HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Ward, J. Articles by Robinson, P. J. Search for Related Content PubMed PubMed Citation Articles by Ward, J. Articles by Robinson, P. J.

Hepatic Lesion Detection after Superparamagnetic Iron Oxide Enhancement: Comparison of Five T2-weighted Sequences at 1.0 T by Using Alternative-Free Response Receiver Operating Characteristic Analysis¹

¹ From the Departments of Clinical Radiology (J.W., J.A.G., P.J.R.), Medical Physics (D.W.), Hepatobiliary Surgery (J.P.A.L.), and Pathology (J.I.W.), St James's University Hospital, Beckett St, Leeds LS9 7TF, United Kingdom; and the Department of Radiology, Nanjing Railway Medical College Hospital, People's Republic of China (F.C.). Received December 2, 1998; revision requested December 30; revision received May 5, 1999; accepted July 30. Address reprint requests to J.W. (e-mail: 113566.2505@compuserve.com).

	Abstract

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
PURPOSE: To compare the accuracy of five T2-weighted sequences in the detection of liver lesions at magnetic resonance (MR) imaging after superparamagnetic iron oxide (SPIO) enhancement.

MATERIALS AND METHODS: Forty-nine candidates for hepatic resection with known colorectal metastases were examined. Before SPIO enhancement, fast spin-echo (SE) images were obtained. After enhancement, the same fast SE sequence and long TR/short TE, short TE, long TR/TE, and T2-weighted fast low-angle shot (FLASH) sequences were used. All images were viewed independently by four observers who were blinded to the results of the other imaging sequences, the results of the other observers, and the findings at surgery and histopathologic examination. Four weeks after the initial reading, the combined long TR/short TE and long TR/TE dual-echo images were also viewed as an additional set. The alternative free response receiver operating characteristic (ROC) method was used to analyze the results, which were correlated with findings at surgery, intraoperative ultrasonography, and histopathologic examination.

RESULTS: Irrespective of lesion size, the accuracy of all sequences after enhancement was significantly greater than that of the nonenhanced fast SE sequence (P < .01). Dual-echo and FLASH sequences were significantly more accurate than the enhanced fast SE sequence (P < .03 or P < .02, respectively). For all lesions, lesions smaller than 1 cm, and lesions 1 cm or larger, mean accuracies were as follows: dual-echo, 0.75, 0.54, and 0.93; FLASH, 0.75, 0.54, and 0.95; and enhanced fast SE, 0.72, 0.49, and 0.92.

CONCLUSION: At 1.0 T, dual-echo and FLASH sequences are the most accurate pulse sequences after SPIO enhancement.

Index terms: Iron, 761.12143 • Liver neoplasms, metastases, 761.3327 • Liver neoplasms, MR, 761.121411, 761.121412, 761.12143, 761.3327 • Magnetic resonance (MR), comparative studies, 761.121411, 761.121412, 761.12143 • Magnetic resonance (MR), contrast enhancement, 761.121411, 761.121412, 761.12143 • Receiver operating characteristic (ROC) curve

	Introduction

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
Superparamagnetic iron oxide (SPIO) is a liver-specific magnetic resonance (MR) contrast agent that is taken up by the Kupffer cells. Numerous comparative studies have shown SPIO-enhanced MR imaging to be more accurate than nonenhanced MR imaging or dual-phase helical computed tomography (CT) in the detection of focal liver lesions (1–8). So far, there is no consensus regarding the optimum pulse sequence for use with SPIO enhancement; several studies have shown conflicting results, although the influence of field strength is clear.

Theoretic considerations suggest that maximum liver-to-lesion contrast would occur with gradient-echo (GRE) sequences due to increased susceptibility from local field inhomogeneity. However, because susceptibility also increases with increasing field strength, the combination of GRE sequences and high field strength after SPIO enhancement may produce a pronounced signal loss; small lesions may be obscured by a blooming effect from the diffuse loss of signal in the adjacent parenchyma.

Consequently, several studies have found long repetition time (TR)/short echo time (TE) sequences to be more sensitive than long TR/TE sequences at 1.5 T (3,9), although GRE sequences have been shown to be more sensitive than spin-echo (SE) sequences at 0.5 T (6,10). In one study, images obtained at 1.0 T with a long TR/TE sequence were superior to those obtained with a long TR/short TE sequence (11). Other investigators (12,13) have evaluated the fast SE sequence and have suggested that the shorter acquisition time compared with that of conventional SE sequences would reduce motion artifact and, therefore, increase lesion conspicuity. But again, conflicting results have emerged.

To our knowledge, to date, three studies (11,12,14) were performed with a field strength of 1.0 T, but none included an analysis of all pulse sequences. We aimed to compare the accuracy of all previously reported sequences (conventional SE, fast SE, and GRE) for the detection of liver lesions after SPIO enhancement in a single study, with a larger patient population and a field strength of 1.0 T. Alternative–free response receiver operating characteristic (ROC) analysis with multiple observers was used to increase the validity of our findings.

	MATERIALS AND METHODS

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
Patients
Eighty-seven consecutive patients who had colorectal liver metastases and who were candidates for hepatic resection underwent MR imaging with SPIO enhancement after a review of imaging findings from the referring hospital or from the initial CT examination. A decision not to perform SPIO enhancement was made in five patients who were referred for surgical consideration because they were found to have unresectable disease at nonenhanced MR imaging. Of the 87 patients in the study, 38 were subsequently found to be unsuitable for surgery because of the extent of their disease or because of poor cardiac status; they were not included in the analysis.

