European Journal of Radiology

Transcription

1 European Journal of Radiology 76 (2010) Contents lists available at ScienceDirect European Journal of Radiology journal homepage: Characterization of small ( 10 mm) focal liver lesions: Value of respiratory-triggered echo-planar diffusion-weighted MR imaging Konstantin Holzapfel a,, Melanie Bruegel a, Matthias Eiber a, Carl Ganter a, Tibor Schuster b, Petra Heinrich b, Ernst J. Rummeny a, Jochen Gaa a a Department of Radiology, Klinikum rechts der Isar, Technische Universitaet Muenchen (TUM), Ismaninger Str. 22, Munich, Germany b Institute of Medical Statistics and Epidemiology, Klinikum rechts der Isar, TUM, Munich, Ismaninger Str. 22, Munich, Germany article info abstract Article history: Received 25 January 2009 Received in revised form 22 March 2009 Accepted 7 May 2009 Keywords: Diffusion-weighted MR imaging Echo-planar imaging Apparent diffusion coefficient Focal liver lesions Liver neoplasms Purpose: To evaluate respiratory-triggered diffusion-weighted MR imaging (DWI) in the characterization of small ( 10 mm) focal liver lesions (FLL). Materials and methods: A total of 185 FLL (76 metastases, 11 HCCs, 71 cysts, 18 hemangiomas, 6 focal nodular hyperplasias, 3 adenomas) were retrospectively analyzed in 77 patients. DWI was performed at 1.5 T using a respiratory-triggered single-shot echo-planar imaging (SSEPI) sequence (b values: 50, 300, 600 s/mm 2 ). Diffusion-weighted images were evaluated by two independent observers and apparent diffusion coefficient (ADC) values were determined. Reference standard of diagnosis was obtained by two other radiologists correlating DWI with histopathologic findings, standard MR sequences and imaging follow-up. Receiver operating characteristic curve analysis was performed to evaluate the utility of ADC values for the diagnosis of malignancy. Results: Accuracy for characterizing FLL was 93.0% for reader 1 and 91.9% for reader 2. Interobserver agreement was excellent between both readers ( = 0.88). Sensitivity and specificity for diagnosing malignancy were 90.8% and 89.9% using a threshold ADC of mm 2 /s. Conclusion: DWI using the respiratory-triggered SSEPI sequence can help to characterize FLL, even when the diameter of lesions is 10 mm or less Elsevier Ireland Ltd. All rights reserved. 1. Introduction Despite recent advances in liver imaging, correct characterization of small focal liver lesions (FLL) remains a challenging task. Particularly in cancer patients the classification of a small FLL as benign or malignant and thus, precise tumor staging is crucial for optimal treatment planning. Whereas local resection of liver metastases in patients with colorectal cancer, if possible, improves 5-year survival and patients with up to three foci of hepatocellular carcinoma (HCC) smaller than 3 cm in diameter are eligible for liver transplantation, patients with irresectable or more extensive disease should instead undergo transarterial chemoembolization, radiofrequency ablation or systemic chemotherapy [1 4]. Several studies have investigated prevalence and importance of FLL deemed too small to be sufficiently characterized by CT [5 8]. These studies reveal that in % of cancer patients small hepatic lesions are found. Although the majority of small ( 10 mm) FLL discov- Corresponding author. Tel.: ; fax: address: holzapfel@roe.med.tum.de (K. Holzapfel). ered incidentally or in patients with established malignancy are benign, % of those lesions finally turn out to be malignant [6 8]. These numbers stress the importance of a correct characterization of even small FLL, particularly, as obtaining tissue diagnosis of small lesions by biopsy is not only invasive but often technically challenging. In addition to conventional MR sequences, diffusion-weighted MR imaging (DWI) has been shown to be a valuable tool in both the detection and characterization of FLL [9 12]. Technical advances such as parallel acquisition techniques and respiratory-triggering have helped to improve image quality, signal-to-noise ratio and spatial resolution of diffusion-weighted images [12 15]. Particularly in the detection of small FLL DWI was superior to T2-weighted sequences in recent studies [10,16,17]. Thus, with DWI of the liver being increasingly used and becoming more and more accepted the number of small FLL seen at MR imaging will further increase. However, to the best of our knowledge, there are no studies published evaluating DWI in the characterization of small FLL. Hence, the purpose of our study was to retrospectively analyze the value of DWI in the characterization of FLL with a diameter of 10 mm or less using a respiratory-triggered single-shot echo-planar imaging (SSEPI) sequence X/$ see front matter 2009 Elsevier Ireland Ltd. All rights reserved. doi: /j.ejrad

