(Resovist)-enhanced Breath-hold MR Imaging

Magn Reson Med Sci, Vol. 7, No. 3, pp. 123–130, 2008

MAJOR PAPER

Detection of Hepatocellular Carcinoma with Ferucarbotran (Resovist)-enhanced Breath-hold MR Imaging: Feasibility of 10 minute-delayed Images Kazuhiro SAITO1*, Hiroaki SHINDO1, Taizo OZUKI2, Aimi ISHIKAWA1, Fumio KOTAKE2, Yoko SHIMAZAKI1, and Kimihiko ABE1 1Department

of Radiology, Tokyo Medical University 6–7–1 Nishi-shinjuku, Shinjuku-ku, Tokyo 160–0023, Japan 2Department of Radiology, Tokyo Medical University Kasumigaura Hospital (Received November 12, 2007; Accepted July 2, 2008)

Purpose: We evaluated the optimal timing for breath-hold MR imaging with bolusinjectable superparamagnetic iron oxide (SPIO) for detecting hepatocellular carcinoma (HCC). Materials and Methods: Twenty patients with 62 HCCs (52 hypervascular, 10 non-hypervascular) underwent MR imaging that included unenhanced and SPIO-enhanced T1weighted gradient echo (GRE) and T2-weighted fast spin echo (FSE) sequences, perfusion study, and SPIO-enhanced T2*-weighted GRE sequences. We obtained SPIO-enhanced T2*-weighted sequences 10 and 30 min after injecting SPIO and made 2 image sets, comprising 10- or 30-min delayed T2*-weighted images. Three observers performed alternative free response receiver operating characteristic (AFROC) analysis, and quantitative evaluation was performed. Results: Only Observers 2 and 3 recognized a signiˆcant diŠerence in the area under the AFROC curve (Az) value in the 10-min delayed images; no signiˆcant diŠerence was observed in the 30-min delayed images. There was no signiˆcant diŠerence in the sensitivity of individual observers between 10- and 30-min delayed images. The contrast-to-noise (C/N) ratio of the 30-min delayed images was signiˆcantly higher than that of the 10-min delayed images. The C/N ratio of hypervascular HCCs in the 30-min delayed images was signiˆcantly higher than in the 10-min delayed images, but that of non-hypervascular HCCs showed no signiˆcant diŠerence. Conclusion: In most cases, 10-min delayed SPIO-enhanced T2*-weighted images are sufˆcient to detect HCCs. Keywords: hepatocellular carcinoma, MRI, optimal timing, superparamagnetic iron oxide of SPIO diŠers with primary and metastatic hepatic lesions. However, several reports mention SPIO's usefulness in detecting HCC,8–10 most reporting use of ferumoxides, which require delivery by intravenous drip infusion for more than 30 min, and most acquiring MR imaging 60 min after initiating infusion. On the other hand, ferucarbotran (Resovist}, Schering, Osaka, Japan) can be given as a bolus intravenous injection and MR images obtained 10 min after injection,11,12 and so, its use is increasingly accepted; information is insu‹cient on the optimal timing for detecting hepatocellular carcinoma arising from liver cirrhosis. Accurate evaluation of previous quantitative results with ferucarbotran-enhanced MR imaging was di‹cult because studies did not distinguish metastatic from hepato-

Introduction Superparamagnetic iron oxide (SPIO) particles are used as a liver-speciˆc contrast agent for magnetic resonance (MR) imaging, particularly for detecting metastatic hepatic malignant tumors.1–3 Liver tissue surrounding hepatic metastases is usually normal pathologically; hepatocellular carcinoma (HCC) usually arises from lesions from chronic viral hepatitis and liver cirrhosis.4 The accumulation of SPIO in KupŠer cells depends on liver function;5,6 in liver cirrhosis, KupŠer cells are fewer in number and less functional.7 Thus, the eŠectiveness *Corresponding author, Phone: ＋81-3-3342-6111, Fax: ＋813-3348-6314 123

124

cellular carcinomas,13 and diŠerent results for optimal timing were reported.11,14 We undertook this study to evaluate the optimal timing for breathhold MR imaging for detecting HCC from the point of view of quantitative and qualitative analysis.

