ESGAR 2014 Book of Abstracts

EDITOR S. Jackson (Plymouth/UK) ESGAR EXECUTIVE COMMITTEE PRESIDENT L. Martí-Bonmatí (Valencia/ES) PRESIDENT-ELECT C. Matos (Brussels/BE) VICE PRESIDENT S. Halligan (London/UK) SECRETARY A. Palkó (Szeged/HU) TREASURER S. Jackson (Plymouth/UK) PAST PRESIDENT F. Caseiro Alves (Coimbra/PT) EDUCATION COMMITTEE R.G.H. Beets-Tan (Maastricht/NL) MEMBERSHIP COMMITTEE T. Helmberger (Munich/DE) WORKSHOP COMMITTEE A. Laghi (Latina/IT) MEETING PRESIDENT G. Mostbeck (Vienna/AT) PRE-MEETING PRESIDENT Y. Menu (Paris/FR) PRE-PRE-MEETING PRESIDENT V. Valek (Brno-Bohunice/CZ) PAST MEETING PRESIDENT C. Ayuso (Barcelona/ES) MEMBERS AT LARGE P. Prassopoulos (Alexandroupolis/GR) J. Stoker (Amsterdam/NL) M. Zins (Paris/FR)

Purpose: A nonlinear compartment model analysis of gadoxetate disodiumenhanced MR imaging can quantify cellular uptake function as intrinsic clearance (CLint) which cannot be evaluated by extracellular MR contrast agents. The purpose of this study was to asses prediction capability of nonlinear compartment model analysis of gadoxetate disodium-enhanced MR imaging for histological grade of hepatocellular carcinoma (HCC). Material and Methods: Twenty-four consecutive patients with histologically proven HCCs who preoperatively underwent gadoxetate disodium-enhanced MR imaging were included. T1-weighted images with fat suppression at precontrast and 12 postcontrast enhanced phases were obtained using K-space weighted image contrast reconstruction technique. Time-intensitycurves (TICs) of abdominal aorta, main trunk of portal vein and HCC were obtained from region of interests on MR images. Pharmacokinetic parameters including distribution volume (Vd) at extracellular fluid space and CLint of gadoxetate disodium in HCC were determined by applying TICs into nonlinear compartment model. Linear discriminant analysis according to Vd and CLint was performed on histological grade of HCC. Results: The mean value and its 95% confidence intervals of area under ROC curve in distinguishing moderately and poorly differentiated HCCs from welldifferentiated HCCs by linear discriminant analysis were 0.772 and 0.768 to 0.777 for Vd alone, 0.795 and 0.791 to 0.800 for Vd and CLint, respectively. Conclusion: Separation of moderately and poorly differentiated HCCs from well-differentiated HCCs can be improved by CLint determined from nonlinear compartment model analysis of gadoxetate disodium-enhanced MR imaging.

SS 2.03
withdrawn by the authors SS 1.10 CT imaging aspects of early local recurrence of pancreatic adenocarcinoma after pancreaticoduodenectomy: a retrospective study 2004-2013 C. Balaj, C. Sellal, I. Petit, X. Orry, A. Ayav, V. Laurent; Vandoeuvre les Nancy/FR Purpose: To describe semiology and topography of pancreatic ductal adenocarcinoma early local recurrence after pancreaticoduodenectomy. Material and Methods: This retrospective study was realised in University Hospital of Nancy (France). Between January 2004 and September 2013, patients who underwent pancreaticoduodenectomy for ductal adenocarcinoma, preoperative CT less than 30 days before surgery and recurrent postoperative CT during 1 year after surgery, were included. After studying postoperative imaging, correlations with clinical, histological data and preoperative imaging were made. Results: Among 123 patients who underwent pancreaticoduodenectomy, 48 patients had sufficient follow-up imaging and were included. 33 had a local early recurrence (Group 1). 15 had no local recurrence (Group 2). Local recurrence consisted of 2 types of anomalies: tissue nodules on surgical clips (94%), and peri arterial invasion (82%). In preoperative imaging, tumor diameter (p = 0.02) and venous borderline resectable tumor (p < 0.001) are predictive of local recurrence. Healthy peri tumor parenchyma seems to be correlated with recurrence. Conclusion: Tissue nodules on surgical clips and peri arterial invasion characterize early local recurrence. This tumor dissemination has probably lymphatic and nervous origin, and is linked to the tumor spread type. Radiologist has an essential role in the therapeutic strategy because of imaging predictive factors and recurrence diagnosis. Moreover, neoadjuvant treatments impact, for downstaging and improving survival, must be evaluated.

SS 2.07
Will slow injection of Gd-EOB-DTPA enable to detect more hypervascular hepatocellular carcinoma at the multiple arterial phases? T.L. Harada, K. Saito, N. Yoshimura, Y. Araki, T. Funatsu, T. Saguchi, M. Kawasaki, K. Sugimoto, S. Akata, K. Tokuuye; Shinjuku-ku/JP Purpose: To intra-individually compare a low-tube-voltage (100 kVp) with a standard 120 kVp CT protocol regarding image quality and radiation dose during arterial phase in patients with hepatocellular carcinoma. Material and Methods: 68 patients with known hepatocellular carcinoma underwent abdominal dual-energy CT. 100 kVp and mixed 120 kVp images during the arterial phase were compared. Signal measurements in the abdominal aorta, portal vein, carcinoma and non-infiltrated liver parenchyma were compared. DLP of 100 kVp acquisition and the complete dual-energy CT arterial phase were compared. Three readers subjectively evaluated image quality, delineation, image sharpness and noise of both image series using a five-point Likert scale. Results: Average DLP of 100 kVp acquisition was reduced by nearly half compared to the complete dual-energy CT examination (189.6 vs. 382.9 mGy cm). Signal of the abdominal aorta, portal vein, hepatic parenchyma and carcinoma was significantly higher with low-tube-voltage technique (P < 0.001). Although readers preferred standard 120 kVp over 100 kVp acquisition regarding image quality (4.86 vs. 4.45), delineation (4.84 vs. 4.71), sharpness (4.87 vs. 4.46) and noise (4.88 vs. 4.26), all image series received a score of at least 4 out of 5. Conclusion: Low-tube-voltage arterial-phase CT at 100 kVp allows for significant radiation dose savings with a high subjective image quality for evaluation of hepatic malignancy and parenchyma in patients with hepatocellular carcinoma.

SS 2.04
Spectral CT and volumetric-iodine uptake in hepatocellular carcinoma E. Bozzi, I. Bargellini, O. Perrone, G. Lorenzoni, F. Turini, C. Bartolozzi; Pisa/IT Purpose: To assess image quality and diagnostic accuracy of monochromatic imaging from spectral CT and reproducibility of volumetric iodine uptake (VIU) in patients with hepatocellular carcinoma (HCC). Material and Methods: Forty-three patients with 108 HCC nodules underwent multiphasic CT with spectral imaging in the late arterial phase, to generate conventional 140-kVp polychromatic images (group A), 70-keV monochromatic images (group B) and iodine-based images (group C). Images were compared in terms of tumour-to-liver contrast-to-noise ratio (CNR), image quality (score 1-5), lesion conspicuity (score 1-3) and number of detected HCCs. VIU was calculated in the iodine-based images. Results: Mean CNR was significantly higher in group C (4.19±2.42) compared to group A (2.17±1.6) and group B (2.79±1.78) (P<0.0001). Mean scores for image quality and lesion conspicuity were 4.72±0.5 and 2.51±0.6, respectively, for group A, 4.74±0.5 and 2.83±0.4 for group B and 4.44±0.7 and 2.83±0.5 for group C; the highest number of HCC nodules were identified in group C (106/108; 98.1%) compared to group A (84.2%) and B (94.4%). Measurement of VIU was fast (mean post-processing time 92±80 seconds) and reproducible (k=0.99), with a mean iodine concentration within the segmented volume of 22.4±6.9 µg/cm 3 . Conclusion: In HCC patients, iodine-based images derived from spectral CT can improve lesion detectability and provide a reproducible quantitative parameter that reflects tumor neoangiogenesis.

SS 2.06
Prospective validation of liver-imaging reporting and data system in nodules smaller than 2 cm detected in cirrhotic patients A. Darnell,A. Forner,J. Rimola,M. Reig,J. Bruix,C. Ayuso;Barcelona/ES SS 3.02 Non-invasive differentiation between simple steatosis and steatohepatitis using gadoxetic acid-enhanced MRI in patients with non-alcoholic fatty liver disease: a feasibility study N. Bastati-Huber 1 , D.S. Feier 2 , C. Balassy 1 , A. Wibmer 1 , D. Tamandl 1 , H. Einspieler 1 , C. Kulinna-Cosentini 1 , A. Ba-Ssalamah 1 ; 1 Vienna/AT, 2 Cluj-Napoca/RO Purpose: To distinguish between simple steatosis and non-alcoholic steatohepatitis (NASH) in non-alcoholic fatty liver disease (NAFLD) patients using gadoxetic acid-enhanced MRI. Material and Methods: We examined 81 patients with suspected NAFLD with gadoxetic acid-enhanced 3-Tesla MRI. Patients were histopathologically classified as simple steatosis or NASH using the SAF score based on the semiquantitative scoring of steatosis, activity and liver fibrosis. The MR images were analyzed using the relative liver enhancement. Univariate and multivariate regression analyses were applied to identify variables associated with relative enhancement measurements and the performance of relative enhancement values was assessed for the differentiation between simple steatosis and NASH using the area under the receiver operating characteristic curve (AUC) analysis. Results: Testing the relationship between enhancement measurements and histology, relative enhancement values correlated strongly with lobular inflammation (r=-0.59, p=<0.001), ballooning (r=-0.44, p=<0.001), and fibrosis (r=-0.59, p=<0.001), but not with steatosis (r=-0.15, p=0.15). In multivariate analysis, all three histological parameters were independently associated with the relative enhancement. The measurements performed well for the differentiation between simple steatosis and NASH, with an AUC of 0.85, [cutoff=1.24, (sensitivity=97.1%, Specificity =63%)]. Conclusion: Gadoxetic acid-enhanced MRI may be used as a non-invasive diagnostic tool to distinguish between simple steatosis and NASH in NAFLD patients.

SS 3.07
The feasibility of texture analysis using susceptibilityweighted magnetic resonance imaging in detecting patients with liver cirrhosis D.S. Feier 1 , T. Knogler 2 , M.E. Mayerhoefer 2 , C. Balassy 2 , N. Bastati 2 , A. Ba-Ssalamah 2 ; 1 Cluj-Napoca/RO, 2 Vienna/AT Purpose: To establish the feasibility of textural features of liver parenchyma obtained on susceptibility weighted (SWI) MRI which will enable the detection of liver cirrhosis in patients with diffuse chronic liver diseases (CLD). Material and Methods: Sixty-five patients (mean age, 51.65 years; 60% males) with CLD and histologically proven liver fibrosis were included retrospectively. Fibrosis was evaluated according to the Metavir system and grouped in patients with (F4 Metavir, n=27, 42%) and without (F0-F3 Metavir, n=38, 58%) liver cirrhosis. All patients underwent 2D multislice breath-hold SWI sequence on a 3T MRI unit and the imagines were used for texture analysis. A region of interest (ROI) was manually defined. Texture features derived from the grey-level histogram, the co-occurrence matrix, the autoregressive model, and the wavelet transform were calculated. Fisher coefficients were calculated to determine which texture features were best suited for distinguishing between no-cirrhosis and cirrhotic liver tissue. The classification accuracy was used as the primary outcome measure. Sensitivity and specificity were used as secondary variables. Results: Six out of ten texture features selected on the basis of Fisher coefficients were derived from grey-level histogram. Of the 65 patients included, 62 (95.38%) were classified correctly by k-NN. Sensitivity was 96.3% and specificity was 94.7%. Conclusion: Texture features extracted from the grey-level histogram calculated form SWI MRI data are feasible to correctly identify cirrhotic changes in liver parenchyma of patients with CLD.

