Nodule-in-nodule appearance of hepatocellular carcinomas: comparison of gadolinium-enhanced and ferumoxides-enhanced magnetic resonance imaging

PURPOSE: To perform comparison of gadolinium-enhanced and ferumoxides-enhanced magnetic resonance imaging (MRI) in the detection of nodule-in-nodule appearance of hepatocellular carcinomas (HCCs). MATERIALS AND METHODS: During a recent 45-month period, we had eight patients (five men and three women; age range, 63-84 years; mean, 71 years) with HCCs with nodule-in-nodule appearance who underwent gadolinium-enhanced MRI, ferumoxides-enhanced MRI, and computed tomography during arterial portography (CTAP) and computed tomography during hepatic arteriography (CTHA), combined and separately, within an interval of two weeks. Two blinded radiologists in consensus retrospectively evaluated three sets of sequences: unenhanced T1- and T2-weighted MR, gadolinium-enhanced MR, and ferumoxides-enhanced MR images in random order of patients and imaging sequences. The depiction degree of nodule-in-nodule appearance of HCC was evaluated in a semiquantitative fashion. The sensitivities of unenhanced T1- and T2-weighted, gadolinium-enhanced, and ferumoxides-enhanced MR images were compared with McNemar's test. RESULTS: The eight HCCs with nodule-in-nodule appearance ranged in size from 16-26 mm (mean, 20.0 +/- 4.0 mm), and there existed nine internal HCC foci ranging in size from 5-14 mm (mean, 7.9 +/- 3.5 mm). On gadolinium-enhanced MR images, the nodule-in-nodule appearance of HCC was typically seen as hypervascular foci in an iso- or hypovascular area: the depiction degree of nodule-in-nodule appearance was distinct in two lesions, equivocal in three, and absent in three. On ferumoxides-enhanced MR images, it was typically seen as hyperintense foci in a hypointense area: the depiction degree was distinct in four, moderate in one, and absent in three. The sensitivities for detection of nodule-in-nodule appearance were 25%, 25%, and 63% on T1- and T2-weighted, gadolinium-enhanced, and ferumoxides-enhanced MR images, respectively, but there was no significant difference in sensitivity. CONCLUSION: Nodule-in-nodule appearance of HCCs can be seen on ferumoxides-enhanced MR images, in some cases more clearly than on gadolinium-enhanced MR images, particularly when the background nodule shows hyperintensity on precontrast T1-weighted images. Ferumoxides-enhanced MRI may be considered when development of malignant foci is suspected during routine examinations.     

Bone marrow blood supply in gadolinium-enhanced magnetic resonance imaging

A retrospective study was performed to determine whether bone blood supply can be assessed on gadolinium-enhanced magnetic resonance imaging. Lumbar spine magnetic resonance imaging (MRI) examinations of 49 patients attending for post-laminectomy examination were reviewed (30 male, 19 female, mean age 46.4 years, age range 23–84 years). Each study included sagittal T1-weighted spin echo sequences before and after gadolinium administration. Regions of interest were drawn within the L3 vertebral body from a parasagittal slice from each sequence. Signal intensity (SI) values were ascertained and the percentage increase in SI was calculated. For each patient, changes in receiver gain for pre and post-gadolinium images were corrected by an image scaling factor. In all cases, a measurable increase in SI was found (mean 15.3%, range 4.4–55.7) due to bone vascularity. The results give no indication of the quantity or timing of blood supply but provide a basis for further work.     

Evaluation of recurring meniscal tears with gadolinium-enhanced magnetic resonance imaging: a randomized, prospective study

BACKGROUND: Magnetic resonance imaging has been shown to be of questionable accuracy in detecting recurrent meniscal tears after previous resection or repair. PURPOSE: To compare the accuracy of nonenhanced magnetic resonance imaging with that of intraarticular contrast-enhanced direct magnetic resonance arthrography and intravenous contrast-enhanced indirect magnetic resonance arthrography for detection of recurrent meniscal tears. STUDY DESIGN: Prospective cohort study. METHODS: Forty-one patients who had previous meniscal tears treated by resection or repair but who were experiencing recurrent knee symptoms were prospectively randomized into one of three groups: conventional magnetic resonance imaging, indirect arthrography, and direct arthrography. The interpretations of two musculoskeletal radiologists were compared with the findings of an arthroscopic procedure performed 2 to 14 weeks later. RESULTS: Conventional imaging had a sensitivity of 57.9%, specificity of 80%, and overall accuracy of 62.5%. Intravenous contrast improved the sensitivity to 90.9%, specificity to 100%, and overall accuracy to 93.8%; intraarticular contrast had a sensitivity of 91.7%, specificity of 100%, and an overall accuracy of 92.9%. CONCLUSIONS: We demonstrated an increased accuracy of intravenous or intraarticular contrast-enhanced magnetic resonance arthrography in detecting recurrent meniscal tears. Both contrast routes demonstrated similar accuracy, a finding not previously reported.     