The final study group comprised 49 patients (33 men, 16 women; average age, 63 years; age range, 50–77 years) who underwent surgical exploration with intraoperative ultrasonography (US). The decision to proceed with resection was made on the basis of findings of disease that was localized to the liver and a minimum of two hepatic segments that were free of disease. The number and size of lesions was not a limiting factor. In eight patients, resection of small superficial nodules from the residual segments was performed in addition to hemihepatectomy (n = 3) or trisegmentectomy (n = 5).

Local ethical committee approval was granted. Informed consent was obtained from each patient prior to entry into the study.

MR Imaging
MR imaging was performed with a Magnetom 42SP (1.0-T) system (Siemens Medical Systems, Erlangen, Germany) by using the body coil for transmission and reception of the signal. Before the injection of SPIO, fast SE images were obtained (4,000–4,673/91, TR msec/TE msec) with an echo train length of eight, two signals acquired, and a matrix size of 192 x 256.

After SPIO enhancement, the same fast SE sequence was used, followed by a conventional dual-echo sequence (2,000/45–90), with two signals acquired and a matrix size of 192 x 256, and a T2-weighted GRE fast low-angle shot (FLASH) sequence (150/10; 15° flip angle with one signal acquired and a matrix size of 128 x 256). Section thickness was 8–10 mm, with a 10% gap, and the field of view was 35–40 cm, depending on the size of the liver. In individual patients, these factors were the same for all sequences before and after SPIO enhancement.

All images were obtained in a transverse plane by using a rectangular field of view with a superior presaturation band, which was applied for all SE acquisitions. For the GRE sequence, six sections were obtained during a breath hold of 22 seconds, so three to four sequential acquisitions were required to encompass the entire liver.

In 39 patients, AMI-25 (Endorem; Guerbet, Roissy, France) was administered at a dose of 15 µmol of iron per kilogram of body weight, diluted in 100 mL of a 5% glucose solution and infused over 30 minutes. Imaging commenced 30–60 minutes from the end of the infusion. In 10 patients, SH U 555A (Resovist; Schering, Berlin, Germany) at a dose of 7.0–12.9 µmol of iron per kilogram of body weight was injected as a rapid bolus, immediately followed by a saline solution flush. The injection procedure was performed in approximately 5 seconds, and imaging commenced 10 minutes from the end of the injection.

Image Analysis
Six sets of images for each patient were analyzed. Initially, images obtained with nonenhanced or enhanced fast SE; long TR/short TE; long TR/TE; and FLASH sequences were randomized and were viewed independently by four observers who were blinded to the results of the other imaging sequences, the results of the other observers, and the findings at surgery and histopathologic examination. Three (J.W., J.A.G., P.J.R.) had extensive and comparable expertise in liver MR imaging, and one (F.C.) was 6 months into a 12-month fellowship in MR imaging.

To minimize any learning bias, after 4 weeks or more, the combined long TR/short TE and long TR/TE images were viewed as an additional set. Each observer recorded the presence and segmental location of all lesions and characterized them as benign, indeterminate, or malignant. Alternative–free response ROC analysis of all lesions was performed for each sequence, and each observer used a four-point confidence scale: "1" was probably not a lesion, "2" was possible lesion, "3" was probable lesion, and "4" was definite lesion.

To achieve an accurate correlation of findings for the scored lesions and those confirmed at surgery and histopathologic examination, each observer recorded the individual image number, the segmental location, and the size of each lesion. If multiple lesions with the same identification parameters were scored, the observers added further comments to distinguish the lesions. At the time of surgery, all of the lesions identified at surgical inspection and intraoperative US were exactly correlated with the MR images by one of the authors. Histopathologic correlation of the resected specimen after dissection in the transverse plane at 1-cm intervals was also performed. All surgery was performed by the same experienced hepatobiliary surgeon (J.P.A.L.) who was aware of the findings at preoperative imaging.

Intraoperative US was performed by one of two experienced sonographers in conjunction with the surgeon. Surgery was performed 1–18 weeks (mean, 5 weeks) after imaging. There was full agreement between the findings at MR and those at surgery in four patients whose operations were performed 10 weeks or longer (maximum, 18 weeks in one patient who underwent heart bypass surgery prior to hepatic resection) after imaging.

Statistical Analysis
Alternative–free response ROC curves were calculated (by using ROC curve analysis [15]) for each observer and for each sequence by plotting the true-positive fraction against the likelihood of obtaining a false-positive image (an image with one or more false-positive lesions) at each confidence level (16). The conventional ROC method does not allow the recording or differentiation of multiple responses per image, whereas alternative–free response ROC is a modified ROC technique that allows multiple responses, enabling all of the observers' responses to be correlated with the actual lesions present. As with the ROC method, the area under each alternative–free response ROC curve was used to compare the overall performance of sequences and observers.

The sensitivity for each observer and each technique was also assessed by using only those lesions allocated confidence rating of 3 or 4. Separate analyses of all lesions, malignant lesions 1 cm or larger, and malignant lesions smaller than 1 cm were undertaken. Finally, we investigated whether any improvements in sensitivity and accuracy could be obtained by reviewing all sequences in combination, compared with individual sequences. In those patients (n = 25) in whom any lesion found at surgery or histopathologic examination was missed by any observer with any sequence, a further blinded review of all combined sequences was performed.

