Caecilia S. Reiner

Value of Retrospective Fusion of PET and MR Images in Detection of Hepatic Metastases: Comparison with 18F-FDG PET/CT and Gd-EOB-DTPA–Enhanced MRI

The purpose of this study was to compare the accuracy of lesion detection and diagnostic confidence between 18F-FDG PET/CT, gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)–enhanced MRI, and retrospectively fused PET and MRI (PET/MRI). Methods: Thirty-seven patients (mean age ± SD, 60.2 ± 12 y) with suspected liver metastases underwent PET/CT and Gd-EOB-DTPA–enhanced MRI within 0–30 d (mean, 11.9 ± 9 d). PET and Gd-EOB-DTPA–enhanced MR image data were retrospectively fused. Images were reviewed independently by 2 readers who identified and characterized liver lesions using PET/CT, Gd-EOB-DTPA–enhanced MRI, and PET/MRI. Each liver lesion was graded on a 5-point confidence scale ranging from definitely benign (grade of 1) to definitely malignant (grade of 5). The accuracy of each technique was determined by receiver-operating-characteristic analysis. Histopathology served as the standard of reference for all patients with malignant lesions. Results: A total of 85 liver lesions (55 liver metastases [65%] and 30 benign lesions [35%]) were present in 29 (78%) of the 37 patients. Twenty-four (65%) of the 37 patients had liver metastases. The detection rate of liver lesions was significantly lower for PET/CT than for Gd-EOB-DTPA–enhanced MRI (64% and 85%; P = 0.002). Sensitivity in the detection and characterization of liver metastases for PET/CT, Gd-EOB-DTPA–enhanced MRI, PET/MRI in reader 1, and PET/MRI in reader 2 was 76%, 91%, 93%, and 93%, respectively; the respective specificity values were 90%, 100%, 87%, and 97%. The difference in sensitivity between PET/CT and PET/MRI was significant (P = 0.023). The level of confidence regarding liver lesions larger than 1 cm in diameter was significantly higher in PET/MRI than in PET/CT (P = 0.046). Accuracy values (area under the receiver-operating-characteristic curve) for PET/CT, Gd-EOB-DTPA–enhanced MRI, PET/MRI in reader 1, and PET/MRI in reader 2 were 0.85, 0.94, 0.92, and 0.96, respectively. Conclusion: The sensitivity of Gd-EOB-DTPA–enhanced MRI and PET/MRI in the detection of liver metastases is higher than that of PET/CT. Diagnostic confidence was significantly better with PET/MRI than with PET/CT regarding lesions larger than 1 cm in diameter. Compared with Gd-EOB-DTPA–enhanced MRI, PET/MRI resulted in a nonsignificant increase in sensitivity and diagnostic confidence.

Reproducibility of Dynamic Contrast-enhanced MR Imaging. Part II. Comparison of Intra- and Interobserver Variability with Manual Region of Interest Placement versus Semiautomatic Lesion Segmentation and Histogram Analysis

PURPOSE To compare the inter- and intraobserver variability with manual region of interest (ROI) placement versus that with software-assisted semiautomatic lesion segmentation and histogram analysis with respect to quantitative dynamic contrast material-enhanced (DCE) MR imaging determinations of the volume transfer constant (K(trans)). MATERIALS AND METHODS The study was approved by the institutional review board and compliant with HIPAA. The requirement to obtain informed consent was waived. Fifteen DCE MR imaging studies of the female pelvis defined the study group. Uterine fibroids were used as a perfusion model. Three varying types of lesion measurements were performed by five readers on each study by using DCE MR imaging perfusion analysis software with manual ROI placement and a semiautomatic lesion segmentation and histogram analysis solution. Intra- and interreader variability of measurements of K(trans) with the different measurement types was calculated. RESULTS The overall interobserver variability of K(trans) with manual ROI placement (mean, 28.5% ± 9.3) was reduced by 42.5% when the semiautomatic, software-assisted lesion measurement method was used (16.4% ± 6.2). Whole-lesion measurement showed the lowest interobserver variability with both measurement methods (20.1% ± 4.3 with the manual method vs 10.8% ± 2.6 with the semiautomatic method). The overall intrareader variability with the manual ROI method (7.6% ± 10.6) was not significantly different from that with the semiautomatic method (7.3% ± 10.8), but the intraclass correlation coefficient for intrareader reproducibility improved from 0.86 overall with the manual method to 0.99 with the semiautomatic method. CONCLUSION A semiautomatic lesion segmentation and histogram analysis approach can provide a significant reduction in interobserver variability for DCE MR imaging measurements of K(trans) when compared with manual ROI methods, whereas intraobserver reproducibility is improved to some extent.