Material and Methods Subjects. We included 20 patients (17 men, 3 women; aged 48–83 years, mean, 70 years) in this study, all of whom were positive for the antibody of hepatitis C virus. We evaluated liver function by technetium–99m-diethylenetriaminepentaacetic acid-galactosyl human serum albumin (99mTcGSA) scintigraphy using the grading of liver damage (A, B, C) described by the Japan Liver Cancer Study Group.15 The degree of liver damage was A in 6 patients, B in six, and C in four. The 4 others were normal. We excluded patients with 16 nodules or more because we considered it di‹cult to identify individual lesions, and we excluded diŠuse lesions according to the image classiˆcation of morphology.15 This examination was in accordance with the Declaration of Helsinki principles. Diagnosis of hepatocellular carcinoma. We deˆnitively diagnosed HCC based on a combination of computed tomography (CT) during arterial portography (CTAP) and CT hepatic arteriography (CTHA). We regarded lesions visualized as hypervascular on CTHA and as perfusion defects on CTAP as hypervascular HCC and lesions visualized as hypo- or isovascular on CTHA and as perfusion defects on CTAP as non-hypervascular HCC.16,17 All patients were followed for at least 6 months. We were prepared to exclude pseudolesions as described in previous reports.18,19 Cases of suspected pseudolesions would have been regarded as HCC if they had enlarged, but we observed no pseudolesions. Thus, we found 52 nodules to be hypervascular and ten to be non-hypervascular HCCs. Cysts were excluded. We treated 9 nodules (post-transcatheter arterial chemoembolization 3; post radiofrequency ablation therapy 6) and conˆrmed the absence of local relapse. No nodule showed local relapse. Two lesions visualized as hypovascular nodules on CTHA and showing blood ‰ow similar to that in the surrounding liver parenchyma on CTAP were regarded as hypovascular hepatocellular nodules.16,17 CTAP and CTHA were performed within 60 days after Resovist-enhanced MR imaging (mean, 17 days). The maximum dimension was measured on CTHA. The mean size of hypervascular HCCs was 17.0±18.2 mm and of non-hypervascular HCCs, 15.3±9.2

K. Saito et al.

mm (mean±standard deviation). The mean size of hypovascular hepatocellular nodules was 8.0±4.4 mm. A coordinator for this study marked deˆnitively diagnosed lesions on a liver map for transverse sections using the Couinaud classiˆcation. Imaging procedures for CTAP and CTHA. For CTAP, 150 mgI/mL of iopamidol (Iopamiron, Schering, Osaka, Japan) was infused through the superior mesenteric artery at 2 mL/s (total volume 90 /mL), and helical CT (Suˆda, Shimadzu, Kyoto, Japan) was performed 30 s later. For scanning, beam width was 5 mm, table speed was 7 mm, and images were reconstructed (7 mm). For CTHA, 300 mgI/mL of ioxilan (Imagenil, Kyowa Hakko, Tokyo, Japan) was infused through the common hepatic artery at 1.5 mL/s (total volume 15 /mL), and imaging was performed 5 s later. If there was branching variation of the hepatic artery, we performed selective catheterization. CTHA was performed ˆrst in half of the liver on the cephalic side, followed by the caudal half. The beam width and reconstruction interval were 5 mm, and table speed was 5 mm/s. Acquisition of MR images. We used an Avanto 1.5T super conduction system (Siemens, Erlangen, Germany). The maximum gradient strength was 45 mT/m. For MR imaging, in-phase and out-ofphase T1-weighted images were taken using the gradient echo (GRE) method prior to contrast enhancement. T2-weighted images were taken using the fast spin echo (FSE) method under breath-holding and the half-Fourier acquisition single-shot turbo spin-echo (HASTE) method (T1-weighted images: repetition time [TR] 133 ms; echo time [TE] ; one acquisi4.8 and 2.4 ms; ‰ip angle [FA] 759 tion; acquisition matrix 115×256 [interpolated matrix 230×512] and 23-s breath-hold. T2-weighted images: TR 3780 ms; TE 84 ms; FA 1809; one acquisition; acquisition matrix 154×256 (interpolated matrix 308×512); echo train length 31; and 15-s breath-hold. T2-weighted HASTE images: TR 1200 ms; TE 197 ms; FA 1809; one acquisition; acquisition matrix 165×256 [interpolated matrix 330 ×512]; and 23-s breath-hold). Subsequently, we administered Resovist} rapidly, at 2 mL/s, by pushing with 20 mL of physiological saline. For perfusion study, we employed the echo-planar (EPI) method (TR 1110 ms; TE 20 ms; FA 909; EPI factor 128; acquisition matrix 128×128 [interpolated matrix 256×256]; bandwidth 2056 Hz per pixel). The ˆeld of view (FOV) ranged from 280 to 350 mm. In the sequence using parallel imaging, we used the generalized autocalibrating partially parallel acquisition (GRAPPA)20 algorithm with an acceleration factor (iPAT factor) of two. We obMagnetic Resonance in Medical Sciences