The Student t test was used to assess the statistical significance of the differences between the mean for all four observers, for each sequence, for sensitivity, and for the areas under the alternative–free response ROC curves.

	RESULTS

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
Lesion Detection
One hundred fifty-two lesions were present in 49 patients. At least one malignant lesion was present in every patient; the maximum number in any patient was nine. Twenty-three of the 152 lesions were benign. Of the 129 malignant lesions, 80 (62%) were 1 cm or larger, and 49 (38%) were smaller than 1 cm. Eighteen of 23 (78%) benign lesions were smaller than 1 cm (Table 1).

View larger version (53K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Graph shows the mean alternative-free response ROC values for each sequence in all patients. Error bars indicate the range of results for the four observers. The major factor that influenced detectability was lesion size. Of the sequences used after enhancement, fast SE (FSE) was the least accurate, irrespective of lesion size. DE = dual echo.

View larger version (48K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Graph shows the mean sensitivity with each sequence in all patients. Error bars indicate the range of results for the four observers. The mean number of lesions depicted with each sequence are indicated above the bars. All sequences used after enhancement had approximately equal sensitivity. DE = dual echo, FSE = fast SE.

View larger version (60K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Graph shows the mean alternative-free response ROC values for each sequence and for all sequences combined in the subset of patients (n = 25) in whom any observer missed any lesion. Error bars indicate the range of results for the four observers. The combination of the sequences improved the accuracy, irrespective of lesion size. DE = dual echo, FSE = fast SE.

View larger version (60K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Graph shows the sensitivities for individual sequences and for all sequences combined in the subset of patients (n = 25) in whom any observer missed any lesion. Error bars indicate the range of results for the four observers. The mean number of lesions depicted with each sequence are indicated above the bars. The combination of the sequences led to an increase in sensitivity, irrespective of lesion size. DE = dual echo, FSE = fast SE.

Sensitivity
With the exception of the long TR/TE sequence for lesions smaller than 1 cm, all enhanced sequences showed significantly better mean sensitivity than that of the nonenhanced fast SE sequence (P < .01). After SPIO enhancement, for all lesions, the dual-echo sequence was significantly more sensitive than the long TR/TE (P < .05) sequence. For lesions 1 cm or larger, the long TR/short TE sequence was more sensitive than the enhanced fast SE or long TR/TE sequences (P < .02). For lesions smaller than 1 cm, there was no significant difference in the sensitivity of any enhanced sequence.

For all lesions, the combination of all sequences was more sensitive than the enhanced fast SE (P < .03); long TR/short TE (P < .04); or long TR/TE (P < .03) sequence. For lesions 1 cm or larger, the combined sequences were more sensitive than the enhanced fast SE (P < .05) or long TR/short TE (P < .02) sequence. For lesions smaller than 1 cm, the combined sequences were more sensitive than the long TR/TE sequence alone (P < .05).

False-Negative Lesions
At confidence levels of 3 or 4, 20 lesions in 11 patients were not detected by any observer with any sequence (six of 20 were detected by at least one observer with at least one sequence, but at only a low confidence level). All 20 lesions were smaller than 1 cm (eight, <5 mm); six were located on or close to the liver surface, and seven were located in the left lobe segments. The other 109 malignant lesions were each detected by at least one observer with at least one sequence.

In four patients, surgical management was unaltered because the location of the lesions did not alter the surgical approach. In six patients, the planned surgical resection with additional metastatectomies was performed after the finding of additional lesions at intraoperative US. In another patient, intraoperative US demonstrated a previously undetected lesion close to the right hepatic vein; this may have altered the surgical approach. However, extrahepatic disease was also present, and the procedure was therefore abandoned. In one further patient with a single large metastasis, preoperative imaging caused underestimation of the involvement in the diaphragm and right and middle hepatic veins and of the peritoneal disease, which precluded resection.

False-Positive Lesions
The false-positive findings for each observer and each technique at a confidence threshold of 3 or 4 are shown in Table 2. On the basis of the number of true-positive findings recorded by each observer at this confidence threshold, the false-positive rates for observers 1–4 were 3.2%, 5.6%, 3.1%, and 8.8%. For each observer, the SPIO-enhanced fast SE sequence resulted in the highest number of false-positive findings, and the FLASH sequence resulted in the lowest. Only one false-positive lesion was identified by more than one observer with more than one sequence (enhanced fast SE and long TR/TE sequences).

	DISCUSSION

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
Although the majority of investigators have reported statistically significant improvements in lesion detection after SPIO enhancement, findings from three studies (9,14,17) showed no statistically significant difference between nonenhanced and SPIO-enhanced MR imaging. The most likely cause for these discrepant results is a combination of differences in field strength and pulse sequence parameters. Although several investigators have attempted to establish the optimum sequence after SPIO enhancement, most studies have been limited by the small numbers of patients (3,9,10,17,18), the inadequate verification at histopathologic examination (1,5,8,9,11,13,14,17–19), the inhomogeneous patient populations (3,5,8–11,14,17–19), or the exclusion of small lesions (6). We have attempted to address these issues in the design of our study by comparing all of the major T2-weighted pulse sequences that are used for liver imaging, by using the alternative–free response ROC method (which allows an observer response for all of the lesions present), and by including every lesion in every patient in our analysis.