CT- and MRI-based volumetry of resected liver specimen: Comparison to intraoperative volume and weight measurements and calculation of conversion factors

OBJECTIVE To compare virtual volume to intraoperative volume and weight measurements of resected liver specimen and calculate appropriate conversion factors to reach better correlation. METHODS Preoperative (CT-group, n=30; MRI-group, n=30) and postoperative MRI (n=60) imaging was performed in 60 patients undergoing partial liver resection. Intraoperative volume and weight of the resected liver specimen was measured. Virtual volume measurements were performed by two readers (R1,R2) using dedicated software. Conversion factors were calculated. RESULTS Mean intraoperative resection weight/volume: CT: 855 g/852 mL; MRI: 872 g/860 mL. Virtual resection volume: CT: 960 mL(R1), 982 mL(R2); MRI: 1112 mL(R1), 1115 mL(R2). Strong positive correlation for both readers between intraoperative and virtual measurements, mean of both readers: CT: R=0.88(volume), R=0.89(weight); MRI: R=0.95(volume), R=0.92(weight). Conversion factors: 0.85(CT), 0.78(MRI). CONCLUSION CT- or MRI-based volumetry of resected liver specimen is accurate and recommended for preoperative planning. A conversion of the result is necessary to improve intraoperative and virtual measurement correlation. We found 0.85 for CT- and 0.78 for MRI-based volumetry the most appropriate conversion factors.

Quantification of liver fat in the presence of iron and iodine: an ex-vivo dual-energy CT study.

Purpose:Iodinated contrast media (CM) and iron in the liver are known to hinder an accurate quantification of liver fat content (LFC) with single-energy computed tomography (SECT). The purpose of this study was to evaluate the feasibility and accuracy of dual-energy CT (DECT) for ex vivo quantification of LFC, in the presence of iron and CM, compared with SECT. Materials and Methods:Sixteen phantoms with a defined LFC of 0%, 10%, 30%, and 50% fat and with varying iron content (0, 1.5, 3, and 6 mg/mL wet weight liver) were scanned with a second-generation dual-source 128-slice CT system. Phantoms were scanned unenhanced and contrast-enhanced after adding 1.0 mg/mL iodine to each phantom. Both SECT (120 kV) and DECT (tube A: 140 kV, using a tin filter 228 mAs; tube B: 80 kV, 421 mAs) data were acquired. An iron-specific dual-energy 3-material decomposition algorithm providing virtual noniron images (VNI) was used to subtract iron and CM from the data. CT numbers (Hounsfield units) were measured in all data sets, including 120 kV from SECT, as well as 140 kV, 80 kV, 50%:50% weighted 80 kV/140 kV, and VNI derived from DECT. The dual-energy index was calculated from 80 kV and 140 kV data. SECT and DECT measurements (Hounsfield units) including the dual-energy index of unenhanced and contrast-enhanced phantoms were compared with the known titrated LFC, using Pearson correlation analysis and Student t test for related samples. Results:Inter-reader agreement was excellent for all measurements of CT numbers in both SECT and DECT data (Pearson r, 0.965–1.0). For fat quantification in the absence of iron and CM, CT numbers were similar in SECT and DECT (all, P > 0.05), showing a linear correlation with titrated LFC (r ranging from 0.981 to 0.999; P < 0.01). For fat quantification in the presence of iron but without CM, significant underestimation of LFC was observed for all measurements in SECT and DECT (P < 0.05), except for VNI. Measurements in VNI images allowed for an accurate LFC estimation, with no significant differences compared with measurements in iron-free phantoms (all, P > 0.25). For fat quantification in the presence of iron and CM, further underestimation of LFC was seen for measurements in SECT and DECT (P < 0.015), except for VNI. Measurements in VNI images showed a high accuracy for estimating the LFC, with no significant difference compared with measurements in iron- and CM-free phantoms (P > 0.2). Conclusions:Our ex vivo phantom study indicates that DECT with the use of a dedicated, iron-specific 3-material decomposition algorithm allows for the accurate quantification of LFC, even in the presence of iron and iodinated CM. VNI images reconstructed from DECT data equal nonenhanced SECT data of liver without CM by eliminating iron and iodine from the images. No added value was seen for DECT as compared with SECT for quantification of LFC in the absence of iron and iodine.