Comparison of different magnetic resonance imaging sequences for assessment of fistula-in-ano

AIM: To assess agreement between different forms of T2 weighted imaging (T2WI), and post-contrast T1WI in the depiction of fistula tracts, inflammation, and internal openings with that of a reference test.METHODS: Thirty-nine consecutive prospective cases were enrolled. The following sequences were used for T2WI: 2D turbo-spin-echo (2D T2 TSE); 3D T2 TSE; short tau inversion recovery (STIR); 2D T2 TSE with fat saturation performed in all patients. T1WI were either a 3D T1-weighted prepared gradient echo sequence with fat saturation or a 2D T1 fat saturation [Spectral presaturation with inversion (SPIR)]. Agreement for each sequence for determination of fistula extension, internal openings, and the presence of active inflammation was assessed separately and blindly against a reference test comprised of follow-up, surgery, endoscopic ultrasound, and assessment by an independent experienced radiologist with access to all images.RESULTS: Fifty-six fistula tracts were found: 2 inter-sphincteric, 13 trans-sphincteric, and 24 with additional tracts. The best T2 weighted sequence for depiction of fistula tracts was 2D T2 TSE (Cohen’s kappa = 1.0), followed by 3D T2 TSE (0.88), T2 with fat saturation (0.54), and STIR (0.19). Internal openings were best seen on 2D T2 TSE (Cohen’s kappa = 0.88), followed by 3D T2 TSE (0.70), T2 with fat saturation (0.54), and STIR (0.31). Detection of inflammation showed Cohen’s kappa of 0.88 with 2D T2 TSE, 0.62 with 3D T2 TSE, 0.63 with STIR, and 0.54 with T2 with fat saturation. STIR, 3D T2 TSE, and T2 with fat saturation did not make any contributions compared to 2D T2 TSE. Post-contrast 3D T1 weighted prepared gradient echo sequence with fat saturation showed better agreement in the depiction of fistulae (Cohen’s kappa = 0.94), finding internal openings (Cohen’s kappa = 0.97), and evaluating inflammation (Cohen’s kappa = 0.94) compared to post-contrast 2D T1 fat saturation or SPIR where the corresponding figures were 0.71, 0.66, and 0.87, respectively. Comparing the best T1 and T2 sequences showed that, for best results, both sequences were necessary.CONCLUSION: 3D T1 weighted sequences were best for the depiction of internal openings and active inflammatory components, while 2D T2 TSE provided the best assessment of fistula extension.     

Role of gadolinium-enhanced magnetic resonance imaging in retroperitoneal fibrosis.

PURPOSE: To determine if the gadolinium-enhancement characteristics of magnetic resonance images (MRI) differ for acute and chronic benign retroperitoneal fibrosis (RPF). METHOD: Seven male subjects, 3 with newly diagnosed nontreated RPF (acute group) and 4 with long-standing stable RPF (chronic group) who had been treated with steroids, ureteric stents or both, underwent MRI examinations with gadolinium enhancement. Patients in the acute group were examined again after treatment. Mean dynamic gadolinium-enhancement ratios (both dynamic and delayed) were calculated for each group. RESULTS: The initial mean dynamic enhancement ratio for the acute group (mean 1.86, range 1.80-1.95) was significantly different (p = 0.005) from that of the chronic group (mean 1.37, range 1.26-1.61). The mean dynamic enhancement ratio for the acute group after treatment (4-8 months duration) was 1.40 (range 1.26-1.51). The mean delayed enhancement ratio (RPF/psoas muscle signal intensity) for the acute group was 1.41 (range 1.38-1.43, data from 2 patients) and for the chronic group was 1.29 (range 1.13-1.44). CONCLUSION: Dynamic gadolinium enhancement was useful in differentiating newly diagnosed RPF from treated chronic disease and may have a role in assessing disease activity, monitoring response to treatment and detecting relapse.     