Although our results confirm a significant increase in sensitivity and accuracy after SPIO enhancement, the results for individual sequences are, at first sight, disappointing. In a previous study in which we used only an enhanced fast SE sequence after SPIO enhancement (2), we achieved an accuracy of 0.85. Seneterre et al (6) found an accuracy of 0.95 with SPIO-enhanced MR imaging, but their analysis excluded the majority of small lesions because the conventional ROC method was used.

Increasingly, aggressive surgery has resulted in the referral of more patients with multiple small lesions for surgical consideration; this was reflected in our current patient population. Over one-third (49 of 129) of the malignant lesions in this study were smaller than 1 cm. When only malignant lesions 1 cm or larger were analyzed, the best individual sequence was the FLASH sequence, which failed to depict only one 2-cm lesion, giving a sensitivity of 99%. The findings from this study again demonstrate the limited ability of preoperative imaging to depict lesions smaller than 1 cm. In the study by Hagspiel et al (3), only 36% of lesions smaller than 1 cm were depicted at SPIO-enhanced MR imaging, which is entirely in accordance with the results of our study.

Improvements in sensitivity and accuracy might be expected with multiple sequences. Since this approach is more analagous to clinical practice, we included an additional analysis of all sequences combined in 25 patients in whom a lesion had been missed by one or more observers with one or more sequences. The combined sequences improved the detection of small lesions for three of four observers. Although this led to an improvement in the mean for all four observers, the difference did not reach statistical significance. The reduced sensitivity of the combined sequences for the other observer probably reflects a change in the level of confidence attached to a lesion depicted with only one sequence. All 20 of the lesions missed at MR imaging were smaller than 1 cm (eight were smaller than 5 mm). In keeping with the other studies, most of the lesions missed were located on or close to the liver surface.

Fat suppression has been recommended as a means of improving the conspicuity of surface lesions (20,21). We did not use fat suppression in this study because suppression of the fat signal was less homogeneous after SPIO enhancement and because, in a previous study (2), we found that fat-suppressed T2-weighted SE images were the least sensitive.

Enhanced fast SE was never the most sensitive enhanced sequence. When the alternative–free response ROC values (which reflect both sensitivity and specificity) were considered, in all categories, the enhanced fast SE sequence was significantly less accurate than the dual-echo or FLASH sequences due to the higher number of false-positive findings. In the recent study by Reimer et al (12), fast SE imaging depicted more lesions than did conventional SE imaging. This difference may be partly due to their use of gradient motion rephasing, which was not available with fast SE imaging with our system. Also, bias may have been introduced into their study because over one-third (62 of 168) of their detected lesions were found in only two patients. By using fast SE imaging with an echo train length of 16, Schwartz et al (13) found no improvement in the lesion-to-liver contrast-to-noise ratio after SPIO enhancement, whereas there was a significant improvement with conventional SE imaging.

Magnetization transfer effects, which result in signal loss, are present on fast SE images (Fig 5). Normal liver undergoes considerable magnetization transfer effects, whereas cysts and hemangiomas are unaffected and show no loss of signal. However, malignant lesions undergo more magnetization transfer effects than does normal liver (22), so lesion-to-liver contrast is reduced. In addition, the effect of SPIO enhancement on signal loss in normal liver tissue is reduced on fast SE images since multiple 180° refocusing pulses diminish the local field inhomogeneities induced by the SPIO particles, making the fast SE sequence relatively insensitive to susceptibility.

View larger version (170K): [in this window] [in a new window] [Download PPT slide]   Figure 5a. Transverse (a) nonenhanced fast SE (4,673/91), (b) enhanced fast SE (4,673/91), (c) long TR/short TE (2,000/45), (d) long TR/TE (2,000/90), and (e) FLASH (150/10) images show a 4-cm metastasis (arrow in c-e) in segment 4, which is well shown in c-e. In a and b, the lesion is isointense with background liver possibly due to magnetization transfer effects.

View larger version (167K): [in this window] [in a new window] [Download PPT slide]   Figure 5b. Transverse (a) nonenhanced fast SE (4,673/91), (b) enhanced fast SE (4,673/91), (c) long TR/short TE (2,000/45), (d) long TR/TE (2,000/90), and (e) FLASH (150/10) images show a 4-cm metastasis (arrow in c-e) in segment 4, which is well shown in c-e. In a and b, the lesion is isointense with background liver possibly due to magnetization transfer effects.

View larger version (175K): [in this window] [in a new window] [Download PPT slide]   Figure 5c. Transverse (a) nonenhanced fast SE (4,673/91), (b) enhanced fast SE (4,673/91), (c) long TR/short TE (2,000/45), (d) long TR/TE (2,000/90), and (e) FLASH (150/10) images show a 4-cm metastasis (arrow in c-e) in segment 4, which is well shown in c-e. In a and b, the lesion is isointense with background liver possibly due to magnetization transfer effects.

View larger version (176K): [in this window] [in a new window] [Download PPT slide]   Figure 5d. Transverse (a) nonenhanced fast SE (4,673/91), (b) enhanced fast SE (4,673/91), (c) long TR/short TE (2,000/45), (d) long TR/TE (2,000/90), and (e) FLASH (150/10) images show a 4-cm metastasis (arrow in c-e) in segment 4, which is well shown in c-e. In a and b, the lesion is isointense with background liver possibly due to magnetization transfer effects.