Computed tomography perfusion imaging of renal cell carcinoma: systematic comparison with histopathological angiogenic and prognostic markers.

PurposeThe aim of this study was to systematically analyze the correlation between computed tomography (CT) perfusion and histopathological angiogenic and prognostic markers in patients with renal cell carcinoma (RCC). Material and MethodsFifteen patients (12 men; mean age, 64.5 ± 9.4 years) with RCC underwent contrast-enhanced CT perfusion imaging (scan range, 10 cm; scan time, 40 seconds; dual-source 128-section CT) 1 day before surgery. The procedure for surgical specimen processing was modified to obtain an exact match with CT images. Microvessel density (MVD) was quantified by CD34 staining, and lymphatic vessel density (LVD) was stained with D2-40 antibodies. The CT perfusion values blood flow (BF), blood volume (BV), and flow extraction product (KTrans) were calculated using the maximum-slope and a delay-corrected modified Patlak approach and were correlated to MVD and LVD. The relationship between CT perfusion and the prognostic markers pT stage, Fuhrman grade, and tumor necrosis was evaluated. ResultsHistopathology revealed varying high MVD but low or absent intratumoral LVD. The BF and BV of RCC, both including and excluding necrotic regions, showed significant correlations with MVD (r = 0.600–0.829, P < 0.05 each). Significant correlations between MVD and KTrans were found only in small tumor areas exhibiting no necrosis (r = 0.550, P < 0.05). No significant correlation was found between BF, BV, and KTrans with intratumoral LVD (P = 0.35–0.82). With higher pT stage and Fuhrman grade, BF, BV, and KTrans were lower, similar to the MVD, but without reaching statistical significance. Blood flow, BV, and KTrans were significantly higher in RCCs with less than 50% necrosis than in those with 50% or grater necrosis (P < 0.05 each). ConclusionOur study indicates that BF and BV from CT perfusion reflect blood vessels of RCC. Computed tompgraphic perfusion parameters differ significantly depending upon the degree of tumor necrosis.

Quantitative perfusion analysis of malignant liver tumors: dynamic computed tomography and contrast-enhanced ultrasound.

Objective:To prospectively analyze the correlation between quantitative parameters of perfusion derived from dynamic contrast-enhanced CT (DCE-CT) and contrast-enhanced ultrasound (DCE-US) in patients with malignant liver tumors. Materials and Methods:Thirty patients (mean age: 59.4 ± 12.3 years) with primary malignant liver tumors or hepatic metastases of various origin underwent DCE-CT (4D spiral mode, scan range, 14.8 cm; 15 scans; cycle time, 3 seconds) and DCE-US (low mechanical index, <0.1, 2.4 mL microbubbles). DCE-CT and DCE-US images were evaluated by 2 radiologists regarding quantitative perfusion parameters including arterial liver perfusion (ALP), portal-venous perfusion (PVP), and total perfusion (P = ALP + PVP) from DCE-CT, as well as blood inflow velocity (B) and the normalized slope within the calculation range (CVan) from DCE-US. Results:Quantitative assessment was possible with DCE-CT in 12/30 (40%) patients before and in all patients after automated motion correction. With DCE-US, quantitative assessment could not be performed in 9/30 (30.0%) patients due to respiratory motion. Interreader agreements for quantitative perfusion analysis were good with DCE-CT (r = 0.640–0.892, each P < 0.001) and DCE-US (r = 0.761–0.909, each P < 0.001). Moderate significant correlations were found between the perfusion parameters from DCE-CT (P, ALP) and DCE-US (B, CVan) (r = 0.446–0.621, each P < 0.05). No significant correlations were found between PVP from CT and perfusion parameters from DCE-US (B, CVan; each P = nonsignificant). Conclusions:Quantitative evaluation of DCE-CT data was feasible in all patients after automated motion correction, whereas DCE-US data could not be quantitatively evaluated in 30% of patients due to respiratory motion and lack of motion correction software. Quantitative arterial perfusion analysis showed moderate significant correlations for blood flow parameters among modalities.