2 90 K. Holzapfel et al. / European Journal of Radiology 76 (2010) Materials and methods 2.1. Patients This study was approved by our institutional review board. As images were analyzed retrospectively, the requirement for informed consent was waived. A query of our image database was performed to identify patients who had undergone DWI of the liver between August 2006 and April 2008 (n = 826) and who had at least one FLL with a diameter of 10 mm or less (n = 307). Patients with hepatic neoplasms who had undergone chemotherapy or radiation therapy within the last 3 months prior to the MR examination were excluded from our analysis in order to ensure that ADC measurements were reflective of the natural state of the liver lesions (n = 134). In addition, patients without sufficient confirmation of the nature of the lesions were excluded (n = 96). Therefore, the final study population consisted of 77 patients (40 men, 37 women; mean age 60.6 years, age range years). There were 58 patients with history of an extrahepatic primary malignancy and known or suspected liver metastases (colorectal cancer [n = 23], breast cancer [n = 10], neuroendocrine carcinoma [n = 8], esophageal and gastric cancer [n = 8], malignant melanoma [n = 3], pancreatic cancer [n = 3], cholangiocarcinoma [n = 1], malignant peripheral nerve sheath tumor [n = 1] and malignant fibrous histiocytoma [n = 1]). There were six patients with chronic liver disease (including chronic hepatitis and cirrhosis related to hepatitis B/C or alcohol abuse). Finally, there were 13 patients with no history of malignancy or chronic liver disease who underwent MR imaging for evaluation of presumably benign or indeterminate, incidentally diagnosed FLL. In patients with multiple FLL, a maximum of three lesions of each lesion type were randomly selected for further analysis by the study coordinator. Thus, our study population encompassed a total of 185 FLL with a maximum diameter of 10 mm or less. Eighty-seven lesions were malignant tumors (11 HCCs, 76 metastases). The primary sites of metastases were colorectal cancer (n = 21), neuroendocrine carcinoma (n = 18), breast cancer (n = 15), pancreatic cancer (n = 7), malignant melanoma (n = 5), esophageal and gastric cancer (n = 5), malignant peripheral nerve sheath tumor (n = 3) and cholangiocarcinoma (n = 2). In total, there were 98 benign liver lesions (71 cysts, 18 hemangiomas, 6 focal nodular hyperplasias [FNHs], 3 adenomas). The standard of reference for characterizing FLL was represented by the consensus reading of two radiologists. Malignant nature of lesions was confirmed by pathologic findings following biopsy or surgery for all HCCs (n = 11) and 12 metastases. The diagnosis of the remaining metastatic lesions was established on the basis of new occurrence of a lesion compared to a prior MR study (n = 17), progression or regression of lesion size on serial cross-sectional imaging studies after the commencement of chemotherapy in patients with known extrahepatic primary malignancies (n = 47). Four FNHs and two adenomas were confirmed pathologically. The remaining benign lesions were diagnosed by using established imaging criteria [18 20] in conjunction with stable appearance and size at follow-up imaging in equivocal cases with a minimum follow-up interval of six months (mean follow-up period: 11 months, range: 6 26 months) MR imaging MR imaging was performed on a 1.5-T system (Magnetom Avanto, Siemens Medical Solutions, Erlangen, Germany) with two Fig. 1. MR images of a 81-year-old woman with gastric cancer. In the left liver lobe there is a lesion that is hyperintense on transverse SSEPI diffusion-weighted images at b = 50 s/mm 2 (a), shows substantial signal loss at b = 600 s/mm 2 (b) and is clearly hyperintense on the ADC map (c). On transverse contrast-enhanced images (3D gradient echo volumetric interpolated breath-hold examination, VIBE) acquired in the portal venous phase (d) the lesion shows no uptake of contrast material, a finding compatible with a liver cyst. The ADC value of this lesion was mm 2 /s.