View larger version (180K): [in this window] [in a new window] [Download PPT slide]   Figure 5e. Transverse (a) nonenhanced fast SE (4,673/91), (b) enhanced fast SE (4,673/91), (c) long TR/short TE (2,000/45), (d) long TR/TE (2,000/90), and (e) FLASH (150/10) images show a 4-cm metastasis (arrow in c-e) in segment 4, which is well shown in c-e. In a and b, the lesion is isointense with background liver possibly due to magnetization transfer effects.

Because of gradient limitations, we were unable to test the more recent breath-hold versions of the fast SE sequence in this study. To our knowledge, there are no published studies in which breath-hold fast SE sequences after SPIO enhancement were evaluated, but several investigators (24,25) have concluded that nonenhanced breath-hold fast SE sequences should not replace conventional SE sequences because of the reduced contrast between the liver and the solid lesions. It is likely that the longer echo train lengths required to facilitate breath-hold imaging will increase magnetization transfer effects and also further reduce signal loss after SPIO enhancement.

Irrespective of lesion size, dual-echo and FLASH sequences were the most accurate pulse sequences. The combination of the long TR/short TE and the long TR/TE sequences was particularly effective. The more pronounced loss of signal intensity in the normal liver on long TR/TE images caused some small lesions, which were well depicted on long TR/short TE images, to be partly obscured by blooming. These lesions were given higher confidence scores on long TR/short TE images. Conversely, a greater loss in liver signal intensity resulted in some lesions being more conspicuous on long TR/TE images than on long TR/short TE images.

Lesions that were uncertain on conventional SE or fast SE images because of motion artifact were frequently detected with greater confidence on FLASH images (Fig 6). Along with Van Beers et al (10), we also found FLASH to be useful in the segmental localization of lesions because of high liver-to-vessel contrast due to increased sensitivity to SPIO enhancement. However, because of blooming, FLASH was marginally less sensitive than was the dual-echo sequence in the detection of small lesions; this difference did not reach statistical significance. While results obtained at a field strength of 1.0 T should be extrapolated with care to those obtained at field strengths of 0.5 and 1.5 T, an SPIO-enhanced MR protocol that includes FLASH and dual-echo sequences takes account of previous studies performed at these field strengths. (The FLASH sequence is the most sensitive sequence at 0.5 T, and the long TR/short TE sequence is most sensitive at 1.5 T.)

View larger version (154K): [in this window] [in a new window] [Download PPT slide]   Figure 6a. Transverse (a) enhanced fast SE (4,673/91), (b) long TR/short TE (2,000/45), and (c) FLASH (150/10) images show one metastasis (large straight arrow), which was assigned a score in a-c, and two additional lesions, which are shown in c. One lesion (small straight arrow) is not distinguished from the vessels, and one (bent arrow in c) is not depicted in a or b. Note the increased blurring due to a reduction in signal intensity from late echoes in a compared with b.

View larger version (174K): [in this window] [in a new window] [Download PPT slide]   Figure 6b. Transverse (a) enhanced fast SE (4,673/91), (b) long TR/short TE (2,000/45), and (c) FLASH (150/10) images show one metastasis (large straight arrow), which was assigned a score in a-c, and two additional lesions, which are shown in c. One lesion (small straight arrow) is not distinguished from the vessels, and one (bent arrow in c) is not depicted in a or b. Note the increased blurring due to a reduction in signal intensity from late echoes in a compared with b.

View larger version (176K): [in this window] [in a new window] [Download PPT slide]   Figure 6c. Transverse (a) enhanced fast SE (4,673/91), (b) long TR/short TE (2,000/45), and (c) FLASH (150/10) images show one metastasis (large straight arrow), which was assigned a score in a-c, and two additional lesions, which are shown in c. One lesion (small straight arrow) is not distinguished from the vessels, and one (bent arrow in c) is not depicted in a or b. Note the increased blurring due to a reduction in signal intensity from late echoes in a compared with b.

The enhanced fast SE sequence resulted in an unacceptably high number of false-positive findings and in the greatest disparity between the findings of the individual observers. Nonenhanced fast SE images and long TR/short TE images were often helpful in distinguishing vessels from small lesions and in distinguishing the liver from the lung bases and bowel (Fig 7). False-positive findings that resulted from motion-related artifacts were reduced on FLASH images, and high liver-to-vessel contrast probably accounts for the low number of false-positive findings due to vascular structures. Combining the sequences also led to fewer false-positive findings (seven of 318 [2.2%]) because most false-positive findings were made with only a single sequence.

View larger version (130K): [in this window] [in a new window] [Download PPT slide]   Figure 7a. Transverse (a) nonenhanced fast SE (4,673/91), (b) enhanced fast SE (4,673/91), (c) long TR/short TE (2,000/45), and (d) long TR/TE (2,000/90) images show the liver is distinguished from the lung bases only in a and in c. Note the benign lesion (large arrow) is clearly visible in a-d, but the metastasis (short arrow in a), which is well shown in a, is not visible in b probably because of blooming from the iron in the background liver.