Wash-in kinetics for gadolinium-enhanced magnetic resonance imaging of carotid atheroma

PURPOSE: To determine the wash-in kinetics of intravenous gadolinium into the fibrous cap and lipid core of carotid atheroma, and identify the time following gadolinium administration that maximizes contrast between the cap and core. MATERIALS AND METHODS: Seven subjects with carotid artery stenosis were studied. Magnetic resonance (MR) images of the atheroma were acquired using a single-inversion-recovery fast-spin-echo (IR FSE) sequence, which was serially repeated during the first 30 minutes following intravenous gadolinium administration. Postcontrast time was divided into three intervals: <10, 11-20, and >21 minutes. Adjusted signal intensity (SI), signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) values for postcontrast images were compared to adjusted precontrast values. RESULTS: The mean SNRs of the cap and core were significantly elevated in each postcontrast interval compared with mean precontrast values. The CNRs of the cap vs. the core increased by 19.8% (3.03 to 4.14, P = 0.03) in the first 10 minutes following gadolinium administration, and remained elevated over the next two intervals with a slight decrease in the final interval. CONCLUSION: Increased signal and contrast between the cap and core can be achieved by imaging up to 30 minutes following gadolinium administration, with peak enhancement occurring in the first 10 minutes.     

Fast magnetic resonance imaging techniques

This article reviews fast magnetic resonance (MR) techniques currently used for body imaging. Improvements in gradient performance have made very short repetition and echo times on clinical scanners feasible, thus enabling subsecond image acquisition. The article provides a fundamental overview of the technical aspects from the concept of k-space and k-space segmentation technique, fast MR imaging techniques including fast spin echo, fast gradient echo with or without magnetization preparation to echo planar and hybrid techniques. The article also addresses the use of different fat suppression techniques in MR imaging of the body and improvements in coil technology to obtain faster images and higher signal-to-noise.     

Serial gadolinium-enhanced magnetic resonance imaging in patients with multiple sclerosis treated with mitoxantrone

Serial gadolinium (Gd)-enhanced magnetic resonance imaging (MRI) was used to monitor the effect of mitoxantrone in ten patients with rapidly deteriorating multiple sclerosis (MS). MRI was performed as a baseline and thereafter at 1,3,6,9,12 and 24 months. The total number of Gd-enhancing lesions diminished from 169 at baseline to 10 after 1 year and to 5 after 2 years. This reduction and the percentage of follow-up MRI studies showing no Gd enhancement were more pronounced than in other MRI studies of the natural course of MS. Measured with quantitative neurological scales, only one patient showed deterioration after 2 years; nevertheless, the changes in MRI were much more marked than those observed clinically. Serial Gd-MRI therefore, seems necessary for documenting efficacy in future therapeutic trails.     

Ultrasound and magnetic resonance imaging in Crohn's disease: a comparison

The objective of this retrospective study was to compare MRI of the abdomen with ultrasound of the abdomen and gastrointestinal tract in patients with Crohn's disease. Forty-six patients were included in the study. We analyzed the localization of Crohn's lesions, the number of affected bowel segments, the number of stenoses, and the presence of abscesses, fistulae, and any additional findings. Findings were verified by means of one or more of the following: enteroclysis; surgical findings; and colonoscopy. The results show that MRI is superior to ultrasound in the localization of affected bowel segments (sensitivity: MRI 97.5%; US 76%) and in recognizing fistulae (sensitivity: MRI 87%; US 31%), stenoses (sensitivity: MRI 100%; US 58%) and abscesses (sensitivity: MRI 100%; US 89%). Magnetic resonance imaging of the abdomen should be obtained to clarify discrepant clinical and sonographic findings. In addition, despite its higher cost, MRI of the abdomen is justified in patients in whom Crohn's lesions are known or suspected in anatomic areas proximal to the terminal or neoterminal ileum and in cases with suspicion of fistulae and abscesses.     

Primary intracranial tumor imaging: a comparison of magnetic resonance and CT

Twenty-six patients with primary intracranial tumors were evaluated by magnetic resonance (MR) and a comparison was made with CT findings. The SE technique with TR = 2,000 ms and TE = 56 ms was most useful in delineating normal anatomy, in differentiating gray and white matter, and in optimizing the visualization of edema. The use of TR = 500 ms optimized sensitivity to T1 relaxation time differences in disparate tissue and best defined the cerebrospinal fluid spaces. Although prolongation of T1 and T2 relaxation values was seen with most malignant lesions, several cases produced no obvious prolongation of T1. Separation of tumor from surrounding edema was possible in several instances. In 16/26 patients information not available on CT was obtained with MR. This included detection of altered tissue characteristics where CT showed only mass effect, more accurate depiction of full extent and location of tumor, and visualization of associated abnormalities. However, a small inner ear epidermoid tumor seen with high-resolution CT (1.5-mm sections) was difficult to identify with certainty on MR, and in two other cases punctate foci of calcification were not identified on MR.     

