demarcation without enhancement surrounded by enhanced liver parenchyma. b Not visible at baseline imaging but revealed on contrast-enhanced sonography and confirmed by CT, MR imaging, or both. c Interpretation of lesion was not possible because Levovist (Schering; erlin, Germany) enhancement of surrounding liver parenchyma was inhomogeneous. ll of these patients had liver cirrhosis. d Difference between patients with benign and malignant focal hepatic lesions was statistically significant ( p<0.001). e Inhomogeneous Levovist uptake was found in lesions of these patients, in distinction to the homogeneous uptake in surrounding liver parenchyma. f Histologic examination of regenerative nodule in this patient showed cirrhotic nodular dysplasia. Table 3 summarizes the two observers consensus classification of the enhancement patterns in the lesions in our study population. Most lesions (58/67) were classified as clearly demarcated without enhancement (n = 41) or as homogeneously enhancing with the surrounding liver (n = 17). mong the nine patients whose lesions showed inhomogeneous enhancement, three patients with hepatoma and one with a regenerating nodule had underlying cirrhosis. The remaining five patients with inhomogeneous enhancement had hemangiomas. ll lesions that showed homogeneous enhancement were benign (p < 0.001, Figs. 2 4). mong the lesions that were clearly demarcated by their lack of enhancement, 90% were malignant (p < 0.001, Figs. 5 7). dditional lesions that were not visible on baseline imaging were detected during contrast-enhanced sonography in five patients with hepatocellular carcinoma and in eight patients with metastatic disease. These lesions were confirmed on CT, MR imaging, or both Fig year-old man with focal hyposteatosis., Native -mode sonogram shows hypoechoic lesion (arrow) ventral to portal vein., Contrast-enhanced late-phase pulseinversion sonogram shows homogeneous enhancement in focal lesion and surrounding liver parenchyma. in 12 patients and at exploratory laparotomy in one patient. No additional lesions were found during contrast-enhanced imaging of patients with benign disease. Thus, the finding on contrast-enhanced sonography of lesions that were occult on baseline imaging was 100% specific for malignancy (p < 0.001). Discussion Our study showed that contrast-enhanced sonography improves sensitivity and specificity 1276 JR:179, November 2002

5 Downloaded from by on 01/28/18 from IP address Copyright RRS. For personal use only; all rights reserved Contrast-Enhanced Sonography of Liver Lesions Fig year-old woman with focal nodular hyperplasia., Native -mode sonogram shows hypoechoic inhomogeneous lesion (arrows) with hypoechoic rim., Contrast-enhanced late-phase pulse-inversion sonogram shows homogeneous enhancement in focal lesion and surrounding liver parenchyma. Fig year-old woman with benign adenoma of liver., Native -mode sonogram shows two hypoechoic lesions (arrows) in right lobe of liver., Contrast-enhanced late-phase pulse-inversion sonogram shows homogeneous enhancement in both focal lesions and surrounding liver parenchyma. in discriminating between benign and malignant focal liver lesions when compared with baseline sonography. Furthermore, we found less interobserver variability in contrast-enhanced sonography than in baseline sonography. Homogeneous parenchymal Levovist uptake in the liver-specific late phase appears to be a useful indicator of benignity in focal hepatic lesions. None of the lesions in our study that showed homogeneous Levovist uptake in the liver-specific late phase were proven to be malignant. Homogeneous uptake was found in focal hypersteatosis and hyposteatosis of the liver, in regenerative nodules in patients with liver cirrhosis, in focal nodular hyperplasia, and in benign adenoma. This uptake is a result of the reticuloendothelial cells in these lesions [21]. However, the exact mechanism of Levovist-specific uptake in the parenchyma of the JR:179, November 2002 liver and spleen is still being investigated [2]. The fundamental mechanism seems to be comparable to the 99m Tc-colloid uptake found on scintigraphy of the liver [21 25] and to the Endorem (ferumoxides; Guerbet, Sulzbach, Germany) or Resovist (ferucarbotran; Schering, erlin, Germany) uptake in the reticuloendothelial cells found on MR imaging [26, 27]. None of the patients with histologically proven malignant liver lesions had a homogeneous Levovist uptake in the late phase. However, absence of Levovist uptake in the late phase was found in three patients with liver hemangioma and in one patient with a regenerative cirrhotic nodule. Surgical resection was performed because hepatocellular carcinoma (HCC) was suspected in this particular patient, and dysplasia was found. In this special case, the development to HCC may have been prevented because contrastenhanced sonography allowed localization of a dysplastic nodule and such nodules have a high incidence of HCC [28]. The fact that hemangiomas are free of liver tissue and reticuloendothelial cells [29] may possibly explain the clear demarcation without Levovist enhancement seen in the three patients with giant hemangiomas (diameters, 8 20 cm). However, an inhomogeneous enhancement in the Levovist late phase was seen in the remaining five patients with hemangiomas. This finding is possibly explained by the vascular architecture in hemangiomas. Hemangiomas contain very small veins surrounded by endothelial cells, which could result in a substantial prolongation of the blood pool phase thus resulting in Levovist accumulation as long as

