SS 301a - Liver: advances in CT and MRI
Transient severe motion artifact related to gadoxetic acid enhanced liver MRI: incidence and risk evaluation at a European institution
Purpose: Varying incidences (5-18%) of contrast-related transient severe motion (TSM) imaging artifacts during gadoxetic disodium-enhanced arterial phase liver MRI have been reported. Since previous reports originated from the United States and Japan, we aimed to determine the incidence of TSM at a European institution and to correlate it with potential risk factors and previously published results.
Methods and Materials: Two age and sex matched groups were retrospectively selected (gadoxetate disodium n=89; gadobenate dimeglumine n=89) from dynamic contrast-enhanced MRI examinations. Respiratory motion related artifacts in non-enhanced and dynamic phases were assessed independently by two readers blinded to contrast agents on a 4 point scale. Scores of ≥ 3 were considered as severe motion artifacts. Severe motion artifacts in arterial phases were considered as TSM if scores in all other phases were < 3. Potential risk factors for TSM were evaluated via logistic regression analysis.
Results: For gadoxetate disodium, the mean score for respiratory motion artifacts was significantly higher in the arterial phase (2.2±0.9) compared to all other phases (1.6±0.7) (p<0.05). Incidence of TSM was significantly higher with gadoxetate disodium (n=19; 21.1%) than with gadobenate dimeglumine (n=1; 1.1%) (p<0.001). The incidence of TSM at our institution is in the upper range of previously published reports. Logistic regression analysis did not show any significant correlation between TSM and risk factors (all p>0.05).
Conclusion: No associated risk factors for TSM could be identified. However, we revealed a high incidence of TSM at a European institution, undermining the importance of a diagnosis-limiting phenomenon.
Liver function is significantly correlated with liver to portal vein contrast ratio during the hepatobiliary phase with Gd-EOB-DTPA-enhanced MR at 3 Tesla
Purpose: To quantitatively evaluate the correlation of the liver-to-portal vein contrast ratio (LPC) and liver function on gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced MR imaging.
Methods and Materials: A total of 102 patients who underwent Gd-EOB-DTPA-enhanced 3Tesla MR imaging (normal liver, n=20; Child-Pugh class A, n=55; B, n=21; and C, n=6) were included in this restrospective study. LPC was defined as the enhancement ratio of hepatic parenchyma and potal vein at hepatobiliary phase(HBP) 20min, and it was compared between normal and cirrhosis livers. The correlation between LPC and hepatic function parameters at HBP after injection was quantitatively analysed as well.
Results: The degree of LPC differed between normal and cirrhosis livers (P<0.001) significantly. LPC constantly and significantly decreased as the severity of cirrhosis increased at HBP imaging (P<0.001). Total bilirubin (P<0.001), albumin (P<0.001), platelet count (P<0.001), and model for end stage liver disease score (P<0.001) were independent predictors of LPC at HBP imaging.
Conclusion: The index of hepatic LPC on Gd-EOB-DTPA HBP imaging can predict the severity of cirrhosis and is correlated with clinical hepatic function parameters.
Magnetic resonance elastography for prediction of radiation-induced liver disease after stereotactic body radiation therapy
Purpose: To evaluate the usefulness of magnetic resonance elastography (MRE) for prediction of radiation-induced liver disease (RILD) after stereotactic body radiation therapy (SBRT) in patients with hepatocellular carcinoma (HCC).
Methods and Materials: Seventeen patients who underwent SBRT (dose: 28-60Gy in 4-10 fractions) for HCC (size: 14-68mm) had pretreatment liver stiffness measured by MRE. Diagnosis of RILD was determined by the following criteria: (i) anicteric elevation of alkaline phosphatase to a level at least twice the upper limit of the normal level, or (ii) elevation of transaminases, to at least five times the upper limit of the normal, or five times the pretreatment level within 16 weeks after SBRT. The following variables were analysed as potential predictors of RILD: age; sex; body weight; etiology of hepatitis; liver stiffness measured by MRE; tumour size blood test results; tumour markers; Child-Pugh score; Karnofsky performance status; indocyanine green retention rate at 15min; and parameters of SBRT (fraction size, total dose, V20 (percentage of normal liver volume that received >20Gy), and VS15 (volume spared from ≥15Gy)).