Comparison study of intraarticular and intravenous gadolinium-enhanced magnetic resonance imaging of cartilage in a canine model

Background: Magnetic resonance (MR) imaging and measurement of glycosaminoglycan (GAG) have potential for characterization of hyaline articular cartilage. Recently, some reports have demonstrated the potential of direct administration of contrast media for MR imaging of cartilage.

Purpose: To prove the feasibility of intraarticular gadolinium-enhanced MR imaging of cartilage (iGEMRIC) and T1 relaxation mapping of the articular cartilage in vivo with intraarticular injection of Gd-DTPA^2-.

Material and Methods: Five healthy beagle dogs underwent MR imaging and T1 relaxation mapping of the knee joints of both hind legs. The delayed gadolinium-enhanced MR imaging of cartilage (dGEMRIC) and iGEMRIC techniques were interchanged with MR imaging. For dGEMRIC, a double routine dose of Gd-DTPA^2- (0.2 mM/kg) was administered intravenously. For iGEMRIC, 2.5 and 1.25 mmol/l saline-diluted Gd-DTPA^2- solutions were separately injected into the right and left knee joints, respectively, prior to MR imaging. Color-coded T1 maps of 20 femoral condyles were obtained from the dGEMRIC and iGEMRIC images. The signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and glycosaminoglycan (GAG) delineation of articular cartilage were compared between the dGEMRIC and iGEMRIC techniques.

Results: The mean SNR was higher with dGEMRIC than with iGEMRIC, but the difference was not statistically significant (P = 0.174). The mean (±SD) CNR was higher with iGEMRIC (-11.6±3.4) than with dGEMRIC (-16.7±4.0; P = 0.000), although the absolute value of the CNR was higher with dGEMRIC. The layering and gradient distribution of GAG were more clearly visualized on the iGEMRIC images. The mean scores of GAG delineation with dGEMRIC and iGEMRIC were 0.7±0.6 and 2.2±1.7, respectively. The iGEMRIC method better visualized GAG distribution (P = 0.001).

Conclusion: Although the SNR did not differ significantly between the iGEMRIC and dGEMRIC techniques, the color-coded T1 map produced with iGEMRIC allowed better cartilage evaluation. Thus, iGEMRIC exhibits the useful features of both MR arthrography and dGEMRIC, and provides a color-coded T1 map that is useful for diagnosing early articular cartilage damage.     

Fat-suppressed dynamic and delayed gadolinium-enhanced volumetric interpolated breath-hold magnetic resonance imaging of cholangiocarcinoma

OBJECTIVE: To determine the enhancement phase providing the highest contrast-to-noise ratio (CNR) between cholangiocarcinoma and liver or portal vein on dynamic and delayed gadolinium-enhanced magnetic resonance imaging (MRI). SUBJECTS AND METHODS: Precontrast, 3-phase dynamic postcontrast, and delayed postcontrast MRI of the liver was performed in 25 patients with cholangiocarcinoma and correlated with surgical findings, pathology, and other imaging studies. Contrast-to-noise ratios for tumor relative to adjacent liver and portal vein were calculated from signal intensities determined from regions of interest obtained for each phase of enhancement. A subjective assessment of the signal intensity of the periportal tissues relative to the portal vein was made for each set of delayed images. RESULTS: A mass was visible in 24 of 25 patients. Tumor masses were hypointense in 92%, 67%, 75%, and 21%; isointense in 8%, 8%, 17%, and 12%; and hyperintense in 0%, 25%, 8%, and 67% of patients relative to liver on precontrast, arterial, portal venous, and delayed images, respectively. No single phase of gadolinium enhancement demonstrated consistently superior tumor-versus-liver CNR. Delayed imaging demonstrated the highest tumor-versus-liver CNR in 25% of patients and the lowest in 33%. The portal venous phase demonstrated the highest tumor-versus-portal vein CNR in 75% of patients. Delayed postcontrast images demonstrated the lowest tumor-versus-portal vein CNR in 38% of patients. Periportal tissues were isointense to portal vein in all but 1 patient on delayed images. CONCLUSION: No single phase of dynamic and delayed gadolinium-enhanced MRI demonstrates superior CNR between cholangiocarcinoma and normally enhancing liver, although the portal phase provides the best CNR between tumor and portal vein in most cases. Although delayed enhancement is typical of cholangiocarcinoma, delayed imaging does not necessarily offer superior contrast between tumor and liver parenchyma compared with other phases of enhancement. Differentiation between tumor and portal vein and periportal tissues may be difficult on delayed images.     