SS 3.08
Magnetization transfer ratio imaging of the liver: could it play a role in quantifying liver cirrhosis? M.H. Martens 1 , S. Noordzij 1 , D.M. Lambregts 1 , N. Papanikolaou 2 , L.A. Heijnen 1 , G. Koek 1 , R.G.H. Beets-Tan 1 ; 1 Maastricht/NL, 2 Heraklion/GR Purpose: The diagnosis of liver cirrhosis is an important diagnostic challenge. So far, imaging techniques have not proven successful in quantifying cirrhosis and invasive liver biopsy remains the golden standard. The contrast in magnetization transfer ratio (MTR) imaging is based on the presence of macromolecules. We hypothesize that cirrhosis contains many macromolecules and will show a higher MTR than normal liver parenchyma. Hence, the aim of this study was to evaluate if MTR is able to differentiate between liver cirrhosis and normal liver parenchyma. Material and Methods: Consecutive patients that have clinically indicated liver biopsy were recruited to perform a 3.0 Tesla MRI examination (multislice MECSh GRE sequence: TR/TE=10/0. 99, 1.69, 2.39, 3.09, 3.79, 4.49, 5.19, 5.89, 6.59, 7.29, 7.99, 8.69, msec), end expiratory phase breath-hold acquisition. Quantification for liver fat fraction (FF) was performed with dedicated software (with phase reconstruction and T2* correction) selecting a ROI within the biopsied liver segment. Liver biopsy was used as gold standard for steatosis grading (0-3). Results: Twenty-five patients followed the inclusion criteria. There were 19 patients with grade 0, 3 with grade 1, and 3 with grade 3 liver steatosis. High significant correlation was obtained between steatosis histological grade and the median of calculated fat fraction (ρ=0.741, p<0.001). Conclusion: Our preliminary results demonstrate a good correlation between non-invasive fat fraction quantification by MECSh GRE MRI sequence with phase reconstruction and T2* correction, and liver biopsy. If this is validated in larger cohorts, this MRI sequence could be used as a fast, non-invasive and quantitative liver fat imaging biomarker.

SS 3.04
Comparison of the controlled attenuation parameter values and 1H-MR spectroscopy determined liver fat percentages in patients with NAFLD or NASH: interim results J.H. Runge, A.J. Nederveen, J. Verheij, U. Beuers, J. Stoker; Amsterdam/NL Purpose: In non-alcoholic fatty liver disease (NAFLD) and non-alcoholic steatohepatitis (NASH) patients, MRI now allows non-invasive, quantitative fat measurement. Due to MRI's costs and relative burden, other methods are soughtafter. In this study, we compared the new controlled attenuation parameter (CAP) on the FibroScan® (FS) with 1 H-MR spectroscopy ( 1 H-MRS) determined fat-% in patients with clinical suspicion of NAFLD or NASH and liver biopsy. Material and Methods: Consecutive adult NAFLD/NASH patients were seen within 6 weeks of biopsy for a ≥8 hr fasting FS and 1 H-MRS measurement. From March 2013 up to January 2014, 16 subjects were included: one had to be excluded as FS was not possible. 1 H-MRS was performed at 3T with a 5-echo STEAM sequence (TE/ΔTE/TR: 10/5/3500ms) in a single, 21s breath hold, allowing for individual T2-correction. CAP-values and fat-% were compared using Spearman's correlation coefficient (rS). Results: Data of 15/16 subjects were eligible. Median (IQR) age and BMI were 53.5 (49.3-57.3) and 26.8 (26.1-30.5), respectively. Median CAP-value in dB/m was 303 (260-337). Median fat-% was 11.9% (4.2-22.3%). rS between CAPvalues and fat-% was 0.91 (95%-CI: 0.73-0.97, P<0.0001). Conclusion: The CAP-value can quantitatively measure liver fat, showing near perfect correlation with 1 H-MR spectroscopy as reference standard. The results thus far appear to support the more widespread use of the CAP-value for liver steatosis assessment, especially as the FibroScan allows simultaneous elasticity measurement.

SS 3.05
Reliability and repeatability of diffusion tensor imaging of the liver R. Girometti, M. Maieron, G. Lissandrello, L. Cereser, C. Zuiani, M. Bazzocchi; Udine/IT Purpose: To evaluate the reliability and repeatability of diffusion tensor imaging of the liver (LDTI) quantitative results. Material and Methods: Ten healthy volunteers (median age 23 years) underwent LDTI on a 3.0 T magnet in two imaging sessions separated by 2 weeks (session-1/-2, respectively). A single-shot Echo-planar sequence with 16 non-collinear diffusion gradient directions and b-values of 0, 1000 sec/mm 2 was used. By averaging multiple measurements, two radiologists in consensus assessed liver apparent diffusion coefficient (ADC) and fraction of anisotropy (FA) values on ADC-and FAmaps at four reference levels, namely: right-upper-level (RUL), right-lower-level (RLL), left-upper-level (LUL) and left-lower-level (LLL), respectively. We then performed: (i) analysis of reliability/repeatability, by calculating the coefficient of variation (CV%) of ADC/FA values on intra-and inter-session basis; (ii) Bland-Altman analysis, to assess the limits of agreement between inter-session measurements. Results: Intra-session CVs% were 9.51% (session 1) and 9.48% (session 2) for ADC, and 12.93% (session 1) and 11.48% (session 2) for FA, respectively. When comparing RUL, RLL, LUL and LLL on an inter-session basis, we showed CVs% of 2.43%, 4.01%, 4.09% and 3.98% for ADC values, and 5.20%, 4.91%, 7.00% and 5.16% for FA values, respectively. We observed limits of agreement no larger than -0.15 to 0.23 x10 -3 mm 2 /sec (ADC values) and -0.05 to 0.07 (FA values Purpose: MRI allows accurate, reproducible, non-invasive measurement of the liver iron concentration (LIC) and is preferable over biopsy. Conversely, serum markers assess whole body iron content, can thus serve as screening tools but may be influenced by concurrent inflammation or infection. As the serum marker ferritin is an acute phase protein, it may be raised without reflecting actual increased LIC. Few studies have thus far compared MRIdetermined LIC with ferritin and non-acute phase (protein) markers such as transferrin and transferrin-saturation. Material and Methods: All MRI LIC-measurements performed between January 2009 and December 2012 in our centre were included retrospectively. The Gandon method was applied to derive the mean LIC over three slices. Serum iron, transferrin, transferrin-saturation and ferritin levels respectively were noted when available. Spearman's correlation coefficient (rS) was used to compare the serum levels with the LIC value. Results: A total of 99 baseline MRI-scans were performed. LIC values could be determined accurately in 82 cases. Ferritin and transferrin correlated substantially with LIC at rS=0.72 (P<0.0001, n=69) and rS=-0.69 (P<0.0001, n=41). These correlations did not differ significantly (P=0.77). Iron and transferrin-saturation showed lower but significant correlations with LIC. Conclusion: Correlations between transferrin and ferritin with MRI-determined LIC did not differ significantly. Based on these results, no preference can be given to either ferritin or transferrin to screen for increased liver iron content.

Material and Methods:
Eighteen patients with proven liver cirrhosis were compared to 5 healthy volunteers. All subjects received a standard liver MRI including one axial slice of the MT sequence at the center of the liver. Regions of interest (ROI) covering the liver parenchyma (excluding large vessels and focal liver lesions) were drawn on the MT-MRI to calculate MTR. Mann-Whitney U test was used to compare the MTR of the liver parenchyma between cirrhosis patients and controls. Results: ROI size ranged between 19.0 and 173 cm 2 . The mean MTR of the cirrhosis group was 53.8% and of the control group was 23.1%. This difference was statistically significant with p<0.001. Conclusion: The magnetization transfer ratio could play a role in the diagnosis of liver cirrhosis. Further research with a larger patient cohort is recommended.

SS 3.09
High field (3T) MR evaluation of cirrhotic patients: comparison of two different injection rates of hepatospecific contrast agent V. Battaglia 1 , A. Grigolini 1 , F. Signorini 1 , G. D'Ippolito 2 , C. Bartolozzi 1 ; 1 Pisa/IT, 2 Sao Paulo/BR Purpose: To evaluate if different injection rates of the hepatospecific contrast agent (Gd-EOB-DTPA) in a selected group of cirrhotic patients determines alterations of liver enhancement in early arterial and arterial phase. Material and Methods: Eighty cirrhotic patients (M-F: 60:20; mean age: 62 ys; range: 41-82 ys) underwent high-field (3T) MR examination. All patients were within Milan criteria, without signs of portal thrombosis (main branch and segmental branches); hepatic function was preserved. Standard study protocol presupposed a fixed time delay for the acquisition of early arterial and arterial phases after the injection of hepatospecific contrast agent (Gd-EOB-DTPA), administered by automatic injector. In 48/80 patients, the contrast agent injection flow rate was 3 ml/sec (group 1); in 32/80 patients, the injection flow rate was 1.5 ml/sec (group 2). For each MR study, in both early arterial and arterial phase, the parenchymal enhancement values were calculated, by means of evaluation of ER (enhancement ratio); in all cases, a standard (150 mm 2 ) ROI (region of interest) was positioned in pre-definite segments. Values obtained in the two groups were compared by means of two-tailed Student's T test. Results: Means (±SD) of parenchymal enhancements on early arterial and arterial phases in 48/80 patients were 0.12 (±0.17) and 0.58 (±0.19); in group 2 were 0.13 (±0.14) and 0.67 (±0.22) (p>0.01). Conclusion: Different injection rates of hepatopecific contrast agent do not influence the parenchymal enhancement in cirrhotic patients in none of arterial phases.

SS 3.10
Interobserver agreement of Gandon-and R2*-methods for liver iron concentration measurement J.H. Runge, E.M. Akkerman, A.J. Nederveen, J. Stoker; Amsterdam/NL Purpose: Liver iron concentration (LIC) can be measured with a number of MRI-methods. As thus far no reasonably sized study has been performed comparing the interobserver agreement of the Gandon-and R2*-methods, we aimed to compare these. Material and Methods: Baseline MRI LIC-measurements performed between January 2009 and December 2012 in our centre were included retrospectively. The Gandon-, and R2*-methods were used to derive LIC in μmol/g and R2* in s -1 , respectively. To assess interobserver variability, two readers independently drew regions-of-interest (ROIs). LIC and R2* values of both reades were assessed using intraclass correlation coefficients (ICCs) and Bland-Altman analysis derived repeatability indices. Results: Gandon and R2* data were acquired in 77 patients. Median (IQR) LIC and R2* for reader 1 were 90 (32.5-220) μmol/g and 44 (105-279) s -1 , respectively. ICCs (95%-CI) between reader 1 and 2 for LIC and R2* were 0.995 (0.992-1.000) and 1.0 (1.0-1.0), respectively. Repeatability indices (95%-CI) for LIC and R2* were 5.14% (4.11-6.17%) and 0.81% (0.65-0.97%), respectively. Conclusion: Interobserver agreement in terms of ICC and repeatability indices for both Gandon and R2* methods were excellent. Given that R2* measurements can be performed in a single breath hold and are sensitive for iron concentrations that are above the Gandon threshold, R2* seems preferable.