Results: Pretreatment liver stiffness measured by MRE in patients with RILD (n=4; median 8.3kPa) was significantly higher than that in patients without RILD (n=13; median 5.0kPa) (P=0.0090). Other variables showed no significant differences between the two groups. No patient died of RILD.
Conclusion: Pretreatment liver stiffness measured by MRE can be a predictor of RILD after SBRT in patients with HCC.
Purpose: To evaluate a free-breathing self-gated volumetric interpolated breath-hold examination (VIBE) with compressed sensing (CS) for continuous dynamic contrast-enhanced (DCE) MR imaging of the liver.
Methods and Materials: 25 patients underwent gadobutrol contrast-enhanced MRI of the liver on a 1.5T scanner. A free-breathing prototype VIBE-sequence with CS and automated respiration gating (VIBECS) was continuously acquired for 128 seconds starting with the administration of the contrast agent. From the acquired raw data, 16 consecutive series with a temporal resolution of 8 seconds and spatial resolution of 1.2x1.2x3 mm were reconstructed. The unenhanced, arterial, portal-venous and venous phase series with the subjectively most satisfying image quality were selected and compared to a clinical routine VIBE-sequence (VIBEstd). Image quality was assessed qualitatively (overall, sharpness, lesion conspicuity, motion/other artefacts; two readers independently; 5-point-Likert scale; 5=excellent) and quantitatively (coefficient-of-variation (CV); mean liver SI).
Results: Overall image quality, lesion conspicuity and motion artefacts showed no significant differences between the sequences (p≥0.07). Image sharpness was significantly higher in the portal-venous (4.9±0.2) and venous phase (4.9±0.3) images of VIBECS as compared to VIBEstd (4.2±0.6; p=0.01). However, VIBECS showed a significant appearance of bow-like reconstruction artefacts (3.8±0.5), which were not present in VIBEstd (5.0±0.0; p≤0.001). Mean liver SI was significantly higher in VIBEstd (200.1±85.1) than in VIBECS (170.3±68.3; p=0.002) whereas CV calculations revealed no significant differences among the sequences (p=0.1).
Conclusion: VIBECS is applicable for continuous self-gated DCE-MRI of the liver with similar overall image quality and lesion conspicuity and improved image sharpness as compared to a VIBEstd.
Purpose: To observe transient dyspnea pattern after gadoxetic acid administration and prove clinical feasibility of free-breathing gadoxetic acid-enhanced liver magnetic resonance imaging (MRI) using golden-angle radial sparse parallel imaging (GRASP) with respiratory gating.
Methods and Materials: In this IRB-approved prospective study, 60 patients have been enrolled and informed consent was obtained from all patients. Dynamic T1-weighted image (T1WI) was obtained using GRASP in free-breathing manner. Development of transient motion after gadoxetic acid administration was monitored using respiratory motion recording. In addition, subjective symptom, oxygen saturation and heart rate were monitored. Early arterial, late arterial and portal venous phases with different temporal resolution (13sec vs. 6sec) and respiratory gating were reconstructed for evaluating image quality and motion artifact.
Results: In 40.0% (23/59), transient motion (mean duration: 21.5±13.0 seconds) was observed and it developed in 15 seconds after gadoxetic acid administration in 73.9% (17/23). There were no significant difference of subjective symptom and vital signs between groups with and without transient motion (P=0.25~0.99). Motion artifact reduced after respiratory gating, and there was no significant difference of motion artifact on early and late arterial phases between the two groups (P=0.42) whereas significantly higher motion artifact was observed on non-gated 13sec-T1WI in patients with transient motion (P=0.001). Gated 6-sec-T1WI provided significantly lower image quality than gated 13-sec-T1WI.
Conclusion: Free-breathing T1WI using GRASP and incoherent undersampling technique provided timely and acceptable arterial phase consistently in patients with transient motion which is potentially influential on arterial phase.
Purpose: To evaluate the feasibility of multiarterial MRI of hepatic arterial dominant phase (HAD) using a free-breathing self-gated spoiled gradient-echo sequence with compressed sensing (CS).