Detection of HCCs in Early-phase SPIO MRI

tained images before administering contrast medium and 8 s or more after starting administration; 18 phases were taken during a breath-holding. It took 2 s to obtain one phase with 19 slices. The initial reference scan required about 8 s, after which we began the image-acquisition scanning. Three minutes after administration, T1-weighted images were obtained (TR 130 ms; TE 2.0 ms; FA 909; acquisition matrix 154×256 [interpolated matrix 308 ×512]; one acquisition; 24-s breath-hold). Ten minutes after administration, we obtained T2weighted FSE images and T2*-weighted images (T2*-weighted images: TR 150 ms; TE 9 ms; FA 30 or 609; acquisition matrix 138×256 [interpolated matrix 275×512]; 7 slices; 21-s breath-hold; bandwidth 60 Hz per pixel). Three scans were needed to obtain T2*-weighted images of the entire liver. In all sequences, slice thickness was 7 mm and slice gap, 1.4 mm. A rectangular FOV with 75z inphase encoding direction and total length between 280 to 350 mm was chosen on T1-, T2-, and T2*weighted images. After these sequences were obtained, the patient left the examination room and reenter the room for repetition of both T2*-weighted images about 30 min after the Resovist administration. The actual times to obtain the 10-min delay images (early T2* images) were 12.6±1.7 min (mean±standard deviation) with ‰ip angle of 309 and 15.4±2.0 min for FA of 609 ; for the 30-min delay images (delayed T2* images), the times were 29.3±6.7 min for FA of 309and 32.4±6.6 min for FA of 609. Qualitative analysis. Three abdominal radiologists in a‹liated hospitals evaluated the images independently. They were informed of the patients' type-C viral hepatitis but not of the presence, absence, or number of tumors. A series of images before and after contrast enhancement, excluding delayed T2* images, were prepared. One of 5 conˆdence levels was assigned to each decision: 1, deˆnitely absent; 2, probably absent; 3, possibly present; 4, probably present; 5, deˆnitely present. Post-treatment lesions with no recurrence were assigned to Category 2. Borderline lesions21 were assigned to Category 3. After 2 weeks or more, a series of images before and after contrast enhancement, excluding early T2* images, were prepared, and all images were evaluated. Based on this assessment, the presence of each tumor was evaluated according to the 5 conˆdence levels. Each radiologist recorded the tumor site on a liver map using the Couinaud classiˆcation, as described above, and the degree of conˆdence. As the nodules were recorded, a coordinator conˆrmed the corresponding nodule on the ˆlm. Vol. 7 No. 3, 2008