View larger version (157K): [in this window] [in a new window] [Download PPT slide]   Figure 7b. Transverse (a) nonenhanced fast SE (4,673/91), (b) enhanced fast SE (4,673/91), (c) long TR/short TE (2,000/45), and (d) long TR/TE (2,000/90) images show the liver is distinguished from the lung bases only in a and in c. Note the benign lesion (large arrow) is clearly visible in a-d, but the metastasis (short arrow in a), which is well shown in a, is not visible in b probably because of blooming from the iron in the background liver.

View larger version (155K): [in this window] [in a new window] [Download PPT slide]   Figure 7c. Transverse (a) nonenhanced fast SE (4,673/91), (b) enhanced fast SE (4,673/91), (c) long TR/short TE (2,000/45), and (d) long TR/TE (2,000/90) images show the liver is distinguished from the lung bases only in a and in c. Note the benign lesion (large arrow) is clearly visible in a-d, but the metastasis (short arrow in a), which is well shown in a, is not visible in b probably because of blooming from the iron in the background liver.

View larger version (163K): [in this window] [in a new window] [Download PPT slide]   Figure 7d. Transverse (a) nonenhanced fast SE (4,673/91), (b) enhanced fast SE (4,673/91), (c) long TR/short TE (2,000/45), and (d) long TR/TE (2,000/90) images show the liver is distinguished from the lung bases only in a and in c. Note the benign lesion (large arrow) is clearly visible in a-d, but the metastasis (short arrow in a), which is well shown in a, is not visible in b probably because of blooming from the iron in the background liver.

We used two different SPIO compounds in this study, but there is now a substantial body of literature on both agents (4,12,27,28) that indicates their effects are similar. Also, in a recent quantitative analysis (29), we found no statistically significant difference between the two agents in their effects on the liver signal intensity and on the contrast-to-noise ratio on T2-weighted images.

In conclusion, SPIO-enhanced MR imaging is significantly more sensitive than nonenhanced MR imaging in the detection of focal liver lesions. At 1.0 T, the most accurate sequences after SPIO enhancement are T2-weighted FLASH and dual-echo sequences. The enhanced fast SE sequence is not recommended since it offers no improvement in sensitivity and since it is associated with a high false-positive rate. Further work should focus on improving the accuracy of SPIO enhancement with state-of-the-art techniques for the detection of small lesions.

	Acknowledgments

 
The authors thank Jane Bates and Sarah Riley for intraoperative US correlation, Jane Howard for preparing the manuscript, the Medical Illustration Department, and the staff of the MR imaging unit.

	Footnotes

 
Abbreviations: FLASH = fast low-angle shot GRE = gradient echo ROC = receiver operating characteristic SE = spin echo SPIO = superparamagnetic iron oxide TE = echo time TR = repetition time

Author contributions: Guarantor of integrity of entire study, J.W.; study concepts and design, J.W., P.J.R.; definition of intellectual content, J.W.; literature research, J.W.; clinical studies, J.A.G., P.J.R., J.P.A.L., J.I.W.; data acquisition, J.W., F.C., J.A.G., P.J.R.; data analysis, J.W.; statistical analysis, D.W., J.W.; manuscript preparation, J.W.; manuscript editing, all authors; manuscript review, J.W., P.J.R.