Scientific Sessions
Authors' index  (GA). The purpose of our study is to evaluate T staging and diffuse type of Lauren's classification as surrogate endpoints (SE) to improve N staging of GA by MDCT. Material and Methods: Between January 2010 and July 2013, 102 patients were retrospectively evaluated according with following criteria: endoscopic diagnosis of GA, gastrectomy associated with retrieval of at least 15 locoregional nodes and state-of-the-art MDCT (three cases with complementary MRI) performed less than 60 days before surgery. Staging was performed in consensus by two radiologists, considering two steps.
Step one (S1): N staging by imaging (NSI) versus N staging by pathology.
Step two (S2): the same with adding one level to NSI in patients with T4 staging by imaging or with diffuse type of Lauren's classification. Absolute agreement and effectiveness analysis of diagnosing stage N2 or higher were compared. Results: Absolute agreement raised from 49% to 60.7% from S1 to S2 (Chisquare p-value 0.091, not statistically significant). From S1 to S2: sensitivity and NPP raised 0,711 to 0.844 and 0.812 to 0.889 respectively, specificity and PPP lowered 0.982 to 0.883 and 0.970 to 0.905, respectively. Conclusion: Aforementioned SE are useful to increase absolute agreement of NSI in GA and increased NPP of NSI N2 or higher. With further SE statistically significant results may be possible.

SS 4.04
Accuracy of radiological staging in identifying poor prognostic factors in colon cancer G. Gossedge, A. Hood, S. Sook-Cheng Chin, L. Saraswat, H. Lambie, R. Hyland, D. Jayne, D.J.M. Tolan; Leeds/UK Purpose: Within the context of the FOxTROT trial, CT can identify 'poor prognosis' colon cancer patients with a sensitivity of 78% and a specificity of 67%. We aimed to undertake a 'pragmatic' study of the diagnostic accuracy of routine radiological staging in colon cancer following completion of the pilot phase of FOxTROT, at a single institution. Material and Methods: Consecutive primary colon cancer patients, discussed at MDT between October 2011 and October 2013, had their pre-operative CT images evaluated according to TNM staging, prior to surgery. Following surgery, radiological staging was correlated with histopathological staging (TNM 5) as the reference standard. Patients with synchronous tumours and distant metastases were excluded. Patients were stratified according to 'good' (T2 or less, N0, and absence of extramural venous invasion (EMVI)) or 'poor' prognosis (T3 or more, N1 or more, or presence of EMVI). Results: Of 172 cases, the sensitivity, specificity, positive and negative predictive values for stratification by CT compared with histological examination were 82% (95% CI: 75-88), 76% (95% CI: 54-90), 95% (95% CI: 90-98) and 42% (95% CI: 27-57), respectively, with a diagnostic odds ratio of 14.73. Accuracy for detecting malignant lymph nodes and EMVI was 68% and 69%, respectively. R0 resections were performed in 92% of cases. Conclusion: Outside the context of a clinical trial, radiological staging using CT is accurate and feasible and can stratify patients for neo-adjuvant therapies.

SS 4.05
Colon cancer lymph node metastases: where are we going wrong? R. Wiles, E. Hall, P. Healey; Liverpool/UK Purpose: Accurately diagnosing lymph node metastases in colon cancer is notoriously difficult. The study aim is to determine our accuracy in nodal staging and, when we overstage, what features of the nodes contribute to this. Material and Methods: Preoperative CT scans of 50 patients with histologically proven colon cancer were independently reviewed by two radiologists (two cases were later excluded due to non-visualisation of the tumour). Data recorded included: suspected number of involved nodes, short axis length, shape and character of the most suspicious node. Histological findings were compared with the recorded data. Results: Reader 1 and 2 agreed on lymph node staging in 62.5% of cases. Reader 1 correctly staged 45.8% of cases, overstaged 47.1% and understaged 12.5%. Values for reader 2 were 52.1%, 29.2% and 18.8%, respectively. In 15 cases, one or both reader overcalled when histological N=0. The readers agreed in 10 of these cases (67%). Nodal characteristics in these cases were

Shanghai/CN
Purpose: When considering local excision or 'wait-and-see' after CRT in rectal cancer, nodal status is crucial. In spite of a good tumour response, any remaining N + -nodes harbour a risk for recurrence. Knowledge on the presence and location of these remaining N + -nodes may help guide treatment intensification (e.g. boost radiotherapy). Aim of this study was to investigate patterns in the location of remaining N + -nodes in good responders after CRT. Material and Methods: 211 locally advanced rectal cancer patients underwent CRT, which resulted in a good response (downstaging to ypT0-2) in 134 patients, who constituted the final study group. For the patients with a ypT0-2N + status, a detailed lesion-by-lesion comparison between restaging-MRI and histology was performed on the location of the individual N + -nodes. Results: 8/134 patients (6%) had a ypT0-2N + status. 47% of the N + nodes was located at the same height as the tumour, 53% was located at a distance of 1.2-6.5 cm above the tumour level (of which 40% within <5 cm from the tumour and 60% >5 cm above). In the axial plane, 71% of the nodes was located near (within <1 cm from) the tumour/rectal lumen. Conclusion: 1. The incidence of remaining N + -nodes in case of good tumour response is very low. 2. Remaining N + -nodes are solely located at the same level or proximal to the tumour. 3. The majority of N + nodes are located near the tumour/lumen.

SS 4.09
How accurate is MRI in predicting a positive/negative surgical margin in patients with recurrent rectal cancer D. Mondal, R. Hyland, H. Lambie, N. Scott, P. Sagar, D.J.M. Tolan; Leeds/UK Purpose: Recurrent rectal cancer has a poor prognosis and curative resection is associated with significant morbidity. MRI detects the site and extent of local recurrence. It may also be able to predict a positive surgical margin, to allow optimal selection of patients for surgical resection. This study assesses whether MRI predicts involvement of the surgical margin in patients with recurrent rectal cancer. Material and Methods: Patients with recurrent rectal carcinoma being considered for surgery in a national recurrent rectal cancer centre were studied over a 7-month period from January 2013. Radiology was reviewed and data collected including site(s) of recurrence, pelvic structures involved and predicted positive/negative surgical margin. Pathological reports were assessed for correlation. Results: 51 patients were eligible. 21/50 patients proceeded to surgical resection; one was excluded with no pathology specimen for correlation. In 17/21 patients, radiological prediction of a positive or negative surgical margin was confirmed. In 4/21 patients, radiology and pathology did not match: one patient had a long time interval (5 months) between MRI and surgery; 2/21 patients had post-chemoradiotherapy complete tumour response with fibrosis and scan to surgery interval of >3 months; in 1/21 patients, a positive margin in the surgical specimen was not predicted on MRI. MRI assessment in recurrent rectal cancer has sensitivity 88.2%, specificity 50%, PPV 88.2% and NPV 50%. Conclusion: MRI can pre-operatively predict margin status in recurrent rectal cancer.

SS 4.10
withdrawn by the authors reviewed. 96% of overcalled nodes measured >5 mm. 88% had at least one feature of round shape, irregular border or short axis diameter >9 mm. 56% had two of these features and 4% had all three. Conclusion: In our practice, as with other studies, nodal staging is inaccurate using CT. Overcalling was more common than undercalling nodal metastases, but both occurred. Abnormal nodal shape, outline or size does not accurately predict metastatic involvement.

SS 4.06
Transrectal ultrasound and 3-Tesla magnetic resonance imaging in preoperative staging of rectal cancer: a performance comparison E. Guidi 1 , R. Scandiffio 1 , L. Faggioni 1 , R. Balestri 1 , G. D'Ippolito 2 , P. Buccianti 1 , C. Moreover, TRUS and MRI were able to demonstrate the levator ani muscles infiltration with an overall agreement of 78% and the lymph nodes involvement with an accordance of 76%. MRI allowed, however, the evaluation of other very important features for the correct staging of rectal cancer, in particular the distance between lesion and mesorectal fascia.

Conclusion:
The excellent agreement between MRI and TRUS in preoperative staging of rectal cancer argues in favor of the use of MRI, because it also allows a more comprehensive local assessment.

SS 4.07
Initial staging of locally advanced rectal cancer and regional lymph nodes: All the patients underwent a whole body 1.5T DWI MRI (b0, b1000) and a whole body FDG-PET/CT within the same week. FDG-PET/CT was considered as the standard reference. In consensus, two independent readers measured maximum and mean FDG standardized uptake values (SUVmax and SUVmean) of the rectal tumour and of the pathologic regional lymph nodes on PET/CT and compared these values to minimum and mean values of apparent diffusion coefficient (ADCmin and ADCmean), displayed on maps generated from DWI.

Villejuif/FR
Purpose: To develop a small bowel CD magnetic resonance index of inflammation severity (CDMR-IS). Material and Methods: Thirteen centres participated in this prospective transversal study. Selected MR-DVDs (6 per centre) were blinded and allocated to others centres. A combined reading was conducted by a pair of radiologist and gastroenterologist in each centre. Total length and number of 20 cmdiseased segments were noted. In each segment, following data were recorded: wall thickness, T2 wall hypersignal, T1 degree of intensity and pattern of enhancement, deep ulceration, comb sign, sclerolipomatosis, lymph nodes, fatty infiltration, fistula and abscess. For each MR examination, global inflammation severity (GIS) was evaluated qualitatively using a linear analog scale. To construct the CDMR-IS, multivariate linear mixed model was used with forward selection through likelihood ratio test. Results: 438 readings were analysed. Mean of GIS was 20.7. The independent predictors of the GIS were the number of segments (n) with the following independent predictors : T1 mild to moderate intensity of enhancement (MM), T1 severe intensity of enhancement (S), deep ulceration (DU), comb sign (CS), any fistula (F), and abscess (A). This leads to build the formula: CDMR-IS=2*n MMT1+3*nST1+nDU+nCS+3*nF+4*Na.

Conclusion:
The CDMR-IS is a standardized index to evaluate the severity of inflammation in patients with small bowel CD. Further studies are mandatory for validation of CDMR-IS in new patients, and to study its sensitivity to change after anti-inflammatory treatment.