Turbo STIR magnetic resonance imaging as a whole-body screening tool for metastases in patients with breast carcinoma: preliminary clinical experience

This study was undertaken to assess the utility of whole-body turbo short tau inversion recovery (STIR) magnetic resonance imaging (MRI) to detect metastases to liver, brain, and bone as a single examination in women with breast cancer. Seventeen patients with biopsy-proven breast cancer and suspected metastatic disease attending over a 12-month period referred for both conventional imaging and whole-body MRI were included in the study. Three patients were found to be free of metastases at both conventional and MR imaging. Appendicular or axial skeletal metastases were identified in 11 of 17 patients, with correlation between findings at whole-body MRI and scintigraphy in 15 of the 17 patients. Five patients had evidence of hepatic metastases on whole-body MRI, of which metastases were identified in only three patients at CT despite contrast enhancement. Four patients had brain abnormalities (metastases in three patients, meningioma in one patient) detected on both whole-body and dedicated brain MRI. Preliminary clinical experience suggests that turbo STIR whole-body MRI may represent a convenient and cost-effective method of total body screening for metastases in patients with breast carcinoma.     

Depiction of the thoracic duct by magnetic resonance imaging: comparison between magnetic resonance imaging and the anatomical literature

Purpose  

The aim of this study was to compare thoracic duct (TD) configuration depicted by magnetic resonance imaging (MRI) with TD configuration described in the anatomical literature.     

Differentiation of normal and neoplastic bone tissue in dynamic gadolinium-enhanced magnetic resonance imaging: validation of a semiautomated technique

Purpose

This study was undertaken to clinically validate the accuracy of a semiautomated software tool for analysing the enhancement curve in focal malignant bone lesions.

Materials and methods

Twenty-three patients affected by cancer with malignant focal bone lesions underwent dynamic gadolinium-enhanced magnetic resonance (MR) imaging using the following protocol: T1-weighted turbo spin-echo sequences (time to repeat [TR] 600 ms, time to echo [TE] 8.6 ms, field of view [FOV] 40×40 cm) before and after intravenous injection of gadolinium-containing contrast agent. Image postprocessing was performed using the software DyCoH. Each region of interest (5×5 pixels), drawn to include the area of the lesion with the highest values of the area under the curve map, was analysed to obtain time-intensity curves and relative perfusion parameters: time to peak (TTP), peak intensity (PI), slope (60-s slope), intensity at 60 s after contrast agent injection (60-s I) and final intensity (FI).

Results

Data were obtained by analysing 86 malignant lesions and 86 apparently normal bone regions. PI, 60-s slope, 60-s I and FI were significantly different between neoplastic and apparently normal (p<0.001) samples. Sensitivity, specificity and accuracy were, respectively, 94%, 93% and 94% at a PI threshold of 100 (signal-to-noise ratio), with positive and negative predictive values of 93% and 94%. At a threshold value of 0.85 for 60-s slope, sensitivity and specificity values were both 91%.

Conclusions

The semiautomated technique we report appears to be accurate for identifying neoplastic tissue and for mapping perfusion parameters, with the added value of a consistent measurement of perfusion parameters on colour-coded maps.     

Evaluation of prostate size: a comparison of ultrasound and magnetic resonance imaging

To determine the relative accuracy of ultrasonography (US) and magnetic resonance imaging (MRI) in the evaluation of prostate volume, we compared US and MR images with surgical findings in 15 patients. Transabdominal US was excellent for determining prostate size in patients with small to moderate enlargement. When compared with surgical specimens, the difference between the weight of the gland as predicted by US and the actual weight was 14% (SD +/- 12). With the transabdominal approach, the length was often inaccurately imaged, but the addition of transrectal scans in the sagittal projection improved results: with combined transabdominal and transrectal US, the average difference in weight was 8% (SD +/- 7). The MRI more accurately predicted prostatic volume (average difference, 6% (SD +/- 6), but the difference between the latter 2 is not significant. In 5 additional patients who had undergone transurethral resection of the prostate, residual prostatic tissue was evaluated by MRI and US. In addition to demonstrating prostate size, transrectal sagittal US showed the relationship among the bladder neck, prostatic urethra, and remaining prostatic tissue. Real-time US also allowed evaluation of dynamic sphincteric contractions, which is a finding not available today with MRI. Tissue differentiation was attempted with both MRI and US. Neither imaging modality could differentiate benign from malignant disease.     