3 K. Holzapfel et al. / European Journal of Radiology 76 (2010) Fig. 2. MR images of a 50-year-old woman with breast cancer. In the dorsal part of the right liver lobe there is a lesion that is hyperintense on transverse SSEPI diffusionweighted images at b = 50 s/mm 2 (a), shows some signal loss at b = 600 s/mm 2 (b) and is moderately hyperintense on the ADC map (c). On transverse dynamic contrast-enhanced images (d) the lesion is hypointense in the arterial phase (upper left), shows peripheral nodular enhancement in the portal venous phase (upper right) and is almost isointense to liver parenchyma in the equilibrium phase (lower left). In addition the lesion was homogeneously hyperechoic in ultrasound (not shown) and showed no growth over a time period of nine months. In summary, the findings were compatible with an hemangioma. The ADC value of this lesion was mm 2 /s. six-channel body phased array coils anterior and two spine clusters (three channels each) posterior. Diffusion-weighted images were acquired using a SSEPI sequence. The gradient factors (b values) were 50, 300 and 600 s/mm 2. The technical parameters were as follows: echo time, 69 ms; echo train length, 58; echo spacing, 0.69 ms; receiver bandwidth, 1736 Hz/pixel; spectral fat saturation; field of view, 263 mm 350 mm; matrix, ; number of signal averages, 3; section thickness, 5 mm; intersection gap, 0.5 mm; sections acquired; 4 6min acquisition time. For shortening of the echo train length, integrated parallel imaging techniques (ipat) by means of generalized autocalibrating partially parallel acquisitions (GRAPPA) with a twofold acceleration factor were used. For respiratory triggering, PACE (prospective acquisition correction) was implemented. Data was acquired during the end-expiratory phase Image analysis Review of all MR images and of all follow-up imaging (MRI, CT, PET-CT) was performed on a commercial PACS workstation (Easy vision, Philips, Best, The Netherlands). FLL to be analyzed were marked by the study coordinator. Subsequently, these lesions were evaluated by two independent readers. Both readers were blinded to MR imaging reports, findings of other MR sequences, clinical history and pathologic results. Readers were asked to qualitatively classify lesions as benign or malignant. Lesion characterization was performed by using an ordinal five-point scale (1: certainly benign, 2: probably benign, 3: indeterminate, 4: probably malignant, 5: certainly malignant). A lesion was considered benign, when it was hyperintense at a low b value (b = 50 s/mm 2 ), showed a substantial (i.e. 50%) decrease in signal intensity at a high b value (b = 600 s/mm 2 ) and had an ADC value higher than surrounding liver parenchyma (Figs. 1 and 2). If a lesion was considered benign readers were asked to further classify the lesion as cyst, hemangioma or solid benign lesion (FNH or adenoma) depending on the decrease in signal intensity between low (b = 50 s/mm 2 ) and high (b = 600 s/mm 2 ) b value images and the visually assessed ADC values with cysts showing the most pronounced signal loss and the highest ADC values followed by hemangiomas and solid benign lesions. Accordingly, a lesion was classified as malignant, when it was (moderately) hyperintense at b = 50 s/mm 2, remained hyperintense at b = 600 s/mm 2 and showed an ADC value that was equal to or lower than that of adjacent liver parenchyma (Fig. 3). If none of the criteria was met, a lesion was classified as indeterminate. In addition, one of both readers graded the quality of diffusionweighted images by using a five-point scale (1: not evaluable, 2: poor, 3 moderate, 4: good, 5: excellent). ADC values of all FLL analyzed in this study were measured by one of both blinded readers by drawing a circular ROI encompassing as much of the lesion as possible on the lesion on the b = 50 s/mm 2 image (where benign and malignant FLL are visible) and transfer-