SS 5.10
Correlation between morphological expansion and impairment of motility in inflammatory small bowel lesions in patients with Crohn's disease: preliminary data S. Bickelhaupt beginning of specific anti-inflammatory treatment. In each patient, the variation (Δ = before -after 6 weeks of therapy) of the area under the time-intensity curve during the wash in, the peak enhancement, and the time to peak enhancement were related to the therapeutic outcome assessed after 12-18 weeks from the beginning of pharmacologic treatment by CDAI and/or endoscopy. 0T platforms, providing a same-day stool sample for fC assay. Enteric signal intensity on high B value DWI images was scored (0-normal 1-probably normal, 2 probably abnormal 3-definitely abnormal by 2 radiologists in consensus. The extent of any DWI abnormally was then classified into small bowel alone (<10 cm or >10 cm), colon alone or both small bowel and colon. fC was compared according to DWI signal intensity and disease extent using Wilcoxon rank sum and Kruskal Wallis analysis of variance. Results: Mean fC score was 398 ug/g (range 0-1970 ug/g), and raised (>120 ug/g) in 44/69 patients. In patients scored probably or definitely abnormal on DWI, fC was significantly higher than in those scored normal or probably normal (492±422 ug/g vs 159±257 ug/g, p<0.0001). DWI had sensitivity 84%, specificity 52% for raised fC. There was a significant difference in fC according to DWI defined disease extent, being lowest in isolated small bowel disease <10 cm and highest in patients with both colonic and small bowel involvement. Conclusion: Qualitative assessment of diffusion MRI is sensitive to the severity and extent of Crohn's disease activity against a global reference standard of calprotectin.

Scientific Sessions
Authors' index SCIENTIFIC SESSIONS / FRIDAY, JUNE 20, 2014 Material and Methods: MR examinations of 39 patients with histology-proven pancreatic neuroendocrine tumors were retrospectively evaluated. Both a qualitative analysis (appearance, site, main pancreatic duct dilation, signal intensity) and a quantitative analysis (dimensions, degree of main pancreatic duct dilation, ADC value) were performed. Results: 69% of lesions were solid, mainly located in the pancreatic head (46%). Mean dimension was 31 mm; mean main pancreatic duct dilation was 7 mm. 79% of lesions did not determine ductal dilation. 74% of lesion were hypointense on T1-weighted images, 56% were hypointense on T1-weighted fat-suppressed images, 57% were hyperintense on T2-weighted fatsuppressed images; 59% of lesions were isointense in pancreatic phase images, 48% were hyperintense during the portal phase. Regarding DWI, 78% of lesions were hyperintense in b=800 images; the majority of T2-isointense lesions showed instead hyperintensity in b=800 images; the majority of isointense pancreatic-phase lesions were hyperintense in b=800 images. Mean ADC value was 1.38±0.2×10 -3 mm 2 /s. Conclusion: DWI depicts a higher number of lesions in comparison with T2weighted fat-suppressed and pancreatic-phase images, being therefore more sensitive for the identification of pancreatic neuroendocrine tumors.

SS 6.03
Patterns of contrast accumulation for differential diagnosis pancreatic solid tumors with homogeneous contrast enhancement at MDCT N.Y. Makeeva-Malinovskaya, G.G. Karmazanovsky; Moscow/RU Purpose: To evaluate the capabilities of contrast-enhanced MDCT for the differential diagnosis of pancreatic solid tumors with homogeneous accumulation of contrast agent. on a 1.5 T MRI scanner. Small bowel motility was examined using a dedicated MR-motility assessment software (Motasso). Motility patterns (contraction frequency, relative occlusion rate and mean diameter) were assessed in correlation to wall thickness, length and prelesionary dilatation of the lesions. Statistical analysis was performed by calculation of the Pearson's-Correlation coefficient. Conclusion: Global motility is not inhibited by inflammatory activity in CD contrasting previous published pilot data. However, a small but significant negative correlation between fC and variance in the motility score suggests inflammatory burden may derange motility coordination, possibly reducing the magnitude of normal intrinsic regional variation.

Material and Methods:
Contrast-enhanced MDCT has been done for 74 patients with solid pancreatic tumors. Delta was calculated as difference of the density of tumor and pancreas for all phases of the examination. The calculation formula for the Delta is: T(p)-P(p), for p= {p1,p2,p3,p4}, T -the absolute value of the relative density of the tumor, P -the absolute value of the relative density of the pancreas, p1-pre contrast phase, p2-arterial phase, p3portal phase, p4 -delayed phase. Results: 24 patterns of Delta were identified, top eight patterns amount 73% of patients (n=54). All tumors were segmented into two classes by accumulation of contrast agents in tumor versus pancreatic parenchyma in the arterial phase: 1. "Adenocarcinoma" -accumulation in tumor is lower than in pancreatic parenchyma. 2. "Not adenocarcinoma" -accumulation in tumor is higher than in pancreatic parenchyma. "Adenocarcinoma" were detected for 65% of patients (n=35). Morphological types of the tumors: adenocarcinoma 85.7% (n=30), neuroendocrine neoplasms 8.6% (n=3), solid pseudopapillary tumor (SPT) 5.7% (n=2). "Not adenocarcinoma" were detected for 35% patients (n=19). Morphological types were: adenocarcinoma 5% (n=1), neuroendocrine neoplasms 80% (n=15), SPT 5% (n=1) Focal liver lesions -non-cirrhotic patient

SS 7.01
To compare the per-lesion sensitivity of detection of focal hepatic lesions before and after administration of hepatocyte specific gadolinium chelate (gadobenate dimeglumine) using diffusion-weighted imaging K. Ganesan, B. Nanda; Mumbai/IN Purpose: To compare the lesion-to-liver signal intensity ratio on DW images acquired before and after intravenous administration of hepatocyte specific gadolinium chelates. Material and Methods: 93 patients underwent MR imaging at 3 T for the evaluation of focal hepatic lesions. In these patients, we assessed 93 focal hepatic lesions: 39 HCC, 21 hemangioma, 10 metastasis, 7 cholangiocarcinoma and 16 other focal liver lesions including the following (cysts, FNH, combined neoplasms and granulomas). All patients underwent free breathing DW imaging (b=0 and b=500) before and 5-minutes after intravenous administration of gadobenate dimeglumine (Gd-BOPTA, MultiHance). Results: Statistical comparison was performed using paired two-sided student t-tests and p=0.05 as threshold for significance. In our study, there was no significant difference in the signal intensity of the lesion at b0 and b500 sec/mm 2 values before and after contrast administration. However, the signal intensity values of liver at b0 and b500 sec/mm² following contrast adminsitration were significantly decreased (p<0.001). Similarly the mean ADC and mean ADC ratios of the lesion to liver were decreased significantly (p<0.001) after contrast administration compared to the precontrast values. This study demonstrated increased conspicuity of the focal liver lesions on the post-contrast DW images.

Conclusion:
In this study, we see a trend that supports our hypothesis that DW imaging following intravenous administration of hepatocyte specific gadolinium-based contrast is more sensitive in detecting focal liver lesions vis-a-vis precontrast DW imaging.

SS 7.02
Hypertrophic scar in small focal nodular hyperplasia evaluated with hepatospecific MR contrast media (Gadoxetic acid): a diagnostic challenge? A. Colleoni, L. Romanini, P. Tessitore, M. Orsatti, F. Laffranchi, L. Grazioli, R. Maroldi; Brescia/IT Purpose: The differential diagnosis of small FNH (<3cm) with hypertrophic scar, hypointense in the portal and delayed phase after hepatospecific MR contrast media (Gd-EOB-DTPA) could be difficult, especially in oncologic patients. The enhancement pattern of small FNH with hypertrophic scar was therefore evaluated. Material and Methods: In this retrospective review from 2007 to 2013, we selected 19 liver lesions, of more than 300 nodules characterized as FNH by MR, smaller than 3 cm with hypertrophic scar (>50% of lesion volume). T2w, DWI and not-enhanced T1w signals, the enhancement pattern after Gd-EOB-DTPA in the dynamic study and in the hepatospecific phase (20') of central scar and parenchymal component were evaluated.

Results:
The scar was hyperintense on T2w and hypointense on not-enhanced T1w sequences in 84% of cases, showed intense arterial enhancement in 79% of cases, venous washout in 67%, hypointensity at the equilibrium and hepatobiliary phases in all the cases. The thin, peripheral, parenchymal component around the scar was isointense on T2w (74%) and T1w (53%); hyperintense in all dynamic phases, with typical "rim-like" pattern in the hepatospecific phase (100%).

Conclusion:
The hypointensity of the hypertrophic scar during the dynamic study with Gd-EOB-DTPA could be misinterpreted, the hyperintensity of the parenchymal rim in hepatobiliary phase has to be the feature to characterize these small FNHs. Nottingham/UK Purpose: The aim of this study was to measure the accuracy of CT and MRI scans in detecting colorectal liver metastases (CRLM) and to determine if patients staged with MRI in addition to CT have a longer liver recurrence-free survival than those staged with CT alone in a UK tertiary referral centre performing routine intra-operative ultrasound.

Material and Methods:
Eight patients with FNH and four with HCA (>2cm) underwent DCE-MRI using breath-hold T1w gradient echo sequence with keyhole imaging and profile sharing after injection of 0.1 ml/kg body weight Gd-BOPTA (Multihance) followed by 20 ml saline injection at 3 ml/s. Rigid liver registration was used. New software was used (OleaSphereTM, Olea Medical SA, La Ciotat, France) in which identical ROIs were placed in parametric maps of Ktrans and Ve for each lesion. This resulted in quantification of Ktrans and Ve using the extended Tofts model. Also, the time intensity curves (TICs) were classified in different types. Ktrans and Ve data were analyzed with independent t-test. Results: The means of Ktrans and Ve of both groups were not significantly different (t=0.711, p=0.493 and t=-1.316, p=0.218, respectively). Seven of eight FNH-TICs and three of four HCA-TICs could be classified as type 1 curve, one of four HCA-TICs as type 2 curve and one of eight FNH-TICs was undefined. Conclusion: No significant difference between the two lesion types could be found based on analysis of Ktrans and Ve. The relative frequence of type 2 curve in HCA might help as an additional tool for differentiating HCA from FNH.

SS 7.04
Subtyping of hepatocellular adenomas using Gd-EOB-DTPA: a qualitative and quantitative analysis S. Kreimeyer, T. Longerich, M. Kieser, L. Grenacher; Heidelberg/DE Purpose: To evaluate if a differentiation between subtypes of hepatocellular adenomas is possible using a hepato-biliary specific contrast agent (Gd-EOB-DTPA) in MRI. Material and Methods: 11 patients with 39 lesions with histologically proven hepatocellular adenomas were evaluated. 34 were inflammatic adenomas (IHCA) and 5 HNF1α adenomas. No β-catenin mutated adenoma was found.
We performed a standard protocol following the guidelines of the international consensus conference of Gd-EOB-DTPA using a 1.5T scanner (Somatom Avanto, Siemens). Beside a qualitative analysis of all sequences, we measured quantitative SI-ratio (signal-intensity ratio: ((SILE -SILVE) / SILVE) * 100) and LLCratio (liver to lesion contrast: (SILE -SILVE) / SIM). Results: Qualitative analysis shows that the best sequences for a differentiation of HNF1α adenomas from IHCA were T1w precontrast (p=0.03), portalvenous phase (p<0.0001) as well as the arterial phase (p=0.002). All adenomas were hypointense in the hepatocyte specific phase (15 min). The quantitative analysis of SI-ratio and of LLC-ratio also show statistically significant differences in T1w precontrast (SI: p=0.039 / LLC: p=0.049) and in portalvenous phase (SI: p=0.002 / LLC: p=0.002) images. Conclusion: Subtyping of hepatocellular adenomas using Gd-EOB-DTPA is possible based on the qualitative and quantitative analysis of the T1w precontrast and portalvenous phase. Beside the SI-ratio and liver to lesion contrast ratio the arterial phase gives additional qualitative information for the differentiation.