Asbestos-related pleural disease: value of dedicated magnetic resonance imaging techniques

OBJECTIVES: We sought to compare respiratory-gated high-spatial resolution magnetic resonance imaging (MRI) and radial MRI with ultra-short echo times with computed tomography (CT) in the diagnosis of asbestos-related pleural disease. METHODS: Twenty-one patients with confirmed long-term asbestos exposure were examined with a CT and a 1.5-T MR unit. High-resolution respiratory-gated T2w turbo-spin-echo (TSE), breath-hold T1w TSE, and contrast-enhanced fat-suppressed breath-hold T1w TSE images with an inplane resolution of less than 1 mm were acquired. To visualize pleural plaques with a short T2* time, a pulse sequence with radial k-space-sampling was used (TE = 0.5 milliseconds) before and after administration of Gd-DTPA. CT and MR images were assessed by 4 readers for the number and calcification of plaques, extension of pleural fibrosis, extrapleural fat, detection of mesothelioma and its infiltration into adjacent tissues, and detection of pleural effusion. Observer agreement was studied with the use of kappa statistics. RESULTS: The MRI protocol allowed for differentiation between normal pleura and pleura with plaques. Interobserver agreement was comparable for MRI and CT in detecting pleural plaques (median kappa = 0.72 for MRI and 0.73 for CT) and significantly higher with CT than with MRI for detection of plaque calcification (median kappa 0.86 for CT and 0.72 for MRI; P = 0.03). Median sensitivity of MRI was 88% for detection of plaque calcification compared with CT. For assessment of pleural thickening, pleural effusion, and extrapleural fat, interobserver agreement with MRI was significantly higher than with CT (median kappa 0.71 and 0.23 for pleural thickening, 0.87 and 0.62 for pleural effusion, and 0.7 and 0.56 for extrapleural fat, respectively; P < 0.05). For detection of mesothelioma, median kappa was 0.63 for MRI and 0.58 for CT. CONCLUSION: High-resolution MR sequences and radial MRI achieve a comparable interobserver agreement in detecting pleural plaques and even a higher interobserver agreement in assessing pleural thickening, pleural effusion, and extrapleural fat when compared with CT.     

Early-enhancing nonneoplastic lesions on gadolinium-enhanced magnetic resonance imaging of the liver following partial hepatectomy

PURPOSE: To assess the frequency, imaging findings, and significance of early-enhancing nonneoplastic (EN) lesions with gadolinium-enhanced magnetic resonance imaging (MRI) of the liver following partial hepatectomy. MATERIALS AND METHODS: We retrospectively reviewed MR images after partial hepatectomy in 30 patients. Postoperative MRI was performed in 1-12 months (mean, 3.7 months) after partial hepatectomy. We defined the EN lesion as a lesion that was ill defined; irregular, wedge shaped, or serpiginous; located along the liver edge; not visible on unenhanced MR images; did not appear hypointense on portal venous- or equilibrium-phase images; or a combination of those imaging findings. RESULTS: A total of 39 EN lesions (size range, 5-60 mm; mean, 25.2 mm) in 19 patients and 17 recurrent tumors (size range, 5-50 mm; mean, 16.8 mm) in 10 patients newly appeared after partial hepatectomy. The EN lesions were diagnosed as pseudolesions by the second postoperative follow-up MRI in 17 patients or contrast-enhanced computed tomography (CT) in two. A total of 13 EN lesions (33%) were located along the liver edge and 20 (51%) were adjacent to the resected area. The shape was circular in 11 (28%), oval in three (8%), irregular in 11 (28%), wedge shaped in five (13%), and serpiginous in nine (23%). No EN lesion showed hypointensity on gadolinium-enhanced portal venous-phase or equilibrium-phase images. A total of 14 EN lesions (36%) showed slight hyperintensity on T2-weighted images. The confidence levels for malignancy probability assigned by blinded radiologists were lower with EN lesions than with recurrent tumors (P < 0.001). CONCLUSION: EN lesions are frequently seen in MRI following partial hepatectomy, and occasionally are slightly hyperintense on T2-weighted images, mimicking malignant tumors. However, most EN lesions can be correctly diagnosed with MRI findings.