4 92 K. Holzapfel et al. / European Journal of Radiology 76 (2010) Fig. 3. MR images of a 64-year-old man with esophageal cancer. In the right liver lobe there is a lesion that is hyperintense on transverse SSEPI diffusion-weighted images at b = 50 s/mm 2 (a), remains hyperintense at b = 600 s/mm 2 (b) and is hypointense on the ADC map (c, arrow). On transverse multi-slice CT images acquired in the portal venous phase (d) the lesion is hypodense and shows substantial growth over a time period of six months, a finding compatible with a metastasis. The ADC value of this lesion was mm 2 /s. ring it to the ADC map. ADC values were measured twice and the two measurements were averaged Statistical analysis Statistical analysis was performed using SPSS (version 16.0, SPSS Inc., Chicago, IL. USA) and Graph-Pad Prism 4 software package for Windows. The ADC values of benign and malignant FLL as well as the ADC lesions of lesion groups were compared using the Mann Whitney-U test. Bonferroni correction was used for multiple pairwise comparisons. A P value less than 0.05 was considered to indicate a statistically significant difference. We used receiver operating characteristic (ROC) curve analysis to evaluate the diagnostic capability of ADC values in the differentiation of benign from malignant FLL and to determine the threshold of ADC values providing the highest accuracy for discriminating benign from malignant liver lesions. The area under the ROC curve was used to evaluate the overall diagnostic performance of DWI. Interobserver agreement for nominal data was determined using Cohen s kappa coefficient () ( , slight agreement: , fair agreement: , moderate agreement: , substantial agreement: , excellent agreement). For ordinal data Kendallǐs tau-b was computed. 3. Results Quality of diffusion-weighted images graded by one reader was categorized as excellent in 61 cases, good in 12 cases, moderate in two cases and poor in two cases. Image quality was assessed as not evaluable in none of the 77 patients. The mean diameter of all 185 FLL was 6.4 ± 1.8 mm. Overall, of the 185 FLL analyzed, 172 lesions (93.0%) were correctly classified as benign or malignant by reader 1, whereas reader 2 correctly classified 170 lesions (91.9%) as benign or malignant. Sensitivity and specificity for diagnosing malignancy were 96.6% and 89.8% for reader 1 and 96.6% and 87.8% for reader 2, respectively. Three cysts in two patients were misclassified as metastases and two metastases in two patients were misclassified as benign lesions (both as hemangiomas) by reader 1. The second reader incorrectly classified three hemangiomas as metastases in one patient, one cyst as a metastasis and one metastasis as a benign lesion (hemangioma). Of the lesions mentioned above, two FLL in two different patients (one metastasis, one cyst) were misclassified by both readers. In addition, both readers incorrectly classified three FNHs and two adenomas. Three lesions in three different patients (one metastasis, one hemangioma, one cyst) were classified as indeterminate by reader 1. Reader 2 classified five lesions in five patients (three hemangiomas, two metastases) as indeterminate. No lesion was deemed indeterminate by both readers. Among the benign lesions reader 1 misclassified seven cysts as hemangiomas and one hemangioma as a cyst. Reader 2 misclassified five cysts as hemangiomas and one hemangioma as a cyst. Overall, interobserver agreement was excellent between both readers ( = 0.88; 95% confidence interval: ). Agreement between both readers remained high when the score on a fivepoint-confidence scale was analyzed (Kendall s tau-b = 0.85) ADC values Mean ADC values determined by a blinded radiologist were (1.09 ± 0.30) 10 3 mm 2 /s for malignant and (2.36 ± 0.62)

5 K. Holzapfel et al. / European Journal of Radiology 76 (2010) Fig. 4. Box and whisker plots showing the ADC values of benign and malignant FLL. The horizontal line in each box is a median (50th percentile) of the measured values, the top and bottom of the boxes represent 25th and 75th percentiles, respectively. Whiskers indicate the range between the smallest and the largest data point observed. Despite some overlap, differences of ADC values between benign and malignant FLL were statistically significant. Fig. 6. Receiver operating characteristic (ROC) curve used for differentiating between benign and malignant FLL based on ADC values. The area under the curve (AUC) is Using a threshold ADC value of mm 2 /s allows differentiating benign from malignant FLL with a sensitivity of 90.8% and a specificity of 89.9%. Fig. 5. Box and whisker plots showing the ADC values of HCCs, metastases, benign solid FLL, hemangiomas and cysts. Mean ADC values ( 10 3 mm 2 /s) of HCCs (1.12 ± 0.28) and