	References

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 

1. Bellin MR, Zaim S, Auberton E, et al. Liver metastases: safety and efficacy of detection with superparamagnetic iron oxide MR imaging. Radiology 1994; 193:657-663.[Abstract/Free Full Text]
2. Blakeborough A, Ward J, Wilson D, et al. Hepatic lesion detection at MR imaging: a comparative study with four sequences. Radiology 1997; 203:759-763.[Abstract/Free Full Text]
3. Hagspiel KD, Neidl KFW, Eichenberger AC, Weder W, Marincek B. Detection of liver metastases: comparisons of superparamagnetic iron oxide-enhanced and unenhanced MR imaging at 1.5 T with dynamic CT, intraoperative US, and percutaneous US. Radiology 1995; 196:471-478.[Abstract/Free Full Text]
4. Kopp AE, Laniado M, Dammann F, et al. MR imaging of the liver with Resovist: safety, efficacy, and pharmacodynamic properties. Radiology 1997; 204:749-756.[Abstract/Free Full Text]
5. Ros PR, Freeny PC, Harms SE, et al. Hepatic MR imaging with ferumoxides: a multicenter clinical trial of the safety and efficacy in the detection of focal hepatic lesions. Radiology 1995; 196:481-478.[Abstract/Free Full Text]
6. Seneterre E, Taourel P, Bouvier Y, et al. Detection of hepatic metastases: ferumoxides-enhanced MR imaging versus unenhanced MR imaging and CT during arterial portography. Radiology 1996; 200:785-792.[Abstract/Free Full Text]
7. Ward J, Naik K, Guthrie JA, Wilson D, Robinson PJ. Hepatic lesion detection: comparison of MR imaging after the administration of superparamagnetic iron oxide with dual-phase CT by using alternative–free response receiver operating characteristic analysis. Radiology 1999; 210:459-466.[Abstract/Free Full Text]
8. Winter TC, Freeney PC, Nghiem HV, et al. MR imaging with i.v. superparamagnetic iron oxide: efficacy in the detection of focal hepatic lesions. AJR 1993; 161:1191-1198.[Abstract/Free Full Text]
9. Marchal G, Van Heche P, Demaerel P, et al. Detection of liver metastases with superparamagnetic iron oxide in 15 patients: results of MR imaging at 1.5 T. AJR 1989; 152:771-775.[Abstract/Free Full Text]
10. Van Beers BE, Lacrosse M, Jamart J, et al. Detection and segmental location of malignant hepatic tumors: comparison of ferumoxides-enhanced gradient-echo and T2-weighted spin-echo MR imaging. AJR 1997; 168:713-717.[Abstract/Free Full Text]
11. Halavaara JT, Lamminen AE, Bondestam S, Standertskjold-Nordenstam CG, Hamberg LM. Detection of focal liver lesions with superparamagnetic iron oxide: value of STIR and SE imaging. J Comput Assist Tomogr 1994; 18:897-904.[Medline]
12. Reimer P, Rummeny EJ, Daldrup HE, et al. Clinical results with Resovist: a phase 2 clinical trial. Radiology 1995; 195:489-496.[Abstract/Free Full Text]
13. Schwartz LH, Seltzer SE, Tempany CMC, et al. Superparamagnetic iron oxide hepatic MR imaging: efficacy and safety using conventional and fast spin-echo pulse sequences. JMRI 1995; 5:566-570.
14. Denys A, Arrive L, Servois V, et al. Hepatic tumors: detection and characterization at 1-T MR imaging enhanced with AMI-25. Radiology 1994; 193:665-669.[Abstract/Free Full Text]
15. McNeil BJ, Hanley JA. Statistical approaches to the analysis of receiver operating charateristic (ROC) curves. Med Decis Making 1984; 4:137-150.
16. Chakraborty DP, Winter LHL. Free-response methodology: alternate analysis and a new observer-performance experiment. Radiology 1990; 174:873-881.[Abstract/Free Full Text]
17. Duda SH, Laniado M, Kopp AF, et al. Superparamagnetic iron oxide: detection of focal liver lesions at high-field-strength MR imaging. JMRI 1994; 4:309-314.
18. Fretz CJ, Stark DD, Metz CE, et al. Detection of hepatic metastases: comparison of contrast-enhanced CT, unenhanced MR imaging, and iron-oxide enhanced MR imaging. AJR 1990; 155:763-770.[Abstract/Free Full Text]
19. Stark DD, Weissleder R, Elizondo G, et al. Superparamagnetic iron oxide: clinical application as a contrast agent for MR imaging of the liver. Radiology 1988; 168:297-301.[Abstract/Free Full Text]
20. Lu DSK, Sarni S, Hahn PF, et al. T2-weighted MR imaging of the upper part of the abdomen: should fat suppression be used routinely?. AJR 1994; 162:1095-1100.[Abstract/Free Full Text]
21. Semelka RC, Chew W, Hricak H, Tomei E, Higgins C. Fat saturation MR imaging of the upper abdomen. AJR 1990; 155:1111-1116.[Abstract/Free Full Text]
22. Outwater E, Schnall MD, Bratman LE, Dinsmore BJ, Kressel HY. Magnetization transfer of hepatic lesions: evaluation of a novel contrast technique in the abdomen. Radiology 1992; 182:535-540.[Abstract/Free Full Text]
23. Listerud J, Einstein S, Outwater E, Kressel HY. First principles of fast spin echo. Magn Reson Q 1992; 8:199-244.[Medline]
24. Carpenter KD, Macaulay SE, Schulte SJ, et al. MR of focal liver lesions: comparison of breath-hold and non-breath-hold hybrid RARE and conventional spin-echo T2-weighted pulse sequences. JMRI 1996; 6:596-602.
25. Yamashita Y, Yamamoto H, Namimoto T, Abe Y, Takahashi M. Phased array breath-hold versus non-breath-hold MR imaging of focal liver lesions: a prospective comparative study. JMRI 1997; 7:292-297.
26. Soyer P, Elias D, Zeitoun G, Roche A, Levesque M. Surgical treatment of hepatic metastases: impact of intraoperative ultrasound. AJR 1993; 160:511-514.[Abstract/Free Full Text]
27. Vogl TJ, Hammerstinl R, Schwarz W, et al. Superparamagnetic iron oxide-enhanced versus gadolinium-enhanced MR imaging for differential diagnosis of focal liver lesions. Radiology 1996; 198:881-887.[Abstract/Free Full Text]
28. Hamm B, Staks T, Taupitz M, et al. Contrast-enhanced MR imaging of liver and spleen: first experience in humans with a new superparamagnetic iron oxide. JMRI 1994; 4:659-668.
29. Chen F, Ward J, Robinson PJ. MR imaging of the liver and spleen: a comparison of the effects on signal intensity of two superparamagnetic iron oxide agents. Magn Reson Imaging 1999; 17:549-556.[Medline]

This article has been cited by other articles:

				M.-J. Kim, J. H. Kim, J. Y. Choi, S. H. Park, J.-J. Chung, K. W. Kim, and D. G. Mitchell Optimal TE for SPIO-enhanced gradient-recalled echo MRI for the detection of focal hepatic lesions. Am. J. Roentgenol., September 1, 2006; 187(3): W255 - W266. [Abstract] [Full Text] [PDF]

				H. J. Kim, K. W. Kim, J. H. Byun, H. J. Won, Y. M. Shin, P. N. Kim, M.-S. Lee, and M.-G. Lee Comparison of mangafodipir trisodium- and ferucarbotran-enhanced MRI for detection and characterization of hepatic metastases in colorectal cancer patients. Am. J. Roentgenol., April 1, 2006; 186(4): 1059 - 1066. [Abstract] [Full Text] [PDF]