6 Downloaded from by on 01/28/18 from IP address Copyright RRS. For personal use only; all rights reserved von Herbay et al. Fig year-old man with metastatic liver disease., Native -mode sonogram shows inhomogeneous parenchyma of liver, suggestive of focal liver lesions., Contrast-enhanced late-phase pulse-inversion sonogram shows clear demarcation of multiple focal lesions without enhancement surrounded by enhanced liver parenchyma. Liver biopsy revealed metastatic liver disease by adenocarcinoma. Diffuse infiltration of liver was confirmed on CT. Fig year-old woman with cholangiocellular carcinoma., Native -mode sonogram shows hypoechoic area (arrows) in right lobe of liver., Contrast-enhanced late-phase pulseinversion sonogram shows clear demarcation of area involved by cholangiocellular carcinoma (arrow). min after injection. Normal vessels are already free of Levovist at this time. Our findings in hemangiomas are counter to the results of ottinelli et al. [30], who found an enhancement similar to liver parenchyma in the late phase. The differences reported in the cited study might be explained by an earlier late phase and smaller diameter of the hemangiomas than in our study. None of the patients with HCC or metastatic liver disease showed a homogeneous uptake of Levovist in the lesion because of the lack of reticuloendothelial cells in HCC and metastases. The absence of late phase Levovist uptake is comparable to the failure of 99m Tc-colloid uptake in hepatocellular carcinoma that was diagnosed on scintigraphy [25] and the failure of Endorem and Resovist uptake found on MR im aging [26, 27]. Calvet et al. [31] found a positive correlation between the uptake of diethylaminodiacetate derivatives on scintigraphy and the grading of HCC. Whether the grading of HCC affects the late phase of Levovist uptake is unknown. In our study, demarcation of HCC without contrast enhancement was found both in patients with highly differentiated HCC and in those with less differentiated HCC. In 32% of the patients in our study who had malignant liver disease, late-phase imaging depicted lesions that were undetected on sonography without Levovist. These findings correlate well with the findings of Dalla Palma et al. [32], Harvey et al. [3], and lbrecht et al. [33], who performed Levovist late-phase sonography using pulse-inversion harmonic imaging in patients with known malignancies. These investigators found additional lesions in 56%, 40%, and 45% of patients, respectively. lthough it appears to be a powerful tool for improving the sonographic diagnosis of liver lesions, contrast-enhanced late-phase sonography has some problems. So far, only high-end sonography machines are equipped for this procedure. In addition, several important pitfalls need to be considered by the investigator if correct results are to be obtained, including the destruction of Levovist by continuous insonation, putative destruction of Levovist during the bolus injection by using a needle smaller than 21 gauge or the valve of the needle [34], and ineffective adaptation of the focus zone or the mechanical index. In conclusion, sonography in the late phase of Levovist enhancement can help in making JR:179, November 2002

7 Contrast-Enhanced Sonography of Liver Lesions Downloaded from by on 01/28/18 from IP address Copyright RRS. For personal use only; all rights reserved Fig year-old man with hepatocellular carcinoma., Native -mode sonogram shows inhomogeneous lesion in right lobe of liver., Contrast-enhanced late-phase pulseinversion sonogram shows clear demarcation of focal lesion (arrows) without contrast enhancement surrounded by enhanced liver parenchyma. the clinical decision of whether a sonographically detected liver lesion will need further investigation. ccording to our results, all patients with homogeneous contrast enhancement in the lesion had benign lesions. Therefore, a more conservative approach might be reasonable in following up these patients. This new technique may help to identify patients in whom examinations that are invasive, expensive, or both such as liver biopsy, CT, or MR imaging may be unnecessary. cknowledgments We thank Reinhart Willers of the Computer Department of the University of Düsseldorf, Germany, for the excellent statistical assistance and Philip May for editorial assistance. References 1. Forsberg F, Goldberg, Liu J, Merton D, Rawool NM, Shi WT. Tissue-specific US contrast agent for evaluation of hepatic and splenic parenchyma. Radiology 1999;210: Schlief R. Galactose-based echo-enhancing agents. In: Goldberg, ed. Ultrasound contrast agents. London: Dunitz, 1997: Harvey CJ, lomley MJ, Eckersley RJ, Heckemann R, utler-arnes J, Cosgrove DO. Pulseinversion mode imaging of liver specific microbubbles: improved detection of subcentimetre metastases. Lancet 2000;355: auer, Hauff P, Lazenby J, et al. Wideband harmonic imaging: a novel contrast ultrasound imaging technique. Eur Radiol 1999;9[suppl 3]:S364 S lbrecht T, Hoffmann CW, Schettler S, et al. - mode enhancement at phase-inversion US with airbased microbubble contrast agent: initial experience in humans. Radiology 2000;216: Dill-Macky MJ, urns PN, Khalili K, Wilson SR. Focal hepatic masses: enhancement patterns with SH U 508 and pulse-inversion US. Radiology 2002;222: Shi WT, Forsberg F. Ultrasonic characterization of the nonlinear properties of contrast microbubbles. Ultrasound Med iol 2000;26: urns PN, Hope Simpson D, verkiou M. Nonlinear imaging. Ultrasound Med iol 2000;26[suppl 1]:S19 S22 9. lbrecht T, Hoffmann CW, Schettler S, Overberg, Ilg M, Wolf KJ. Improved detection of liver metastases with phase inversion imaging during the liver-specific phase of the ultrasound contrast agent Levovist. Eur Radiol 1999;9[suppl 3]:S lbrecht T, arr R, lomley M, et al. Seeking consensus: contrast ultrasound in radiology. Invest Radiol 2002;37: World Medical ssociation. Declaration of Helsinki: ethical principles for medical research involving human subjects, revised. Edinburgh: World Medical ssociation, Campani R, Calliada F, ottinelli O, et al. Contrast media in ultrasonography: hepatic focal lesions. Radiol Med 1998;95[suppl 1]: Goldberg, Liu J, urns PN, Merton D, Forsberg F. Galactose-based intravenous sonographic contrast agent: experimental studies. J Ultrasound Med 1993;12: Kedar RP, Cosgrove D, McCready VR, amber JC, Carter ER. Microbubble contrast agent for color Doppler US: effect on breast masses. Radiology 1996;198: Maruyama H, Matsutani S, Sato G, et al. Enhanced color flow images in small hepatocellular carcinoma. bdom Imaging 2000;25: Strobel D, Krodel U, Martus P, Hahn EG, ecker D. Clinical evaluation of contrast-enhanced color Doppler sonography in the differential diagnosis of liver tumors. J Clin Ultrasound 2000;28: Cohen. Comparison of correlated correlations. Stat Med 1989;8: renner H, Kliebsch U. Dependence of weighted kappa coefficients on the number of categories. Epidemiology 1996;7: Swensson RG, King JL, Gur D. constrained formulation for the receiver operating characteristic (ROC) curve based on probability summation. Med Phys 2001;28: Dorfman DD, erbaum KS, Metz CE. Receiver operating characteristics rating analysis: generalization of the population of readers and patients with the jackknife method. Invest Radiol 1992;27: Rogers JV, Mack L, Freeny PC, Johnson ML, Sones PJ. Hepatic focal nodular hyperplasia: angiography, CT, sonography, and scintigraphy. JR 1981;137: Rubin R, Lichtenstein GR. Hepatic scintigraphy in the evaluation of solitary solid liver masses. J Nucl Med 1993;34: Piers D, Houthoff HJ, Krom R, Schuur KH, Sikkens H, Weits J. Hot spot liver scan in focal nodular hyperplasia. JR 1980;135: Welch TJ, Sheedy PF 2nd, Johnson CM, et al. Focal nodular hyperplasia and hepatic adenoma: comparison of angiography, CT, US, and scintigraphy. Radiology 1985;156: Kotzerke J, Schwarzrock R, Krischek O, Wiese H, Hundeshagen H. Technetium- 99m DISID hepatobiliary agent in diagnosis of hepatocellular carcinoma, adenoma, and focal nodular hyperplasia. J Nucl Med 1989;30: Hundt W, Petsch R, Helmberger T, Reiser M. Signal changes in liver and spleen after Endorem administration in patients with and without liver cirrhosis. Eur Radiol 2000;10: Mintorovitch J, Shamsi K. Eovist injection and Resovist injection: two new liver-specific contrast agents for MRI. Oncology 2000;14[suppl 3]: Orsatti G, Theise ND, Thung SN, Paronetto F. DN image cytometric analysis of macroregenerative nodules (adenomatous hyperplasia) of the liver: evidence in support of their preneoplastic nature. Hepatology 1993;17: Tung G, Cronan JJ. Percutaneous needle biopsy of hepatic cavernous hemangioma. J Clin Gastroenterol 1993;16: ottinelli O, Callida F, Campani R, et al. Specific features of hepatic lesions on pulse inversion and intermittent imaging after ultrasound contrast agent injection. (abstr) Ultrasound Med iol 2000;26[suppl 2]: Calvet X, Pons F, ruix J, et al. Technetium- 99m DIS- ID hepatobiliary agent in diagnosis of hepatocellular carcinoma: relationship between detectability and tumor differentiation. J Nucl Med 1988;29: Dalla Palma L, ertolotto M, Quaia E, Locatelli M. Detection of liver metastases with pulse inversion harmonic imaging: preliminary results. Eur Radiol 1999;9[suppl 3]:S382 S lbrecht T, Hoffmann CW, Schmitz S, et al. Phaseinversion sonography during the liver-specific late phase of contrast enhancement: improved detection of liver metastases. JR 2001;176: Kratzer W, Kachele V, Merkle E, et al. Contrast enhanced power Doppler sonography: comparison of various administration forms of the ultrasound contrast agent Levovist [in German]. Rofo Fortschr Geb Rontgenstr Neuen ildgeb Verfahr 2000;172: JR:179, November