Methods and Materials: 40 patients who underwent contrast-enhanced liver MRI with gadobutrol were included. Each 20 patients were examined in 1.5 and 3T (Magnetom Aera and Skyra, Siemens Healthineers), respectively. Multiarterial imaging was performed for 64s beginning with contrast injection using a rapid free-breathing prototype volume-interpolated-breath-hold examination (VIBE) sequence with CS and automated respiration gating (iVIBE, 1,2x1,2x3mm3). 8 consecutive arterial subphases were reconstructed (8s/phase). Two readers selected and assessed the best HAD phase on a five-point Likert scale (5=excellent) regarding image quality, respiratory artefact compensation, enhancement of abdominal aorta and hepatic artery, lesion detectability and diagnostic confidence. Signal intensity (SI) and SNR were measured in the liver and in the abdominal aorta.
Results: HAD phase varied among patients (3rd - 7th phase). Image quality and respiratory artefact compensation were good (median=4, 3-5) without significant differences among field strengths (p≥.089). Lesion detectability, contrast enhancement of aorta and diagnostic confidence were excellent (median 5, 3-5) without significant differences among field strengths (p≥.165). Signal intensity and liver SNR were significantly higher in 3T (p≤.015). SNR in abdominal aorta was high in both field strengths without significant differences (p.=637).
Conclusion: Acquisition of continuous multiarterial MRI of hepatic arterial dominant phase during free-breathing using iVIBE is feasible and yields good image quality and excellent arterial enhancement with good compensation of respiratory artefacts resulting in excellent diagnostic confidence in both 1.5T and 3T.
Gadoxetic acid-enhanced MR imaging of transient hepatic enhancement difference: another cause of hypointense observation on hepatobiliary phase
Purpose: To determine the frequency, natural history and factors associated with the presence of transient hepatic enhancement difference (THED) showing hypointensity in the hepatobiliary phase (HBP) of gadoxetic acid-enhanced MR imaging.
Methods and Materials: Gadoxetic acid-enhanced MR imaging of 125 patients (91 males; mean age: 68 years; range 26-84 years) with THED were retrospectively reviewed. Three readers qualitatively and quantitatively evaluated MR imaging features of THED and evolution at follow up. Fisher’s exact test and Mann-Whitney-Wilcoxon test were used for statistical analysis.
Results: THED were wedge-shaped (n=106 (85%)), hypervascular on HAP (125 of 125; 100%) and isointense on both HBP (105 of 125; 84%) and fat-suppressed T2-weighted imaging (n = 100 (80%)). Hypointense THED on hepatobiliary phase were more commonly wedge-shaped (P < .05), hyperintense on T2-weighted imaging (P < .05), and were associated with previous locoregional treatment (P < .05). Of 12 patients with hypointense THED on HBP who had follow-up MR, nine showed reduction in size.
Conclusion: Hypointensity on HBP is not uncommonly observed in THED, and might be explained with chronic portal hypoperfusion and or biliary obstruction resulting from previous locoregional treatment of adjacent tumour and leading to hepatocellular dysfunction, fibrotic changes and parenchymal atrophy.
Intravoxel incoherent motion diffusion-weighted imaging of hepatic warm ischemia-reperfusion injury in a rabbit model
Purpose: To investigate the diagnostic efficacy of intravoxel incoherent motion (IVIM) MRI in the grade of hepatic warm ischemia-reperfusion injury (WIRI) in rabbit models.
Methods and Materials: Fifty rabbits were randomly divided into sham-operation group (n=10) and four test groups (n=10 for each group) according to different hepatic warm ischemia time. IVIM was performed on a 3 T MR scanner with 11 b values of 0 to 800 s/mm2. Rabbits were sacrificed for biochemical and histomorphological analysis. Interobserver variability, the parametric, non parametric methods and receiver operating characteristic (ROC) curves were used to determine diagnostic efficacies.
Results: Apparent diffusion coefficient (ADC), true diffusion (Dslow), pseudodiffusion (Dfast) and perfusion fraction (PF) showed moderate to excellent interobserver reproducibility. Significant differences were found between different grades of hepatic WIRI by ADC, Dslow, Dfast, and PF (P <0.001). IVIM parameters corresponded well with biochemical parameters (P <0.01). ROC analysis showed that the differences in ADC, Dslow, Dfast, and PF among different grades of hepatic WIRI were significant (P <0.05), and the area under the ROC curve (AUC) of PF was the largest.
Conclusion: The grades of hepatic WIRI could be effectively evaluated by using IVIM in a rabbit model, with PF being the most accurate parameter.