metastases (1.08 ± 0.32) were not significantly different (P = 0.82). However, both were significantly lower than ADC values of benign solid FLL (1.43 ± 0.22), hemangiomas (1.69 ± 0.34) and cysts (2.61 ± 0.57) mm 2 /s for benign liver lesions. The ADC values of malignant lesions were significantly lower than those of benign lesions (P < 0.001) (Fig. 4). The mean distribution of lesion ADCs was as follows ( 10 3 mm 2 /s): HCCs, 1.12 ± 0.28; metastases, 1.08 ± 0.32; FNHs and adenomas 1.43 ± 0.22; hemangiomas, 1.69 ± 0.34; cysts, 2.61 ± 0.57 (Fig. 5). There was no significant difference between ADC values of HCCs and metastases (P = 0.82). ADC values of HCCs were significantly lower than those of FNHs and adenomas (P < 0.05), hemangiomas (P < 0.001) and cysts (P < ). Despite considerable overlap, ADC values of metastases were significantly lower than ADC values of FNHs and adenomas (P < 0.05), hemangiomas (P < 0.001) and cysts (P < 0.001). Using a threshold ADC value of mm 2 /s the area under the receiver operating characteristic curve was 0.965, with a sensitivity of 90.8%, a specificity of 89.9%, a positive predictive value of 90.0%, and a negative predictive value of 90.7% (Fig. 6). 4. Discussion In the present study DWI using a respiratory-triggered SSEPI sequence allowed to characterize small (diameter 10 mm) FLL as benign or malignant with high accuracy and excellent interobserver agreement. In several studies respiratory-triggered DWI was found to show overall better image quality compared to breath-hold DWI and was superior to breath-hold DWI in the detection of FLL [10,21,22]. In accordance with this observation image quality in this study using a respiratory-triggered SSEPI sequence was good or excellent in 94.8% of cases and was assessed as not evaluable in none of the patients. However, it has to be mentioned that respiratory-triggered DWI can result in pseudo-anisotropy artefacts leading to spuriously calculated ADC values [23]. DWI has been shown to be superior to T2-weighted MR sequences [10,16,17] and to SPIO-enhanced MR imaging [12,24] in the detection of small FLL. However, with an increasing number of subcentimeter FLL being detected by DWI, correct characterization of small FLL becomes increasingly important and has to be re-evaluated. In a study conducted by Coenegrachts et al. SPIO-enhanced imaging combined with DWI allowed improved characterization of FLL [24]. Thus, if a lesion can be seen on post contrast images, these images may supply additional information about the nature of a lesion. Unfortunately, those lesions often are >10 mm in diameter. In the present study small FLL could be classified as benign or malignant with a high accuracy and excellent interobserver agreement. Applying a threshold ADC value of mm 2 /s allowed differentiating benign from malignant FLL with a sensitivity of 90.8% and a specificity of 89.9%. Sensitivity and specificity for characterizing subcentimeter lesions using traditional T1-, T2-weighted and dynamic contrast-enhanced T1-weighted MR sequences have been reported to be % and %, respectively [8,25]. The high sensitivity (83.3%) and specificity (97.5%) described in the study conducted by Holalkere et al. may in part be caused by the low percentage of malignant FLL (27.5%) that were analyzed [8]. Only 71.4% of malignant lesions were correctly diagnosed in that study, whereas 94.4% of benign lesions were correctly classified. Thus, the performance of gadolinium-enhanced MR imaging may have been substantially worse in the present study population with 49.7% of FLL being malignant. In contrast, using DWI 96.6% of the malignant lesions analyzed in the present study were correctly classified by both readers. However, although, given these numbers, DWI seems to perform equal or even better than gadolinium-enhanced MR imaging, a potential benefit of DWI when used instead of or in combination with dynamic contrast-enhanced MR imaging remains to be investigated. Although metastases showed significantly lower ADC values than cysts and hemangiomas in the present study, there was considerable overlap, in particular between metastases and solid benign FLL (FNHs, adenomas) and between metastases and hemangiomas. Hence, it may be difficult to prospectively differentiate metastasis from solid benign FLL or hemangiomas in an individual case.