				K. Nasu, Y. Kuroki, S. Nawano, S. Kuroki, T. Tsukamoto, S. Yamamoto, K. Motoori, and T. Ueda Hepatic Metastases: Diffusion-weighted Sensitivity-encoding versus SPIO-enhanced MR Imaging Radiology, April 1, 2006; 239(1): 122 - 130. [Abstract] [Full Text] [PDF]

				J. Ward, P. J. Robinson, J. A. Guthrie, S. Downing, D. Wilson, J. P. A. Lodge, K. R. Prasad, G. J. Toogood, and J. I. Wyatt Liver Metastases in Candidates for Hepatic Resection: Comparison of Helical CT and Gadolinium- and SPIO-enhanced MR Imaging Radiology, October 1, 2005; 237(1): 170 - 180. [Abstract] [Full Text] [PDF]

				S. Bipat, M. S. van Leeuwen, E. F. I. Comans, M. E. J. Pijl, P. M. M. Bossuyt, A. H. Zwinderman, and J. Stoker Colorectal Liver Metastases: CT, MR Imaging, and PET for Diagnosis--Meta-analysis Radiology, October 1, 2005; 237(1): 123 - 131. [Abstract] [Full Text] [PDF]

				M. Matsuo, M. Kanematsu, K. Itoh, T. Murakami, Y. Maetani, H. Kondo, S. Goshima, N. Kako, H. Hoshi, J. Konishi, et al. Detection of Malignant Hepatic Tumors with Ferumoxides-Enhanced MRI: Comparison of Five Gradient-Recalled Echo Sequences with Different TEs Am. J. Roentgenol., January 1, 2004; 182(1): 235 - 242. [Abstract] [Full Text] [PDF]

				S. Kumano, T. Murakami, T. Kim, M. Hori, A. Okada, T. Sugiura, Y. Noguchi, S. Kawata, K. Tomoda, and H. Nakamura Using Superparamagnetic Iron Oxide-Enhanced MRI to Differentiate Metastatic Hepatic Tumors and Nonsolid Benign Lesions Am. J. Roentgenol., November 1, 2003; 181(5): 1335 - 1339. [Abstract] [Full Text] [PDF]

				J. Ward, J. A. Guthrie, D. Wilson, P. Arnold, J. P. Lodge, G. J. Toogood, J. I. Wyatt, and P. J. Robinson Colorectal Hepatic Metastases: Detection with SPIO-enhanced Breath-hold MR Imaging—Comparison of Optimized Sequences Radiology, September 1, 2003; 228(3): 709 - 718. [Abstract] [Full Text] [PDF]

				Y. Noguchi, T. Murakami, T. Kim, M. Hori, K. Osuga, S. Kawata, S. Kumano, A. Okada, T. Sugiura, and H. Nakamura Detection of Hepatocellular Carcinoma: Comparison of Dynamic MR Imaging with Dynamic Double Arterial Phase Helical CT Am. J. Roentgenol., February 1, 2003; 180(2): 455 - 460. [Abstract] [Full Text] [PDF]

				K. Mori, J. Scheidler, T. Helmberger, N. Holzknecht, R. Schauer, C. A. Schirren, I. Bittmann, M. Dugas, and M. Reiser Detection of Malignant Hepatic Lesions Before Orthotopic Liver Transplantation: Accuracy of Ferumoxides-Enhanced MR Imaging Am. J. Roentgenol., October 1, 2002; 179(4): 1045 - 1051. [Abstract] [Full Text] [PDF]

				A. R. van Erkel, M. E. J. Pijl, A. A. van den Berg-Huysmans, M. N. J. M. Wasser, C. J. H. van de Velde, and J. L. Bloem Hepatic Metastases in Patients with Colorectal Cancer: Relationship between Size of Metastases, Standard of Reference, and Detection Rates Radiology, August 1, 2002; 224(2): 404 - 409. [Abstract] [Full Text] [PDF]

				S. Kawata, T. Murakami, T. Kim, M. Hori, M. P. Federle, S. Kumano, E. Sugihara, S. Makino, H. Nakamura, and M. Kudo Multidetector CT: Diagnostic Impact of Slice Thickness on Detection of Hypervascular Hepatocellular Carcinoma Am. J. Roentgenol., July 1, 2002; 179(1): 61 - 66. [Abstract] [Full Text] [PDF]

				T. Murakami, T. Kim, M. Takamura, M. Hori, S. Takahashi, M. P. Federle, K. Tsuda, K. Osuga, S. Kawata, H. Nakamura, et al. Hypervascular Hepatocellular Carcinoma: Detection with Double Arterial Phase Multi-Detector Row Helical CT Radiology, March 1, 2001; 218(3): 763 - 767. [Abstract] [Full Text]

D. Choi, S. H. Kim, J. H. Lim, W. J. Lee, H.-J. Jang, S. J. Lee, and H. K. Lim Preoperative Detection of Hepatocellular Carcinoma: Ferumoxides-Enhanced MR Imaging Versus Combined Helical CT During Arterial Portography and CT Hepatic Arteriography Am. J. Roentgenol., February 1, 2001; 176(2): 475 - 482. [Abstract]