Accuracy and confidence of Gd-EOB-DTPA enhanced MRI and diffusion-weighted imaging alone and in combination for the diagnosis of liver metastases

PURPOSE To evaluate the accuracy and confidence in diagnosing liver metastases using combined gadolinium-EOB-DTPA (Gd-EOB-DTPA) enhanced magnetic resonance imaging (MRI)/diffusion-weighted imaging (DWI) in comparison to Gd-EOB-DTPA enhanced MRI and DWI alone. MATERIALS AND METHODS Forty-three patients (age, 58 ± 13 years) with 89 liver lesions (28 benign, 61 malignant) underwent liver MRI for suspected liver metastases. Three image sets (DWI, Gd-EOB-DTPA and combined Gd-EOB-DTPA/DWI) in combination with unenhanced T1- and T2-weighted images were reviewed by three readers. Detection rates of focal liver lesions were assessed and diagnostic accuracy was evaluated by calculating the areas under the receiver-operating-characteristics curve (AUC). Confidence in diagnosis was evaluated on a 3-point scale. Histopathology and imaging follow-up served as the standard of reference. RESULTS Detection of liver lesions and confidence in final diagnosis for all readers were significantly higher for the combined Gd-EOB-DTPA/DWI dataset than for DWI. The combination of DWI and Gd-EOB-DTPA rendered a significantly higher confidence in final diagnosis (2.44 vs. 2.50) than Gd-EOB-DTPA alone for one reader. For two readers, accuracy in diagnosis of liver metastases was significantly higher for Gd-EOB-DTPA/DWI (AUCs of 0.84 and 0.83) than for DWI datasets (AUCs of 0.73 and 0.72). Adding DWI to Gd-EOB-DTPA did not significantly increase diagnostic accuracy as compared to Gd-EOB-DTPA imaging alone. CONCLUSION Addition of DWI sequences to Gd-EOB-DTPA enhanced MRI did not significantly increase diagnostic accuracy as compared to Gd-EOB-DTPA enhanced MRI alone in the diagnosis of liver metastases. However, the increase in diagnostic confidence might justify acquisition of DWI sequences in a dedicated MRI protocol.

CT perfusion of renal cell carcinoma: Impact of volume coverage on quantitative analysis

Purpose:To assess the feasibility, image quality, and radiation dose of computed tomography (CT) renal perfusion imaging in the adaptive 4-dimensional (4D)-spiral mode in patients with renal cell carcinoma (RCC), and to compare quantitative measurements between 2-dimensional regions-of-interest (2D-ROI) and 3-dimensional volumes-of-interest (3D-VOI). Materials and Methods:Twenty-one patients (13 male; age, 67.4 ± 9.5 years) with 24 histologically proven RCCs underwent CT perfusion imaging (100 kV, 100 mAs/rotation, scan range 10 cm, examination time 40.17 seconds) in a 4D-spiral mode with dual-source 128-slice CT. The ability to suspend respiration during CT perfusion imaging was visually monitored. Two independent readers assessed motion artifacts of CT perfusion imaging data sets on a 4-point scale before and after automated motion correction. Qualitative (enhancement pattern) and quantitative perfusion analysis (blood flow [BF], blood volume [BV], flow extraction product [KTrans]) were performed in the tumor and in healthy ipsi- and contralateral renal cortex applying the maximum-slope and a modified Patlak approach for quantitative analysis in 2D-ROI and 3D-VOI, the latter including the entire RCCs. Results:Of the 21 patients, 8 (38%) could suspend respiration throughout the perfusion scan. Of 21 RCCs, 18 (86%) were completely included in the scan range. Motion artifacts were significantly reduced after automated motion correction (P < 0.001). All 24 RCCs could be included in the qualitative perfusion analysis, and 22 of 24 (92%) were eligible for quantitative perfusion analysis. Enhancement was homogenous in 4 (17%), peripheral in 4 (17%), and heterogeneous in 16 (66%) tumors (good interobserver agreement, &kgr;=0.74). A high correlation was found between the 2 readers regarding quantitative perfusion parameters (r=0.93–0.94, P < 0.01). Quantitative measurements in 3D-VOIs revealed significantly lower BV, BF, and KTrans in RCCs than in normal renal cortex (P < 0.001). In solid tumor periphery, BV was similar to the renal cortex (P=0.299), while BF and KTrans were significantly lower (P < 0.01 and <0.001) in tumor tissue. Comparison of tumor measurements in 3D-VOIs with those obtained from 2D-ROIs revealed considerable differences in perfusion parameters beyond the 95% confidence limits in 46% to 68% of the tumors. KTrans was significantly higher in the contralateral than in healthy ipsilateral renal cortex (P < 0.01). Estimated effective radiation dose of the CT perfusion protocol was 16.3 mSv. Conclusion:CT perfusion imaging using an adaptive 4D-spiral mode is feasible and enables, after use of automated motion correction, the reliable analysis of renal perfusion in patients with RCCs. Considerable tumor heterogeneity was found, with differences in perfusion parameters between 2D-ROI and 3D-VOI analysis, reinforcing the use of volumetric techniques for perfusion imaging and analysis. Differences between ipsi- and contralateral healthy renal cortex KTrans suggest a compensatory increase in glomerular filtration rate in the healthy contralateral kidney.