6 94 K. Holzapfel et al. / European Journal of Radiology 76 (2010) The fact that despite a statistically significant difference between ADC values of metastases and solid benign FLL five out of nine solid benign FLL were incorrectly classified as malignant by visual scoring of readers 1 and 2 underlines this statement and emphasizes the importance of combining quantitative ADC measurement and visual assessment of small FLL when using DWI. Whereas DWI was superior to breath-hold T2-weighted MR sequences in the detection of small FLL in a study conducted by Parikh et al., it was equal to T2-weighted imaging for FLL characterization [10]. The diagnostic accuracy of DWI in Parikh s study was 89.1% which is faintly lower than that in our study, despite only lesions with a diameter 10 mm were analyzed by Parikh et al. However, as different DWI sequences including breath-hold and respiratory-triggered sequences were used in that study and as breath-hold DWI has been shown to be significantly inferior to respiratory-triggered DWI in some studies, the slightly higher accuracy in our study may, in part, be explained by the differences of the sequences used in both studies. Furthermore, Parikh et al. did not explicitly exclude patients receiving chemotherapy at the time or prior to MR imaging from their study. Anticancer treatment results in tumor lysis, loss of cell membrane integrity, increased extracellular space, and, therefore, an increase in water diffusion [26,27]. Consequently, an increase of ADC values can be observed in patients receiving chemotherapy who at least partially respond to treatment [28]. Thus, the differentiation of benign from malignant FLL on the basis of ADC values may become more difficult as overlap of ADC values will be more pronounced and hence, the diagnostic accuracy will decline. The fact that DWI was not superior to T2-weighted MR imaging in characterizing FLL in Parikh s study is not contradictive to the results of the present study as Parikh et al. only analyzed FLL with a diameter of more than 10 mm [10]. However, as DWI has proven to be significantly superior to T2-weighted imaging in the detection of subcentimeter FLL, the comparison of DWI and T2-weighted sequences in the characterization of lesions 10 mm is not reasonable, as about 50% of lesions seen by DWI will not be detected using T2-weighted sequences [16,17]. In a study conducted by Coenegrachts and co-workers, DWI was of no use for characterizing FLL including subcentimeter lesions, as all hepatic lesions appeared hyperintense on all b value images [24]. However, the b values used in Coenegracht s study were between 0 and 400 s/mm 2. Thus, the maximum diffusion gradient of 400 s/mm 2 seems to be too low to observe a substantial decline of signal intensity within cysts and hemangiomas compared to low b value images that could be seen in the present study where the maximum b value was 600 s/mm 2. Our study has few potential limitations. Firstly, this is a retrospective analysis of patients with at least one FLL 10 mm undergoing MR imaging including DWI and follow-up imaging. Thus, there is the possibility of selection bias and, consequently, overestimation of the accuracy of DWI in the characterization of small FLL. In patients with multiple subcentimeter FLL, lesions to be analyzed were randomly selected by the study coordinator, an approach entailing another potential source of a selection bias. Secondly, like in most studies dealing with small FLL, histopathologic confirmation was only available in a relatively small number of cases. Obtaining histopathologic diagnosis in all patients would