SS 7.05
Preoperative assessment of colorectal liver metastases To compare the potential of CT texture analysis, CT tumor volumetry and unidimensional (RECIST) measurements for response assessment of colorectal liver metastases undergoing preoperative chemotherapy. Material and Methods: Twenty-one patients with known colorectal liver metastases who underwent CT (portal phase) both before and after preoperative chemotherapy were included. Texture analysis of the largest metastatic lesion was performed. Mean intensity (M), entropy (E) and uniformity (U) were calculated with no filtration (nf) and using different filter values (fine=0.5, medium=1.5, coarse=2.5). Total tumor volume (cm 3 ) of all metastatic lesions and the largest lesion diameters (RECIST 1.1) were determined. Pre-, post-chemotherapy and the relative change (Δ%) of texture parameters, volume and RECIST-diameters were compared between good responders (n=9; tumor regression grade 1-2) and poor responders (n=12; TRG 3-5). ROCanalyses were performed to establish diagnostic performance for the significant parameters. Results: Pre-chemotherapy texture parameters were not significantly different between good and poor responders. Post-treatment Enf, Unf as well as the Δ%-Enf,0.5,1.5 and Unf,0.5,1.5 were significantly different between good and poor responders (p=0.003-0.04), with AUCs ranging between 0.73 and 0.88. Δvolume was significantly different between good and poor responders (p=0.047, AUC=0.77). No significant differences were found for pre-and postchemotherapy volumetry and RECIST-diameters. with resectable AGC, clinical stage >/= T2 N1, who had been treated with neoadjuvant chemotherapy and gastric resection with D2 lymph nodal dissection, were prospectively enrolled in this study. Pre and postchemotherapy abdominal CT were analysed by two radiologists in consensus, measuring the Dmax and depth invasion of the primary gastric lesion. Lymph nodal status were also evaluated, according to the III Japanese Classification of Gastric Carcinoma, using two dimensional criteria: a) the measurement of the short axis, with a cut-off of 5 mm for station in group 1, and 8 mm for station in group 2 and 3, and b) the calculation of the reduction rate between the two CT examinations. The sensitivity, specificity, and overall accuracy for both T and N staging of AGC were calculated.

Results:
The overall accuracy of MDCT in the T staging was 77.7% (14/18). Regarding the N staging, the dimensional criteria based on the evaluation of reduction rate was more accurate in comparison with the other one. Conclusion: MDCT may be an accurate tool in the prediction of pathologic response following neoadjuvant chemotherapy in patients with resectable AGC, in spite of the problems related to fibrous changes after chemotherapy.

SS 9.04
Ultrasound scan in diagnosing acute appendicitis Y. Nakhuda 1 , Z. Al-Ani 2 , A. Tokala 1 , I. Harris 1 ; 1 Preston/UK, 2 Manchester/UK Purpose: To assess the accuracy of ultrasound scan (USS) in the assessment of patients presenting acutely with suspected appendicitis. The influence of USS findings on patient's management in terms of reimaging, conservative treatment or surgical intervention.

Material and Methods:
Retrospective study within the settings of a teaching hospital and included patients investigated with USS for suspected acute appendicitis over a period of 10 months. All the USS were performed by specialist gastrointestinal radiologists. Patients who were managed conservatively were followed up for a minimum of 5 months. Ethical approval obtained. Results: A total of 142 patients had an USS for suspected acute appendicitis. USS showed no evidence of appendicitis in 58% (USS negative group) while suggested acute appendicitis in 29% (USS positive group). In 10% of patients, USS findings were equivocal (USS equivocal group). An alternative diagnosis was suggested by USS in 3%. The specificity and sensitivity for diagnosing acute appendicitis using USS were 96.4% and 92.5%, respectively, with 95% confidence interval. Final outcomes for these 3 groups will be highlighted. Conclusion: USS is highly sensitive and specific in the hands of gastrointestinal radiologists for assessment of suspected acute appendicitis and recommending management in terms of surgery vs conservative approach. When USS is equivocal, an integrated approach between radiologists and surgeons is necessary.

SS 9.05
Mandatory imaging cuts costs and reduces unnecessary surgery in the diagnostic work-up of patients suspected for appendicitis M.

Material and Methods:
We searched our reporting database for comments on mesenteric pathology over 10 years. We reviewed the images and graded the mesenteric findings using various criteria; e.g. density of stranding, lymph nodes dimensions and morphology. We added to this to non-imaging findings such as systemic symptoms and biochemical markers. This grading was correlated with the outcome found through electronic patient records and follow-up imaging. We have created a diagnostic algorithm which provides follow up recommendations.

Results:
The majority of mesenteric abnormalities will represent benign aetiology. Only 15% had significant pathology; lymphoma or metastatic cancer the most common of these. The remaining patients had no follow up or stable or resolving findings on repeat CT. Our algorithm provides a practical tool which objectively graded risk and triaged patients into appropriate management pathways. Conclusion: The majority of patients will have a benign aetiology underlying the mesenteric findings, however significant pathology can be overlooked. Our algorithm provides a useful tool which grades risk and guides radiologists towards appropriate follow up recommendations.

SS 9.09
Straight to surgery! When contrast-enhanced ultrasound can spare blunt abdominal trauma patients the delay of an emergency preoperative CT scan D.D. Cokkinos All CTA images were wirelessly transferred on the iPad in JPEG losslessformat using the BonjourTM protocol. The time needed to complete reading of every CTA examination on the iPad and the workstation, detection of arterial blush, and suspected bleeding arteries as assessed on both devices were recorded.

Results:
The time needed to complete reading of every CTA study was significantly shorter on the iPad than on the workstation (171±84 vs 227±91 seconds, respectively; p<0.002). Sensitivity and specificity for arterial blush detection were 85% and 60% for the workstation and 90% and 80% for the iPad, respectively. Inter-reader agreement for workstations and iPads was good (Cohen k=0.70) and very good (k=0.88), respectively. Agreement of workstation and iPad vs DSA on bleeding arteries was very good (k=0.82 and 0.89, respectively). Conclusion: iPad-based preliminary 2D reading of CT angiography studies of patients with suspected acute gastrointestinal bleeding is feasible and significantly faster compared with a conventional workstation.

SS 9.07
Large bowel obstruction: diagnostic performance of MDCT and added value of multiplanar reformations A. Bonnard  Purpose: CT-guided core needle biopsy (CNB) can be impaired due to streak artefact obscuring needle tip visualization. Material and Methods: Eight biopsy needles in two sizes from two manufacturers with and without stylets were imaged in a CT phantom using 64-slice CT reconstructed with ASIR and FBP (n=16). CNB-related streak artefact was quantified with profile analysis using ImageJ and Microsoft Excel. Hounsfield unit (HU) differences between the maximum HU at the needle tip and the minimum HU in the streak artefact beyond were compared for each variable. Image acceptability and streak artefact were qualitatively assessed on each phantom image after randomization and on 40 CT CNB procedures with and without the central stylet by three blinded reviewers. Statistical analysis (Mann-Whitney U-test, Kappa and T-test) was performed using SPSS (p-value <0.05 considered statistically significant).  Material and Methods: This retrospective study was approved by our institutional review board and informed consent was waived. A total of 105 liver donor candidates (M:F=73:32, age: 18-61 years) who underwent preoperative liver CT were enrolled. Volumes of the whole liver (VT), four hepatic segments, and predicted volume of the hepatic graft (Vpred) were obtained using a semiautomatic analysis program by radiologists and a conventional volumetry program by surgeons. Intraobserver and interobserver agreements of VT and Vpred were assessed using intraclass correlation coefficients (ICCs). Vpred was also compared to the actual graft weight (Wact) and the analysis time was recorded. In addition, possible vascular complications were assessed using the surgical planning function of the software. Results: The mean processing time of hepatic volumetry, segmentation and surgical planning using software (175.9±46.6 seconds) was significantly shorter than manual volumetry (916.6±52.8 seconds). VT and Vpred showed significant intra-and inter-observer agreements and Vpred showed strong correlation with Wact (r=0.83-0.86, P<0.0001) albeit with a tendency toward overestimation. Furthermore, image review using the liver analysis program detected 4 of 5 vascular complications (80%). Conclusion: Semiautomatic liver analysis software provided good accuracy and excellent reproducibility to estimate liver volume in a short time and it was useful for predicting vascular complications. To evaluate the potential of grating-based phase contrast computed tomography (gbPCCT) for the quantification of liver fibrosis in human specimens and to correlate with histological diagnosis. Material and Methods: IRB approval was obtained. Thirty human liver specimens with varying degrees of fibrosis were prospectively collected during autopsy at the institute of forensic medicine. Tissue samples were fixed in 4% formalin solution and imaged using a Talbot-Lau interferometer with a rotatinganode X-ray tube and a photon-counting detector. Phase-contrast and attenuation-contrast image sets were visually graded according to fibrotic stage using a 5-point scale. Additionally, a software-based method was used for quantfication using thresholds. Specimens were then sliced and stained for histological diagnosis including classification of fibrosis. Results of visual and software-based staging of the severity of fibrosis were compared to histology. Results: In phase-contrast images, strands of fibrous tissue show high signal intensity and delineate from surrounding liver tissue. When the extent of fibrosis was addressed visually or software-based using a threshold, the results showed strong correlation (r = 0.84) with the histological grading. On the contrary, attenuation-contrast images did not allow quantification of liver fibrosis. Conclusion: Grating-based phase contrast computed tomography allows ex vivo quantification of liver fibrosis in human specimens. In a clinical setup, a gbPCCT scanner may have the potential to improve diagnostics and therapy monitoring of chronic liver disease. The most frequently involved organs were the lungs in 13 (52%) cases and the bones in 9 (36%) cases. Recurrent HCC involved >1 organ in 11 (44%) patients. Recurrences were limited to the liver in one (4%) patient, were exclusively extrahepatic in 18 (72%) patients and were both intrahepatic and extrahepatic in 6 (24%) patients. Conclusion: Routine US examination alone failed to detect most HCC recurrences. Because HCC recurrence may occur even in patients with favourable prognostic factors, is mainly extra-hepatic, may be delayed, and is rarely identified using US examination, we suggest performing regular wholebody imaging during follow-up visits 2 years or more after LT.

SS 11.04
Withdrawn by the authors SS 11.05 All initial 115 records were evaluated by one author resulting in 8 relevant articles. These were critically appraised using evidence-based methods, subtracting information about test characteristics of CEUS compared to a golden standard (CTA, angiography, surgery and/or follow up).
Results: All appraised studies show a sensitivity of CEUS of 100%, a specificity of CEUS of 92-100% and negative predictive value of CEUS of 100%. CEUS is a useful noninvasive technique for detection of vascular complications, it helps improve flow visualization in the hepatic arteries and portal veins, and can differentiate thrombosis from a patent artery. CEUS decreases scanning time and the level of observer confidence increases considerably. CEUS has a sensitivity and specificity for detection of hepatic artery thrombosis similar to CTA, but CEUS is available at bedside in the ICU. Conclusion: CEUS has excellent test characteristics and offers substantial benefits for both patient and clinical team in terms of time, cost and potential risks compared to CDUS. CEUS can be used as a primary test in the evaluation of vascular complications in the early phase after OLTx.

SS 11.06
Acoustic structure quantification using probability density function of echo amplitude for the assessment of hepatic steatosis in living liver donors J.Y.