125

We conducted alternative free response receiver operating characteristic (AFROC) curve analysis on a tumor-by-tumor basis. The conventional receiver operating characteristic (ROC) method allows for only one response per liver segment, but the AFROC curve method allows for multiple responses per segment. We prepared AFROC curves using a maximum likelihood estimation program (ROCKIT 0.9B; Metz CE, University of Chicago, Il 1998) and the degree of conˆdence reported by each observer. We calculated the area under the AFROC curve (Az) and evaluated the diagnostic results for each observer with respect to the inclusion of early or delayed T2* images. DiŠerences in the mean Az value related to the inclusion of early or delayed T2* images were investigated in the same observer using the univariate Z-score test. Pº0.05 was considered statistically signiˆcant. Concerning the sensitivity of each observer with respect to the inclusion of early or delayed T2* images, a conˆdence degree of 3 or more was regarded as positive and so calculated. Sensitivity was compared using the McNemar test with respect to the early or delayed T2* images. Pº0.05 was considered statistically signiˆcant. Interobserver variability of the early or delayed * T2 images was evaluated using the kappa value to measure the degree of agreement. A kappa value of 0.2 or less was regarded as slight agreement, 0.21 to 0.40 as fair, 0.41 to 0.60 as moderate, 0.61 to 0.80 as substantial, and 0.81 or more as almost perfect. Quantitative analysis. We performed qualitative analysis for 64 lesions (52 hypervascular HCCs, 10 non-hypervascular HCCs, and 2 hypovascular hepatocellular nodules). We calculated signal-tonoise (S/N) and contrast-to-noise (C/N) ratios by the formulas: S/N＝LI/SD and C/N＝(TI-LI)/ SD, in which LI represents the intensity of the liver parenchymal image; TI, the intensity of the tumor image; and SD, the standard deviation of the background noise. The intensity of the liver parenchyma was measured in a circular region of interest (ROI) near the liver tumor that avoided vascular structures and artifacts. Tumor intensity was measured in an ROI that avoided areas of necrosis. Standard deviation of background noise was measured in an ROI outside the body to avoid phase direction artifact. The S/N and C/N ratios were calculated on early and delayed T2* images, and diŠerences were evaluated by Student's-t test. Pº0.05 was considered statistically signiˆcant.

126

K. Saito et al.

Results Qualitative analysis. The Az values of individual observers in the sets including early T2* images were 0.82, 0.78 and 0.90, respectively (Table 1, Fig. 1). A signiˆcant diŠerence was recognized between Observers 2 and 3, but not among the other observers. The respective Az values of individual observers in the sets including delayed T2* images were 0.86, 0.89 and 0.91 (Table 1, Fig. 2). There were no signiˆcant diŠerences among observers. The Az values of individual observers were not signiˆcantly diŠerent between the sets including early and delayed T2* images (Table 1). Table 2 shows interobserver variability. Moderate agreement was noted in both sets including early and delayed T2* images. The sensitivity level of Observer 1 was 67 for both early and delayed T2* images; of Observer 2, 64 for early and 66 for delayed T2* images; and of Observer 3, 66 for both early and delayed images. There was no signiˆcant diŠerence in the sensitivity of individual observers between early and delayed

Table 1. The Az values of individual observer in early and delayed T2* images Delayed Early T2* images T2* images Observer 1 Observer 2 Observer 3

0.82 0.78* 0.90*

0.86 0.89 0.91

T2* images (Table 3). Quantitative analysis. The S/N ratios of T2*weighted images with ‰ip angles (FA) of 309were 72.8±32.1 for early T2* images and 57.8±28.7 for delayed images and with FA of 609, 69.3±29.7 for early and 56.3±26.8 for delayed T2* images. Signiˆcant diŠerences were recognized between early (Pº0.01) and delayed T2* images (Pº0.01) of both 309and 609(Fig. 3). The S/N ratio decreased in delayed compared with early T2* images regardless of liver function (Fig. 4a, b). The C/N ratios of the early T2* images were 65.4 ±58.5 with FA of 309and 65.3±50.8 with FA of 609. Those of delayed T2* images were greater, 88.0 ±76.6 for FA of 309(Pº0.01) and 77.6±66.0 for FA of 609(Pº0.01) (Fig. 5).

Table 2. Interobserver agreement on hepatocellular carcinoma detection (kappa value) Delayed Early T2* images T2* images Observer 1 versus Observer 2 Observer 1 versus Observer 3 Observer 2 versus Observer 3

0.43 0.47 0.42

0.55 0.48 0.49

Early versus delayed T2* images

Table 3. Sensitivity of detection by individual observers on early and delayed T2* images.

N.S. N.S. N.S.

Early Delayed T2* images T2* images

* A signiˆcant diŠerence was recognized between Observers 2 and 3 on early T2* images. Az, area under the alternative free response receiver operating characteristic (AFROC) curve; N.S., no signiˆcant diŠerence.