Liver function evaluation at magnetic resonance imaging: comparison between liver enhancement and MELD
Purpose: To evaluate the enhancement of the liver parenchyma in patients with different grade of liver cirrhosis.
Methods and Materials: Seventy-five patients scheduled for liver transplantation were divided according to the MELD ( < 15 and ≥ 15) and Child-Pugh (A, B, C) classifications for liver cirrhosis. Fourty-one patients reported also hepatocellular carcinoma with 63 lesions (1-4 per patients). All subjects underwent MRI at 1.5T (Avanto, Siemens). MR acquisitions comprised unenhanced breath-hold T2W images and volumetric 3D gadoxetic acid enhanced (0.1 mmol/kg bodyweight) T1W GRE images acquired at 25s, 60s, 180s (dynamic phases) and 20 minutes (hepatobiliary phase). Quantitative analysis comprised: enhancement ratio (ER) of liver parenchyma, signal-to-noise ratio of liver parenchyma (SNR) and contrast-to-noise ratio (CNR) between signal intensity of lesions and liver parenchyma. Comparisons among different classes were performed with ANOVA or Kruskall-Wallis test and Mann-Whitney U and Student's t test. P values of < 0.05 were considered statistically significant.
Results: Patients with MELD score lower than 15 reported a significant higher ER and CNR (0.47 vs 0.35 and -11.6 vs -0.09). The same result was observed between Child-Pugh groups A - B and C (0.46 - 0.52 and 0.31). Regarding CNR a significant difference was found between class B and C (-11.8 vs. -0.929; p=0.03). No significant differences were reported on SNR.
Conclusion: Liver enhancement during the hepatobiliary phase should be useful to distinguish different grade of liver function.
Purpose: To develop a prediction method using liver surface analysis based on the computed tomography (CT) for liver stiffness measurement in patients with liver cirrhosis.
Methods and Materials: This retrospective study was approved by the institutional review board, and the requirement for informed consent was waived. 47 patients with liver cirrhosis who had received transient elastography for liver stiffness measurement and liver CT were retrospectively analyzed. We assessed the inflection/length of liver surface (left lateral segment) using auto-segmentation method based on the CT images to evaluate the degree of surface serration. The correlations between inflection/length of liver surface and the degree of fibrosis stage F1 from F4 on the METAVIR scale were assessed using simple regression analysis.
Results: Median liver stiffness values measured using transient elastography was 8.9 kPa (range, 7.2-52.3 kPa). The degrees of liver stiffness according to the fibrosis stage in the patients included were 2.4-7.1 in F1, 7.2-8.0 kPa in F2, 8.1-10.9 kPa in F3, and 11.0-52.3 kPa in F4, respectively. The inflection/length of liver surface using auto-segmentation method showed a stepwise increase with increasing degree of fibrosis stage (ρ = 0.43; p = 0.018).
Conclusion: Liver surface analysis using CT auto-segmentation method could stratify the degree of liver fibrosis in patients with liver cirrhosis.
Purpose: Assessment of baseline values in CT perfusion (CTp) studies is as much difficult as it plays a crucial role for an accurate computation of perfusion parameters. Besides that, the unenhanced stage of CTp provides repeated measures of baseline. This study aims at computing and analysing the voxel-based baseline values of normal liver in a set of patients with colorectal cancer.
Methods and Materials: 40 patients with colorectal cancer, free from any hepatic diseases, underwent axial CTp examinations of liver at colorectal cancer diagnosis. A large region of interest (ROI) was drawn on one central section of the liver, avoiding large vessels. Baseline portion (BP) of each TCC of the ROI was determined through the use of an adaptive 3D filtering and the relative baseline value (HU) was computed as the mean of the TCC concentration values over BP. Mean, median, and standard deviation values were computed for each patient as well as for the entire cohort.
Results: For each patient, mean and median baseline values coincide, thus suggesting quite symmetric distributions. The average and standard deviation (std) of baseline mean values of all patients are 61.7HU (range 32-79HU), and 10.1, with 73% of values within one std interval (51-71HU). Moreover, baseline colorimetric maps showed a high spatial correlation.
Conclusion: Baseline values of normal liver of patients with colorectal cancer are compliant with values of normal liver in healthy subjects reported in the literature. The spatial coherence hints at a local similarity of tissue features.