SS 12.02
Eye-tracking at endoluminal CT colonography: differences between false-positives diagnosed by experienced and novice readers A. Plumb 1 , E. Helbren 1 , P. Phillips 2 , T. radiologists each viewed 15 endoluminal CTC videos with simultaneous monitoring of eye movements with an eye-tracker. Readers signaled when they suspected a polyp with a mouse click. False-positives were defined as mouse clicks that occurred when no polyp was visible, with the reference standard being a composite of 3 expert radiologists and unblinded colonoscopy. Features provoking at least 3 readers to make a false-positive diagnosis were further interrogated by an experienced colonographer, who compared reader eye positions at the time of their mouse click with the endoluminal CTC video. The true nature of the perceived endoluminal abnormality was determined by 2D correlation. Results: Overall, 22 different features provoked more than 3 false-positive mouse clicks (mean=15 FP clicks/feature). 95% of these features were either diverticula (7 instances: 67 FP clicks), faecal residue (8 instances: 133 FP clicks) or artefacts (6 instances: 103 FPs). Experienced readers were significantly less likely than novices to generate false-positive clicks for diverticula (p=0.03). There was no significant difference in false-positives between experienced and novice readers for faecal residue or artefacts. Conclusion: Diverticula, faecal residue and artefacts commonly provoke false-positive diagnoses at endoluminal CTC. Experienced readers seem more capable than novices at distinguishing diverticula from polyps.

SS 12.03
Impact of post-hysterectomy changes on distal colonic morphology and endoscopy: quantitative assessment at CTC with endoscopic performance correlation C.N. Weber 1 , H.M. Zafar 1 , M.S. Levine 1 , B. Geiger 2 , A.S. Lev-Toaff 1 ; 1 Philadelphia, PA/US, 2 Princeton, NJ/US Purpose: The aim of this study is to quantitatively assess distal colonic morphology using novel software, and evaluating differences by hysterectomy status which may contribute to higher rates of incomplete/difficult endoscopic exams in post-hysterectomy women.

SS 11.09
Comparison between ElastPQ and Fibroscan for evaluation of fibrosis in chronic liver diseases M.-A. Van Caulaert, N. Michoux, Y. Horsmans, P. Starkel, E. Danse; Brussels/BE Purpose: Shear wave ElastPQ (PhilipsHealthcare) is a new technique investigating the elasticity of liver tissue. However, its interest, compared to already validated Fibroscan (Echosens) remains to be fully evaluated. Therefore, we sought to compare ElastPQ to Fibroscan in liver fibrosis. Material and Methods: 132 patients underwent liver elasticity assessment using both ElastPQ and Fibroscan on the same day. 70 patients, who fulfilled the validity criteria for Fibroscan, were included in the analysis and divided into 4 groups. 33 hepatitis C (HCV) patients, 12 hepatitis B (HBV), 5 non-alcoholic steatohepatitis (NASH) patients and 20 patients with other diseases. Intertechnique agreement on fibrosis gradation was assessed with Cohen's weighted kappa. Inter-technique agreement on elasticity measurement (in kPa) was studied with Bland-Altman plot.
Results: Inter-technique agreement for fibrosis gradation was almost perfect in HCV-related and "other chronic liver disease" group. Results were poorly concordant in the HBV and NASH groups. Bland-Altman plot showed that fibroscan slightly overestimates elasticity measurements compared to ElastPQ (regardless of patients group) without influencing the statification.

Conclusion:
This study sought to compare ElastPQ and Fibroscan in 4 subgroups of patients. Our preliminary results, although based on a small number of patients, seem to indicate that ElastPQ cannot be considered as a suitable alternative for fibrosis assessment in HBV-related cirrhosis and NASH. However, it revealed accurate concerning HCV-related cirrhosis and "other causes"-related cirrhosis.

SS 11.10
Multiple To determine the feasibility of multiphasic dynamic CT including multiple arterial phases for evaluation of liver tissue perfusion characteristics using the maximum slope model in humans compared with perfusion CT (PCT) as the standard of reference. Material and Methods: PCT was performed in 23 patients with chronic liver diseases using Xenetix 370. Five-phase, dynamic CT including unenhanced, triple-arterial phases including information regarding the peak aortic and splenic enhancement and the portal phase were selected in order to obtain perfusion parameters of liver parenchyma using a maximum slope method. Those selected CT datasets and the whole PCT data sets were analyzed using dedicated perfusion software (VPCT body; Siemens Healthcare) for estimating the perfusion parameters. Comparison between the perfusion parameters calculated from the multiphasic dynamic CT datasets and those of PCT was made using the intraclass correlation coefficient. Results: All of the perfusion parameters of patient liver parenchyma including ALP, PVP, and HBPi obtained by five-phase images in the 23 patients did not differ significantly compared with those of PCT. They showed very high agreement with PCT (ICCs > 0.80, P-value < 0.01) in both the validation and the evaluation groups. Conclusion: It was feasible to obtain perfusion parameters of the liver using multiphasic dynamic CT scans, and the perfusion parameters using the dynamic CT scans were comparable to those of perfusion CT.

Scientific Sessions
Authors' index SCIENTIFIC SESSIONS / SATURDAY, JUNE 21, 2014 S486 Conclusion: In conclusion, high-quality diagnostic CTC can be obtained without the need for laxative in bowel preparations, also the MP regime is of benefit as it is a shorter regime with reasonably fewer symptoms, and therefore, more manageable for the patient.

SS 12.06
The effect of computer-aided detection markers on visual search and reader performance in CT colonography E. Helbren 1 , A. Plumb 1 , T. Colchester/UK, 5 Loughborough/UK Purpose: To identify the effect of computer-aided detection (CAD) on visual search and performance in CT colonography (CTC) and to identify differences between novice and experienced readers. Material and Methods: 15 videos were recorded from CTC fly-through examinations, each containing a single polyp. Two versions of each video were generated, with without a CAD prompt. Videos were viewed in a randomised order by 42 readers (17 experienced [>200 cases] and 25 novices [<100 cases]). Visual search was monitored via eye-tracking. Both CAD markers and polyps were treated as regions of interest in data processing. Statistical analysis using multilevel modelling was applied to allow for the cross-classified data structure. Results: CAD drew readers' attention to polyp location faster, leading to quicker identification times. For example median 'time to first pursuit' was 0.48s (IQR: 0.27-0.87s) with CAD versus 0.58s (IQR: 0.35-1.06s) without. CAD also held readers' visual attention on the polyp for longer. A significant increase in the number of correct polyp identifications across all readers was seen when using CAD (74% without-CAD, 87% with-CAD). Conclusion: CAD marks significantly alter the visual search patterns of readers viewing endoluminal CTC by all quantitative metrics we measured. CAD exerted a larger effect on novice readers.

SS 12.07
Diagnostic accuracy of low-dose CT colonography using adaptive statistical iterative reconstruction D. Bellini, D. Caruso, M. Rengo, T. Biondi, A. Laghi; Latina/IT Purpose: The purpose of this study was to evaluate the diagnostic accuracy and image quality of CT colonography (CTC) using a low-dose acquisition protocol with adaptive statistical iterative reconstruction (ASIR). Material and Methods: From our dataset, 60 cases of CTC were selected, 30 acquired using a low-dose protocol reconstructed with ASIR (Group I, 100 kV, 50-100 mAs, 50%DR) and 30 acquired using a standard CTC protocol reconstructed with FBP (Group II, 120 kVs, 50-100 mAs). Prevalence of polyps, quality of bowel preparations and colon distension are homogenous between two groups. Two expert radiologists, blinded to the scanning techniques, reported the examinations and assessed image quality. Per patient sensitivity, specificity, negative predictive value and positive predictive value were evaluated for each group. Comparative analyses of image quality score, diagnostic confidence and interpretation time were also performed. Results: Diagnostic accuracy is similar between two groups (Group I: Se 92%, Sp 95%; Group II: Se 93%, Sp 94%). Radiation dose is significantly lower in group I (2.5 mSv vs 5.6 mSv, mean values). Reporting time and diagnostic confidence are similar between two groups (P> 0.05). Conclusion: ASiR showed similar diagnostic performance with good image quality compared with FBP reconstruction. Our study confirmed feasibility of low-dose CTC with iterative reconstruction as a promising screening tool.

SS 12.08
Polyp classification error at endoluminal CT colonography: characteristics of polyps viewed but incorrectly dismissed by radiologists A. Plumb 1 , E. Helbren 1 , P. Phillips 2 , T. Classification errors were deemed to occur when readers looked at a polyp but Material and Methods: CTC datasets from 73 women (37 posthysterectomy/36 control) were compared. Inclusion criteria: incomplete endoscopy. Exclusion criteria: severe diverticulosis, large uterine fibroids, colonic resection, colonic masses. Length, volume, tortuosity (number of high curvature points (hcp)), compactness (boxed volume containing centerline divided by centerline length) of the rectum, sigmoid and combined rectosigmoid were assessed. Height of the sigmoid apex (HSA) relative to the lumbosacral junction was assessed. Each dataset was quantified twice by a single reader on separate occasions. Groups were compared using student's t-test. Relative risk associated with post-hysterectomy status and failure to clear the sigmoid at endoscopy was calculated.  (c) pathologists exhibit bias when measuring polyps via "rounding" to the nearest 5 mm. Material and Methods: Polyp diameters measured at CT colonography (CTC), endoscopy and histopathology were collated from the English bowel cancer screening programme (BCSP) and two randomised trials. Log-normal models were fitted to estimate the expected number of polyps at 1 mm increments. The difference between the expected and observed number of polyps was calculated for each terminal digit (1 mm, 11 mm, 21 mm…;2 mm, 12 mm, 22 mm…). Following statistical advice, significance testing was not performed due to over-dispersion. Results: In the BCSP, at CTC the number of polyps with sizes ending in zero was 75% greater than expected (217 observed, 123.7 expected) and there was a 33% excess of polyps ending in 5 mm (284 observed, 213.1 expected). Similar patterns were demonstrated at endoscopy for polyps with sizes ending in zero (145% greater than expected) and 5 mm (44% excess), and at histopathology for polyps with sizes ending in zero (152% excess). An identical phenomenon was observed in the randomised trials for CTC (155% overrepresentation of terminal digit zero), endoscopy (285% over-representation) and histopathology (232% over-representation), with similar but lesser effects for sizes ending in 5 mm. Conclusion: Radiologists, endoscopists and pathologists exhibit terminal digit preference bias when measuring polyps. This may influence trial data, referrals for further testing, polyp surveillance regimes and comparisons between tests.