MR defecography in patients with dyssynergic defecation: spectrum of imaging findings and diagnostic value

OBJECTIVES We describe the spectrum of findings and the diagnostic value of MR defecography in patients referred with suspicion of dyssynergic defecation. METHODS 48 patients (34 females, 14 males; mean age 48 years) with constipation and clinically suspected dyssynergic defecation underwent MR defecography. Patients were divided into patients with dyssynergic defecation (n = 18) and constipated patients without dyssynergic defecation (control group, n = 30). MRIs were analysed for evacuation ability, time to initiate evacuation, time of evacuation, changes in the anorectal angle (ARA-change), presence of paradoxical sphincter contraction and presence of additional pelvic floor abnormalities. Sensitivity, specificity, positive and negative predictive values and accuracy for the diagnosis of dyssynergic defecation were calculated. RESULTS The most frequent finding was impaired evacuation, which was seen in 100% of patients with dyssynergic defecation and in 83% of the control group, yielding a sensitivity for MR defecography for the diagnosis of dyssynergic defecation of 100% (95% confidence interval (CI) 97-100%), but a specificity of only 23% (95% CI 7-40%). A lower sensitivity (50%; 95% CI 24-76%) and a high specificity (97%; 95% CI 89-100%) were seen with abnormal ARA-change. The sensitivity of paradoxical sphincter contraction was relatively high (83%; 95% CI 63-100%). A combined analysis of abnormal ARA-change and paradoxical sphincter contraction allowed for the detection of 94% (95% CI 81-100%) of the patients with dyssynergic defecation. CONCLUSION MR defecography detects functional and structural abnormal findings in patients with clinically suspected dyssynergic defecation. Impaired evacuation is seen in patients with functional constipation owing to other pelvic floor abnormalities than dyssynergic defecation.

Diagnostic performance and accuracy of 3-D spoiled gradient-dual-echo MRI with water- and fat-signal separation in liver-fat quantification: comparison to liver biopsy.

Objectives:To prospectively evaluate a 3-dimensional spoiled gradient-dual-echo (3D SPGR-DE) magnetic resonance imaging (MRI) sequence for the qualitative and quantitative analysis of liver fat content (LFC) in patients with the suspicion of fatty liver disease using histopathology as the standard of reference. Materials and Methods:Thirty-four adult patients (15 women; mean age, 67 ± 13 years) underwent hepatic 1.5-Tesla 3D SPGR-DE MRI including in-/out-of-phase (IP/OP) and fat-only sequences prior to hepatic surgery with biopsy. Histopathological analyses of total and macrovesicular LFC could be made from biopsies of 39 segments in 23 patients. Two radiologists independently classified steatosis visually using a 4-point scale for IP/OP and fat-only images. Additionally, fat signal fractions (FSF) were calculated from signal intensities on IP/OP (FSFip/op) and fat-only images (FSFfat-only). Pearson correlation analysis and Student t test were used to study the relationship between the FSF and LFC as determined by histopathology. The accuracy of MRI in detecting pathologically elevated LFC was assessed by receiver operating characteristic analysis. Results:Histopathology revealed steatosis in 29/39 (74%) segments in 15/23 patients (65%), with a total LFC ranging from <5% to 90%. Inter-reader agreement for visual steatosis grading was moderate (k = 0.53) for IP/OP images and good (k = 0.68) for fat-only images. FSF calculated from IP/OP and fat-only images significantly correlated with LFC from histopathology (both, P < 0.0001). Mean FSFfat-only and FSFip/op values significantly differed from total LFC (both, P < 0.0001), whereas mean FSFfat-only showed no significant differences to macrovesicular LFC (P = 0.46). Both FSFfat-only and FSFip/op performed accurately in discriminating between normal LFC and elevated LFC according to histopathology with good diagnostic accuracy (AUC: 0.85; 95% CI: 0.73–0.89 vs. AUC 0.90, 95% CI: 0.7–1.00). Conclusions:FSFfat-only and FSFip/op derived from 3D SPGR-DE MRI mostly reflect the macrovesicular as opposed to the total LFC from histopathology, whereas both discriminate healthy and fatty liver. Analysis of fat-only images improves interreader-agreement for visual liver steatosis grading.