be desirable, but this is not clinically appropriate in most cases. However, a thorough review of all MR sequences and of cross-sectional follow-up imaging was performed by two readers, leading to a low probability of misclassification of lesions. Thirdly, 79.5% (147/185) of FLL analyzed in the present study were metastases or cysts, i.e. FLL with very low or very high ADC values. With a higher number of FLL with intermediate ADC values (e.g. hemangiomas, solid benign FLL) the accuracy of DWI in characterizing small FLL may have been worse. In particular, very high ADC values seen in cysts may falsely suggest the method s ability to differentiate between benign and malignant lesions. However, as metastases and cysts are the most common FLL encountered in oncologic patients we did not exclude patients with cysts from our study. Finally, in our study FLL with a mean diameter of 6.4 mm were analyzed using a section thickness of 5 mm. Therefore, partial volume effects probably have influenced the results to a certain extent. This may be an explanation why the mean ADC values of metastases, hemangiomas and cysts seen in the present study were lower than those observed in another study using a similar DWI protocol [11]. 5. Conclusion Respiratory-triggered DWI using a SSEPI sequence has the potential not only to improve the detection of FLL 10 mm, but also allows to differentiate benign from malignant lesions in most cases. However, as there is overlap of ADC values between different types of lesions, DWI should rather be considered as a supplementary method than a stand-alone sequence for lesion characterization. In particular, DWI is of limited value in diagnosing solid benign FLL. In addition, a prospective study is required for validation of the method in characterizing subcentimeter lesions. References [1] Jamison RL, Donohue JH, Nagorney DM, Rosen CB, Harmsen WS, Ilstrup DM. Hepatic resection for metastatic colorectal cancer results in cure for some patients. Arch Surg 1997;132: [2] Heslin MJ, Medina-Franco H, Parker M, Vickers SM, Aldrete J, Urist MM. Colorectal hepatic metastases: resection, local ablation, and hepatic artery infusion pump are associated with prolonged survival. Arch Surg 2001;136: [3] Mazzaferro V, Regalia E, Doci R, et al. Liver transplantation for the treatment of small hepatocellular carcinomas in patients with cirrhosis. N Engl J Med 1996;334: [4] Llovet JM, Bruix J. Systematic review of randomized trials for unresectable hepatocellular carcinoma: Chemoembolization improves survival. Hepatology 2003;37: [5] Jones EC, Chezmar JL, Nelson RC, Bernardino ME. The frequency and significance of small (less than or equal to 15 mm) hepatic lesions detected by CT. AJR Am J Roentgenol 1992;158: [6] Schwartz LH, Gandras EJ, Colangelo SM, Ercolani MC, Panicek DM. Prevalence and importance of small hepatic lesions found at CT in patients with cancer. Radiology 1999;210:71 4. [7] Khalil HI, Patterson SA, Panicek DM. Hepatic lesions deemed too small to characterize at CT: prevalence and importance in women with breast cancer. Radiology 2005;235: [8] Holalkere NS, Sahani DV, Blake MA, Halpern EF, Hahn PF, Mueller PR. Characterization of small liver lesions: added role of MR after MDCT. J Comput Assist Tomogr 2006;30: [9] Gourtsoyianni S, Papanikolaou N, Yarmenitis S, Maris T, Karantanas A, Gourtsoyiannis N. Respiratory gated diffusion-weighted imaging of the liver: value of apparent diffusion coefficient measurements in the differentiation between most commonly encountered benign and malignant focal liver lesions. Eur Radiol 2008;18: [10] Parikh T, Drew SJ, Lee VS, et