SS 12.05
Full purgation bowel preparation versus faecal tagging alone prior to CTC: does it really make a difference? G. Ayub, A. Lowe, C. Kay, A. Williams; Bradford/UK Purpose: To see if there was a significant difference in the quality of CTcolonography produced between patients prescribed bowel clearance preparation containing laxative to those not containing laxative. In addition, to also get patient feedback on bowel preparation experience. Material and Methods: Retrospective study was performed looking at 300 CTC investigations. This included 138 investigations in which full preparation (FP) was prescribed (3-day regime with laxative), and 162 investigations in which minimum preparation (MP) was prescribed (2-day regime without laxative). Both preparations also involved a low-residue diet and gastrografin for faecal tagging. CTC investigations were analysed for degree of bowel clearance and bowel insufflation, graded as either poor, moderate or good. Questionnaires were given to patients attending for the investigation to get feedback on bowel preparation experience.
Results: There was no significant difference in the quality of CTC produced with respect to bowel clearance (p=0.23) and bowel insufflation (p=0.81) between the two regimes. 45 patient feedback questionnaires (21 had laxative and 24 did not) showed a statistically insignificant (p>0.05) increase in faecal incontinence, urgency, diarrhoea, nausea and disturbed sleep in patients taking a laxative.
SCIENTIFIC SESSIONS / SATURDAY, JUNE 21, 2014 S487 provoked by CTC. Colonoscopy complication rates were similar (n=779; 1.0%). Conclusion: Whilst overall satisfaction with CTC was high, it was judged unexpectedly uncomfortable more frequently than colonoscopy. Understanding of risks was also greater for colonoscopy than CTC. Clear communication of the risks, benefits and procedural experience of CTC is required.
did not confirm it genuine via a mouse click. "Difficult-to-classify" polyps were defined as those with >15% classification error; others were considered "easy". Polyp diameter, height, subjective conspicuity and proportion of time looked at by radiologists were compared between "easy" and "difficult" polyps using the Mann-Whitney U test for numerical data and ordinal logistic regression for ordinal data. Results: "Difficult" polyps were significantly smaller than "easy" polyps (mean diameter 5.4 mm versus 8.4 mm, p=0.01) and subjectively less conspicuous (median conspicuity score 4 versus 2, p=0.032). Expert readers spent less time viewing "difficult" polyps during classification errors; 30.2% of the time "difficult" polyps were on screen was spent viewing them vs 54.0% for "easy" polyps (p=0.009), irrespective of whether or not CAD was available (

Material and Methods:
Using structure-from-motion, we used the perspective shift of a colonoscopic video camera to build a 3D image; using the relative translation and rotation of the camera from different projections we triangulated 3D points to reconstruct the endoluminal surface via a dense 3D reconstruction algorithm (plane sweep approach) to extract surface point clouds in 3D. Validation used a 3D printed colon model generated from 25cm of real CTC data that contained two polyps (8mm, 15mm). The 3D model was then painted to resemble human colon, examined using an endoscope (1920x539 resolution), and the 15-mm polyp used as the input sequence for the reconstruction algorithm. We then used a point set registration algorithm (coherent point drift) to match polyp point clouds between the colonoscopic and CTC images and computed the matching cost to assess the accuracy of registration. Purpose: Invasive techniques to assess intestinal motility in animal-models should be avoided by application of non-or less-invasive radiological examinations in compliance with the rule of "three-Rs" of humane animalexperimentation (reduction-refinement-replacement). Thus, our aim was to investigate the feasibility of in vivo small-bowel-motility-analyses in mice using dynamic MRI acquisitions. Material and Methods: All experimental procedures were approved by the institutional animal-care-committee. Six C57BL/6-mice underwent MRimaging without additional preparation after isoflurane-anaesthetization in prone-position on a 4.7T-small-animal-imager using a birdcage-whole-bodymouse-coil. Motility was assessed using a true-fast imaging in steadyprecession-sequence in coronal-orientation (acquisition-time per slice 512ms, in plane resolution 234x234 µm, matrix size 128x128, slice-thickness 1 mm) over 30 s (60 acquisitions). Motility was manually assessed measuring the small-bowel diameter-change over time. The resulting motility curves were analyzed for the following parameters: contraction-frequency-per-minute, maximal contraction-amplitude, luminal diameter and luminal-occlusion-rate. Material and Methods: 20 healthy volunteers underwent magnetic resonance enterography following oral 2.5% mannitol. "Cine" BTFE free-breathing (60 second acquisition) volumes were acquired at 3 Tesla with a 1 second/volume temporal resolution. Enteric motility was quantified using the number of contractions per minute (CPM) by measuring luminal diameter on each image: A contraction was defined as a reduction of >10%. Subsequently an optic-flow algorithm was used to generate a deformation field measure (Jacobian standard deviation, JSD), which acts as a surrogate for motility, measured in arbitrary units (AU). Motility was compared in four abdominal quadrants and at two time points (4 weeks apart) using Bland-Altman limits of agreement (LoA). Results: CPM and JSD showed moderate positive correlation (r=0.51, p<0.001).

Results
Mean motility across all quadrants was 6CPM and 0.34AU. There was substantial variation in motility in the same individual, with the mean difference between the most motile and least motile quadrants being 4.4 CPM and 0.21AU. Temporal variability was even greater, with the Bland-Altman LoA being 6CPM and 0.32AU. Conclusion: SB motility is highly variable between enteric segments and at different times, irrespective of the quantitative metric used. Segmental SB motility measurements as a disease biomarker may be limited by high spatial and temporal variation. Purpose: Advanced imaging and increasing endoscopy have produced a decline in performed fluorosocpy. However, fluoroscopic skills are necessary for successful practice of radiology. We designed and integrated an instructional video into the first-year radiology resident orientation to evaluate knowledge and competency in performing an esophagram before and after watching the video. Material and Methods: Prior to any fluoroscopy or instructional video exposure, 15 first-year radiology trainees completed an 11-item written test assessing knowledge and comfort in performing esophagram. One week later, the trainees were provided access to our instructional esophagram video and then asked to complete the same written test. Validity of the exam and statistical significance of the results were analyzed. Results: Ten objective items were analyzed with the pre and post-test. An average of 6.2 questions were answered correctly per resident on the pre-test while 9.5 questions were answered correctly per resident on the post-test, a 36% improvement (p<.01). A 1-5 scale (5 indicating most comfort) was used to assess comfort in performing an esophagram before and after video intervention. Average comfort level rose from 1 to 2.7, a 60% increase (p<.01). Conclusion: Our esophagram video improved the comfort level of novice radiology trainees performing an esophagram. Their knowledge and competency regarding procedural details improved. Other potential audiences benefiting from this video include radiology technology students or practicing radiologists needing fluoroscopy refreshers and continuing education.

SS 13.02
MR fluoroscopy in diagnosis of postoperative complications following laparoscopic Nissen-Fundoplication for GERD C. Kulinna-Cosentini, W. Schima, A. Ba-Ssalamah, E. Cosentini; Vienna/AT Purpose: The aim of this study was to determine the feasibility for detecting postoperative complications in symptomatic patients who underwent Nissen Fundoplication for GERD with dynamic magnetic resonance imaging. Material and Methods: 30 patients (22 patients with recurrent/persistent symptoms and 8 symptomless patients as control group) underwent MRI at 1.5 T. Bolus transit of a buttermilk-spiked gadolinium mixture was evaluated with T2-weighted, half-Fourier acquisition single-shot turbo spin-echo (HASTE) and dynamic gradient echo sequences (B-FFE) in 3 planes. The results of MRI were compared with intraoperative findings, or, if the patients were treated conservatively, with endoscopy, manometry, pH-metry, and barium swallow. Results: MRI was able to identify the correct location of fundoplication wrap in 28/30 cases (93% overall accuracy) and the malpositions of the fundoplication wraps in 4/6 cases (67%), as well as all wrap disruptions (4/4 cases). All six stenoses in the GEJ were identified and could be confirmed intraoperatively or during dilatation. MRI correctly visualized three cases with motility disorders, which were manometrically confirmed as secondary achalasia. Three patients showed signs of recurrent reflux without anatomical failure. Conclusion: Our preliminary results demonstrate that dynamic MRI is helpful in diagnosing postoperative complications in patients having undergone antireflux surgery and may be a useful tool in clarifying dysphagia in such patients additional to manometry and ph-metry.

SS 13.03
Morphological  To evaluate the accuracy of MDCT in the preoperative evaluation of peritoneal carcinomatosis (PC) of bowel loops in patients with advanced ovarian cancer who underwent peritonectomy and hyperthermic intraperitoneal chemotherapy (HIPEC) to obtain an optimal cytoreduction surgery. Material and Methods: Pre-surgery abdominal CT examinations of 24 patients with advanced ovarian cancer after neoadjuvant chemotherapy were blindly and prospectively analysed by a radiologist with expertise in the oncologic field. The peritoneal carcinomatosis index (PCI) was scored according to the Sugarbaker classification, based on lesion size and distribution, with particular attention to the abdomino-pelvic regions refer to small bowel. The results were compared with macroscopic and histologic data after surgery. Results: Considering the patient-level analysis (the capacity to detect PC), the sensibility, specificity, positive predictive value (PPV), negative predictive value (NPP) and accuracy of MDCT, were 75%, 67%, 69%, 73% and 71%, respectively. Considering the regional level analysis (the capacity to localise PC), a sensitivity, specificity, PPV, NPV and accuracy of 63%, 87%, 55%, 91%, and 82%, respectively were obtained for the correlation between CT and histology.

Conclusion:
Our results encourage the use of MDCT as the only technique sufficient to select patients with PC for cytoreductive surgery and HIPEC on the condition that a CT examination will be performed using a dedicated protocol, optimised to detect minimal peritoneal disease and CT images will be analysed by an experienced reader.

Results:
The "layered"-type of involvement corresponded to myriad cancerous nodules and/or plaques covering the intestinal wall; the stenotic loops with irregular borders to extensive mural infiltration; mural enhancement without stenosis to micronodular superficial infiltration; affixed loops, even distented, to small indolent implants in their curves or angulations; or distensible loops adjacent to an "omental cake" mass to extensively infiltrated intestine requiring partial enterectomy. The stranding appearance of the mesentery on CTE corresponded to superficial nodular infiltration along with thickening and rigidity of the mesentery, requiring local thermocoagulation; the "frozen mesentery" (shortened, distorted, rigid) to extensive malignant infiltration precluding cytoreductive surgery; the thickened, enhanced peritoneal lining to covering malignant plaques requiring extensive peritonectomy. Conclusion: Lesion-by-lesion correspondence between imaging and surgical findings is important for preoperative assessment of disease burden, for the selection of candidates for cytoreductive surgery and for the development of common terminology and classification systems for use by both radiologists and surgeons.

Conclusion:
The intraductal PDT is safe and an effective strategy in nonsurgical hilar cholangiocarcinoma patient management, increasing both survival rate and quality of life.

SS 14.06
Estimation of the malignant potential of gastrointestinal stromal tumors: the value of multislice computed tomography and magnetic resonance imaging T.V. Bartolotta We evaluated retrospectively 20 patients (8 women, 12 males, mean age 57). All underwent F-18FdG contrast-enhanced PET-CT, before, after surgery and along the follow-up, between 2008 and 2013. All patients were operated, the molecular analysis showed exon11 mutation (11 cases), Pdgfra mutation (6), and 3 wild type (not considered in statistics). In each case, we evaluated: 1) tumor's maximum diameter, 2) proliferation index, 3) SUVmax (standardised uptake value). 11 patients had preoperative SUVmax>3; 9 had SUVmax<3. SUV >3 proves that PET registration is pathologic without any clinical doubt. We calculated statistical correlation between: 1) mitotic index and SUVmax, 2) SUVmax and tumor maximum diameter (Pearson's test), 3) SUVmax and genotypic mutation. referred for CTE were included. Images were reconstructed with MBIR in addition to standard department protocol (60%FBP/40%ASiR, filtered back projection, adaptive statistical iterative reconstruction). Image quality was assessed objectively and subjectively at 6 anatomical levels. Clinical interpretation was undertaken in consensus by 2 blinded radiologists along with 2 non-blinded readers ('gold standard'). Results were analysed using SPSS. Results: Mean estimated radiation dose was 6.05±2.84 mSv (SSDE 9.25 ± 2.9 mGy). Objective and subjective assessment yielded 6106 datapoints. MBIR images significantly outperformed those using standard reconstruction techniques across all subjective (p<0.001 for all comparisons) (noise, contrast resolution, spatial resolution, streak artefact, axial diagnostic acceptability, coronal diagnostic acceptability) and objective (p<0.004) (noise, signal-tonoise ratio) parameters. Clinical reads of the MBIR images agreed more closely with the gold standard reads than the standard 40% ASiR image reads in terms of overall Crohn's activity score (κ=0.630, 0.308) and detection of acute complications (κ=1.0, 0.896). Results were comparable for bowel wall disease severity assessment (κ=0.523, 0.593). Conclusion: MBIR considerably improves image quality of conventional dose CTE images and therefore should be considered for routine use in CT imaging.

SS 15.02
Comparison of different iterative reconstruction algorithms in dedicated CT of the liver and pancreas A.J. Ruppert-Kohlmayr, M. Fauster, T. Maier, M. Uggowitzer; Leoben/AT Purpose: To compare diagnostic impact of two different iterative reconstruction algorithms, iDose4® and iMR®, compared with standard-CT of liver and pancreas using filter back projection (FBP). Will iterative reconstruction improve diagnostic impact and which algorithm proves to be better? Material and Methods: In a prospective clinical study, 30 patients with jaundice or liver lesions were examined with liver or pancreatic CT. The used 256-slice CT-scanner (iCT-Elite®, Philips) enabled routine and simultaneous use of different iterative reconstruction algorithms, iDose4 and iMR. Two experienced staff-radiologists and two residents independently evaluated in a blinded manner CT-exams with two different algorithms with standard slice thickness of 1 mm for lesion conspicuity, lesion margins, and signal-to-noise. Density measurements in lesions were performed and densities and standard deviations were compared. Results: In 30 patients, 52 liver lesions and 8 pancreatic tumors were found. Lesion conspicuity was significantly better, lesions had sharper margins and density measurements showed significantly lower standard deviations in iMR than in iDose4 and in standard-CT for all 4 radiologists. With iMR reconstruction examination with slice thickness of 1 mm had much less noise than with iDose4 or standard-CT. Conclusion: Iterative reconstruction of modern CT-scanners are responsible for better image quality with less noise and more accurate density measurements. In comparison to iDose4 the recently available reconstruction algorithm iMR delivers images without significant noise. This might improve exam quality with better lesion detection and conspicuity.

SS 14.10
Accuracy of MDCT in the preoperative definition of Peritoneal Cancer Index in patients with advanced ovarian cancer who underwent peritonectomy and hyperthermic intraperitoneal chemotherapy: radiological and surgical correlation C. Cavallini 1 , F. Iafrate 1 , M. Ciolina 1 , D. Bellini 2 , A. Laghi 1 ; 1 Rome/IT, 2 Latina/IT Purpose: To evaluate the accuracy of MDCT in the preoperative definition of Peritoneal Cancer Index (PCI) in patients with advanced ovarian cancer who underwent a peritonectomy and hyperthermic intraperitoneal chemotherapy (HIPEC).To compare the radiological PCI with surgical PCI in three readers with different level of experience. Material and Methods: Pre-HIPEC, CT examinations of 60 patients with advanced ovarian cancer were prospectively analyzed by three radiologists with different levels of experience (Reader1>100 cases, Reader2>25 cases, Reader3 inexperienced). PCI score was evaluated according to Sugarbaker system, based on lesion size and distribution and radiological results were compared with surgical PCI. Sensibility for lesion < 0.5 cm as well as per patient and per anatomic site analysis was evaluated. Results: Concerning the patient-level analysis, MDCT showed a sensitivity of 75%, 66% and 25%, respectively for the three readers according to surgery. Reader1 and Reader2 overestimated 25% of lesions, Reader3 underestimated 75% of lesions. Concerning lesions < 0.5 cm sensitivity of 66% was observed for Reader1 and Reader2 and sensitivity of 12% was obtserved for Reader3. Reader1 had overestimated 55% of lesion less than 0.5 cm, missing 37%, whilst Reader2 had overestimated 65% of 0.5 cm lesions, missing 25% of the lesions. Conclusion: Sensitivity of MDCT in detecting peritoneal metastasis appeared to be strongly related to lesion size and location. For more experienced readers overestimation is observed, whilst underestimation is more common in lesser experienced readers.

S493
patients referred for CT. The SP images were reconstructed using manufacturer recommendations (60% FBP, filtered back projection; 40% ASiR, Adaptive Statistical iterative Reconstruction, SP-ASiR40). The MP datasets were reconstructed in 4 ways (100% FBP; 40% ASiR; 70% ASiR; 100% MBIR). 3 observers measured tissue volumes using Hounsfield unit thresholds for fat, soft tissue and bone and contrast via Osirix (Ver 4.1.1) on each dataset. Data analysis was performed with SPSS. Results: Inter-observer agreement for tissue volume measurement was very strong for the 1530 datapoints (rs>0.9 for all comparisons). Tissue volume measurement on the MP-MBIR images was significantly superior to all other MP reconstructions and closely correlated with the 'gold standard' SP-ASiR40 images for all tissue types. The superiority of MP-MBIR was most marked in fat volume calculation where Bland-Altman plots showed closest correlation between SP-ASIR40 and MP-MBIR with the lowest average difference (336 cm 3 ) when compared with other MP reconstructions. Conclusion: Low-dose CT images reconstructed with MBIR maintain accurate tissue and hence lesion characterisation is superior to other reconstruction techniques when compared with conventional dose images.

SS 15.06
Multi-arterial phase for achieving optimal arterial phase imaging on Gd-EOB-DTPA enhanced magnetic resonance imaging

Material and Methods:
Total 169 consecutive patients underwent Gd-EOB-DTPA liver MRI at 3.0T scanner. In liver MRI, AP was obtained using fourdimensional (4D) contrast-enhanced timing robust acquisition order (CENTRA)keyhole technique. Gd-EOB-DTPA was injected at a rate of 1.5 mL/sec followed by 20 mL saline chaser, using a real-time MR fluoroscopy technique. Three phases of AP (early, middle and late) were obtained in a breath-hold (<25 sec). Two radiologists assessed appropriateness of AP bolus timing and image quality. Results: At least one optimal AP was achieved in 95.9% (162/169). Twelve of 169 patients did not hold their breaths (n=11) or vomited (n=1) during AP. In 9 of 12 patients, at least one AP with optimal bolus timing and diagnostic acceptable quality was obtained among the three phases: one optimal AP (n=6); two optimal AP (n=1); and three optimal AP (n=2). The remaining three patients showed unacceptable APs due to severe motion artifact (n=2) or too delayed AP timing (n=1). In remaining 158 patients, two patients (1.2%) showed diagnostically unacceptable image quality due to severe ringing artifacts. In six patients (3.8%), all of three APs were either too late (n=3) or too early (n=3 referred for CTE were included. Low-dose modified-protocol (MP) and conventional-protocol (CP) CT datasets were contemporaneously acquired. CP-ASiR image formation used 40% adaptive statistical iterative reconstruction. MP data were reconstructed with 100% MBIR (MP-MBIR) and 40% ASiR (MP-ASiR). Image quality was assessed objectively and subjectively at 6 levels. Clinical interpretation was undertaken independently by 2 blinded radiologists along with 2 non-blinded readers in consensus ('gold-standard').
Results: A 75% average radiation dose reduction was seen: MP effectivedose (ED) 1.61±1.18mSv (size-specific-dose-estimate (SSDE) 2.47±1.21mGy); CP ED 6.05±2.84mSv (SSDE 9.25±2.9mGy). Image-quality assessment yielded 9372 datapoints. Objective noise in MP-MBIR images was superior (p<0.05) at 3/6 levels and comparable in the remainder. MP images were superior to CP-ASiR (p<0.05 in all cases) for subjective noise, spatial resolution, contrast resolution, streak artefact and coronal diagnostic acceptability. CP-ASiR axial diagnostic acceptability was superior (p=0.76). MP-MBIR clinical reads agreed more closely with gold-standard reads than CP-ASiR reads regarding bowel wall disease assessment (κ=0.589/0.700 vs 0.583/0.564) for both readers whereas overall Crohn's activity score (  were scanned with standard-dose protocol (group 2) and 101 with low-dose protocol with IR (group 1). MDCT includes a weight-based i.v. contrast protocol and mAs Modulation dose (Care Dose) for all patients. Group 1 used the following parameters: 120 kVp, mAs reference: 210, 2-mm thickness, 128×0.6 mm detector configuration. Group 2 used 100 kVp, mAs: 140, 0,6-mm thickness, 128×0.6 mm detector configuration. One blinded reader reviewed the randomized CTE scans of all patients to assess image quality. S494 was realized during internal meetings through a consensus process (first by the working groups, then by the entire panel in plenary session). Staff radiologists rated independently each template on a 5-point Likert scale. All ratings for each template were analysed with descriptive statistics. A mean score >4 indicated good agreement between staff radiologists, while 5 indicated complete agreement. Results: 16 structured reports were elaborated in the following applications: cirrhotic liver CT/MR, MR-cholangiopancreatography, CT/MR enterography, MR in rectal tumors and fistulas, CT in acute/chronic pancreatitis and pancreatic cancer, defecography, US in TIPS placement. The mean level of agreement between the staff radiologists was 4.2. Conclusion: We achieved to compose a first group of templates usable in the clinical practice. Further new structured reports will be produced and uploaded on the RSNA website making them available to the radiological community. and layered mural post-contrast enhancement (test properties could not be calculated). These results were then incorporated into a structured MRE report, which will be presented.

Conclusion:
The MRE signs that most usefully classify SBCD can be used to create a structured report.

SS 15.10
Structured reporting of hepatic CT and MRI: initial experience L. Van Hoe 1 , B. Pilet 1 , V. De Grove 1 , S. T'Seyen 1 , P. Dewachter 1 , L. Hermie 1 , L. Braeye 2 , A. Vanhoyweghen 1 , A.-S. Vanhoenacker 3 ; 1 Aalst/BE, 2 Halle/BE, 3 Leuven/BE Purpose: To assess the results obtained with a software system designed for structured reporting of hepatic CT and MRI. Material and Methods: An electronic platform for structured reporting in radiology was designed and applied to the evaluation of liver disease using CT and/or MRI. The system contained both a structured approach towards the evaluation of liver lesions and specific information guiding the user towards a correct diagnosis in individual patient cases. Ten observers (five radiologists and five radiology residents) used the platform for reporting in 10 randomly assigned cases. Diagnostic accuracy and reporting times were measured. Subjective assessment of user friendliness and report quality (evaluation by reporting and referring physicians) was obtained using a five-point scoring system. Results: Diagnostic accuracy was 92%, without significant difference between both groups of observers. Average reporting time was 3.5 min. Average score for user friendliness was 4.4. Referring physicians liked the structured reports (average score 4.6).

Conclusion:
The system tested in this study showed promising results for structured reporting of hepatic CT and MRI.