Fig. 1. Receiver operating characteristic (ROC) curves of early T2* images. The only signiˆcant difference was recognized between Observers 2 and 3.

Observer 1 Observer 2 Observer 3

67 64 66

67 66 66

Early versus delayed T2* images N.S. N.S. N.S.

N.S., no signiˆcant diŠerence.

Fig. 2. Receiver operating characteristic (ROC) curves in delayed T2* images. There were no significant interobserver diŠerences. Magnetic Resonance in Medical Sciences


Fig. 3. A 77-year-old woman with hepatocellular carcinoma with a Child-Pugh liver function ; (b) early T2* image, FA 309 . The liver parenscore of C. (a) delayed T2* image, ‰ip angle (FA) 309 chyma signal intensity is lower in the delayed T2* image than in the early image. The peripheral region of the tumor (arrow) shows hypointensity, indicating the accumulation of superparamagnetic iron oxide (SPIO) in this region. This peripheral region is more obscure on the delayed than the early T2* image.

Fig. 4. (a) signal-to-noise (S/N) ratio of in early and delayed T2* images (FA 30)9 ; (b) S/N ratio of early and delayed T2* images (FA 609 ). The S/N ratio decreased in delayed T2* images compared with early images regardless of liver function.

Fig. 5. An 83-year-old man with hepatocellular carcinoma with a Child-Pugh liver function score of B. (a) delayed T2* image, ‰ip angle (FA) 309(signal-tonoise [S/N] ratio: 50.1); (b) early T2* image, FA 309(S/N ratio: 89.0). The lesion (arrow) is clearer on the delayed than the early T2* image.

Vol. 7 No. 3, 2008

127

128

K. Saito et al.

Fig. 6. (a) contrast-to-noise (C/N) ratio of early and delayed T2* images (FA 309); (b) C/N ratio of early and delayed T2* images (FA 609). The C/N ratios of hypervascular hepatocellular carcinomas (HCCs) in early and delayed T2* images also showed statistically signiˆcant diŠerence, but those of non-hypervascular HCCs showed no signiˆcant diŠerences.

The C/N ratios of hypervascular HCCs in early and delayed T2* images were also signiˆcantly diŠerent, 74.3 (early) and 99.0 (delayed) in FA 309 images (Pº0.05) and 70.0 (early) and 85.9 (delayed) in FA 609images (Pº0.05) (Fig. 6a). However, no signiˆcant diŠerences were seen in the C/N ratios of non-hypervascular HCCs in early and delayed T2* images (Fig. 6b, P0.05).

Discussion We were unable to determine optimal delay time in this study. The merit of bolus-injectable SPIO is the ability to obtain both MR images pre- and post enhancement in one examination, which improves examination throughput. Because long delay is not appropriate for this examination, we think 30 min or less is permissible. Several investigators reported decreased SPIO accumulation with a heterogeneous enhancement pattern in severe cirrhosis.5,22 Heterogeneity in‰uences the accurate detection of liver lesions.10 In our study, regardless of liver function, the delayed T2* liver parenchyma signal image was lower than that of the early image. When SPIO-enhanced MR imaging shows slightly decreased parenchymal signal because of severe cirrhosis, we speculate that a su‹ciently decreased signal might be obtained if SPIO-enhanced MR imaging is obtained again after 20 min. The C/N ratio of hypervascular HCC was signiˆcantly greater on the delayed T2* image than the early image, but not that of non-hypervascular HCC. This result re‰ects the partial accumulation of SPIO in non-hypervascular HCC. Previous reports suggested that non-hypervascular HCCs included well diŠerentiated HCCs and that SPIO accumulated in them.23

In the few hypovascular hepatocellular nodules in this study, C/N ratios increased, and values were at the minus level. This re‰ects only slight contrast between liver parenchyma and tumor that might obscure the lesions; thus, long delayed T2* images may miss detecting hypovascular hepatocellular nodules. However, these ˆndings may indicate characteristics of the lesion23 and contribute to the qualitative diagnosis. We used T2* images in this study because Ward and colleagues reported that optimized T2* GRE images were suitable for detecting hepatic lesions.24 Kim's group stated that optimal TE could be chosen between 9 to 18 milliseconds and that in-phase images could reduce ring cancellation artifacts caused by chemical shift eŠects.25 We therefore selected an echo time of 9 milliseconds. Moreover, we used a 309‰ip angle to reduce T1 eŠect25 and higher matrices (275×512) to improve spatial resolution.26 Bolus-injectable SPIO enables image acquisition 10 min after injection. Shamsi and associates11 reported no apparent diŠerence in the quality of images obtained 10 and 40 min after injection of SPIO. Kopp's team14 contended that MR imaging obtained 40 min after injection of SPIO was better than at 10 min after injection. However, these studies included a variety of subjects with metastatic liver tumors as well as HCC. We restricted our study to HCCs. The background liver parenchyma in HCC is diŠerent from that of metastatic liver tumor; with HCC, there is chronic liver damage and cirrhosis, whereas in the case of metastastic tumor, the background is normal liver. Accumulation of SPIO in KupŠer cells depends on the quantity and function of those cells.5,6 Because KupŠer cells are fewer and function less in cirrhosis, SPIO could not decrease the signal of the liver parenchyMagnetic Resonance in Medical Sciences


ma as in normal cases.22 Thus, on SPIO-enhanced MR images, detection of HCC was inferior to that of metastatic tumors.3,9,10 In this study, the liver parenchyma signal was signiˆcantly lower in delayed than in early phase. This fact might hold a potential to improve HCC detection. Although the detection of HCC did not improve in this study, diŠerences in diagnostic accuracy among observers disappeared, so the delayed phase might be useful. In conclusion, early T2* images are su‹cient for detecting HCC in most cases.

Acknowledgement The authors are indebted to Professor J. Patrick Barron of the International Medical Communications Center of Tokyo Medical University for his review of this manuscript.

129

9.

10.

11.

12.

References 1.

2.

3.

4.

5.

6.

7.

8.

Ward J, Guthrie JA, Wilson D, et al. Colorectal hepatic metastases: detection with SPIO-enhanced breath-hold MR imaging-comparison of optimized sequences. Radiology 2003; 228:709–718. Bipat S, van Leeuwen MS, Comans EF, et al. Colorectal liver metastases: CT, MR imaging, and PET for diagnosis–meta-analysis. Radiology 2005; 237:123–131. Kim YK, Ko SW, Hwang SB, Kim CS, Yu HC. Detection and characterization of liver metastases: 16-slice multidetector computed tomography versus superparamagnetic iron oxide-enhanced magnetic resonance imaging. Eur Radiol 2006; 16: 1337–1345. Oikawa T, Ojima H, Yamasaki S, Takayama T, Hirohashi S, Sakamoto M. Multistep and multicentric development of hepatocellular carcinoma: histologic analysis of 980 resected nodules. J Hepatol 2005; 42:225–229. Yamashita Y, Yamamoto H, Hirai A, Yoshimitsu S, Baba Y, Takahashi M. MR imaging enhancement with superparamagnetic iron oxide in chronic liver disease: in‰uence of liver dysfunction and parenchymal pathology. Abdom Imaging 1996; 21: 318–323. Kato N, Ihara S, Tsujimoto T, Miyazawa T. EŠects of resovist on rats with diŠerent severities of liver cirrhosis. Invest Radiol 2002; 37:292–298. Tanimoto A, Yuasa Y, Shinmoto H, et al. Superparamagnetic iron oxide-mediated hepatic signal intensity change in patients with and without cirrhosis: pulse sequence eŠects and KupŠer cell function. Radiology 2002; 222:661–666. Choi D, Kim S, Lim J, et al. Preoperative detection of hepatocellular carcinoma: ferumoxides-enhanced MR imaging versus combined helical CT

Vol. 7 No. 3, 2008

13.

14.

15.

16.

17.

18.

19.

20.

during arterial portography and CT hepatic arteriography. AJR Am J Roentgenol 2001; 176:475– 482. Kim SK, Kim SH, Lee WJ, et al. Preoperative detection of hepatocellular carcinoma: ferumoxidesenhanced versus mangafodipir trisodium-enhanced MR imaging. AJR Am J Roentgenol 2002; 179: 741–750. Mori K, Scheidler J, Helmberger T, et al. Detection of malignant hepatic lesion before orthotopic liver transplantation: accuracy of ferumoxides-enhanced MR imaging. AJR Am J Roentgenol 2002; 179:1045–1051. Shamsi K, Balzer T, Saini S, et al. Superparamagnetic iron oxide particles (SH U 555 A): evaluation of e‹cacy in three doses for hepatic MR imaging. Radiology 1998; 206:365–371. Reimer P, Balzer T. Ferucarbotran (Resovist): a new clinically approved RES-speciˆc contrast agent for contrast-enhanced MRI of the liver: properties, clinical development, and applications. Eur Radiol 2003; 13:1266–1276. Reimer P, J äahnke N, Fiebich M, et al. Hepatic lesion detection and characterization: value of nonenhanced MR imaging, superparamagnetic iron oxide-enhanced MR imaging, and spiral CT-ROC analysis. Radiology 2000; 217:152–158. Kopp A, Laniado M, Dammann F, et al. MR imaging of the liver with Resovist: safety, e‹cacy, and pharmacodynamic properties. Radiology 1997; 204: 749–756. Liver Cancer Study Group of Japan. The general rule for the clinical and pathological study of primary liver cancer. 4th ed. Tokyo Kanehara and Co, 2000. Hayashi M, Matsui O, Ueda K, et al. Correlation between the blood supply and grade of malignancy of hepatocellular nodules associated with liver cirrhosis: evaluation by CT during intraarterial injection of contrast medium. AJR Am J Roentgenol 1999; 172:969–976. Tajima T, Honda H, Taguchi K, et al. Sequential hemodynamic change in hepatocellular carcinoma and dysplastic nodules: CT angiography and pathologic correlation. AJR Am J Roentgenol 2002; 178:885–897. Matsui O, Takahashi S, Kadoya M, et al. Pseudolesion in segment IV of the liver at CT during arterial portography: correlation with aberrant gastric venous drainage. Radiology 1994; 193:31–35. Yoon KH, Matsui O, Kadoya M, Yoshigawa J, Gabata T, Arai K. Pseudolesion in segments II and III of the liver on CT during arterial portography caused by aberrant right gastric venous drainage. J Comput Assist Tomogr 1999; 23:306–309. Griswold MA, Jakob PM, Heidemann RM, et al. Generalized autocalibrating partially parallel acquisitions (GRAPPA). Magn Reson Med 2002; 47:1202–1210.

130

21.

22.

23.

K. Saito et al.

Ashahina Y, Izumi N, Uchimura M, et al. Assessment of KupŠer cells by ferumoxides-enhanced MR imaging is beneˆcial for diagnosis of hepatocellular carcinoma: comparison of pathological diagnosis and perfusion patterns assessed by CT hepatic arteriography and CT arterioportography. Hepatol Res 2003; 27:196–204. Elizondo G, Weissleder R, Stark DD, et al. Hepatic cirrhosis and hepatitis: MR imaging enhanced with superparamagnetic iron oxide. Radiology 1990; 174:797–801. Kato H, Kanematsu M, Kondo H, et al. Ferumoxide-enhanced MR imaging of hepatocellular carcinoma: correlation with histologic tumor grade and tumor vascularity. J Magn Reson Imaging 2004;

24.

25.

26.

19:76–81. Ward J, Robinson PJ, Guthrie JA, et al. Liver metastases in candidates for hepatic resection: comparison of helical CT and gadolinium- and SPIOenhanced MR imaging. Radiology 2005; 237:170– 180. Kim MJ, Kim JH, Choi JY, et al. Optimal TE for SPIO-enhanced gradient-recalled echo MRI for the detection of focal hepatic lesions. AJR Am J Roentgenol 2006; 187:W255–266. Kanematsu M, Hoshi H, Murakami T, et al. Detection of hepatocellular carcinoma: comparison of low- and high-spatial-resolution dynamic MR images. AJR Am J Roentgenol 1999; 173:1207– 1212.

Magnetic Resonance in Medical Sciences