al. Focal liver lesion detection and characterization with diffusion-weighted MR imaging: comparison with standard breath-hold T2-weighted imaging. Radiology 2008;246: [11] Bruegel M, Holzapfel K, Gaa J, et al. Characterization of focal liver lesions by ADC measurements using a respiratory triggered diffusionweighted single-shot echo-planar MR imaging technique. Eur Radiol 2008;18: [12] Nasu K, Kuroki Y, Nawano S, et al. Hepatic metastases: diffusionweighted sensitivity-encoding versus SPIO-enhanced MR imaging. Radiology 2006;239: [13] Yoshikawa T, Kawamitsu H, Mitchell DG, et al. ADC measurement of abdominal organs and lesions using parallel imaging technique. AJR Am J Roentgenol 2006;187: [14] Koh DM, Scurr E, Collins DJ, et al. Colorectal hepatic metastases: quantitative measurements using single-shot echo-planar diffusion-weighted MR imaging. Eur Radiol 2006;16: [15] Taouli B, Martin AJ, Qayyum A, et al. Parallel imaging and diffusion tensor imaging for diffusion-weighted MRI of the liver: preliminary experience in healthy volunteers. AJR Am J Roentgenol 2004;183: [16] Coenegrachts K, Delanote J, Ter Beek L, et al. Improved focal liver lesion detection: comparison of single-shot diffusion-weighted echoplanar and single-shot T2 weighted turbo spin echo techniques. Br J Radiol 2007;80:

7 K. Holzapfel et al. / European Journal of Radiology 76 (2010) [17] Bruegel M, Gaa J, Waldt S, et al. Diagnosis of hepatic metastases: comparison of respiration-triggered diffusion-weighted echo-planar MRI and five T2- weighted turbo-spin echo sequences. AJR Am J Roentgenol 2008;191: [18] Semelka RC, Brown ED, Ascher SM, et al. Hepatic hemangiomas: a multi-institutional study of appearance on T2-weighted and serial gadoliniumenhanced gradient-echo MR images. Radiology 1994;192: [19] Bartolozzi C, Cioni D, Donati F, Lencioni R. Focal liver lesions: MR imagingpathologic correlation. Eur Radiol 2001;11: [20] Horton KM, Bluemke DA, Hruban RH, Soyer P, Fishman EK. CT and MR imaging of benign hepatic and biliary tumors. Radiographics 1999;19: [21] Sandberg A, Parikh T, Johnson G, Stemmer A, Taouli B. Feasibility of a respiratory-triggered SSEPI diffusion-weighted sequence for liver imaging using navigator echo technique: comparison with breath-hold diffusionweighted sequence. Proc Int Soc Mag Reson Med 2006;14 [abstract number 40]. [22] Nasu K, Kuroki Y, Sekiguchi R, Nawano S. The effect of simultaneous use of respiratory triggering in diffusion-weighted imaging of the liver. Magn Reson Med Sci 2006;5: [23] Nasu K, Kuroki Y, Fujii H, Minami M. Hepatic pseudo-anisotropy: a specific artifact in hepatic diffusion-weighted images obtained with respiratory triggering. Magma 2007;20: [24] Coenegrachts K, Orlent H, ter Beek L, et al. Improved focal liver lesion detection: comparison of single-shot spin-echo echo-planar and superparamagnetic iron oxide (SPIO)-enhanced MRI. J Magn Reson Imaging 2008;27: [25] Mueller GC, Hussain HK, Carlos RC, Nghiem HV, Francis IR. Effectiveness of MR imaging in characterizing small hepatic lesions: routine versus expert interpretation. AJR Am J Roentgenol 2003;180: [26] Moffat BA, Chenevert TL, Lawrence TS, et al. Functional diffusion map: a noninvasive MRI biomarker for early stratification of clinical brain tumor response. Proc Natl Acad Sci USA 2005;102: [27] Moffat BA, Hall DE, Stojanovska J, et al. Diffusion imaging for evaluation of tumor therapies in preclinical animal models. Magma 2004;17: [28] Koh DM, Scurr E, Collins D, et al. Predicting response of colorectal hepatic metastasis: value of pretreatment apparent diffusion coefficients. AJR Am J Roentgenol 2007;188: