ESGAR Book of Abstracts

EDITOR S. Jackson (Plymouth/UK) ESGAR MEETING PRESIDENT Prof. Carmen Ayuso University of Barcelona Hospital Clinic Department of Radiology Villaroel 170 ES – 08036 Barcelona, Spain ESGAR EXECUTIVE COMMITTEE PRESIDENT F. Caseiro Alves (Coimbra/PT) PRESIDENT-ELECT L. Martí-Bonmatí (Valencia/ES) VICE PRESIDENT C. Matos (Brussels/BE) SECRETARY A. Palkó (Szeged/HU) TREASURER S. Jackson (Plymouth/UK) PAST PRESIDENT & PROGRAMME COMMITTEE CHAIRMAN Y. Menu (Paris/FR) BY-LAWS COMMITTEE P. Prassopoulos (Alexandroupolis/GR) EDUCATION COMMITTEE H. Fenlon (Dublin/IE) MEMBERSHIP COMMITTEE R.G.H. Beets-Tan (Maastricht/NL) WORKSHOP COMMITTEE A. Laghi (Latina/IT) MEETING PRESIDENT C. Ayuso (Barcelona/ES) PRE-MEETING PRESIDENT G.H. Mostbeck (Vienna/AT) PRE-PRE-MEETING PRESIDENT Y. Menu (Paris/FR) FELLOWS REPRESENTATIVES T. Helmberger (Munich/DE)

The DECT data were reconstructed as monochromatic 70 keV images, grayscale iodine density and color-coded iodine density images. Two independent readers were asked to rate their confidence in the presence or absence of residual tumor. The results were compared with the follow-up imaging after 8-10 weeks. Multireader multicase receiver operating characteristic (ROC) analysis was performed with the area under the curve and 95% confidence intervals (CI) as metric for diagnostic accuracy. Results: Ten out of 38 (26.3%) lesions showed LTP at 8-10 weeks. The mean area under the curve was 0.84 (CI: 0.75-0.93) for 70keV, 0.80 (CI: 0.67-0.93) for grayscale iodine density and 0.78 (CI: 0.65-0.90) for color-coded images. There was no significant difference observed between the three reconstruction techniques (p=0.08). Conclusion: DECT can be used to make an early evaluation within 24 h after RF ablation. Results for interpretation of LTP are equal to the reconstruction techniques used in DECT. To determine whether multi-echo Dixon (mDixon) 3D T1-weighted (T1W) gradient-recalled-echo (GRE) technique with a high-acceleration parallel acquisition can provide better image quality than conventional 3D-fat suppressed (FS)-T1W-GRE for gadoxetic acid-enhanced liver MRI at 3T. Material and Methods: This retrospective study was approved by our institutional review board, and informed consent was waived. 138 patients with suspected hepatic focal lesions underwent gadoxetic acid-enhanced liver MRI at 3T, including dynamic imaging using either high-resolution mDixon 3D technique (n=70). Hepatobiliary phase (HBP) imaging was obtained using standard eTHRIVE with acceleration factor (AF) of 2.6, high-resolution eTHRIVE and mDixon with AF of 5. The image quality of the arterial, portal and HBP image sets of both groups was graded independently using a five-point scale by two radiologists and they reached a final consensus. Results: High-resolution mDixon images with AF 5 provided better overall image quality than eTHRIVE with AF 2.6 and AF 5 (p<0.05). In addition, among the three HBP image sets, mDixon showed significantly less pixel graininess and better fat suppression than eTHRIVE with AF 5 (p<0.05), despite of similar performance in terms of lesion or anatomic structure conspicuity compared with eTHRIVE with AF 5. Conclusion: Use of high-acceleration parallel acquisition factor and mDixon sequence can provide high-resolution T1WI with better image quality and fat suppression than that of conventional 3D-FS-T1GRE sequence.

SS 3.02
Inter and intra-observer reproducibility of MRI for the measurement of NET liver metastases J. Arfi Rouche, C. Caramella, S. Foulon, A. Laplanche, E. Baudin, C. Dromain; Villejuif/FR Purpose: Assessment of treatment is based on metastases size measurement.
Our goal was to determine the intra and inter-observer reproducibility of each MRI sequence for the measurement of NET liver metastases. Material and Methods: 32 patients with NET liver metastases underwent 1.5-T MR imaging of the liver [T2W and T1W sequences before and after the injection of gadolinium on hepatic arterial (HAP) and portal venous phase (PVP)]. A maximum of 5 target lesions by patient was chosen by one reviewer on the MR sequence allowing the better depiction. Then measurements of target lesions diameter were performed by 3 reviewers blindly and independently in two sessions with a delay of one month. Results: 3384 measurements were made (141 metastases, mean 4.4 per patient). The MR sequence allowing the better depiction of the target lesions was T2W in 81% of patients. Non-measurable lesions were significantly higher on HAP and PVP compared to unenhanced T1 and T2W (p<0.001). T2W sequence had the higher overall intra-and inter-observer inter-class correlation coefficient (0.99 and 0.98) but without significant difference compared to other sequences. No statistical difference was found between the senior and the junior radiologists. Conclusion: MRI allows reproducible measurements with low inter and intraobserver variability. Unenhanced T2W sequence should be considered as the sequence of choice for the size measurement of NET liver metastases.
Authors' index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 5, 2013 S481 SS 3.08 Acoustic radiation force impulse imaging in pediatric population: normal liver values? L.B. Barbosa, H. Matos, M.J. Noruegas; Coimbra/PT Purpose: Until now, the range of shear wave values (SWV) of the normal liver, obtained using the ARFI technique, in a pediatric population were not defined. We aimed to define the range of normal values in different age groups and gender evaluating differences resulting from the methodological approach. Material and Methods: Between July and December 2011, we prospectively evaluated 90 healthy children (54 girls, 36 boys), with an age range of 2 months to 17 years. All studies were performed using the Acuson S2000 equipped with ARFI and SWV quantification. Age, gender, ROI depth, liver lobe were analyzed using different methodological approaches. Results: Mean SWV value was 1.09±0.18 m/s (0.57 -1.50 m/s). Significant differences were found between different age groups, respectively: 1.10±0.09 m/s to 0-6 years; 1.09±0.09 m/s to 6-12 years and 1.07±0.11 m/s to 12-17 years. The mean SWV for right lobe was 1.06±0.10 m/s and for left lobe 1.18±0.15 m/s (p<0.05). Differences between superficial (1.17m/s) and deeper interrogation (1.10 m/s, 1.07 m/s and 1.02 m/s -with increasing depth) were significant (p<0.05). Conclusion: We found a mean liver SWV value of 1.09±0.18 m/s in a normal pediatric population, ranging between 0.57 and 1.50 m/s. Depth of tissue interrogation and the ROI placement for SWV calculation (right/left lobe) may influence the results. For standardization, we propose that the SWV should be measured at right lobe, avoiding superficial and deep locations.

SS 3.09
Low-voltage CT of the abdomen: to identify cutoff patient diameters for patient selection through the analysis of the correlation between patient diameters and subjective image quality M.C. Ambrosetti, G.A. Zamboni, F. Lombardo, R. Pozzi Mucelli; Verona/IT Purpose: To identify cutoff patient diameters optimal for low-voltage scans through the correlation between patient diameters and subjective image quality in low-voltage and standard-voltage CT of the upper abdomen in the same patient population. Material and Methods: 32 patients underwent MDCT of the abdomen with arterial phase at 80 kV with angular dose modulation on 64-row MDCT (test group). These examinations were compared with a previous 120kV scan on the same scanner. Patient transverse and sagittal diameters were measured at celiac axis level, and the mean was calculated. Two radiologists by consensus graded image quality on a 5-point scale (5=excellent; 1=non-diagnostic; 3 was the chosen cutoff quality). Image quality was correlated to the transverse, sagittal and mean diameter with ANOVA test. Results: Patient diameters did not change across examinations (all p=ns). In low-voltage scans, image quality was significantly correlated to sagittal (p=0.034) and mean diameters (p=0.025), and only a trend to significance was observed for transverse diameter (p=0.053). In 120 kV scans, there was no significant correlation between image quality and patient diameters, and all patients received grade 4 or 5. In 80 kV scans, a subjective grade 3 corresponded to a transverse diameter of 329 mm and a sagittal diameter of 267 mm. Conclusion: Patient size appears to influence subjective image quality more in low-voltage scans than in standard-voltage scans. For our protocol, cutoff diameters for diagnostic image quality are transverse 329 mm and sagittal 267 mm.

SS 3.07
Performance of shear wave elastography for assessment of liver fibrosis: correlation with histological staging and quantification T. Lefort, G. Renosi, O. Guillaud, J. Dumortier, J. Scoazec, A. Guibal; Lyon/FR Purpose: To evaluate the performance of shear wave elastography (SWE) for the quantification of liver fibrosis. Liver biopsy staging and quantification was used as the standard reference. Material and Methods: Consecutive patients with various chronic liver diseases were prospectively included. On the same day, ultrasound SWE (Aixplorer, Supersonic Imagine, Aix-en-Provence, France), transient elastography (TE) using Fibroscan (Echosens, Paris, France), and subsequent liver biopsy were performed. Histologic analysis included staging according to the METAVIR score and digital quantification of the fibrosis area. Results: Fifty-five patients were included for analysis. Chronic liver conditions included liver transplantation follow-up (n=21), non-alcoholic steatohepatitis (n=13), viral chronic hepatitis B or C (n=9), alcoholic liver disease (n=2), autoimmune hepatitis (n=2), chronic biliary disease (n=2), and others (n=6 , the complication rate of the various portal occlusion techniques were 4.5%, 3.6%, 11.5%, respectively. At least 4 segments were resected during hepatectomy. Overall postoperative morbidity and mortality rates were 13% and 2.5%, respectively. Postoperative complication and mortality rates did not differ significantly in the 3 groups. Conclusion: Patients with previously unresectable liver tumors can benefit from resection after all kind of portal occlusion techniques. Although complication rate of portal occlusion combined with hepatic artery cannula implantion is higher, more patients become resectable due to higher increase rate of FLR in this group.

SS 4.02
Problem solving and aberrant anatomy in selective internal radiotherapy D. Mullan, N. Kibriya, H. Laasch, P. Manoharan, J.A. Lawrance; Manchester/UK Purpose: To discuss the incidence, consequences and outcomes of abnormal arterial anatomy in selective internal radiotherapy (SIRT) for liver metastases. Material and Methods: 106 patients underwent SIRT work-up angiography between 21/9/5 and 14/6/12 at our institution. Angiograms and postangiography Tc-99 uptake scans were retrospectively reviewed to record arterial anatomical variants, radio-isotope uptake, and the effect these had on planning and performing treatment. Procedural and periprocedural complications were recorded. Results: 57/106 patients had abnormal anatomy. 7/106 were deemed unsuitable for SIRT treatment following work-up angiography due to the effects of abnormal anatomy. Chronologically, 4 of these were within the first 19/106 procedures at this institution, suggesting a learning curve when dealing with aberrant vasculature. 51/57 proceeded to treatment. 5/51 required an additional work-up and coil embolisation due to varied anatomy or initial extrahepatic uptake. 2 patients required a two stage treatment to right and left lobes due to hepatic tumour burden and the risk of radiation hepatitis. 1 patient died prior to treatment. Conclusion: Abnormal is the new 'normal' when considering hepatic arterial anatomy. Vascular variants will directly influence all cases and can preclude treatment. There appears to be a learning curve involved when dealing with abnormal anatomy, and understanding the effects aberrant anatomy can have will increase the probability that a patient will be deemed suitable for SIRT. Efficacy of ablation was assessed with CT at 1 month and every 6 months. The mean number of liver tumours per patient was 1.5 and mean tumour diameter was 2.3 cm (range, 0.6-5.8 cm). In 37.5% of the cases, lesions had a subcapsular location and 34% were close to a vascular structure. Results: The morbidity rate was 18.7 %. Extrahepatic recurrence appeared in 50% of the patients. Local recurrence at the site of ablation appeared in 18% of the lesions. Local recurrence rate was 7% in lesions less than 3 cm and 52% in lesions larger than 3 cm. Actuarial survival rates at 1, 3 and 5 years were 94.5%, 65.3% and 21.7%. Conclusion: Although our series include a selected population with poor prognosis, overall survival is comparable to surgical series. Local recurrence rate is higher than surgical resection, but it improves significantly in lesions less than 3 cm (7%). In our experience, RFA is an effective treatment in recurrent liver metastases of CRC and could be proposed as first-line treatment in lesions less than 3 cm. The purpose of this study is to evaluate error rates when using a voice recognition (VR) dictation system, to identify if there is a reduction in error rates over time with familiarity, and to assess if dual reading reduces errors.

Material and Methods:
A total of 350 finalised reports of CT studies for abdominal oncological follow up performed between June 2008 and December 2012 were randomly selected for analysis in this study. Reports were individually scrutinized for errors which were divided into two categories: (1) significant but unlikely to alter patient management and (2) very significant with the meaning of the report affected, thus potentially negatively affecting patient management. Results: 12% of the selected reports contained errors. In subgroup analysis, this fell from 14% in 2008 to 6% in 2012. 3% of reports contained errors that could significantly alter patient management. 60% of errors in reports occurred in dual reader studies. Two thirds of reports containing errors occurred when the report was finalised between 16:00 and 18:00 hrs. Conclusion: Whilst there are many benefits to VR systems, there are also many pitfalls. By raising awareness of the learning curve related to error rates for radiology departments using VR lessons will be learnt. This is important in the context of ever-increasing demand on abdominal CT imaging workflow and for preventing potential detriment to our patients as a result of non-clinical errors.

SS 5.03
Intraoperative contrast-enhanced ultrasound and color-coded elastography for characterization of liver lesion before surgical resection J. Rennert, C. Stroszczynski, E. Jung; Regensburg/DE Purpose: To evaluate if IO-CEUS and CCE allow a differentiation between malignant and benign liver lesions in comparison to histopathology. Material and Methods: Retrospective evaluation of digitally stored intraoperative CEUS and elastography. IO-CEUS and CCE of 49 liver lesions were compared to histopathology following surgical resection. Examinations were performed using a multifrequency linear probe (6-9 MHz). CEUS was evaluated during the arterial, the portal venous and the late-venous phase. Characterization of the CCE quality using cine-loops >10 s, based upon a color-coding system. Semi-quantitative evaluation of the lesions' stiffness based upon a specified scaling of 0-6 (0 low up to 6 high) using 6 ROIs (1 central, 5 peripheral was approved by the local ethics committee. MRI was conducted on the first postoperative day at a 1.5T scanner using three conventional gradient echo (GRE1-3) and one T2-weighted turbospin echo (TSE) sequences. Three radiologists independently assessed the following criteria using a 4-pointscale: visual contrast-to-noise ratio, conspicuity to air artifacts, and diagnostic quality rating with respect to the mesh and to the surrounding anatomy. In addition, mesh's crease formation and localisation in relation to the hernia were rated. Wilcoxon signed-rank test was used for statistical analysis. Results: All GRE sequences facilitated a good delineation of the mesh implant. GRE1 was rated best (3.6, p<0.05) for diagnostic quality with respect to the mesh whereas both GRE2 and GRE3 were suited best for evaluation of mesh localisation in respect to the hernia (3.3, p<0.05). TSE was preferred for evaluation of the anatomy (3.8, p<0.05) but insufficient in mesh delineation. Conclusion: Using a combination of different MRI sequences, iron-loaded mesh implants can be clearly visualised for localization and configuration after hernia repair. Their use could help to identify mesh-related problems and reduce the need for surgical revision. Females; mean age 60 years) with pathological proven liver metastases. 23 patients had a tumor in the head (Group A) and 28 in the bodytail (Group B). We analyzed site, diameter and vascular invasion of the pancreatic adenocarcinoma and number of metastases in each lobe of the liver using Cantlie's line. Total number of metastases was compared between the two groups with unpaired t-test while Fisher's test was used to compare the number of metastases in the two lobes. Results: As expected, the number of liver metastases was significantly higher in group B than in group A (p<0.05). The ratio of metastases in the right and left lobes was 6.5:1 for group A compared to 3.3:1 for group B (p=0.0018). Conclusion: Although liver metastases are more numerous in the right than in the left lobe in both groups, there is a significant difference in the ratio of metastases between the right and the left hemiliver. This can support the existence of a "fast track" to the left liver lobe when the carcinoma invades the splenic vein, and may help in detection of liver metastases. To discuss the correlation between CT and pathology in T4 pancreatic cancer. Material and Methods: 31 arteries (6 celiac axis, 10 hepatic arteries, 8 splenic arteries-SA, 7 superior mesenteric artery), resected in 17 patients were analyzed. The relationship between tumor and artery was graded as: 0, no contact; I, focal contiguity; II, tumor surrounding partially (a) or completely (b) the artery; III tumor surrounding (a) or (b) the vessel with lumen reduction. For each artery, the contact length with the tumor was reported. The neural plexus (NP) and celiac ganglia (CG) involvement was graded with a score between 1 and 3. Results: At CT, 1 resected artery was graded as 0, 2 as I, 17 as IIa, 3 as IIb, 5 as IIIa and 3 as IIIb (all SA). At pathology, none of 0, I and IIa arteries was infiltrated, irrespective to the NP and CG score. Adventitial infiltration was identified in 2 IIb, 2IIIa and all IIIb arteries. Arterial infiltration was significantly related to CG involvement (p=0.014), while no correlation with NP grading was found (p=0.37). No correlation with survival existed considering arterial infiltration (p=0.66), while significance (p=0.0097) was reached considering CG involvement. Conclusion: Radiological T4 poorly correlates with pathology. Current CTgrading system should be revised because of the structural differences between arteries and veins, and the key role of CG involvement in predicting arterial infiltration and prognosis. Computed tomography (CT) is the gold standard for PC detection and its resectability evaluation. Material and Methods: Preoperative radiology data were compared with findings at 51 standard, 58 extended, 17 total pancreaticoduodenectomies (PDs), 9 distal resections with CA excision (DPCA) and 28 palliative bypasses for PC. Survival of 11 patients with controversial data of CT and EUS in regard to arterial invasion, after R0/R1 procedures, was compared to survival of 8 patients after R2 resection (Group B) and of 12 patients with locally advanced cancer after bypass surgery (Group C). Results: CT showed peripancreatic arteries' encasement in 11 cases. In all cases an operative exploration was performed, basing on equivocal yield of endoUS, and no invasion of the arterial wall was revealed. The 1-year survival in Group A was 88.9%, 2-year -26.7% with 22-month median follow-up. Oneyear survival was not attained in Groups B and C. Survival difference between groups was significant (Pa-b = 0.0029,Pb-c = 0.003) Conclusion: Peripancreatic arterial encasement on CT does not necessarily signify arterial wall invasion, which means that PC can still be radically removed. Whenever PC is considered unresectable endoUS should be used; without recourse to an extended pancreatectomy with skeletalisation of the SMA and CA, PC resectability cannot be reliably appraised based on CT data.  The diagnosis of pancreatic tumour was suspected following non-specific abdominal pain (n=24) or fortuitously (n=12). It was located in the head (n=19), the neck (n=10) or the body/tail (n=7). All patients underwent CT scan, MRI and EUS. In doubtful cases, EUS-biopsy or somatostatin-receptors scintigraphy (SRS) were performed in 12 and 6 patients, respectively. Results: In 9 cases (25%), there was a differential diagnosis with a neuroendocrine tumour (NET) (n=6) or a mucinous cystadenoma (n=3), underlying two atypical patterns, a solid well-circumscribed lesion pattern, and a purely cystic pattern some with large shell-like calcifications. A peripheral fibrous capsule was present in 22 cases (61%) and this pattern was not described with other cystic and solid lesion. Conclusion: This series suggests that the presentation of SPT has changed with half of patients aged >30 years and/or with small solid lesions (<3 cm), 1/3 male. Atypical imaging features (entirely cystic or NET-like lesions) can be encountered as well. Material and Methods: Twenty-two consecutive patients with pancreatic adenocarcinoma who underwent MRI (1.5 or 3T) before surgical resection were included. Fat-suppressed (FS) T1 and T2-weighted sequences, 3D FS dynamic T1-weighted gadolinium-enhanced gradient-echo (GRE) during arterial, portal and delayed phases, and diffusion-weighted imaging (DWI) with b values of 600-800 s/mm 2 were obtained. Lesion conspicuity was assessed on each sequence qualitatively (three-point rating scale) and quantitatively (tumour-toproximal and distal pancreas contrast), and compared using paired Wilcoxon tests. Histological characteristics were correlated with MRI features. Results: 95% of pancreatic adenocarcinomas were hypointense on 3D FS T1 GRE arterial phase, which was the best sequence for tumour conspicuity (p≤0.02). DWI was not useful for delineating 25% of tumours. Maximum diameter at pathological examination was 33±10 mm. It was best correlated with MR tumour size on DWI. Progressive enhancement curve was associated with extensive and dense fibrous stroma (p≤0.03). No correlation was found between ADC (mean value: 1.76x10 -3 mm 2 /s) and differentiation, fibrosis or necrosis. Conclusion: 3D FS T1 GRE arterial phase sequence is superior to DWI for pancreatic adenocarcinoma conspicuity but it underestimates the size of the tumour. DWI could be the best sequence for size evaluation when the tumour is correctly delineated.

Oxford/UK
Purpose: ADC measurements have been proposed to characterize focal liver lesions in the basis of restricted diffusion pattern due to hypercellularity. The aim of this study was to explore whether whole liver ADC histogram metrics may be used to differentiate between patients with colorectal liver metastasis and controls with no liver disease. Material and Methods: Ten patients with colorectal liver metastasis and 10 controls with no focal or diffuse liver disease were included. Whole liver segmentation was performed on ADC maps and a corresponding histogram was generated. In addition, quantitative histogram metrics were calculated and compared between the patients and normal controls. P-value <0.05 was considered statistically significant. Results: Mean ADC value of patient group was significantly lower than that of normal controls (0.958x10 -3 and 1.22x10 -3 , respectively, p<0.05), while both skewness and kurtosis were not significantly different (p=0.176 and p=0.056, respectively). In the patient group mode, 5% percentile and 95% percentile were significantly lower to the normal controls (0.936x10 -3 and 1.140x10 -3 , p<0.001, 0.475x10 -3 and 0.7x10 -3 , p<0.05, 1.466x10 -3 and 1.833x10 -3 , p<0.05). Conclusion: Whole liver histogram ADC analysis revealed a significant shift towards lower ADC values in patients with colorectal liver metastasis compared to controls without liver disease. These results show that ADC histogram analysis could be a promising tool for detecting occult colorectal liver metastasis. In the initial patient cohort, none of the evaluated parameters was found to be a statistically significant predictor of TTP. This patient cohort included 14 patients who had received treatment with SSA. Since treatment with SSA was associated with an increased TTP (p=0.01), we also analyzed a subgroup of 30 patients without any antitumoral therapy. In this subgroup of patients, hypoenhancement of hepatic metastases during early contrast phases was found to be an independent prognostic factor predicting rapid tumor progression (p=0.018). Conclusion: Hypovascularization of hepatic metastases from G1 and G2 pNET reflected by hypoenhancement during the early contrast phases seems to be associated with rapid tumor progression. In patients with hypoenhancing metastases, early initiation of antitumoral therapy as well as repeated biopsy for grading of these metastases should be considered.

SS 7.08
The preoperative assessment of hepatic tumours: evaluation of UK regional multidisciplinary team performance M.G. Wiggans, E.M. Armstrong, D.A. Stell, S. Jackson; Plymouth/UK Purpose: In the UK, all patients where liver resection is contemplated are discussed at hepatobiliary multi-disciplinary team (MDT) meetings. The Peninsula HPB unit is a tertiary referral facility in the South-West of England. Our aim was to assess MDT performance by identification of patients where radiological and final pathological diagnoses differed. Material and Methods: A prospective database of cases has been maintained since the inception of the unit. The presumed diagnosis as a result of radiological investigation and MDT discussion is recorded at the time of surgery. Imaging was reviewed by specialist gastrointestinal radiologists and results agreed by consensus. A review of patients undergoing surgery from Mar 2006 to Jan 2012 was performed. Results: We identified 417 patients who underwent hepatic resection. There was a significant increase in the use of preoperative imaging modalities (p≤0.01) but no change in the rate of discrepant diagnosis over time. 42 individuals were identified whose final histological diagnosis differed to the outcome of the MDT discussion (9.6%). These included 30% of patients diagnosed pre-operatively with hepatocellular carcinomas and 25% diagnosed with cholangiocarcinoma of a major hepatic duct.

Conclusion:
The highest rate of discrepancies occurred in patients with focal liver lesions without a history of chronic liver disease or primary cancer, where hepatoma was over-diagnosed during MDT discussion and peripheral cholangiocarcinoma under-diagnosed. Additional care should be taken in these groups and high-quality preoperative multi-modality imaging considered.  (n=10) was performed with motility MRI at baseline and repeated at mean 4 weeks (range 2-7). Two readers drew regions of interest around small bowel independently, and motility was quantified using a registration algorithm that provided a global motility metric in arbitrary units (AU). Repeatability of mean baseline AU was assessed using Bland Altman (BA) and within-subject coefficient of variation (wCV). Changes in mean AU following drug administration was compared to placebo using paired t-testing. Results: Repeatability between baseline measurements of motility was high; BA mean difference -0.0025, range 0.28 to 0.4, 95% Limits of Agreement at ±0.044AU and wCV of 4.9%. Measured motility following neostigmine (mean 0.39AU) was significantly higher than placebo (mean 0.34AU), p<0.001, whilst that following Butylscopolamine (mean 0.13AU) was significantly less than placebo (mean 0.30AU) p<0.001. Conclusion: MRI-generated software-quantified small bowel motility in healthy volunteers is repeatable and sensitive to changes induced by pharmacological manipulation.

SS 8.02
Validation  Material and Methods: CD-CT and LD-CT datasets were acquired in 34 consecutive patients with active CD. LD-CT images were reconstructed with MBIR, 40% and 70% adaptive statistical iterative reconstruction (ASIR). Image quality parameters were subjectively scored and image noise was objectively measured. LD-CT images were clinically interpreted by 2 radiologists in random order with interpretation of the CD-CT images after a 6-week delay. Results: Mean ED was 1.27±0.87 mSv for LD-CT compared with 4.8±2.99 mSv for CD-CT. Reconstructing LD-CT images with MBIR resulted in a significant decrease in mean objective noise when compared with ASIR-40 and ASIR-70 images (p<0.001). No acute complication of CD was missed on the LD-CT MBIR images with excellent statistical agreement between the findings identified on LD-CT MBIR and CD-CT images (k= 0.871). There was also substantial interrater agreement with LD-CT MBIR interpretations (k=0.715). Conclusion: Abdominopelvic CT at doses close to 1 mSv reconstructed with MBIR detected all acute complications in patients with active CD. This protocol has potential to effect substantial cumulative effective dose reductions in this cohort. Results: MDCT revealed hepatomegaly in 6/11 patients, 5 confirmed by macroscopic evaluation. All patients showed smooth liver surface at MDCT imaging and at macroscopic examination. A diffuse hepatic parenchymal hypoattenuation was observed in 10/11 patients, correlating with the diffuse hepatocytes necrosis at histopathology. In all exams, we found periportal hypoattenuation and in 7 of them diffused peribiliary THADs corresponding to periportal inflammatory infiltration and ductular proliferation. All patients had thickened gall-bladder wall corresponding to inflammatory infiltration. Ascite was present in 9/11 patients, splenomegaly in 6/11 and thickened colon wall in 3 patients. Conclusion: A diffuse hepatocytes necrosis and an inflammatory cell infiltration were the main histopathological findings on explanted liver for fulminant hepatic failure. They were the pathological basis of the main features founded on MDCT imaging: diffuse parenchymal and periportal hypoattenuation and thickened gall bladder wall. The regenerative liver processes might be at the basis of parenchymal heterogeneous density, such as THAD.

SS 9.10
Opportunistic screening for hereditary hemochromatosis using unenhanced CT: determination of an optimal liver attenuation threshold P.J. Pickhardt The purpose of this study was to clarify whether quantitative color map of arterial enhancement fraction (AEF) generated with arterial and portal phase images of contrast-enhanced dual energy CT (DECT) can reflect hemodynamic differences of hepatitis C viral (HCV)-related chronic liver diseases. Material and Methods: 22 patients who underwent abdominal multiphasic DECT for closer examination of HCV positive hepatic disease (chronic hepatitis, CH-C n=12; cirrhosis without major portosystemic collateral, LC-C PS (-) n=7; cirrhosis with major portosystemic collateral, LC-C PS (+) n=5), and normal control liver (NL n=8) were included in this study. AEF color map was generated with prototype software. The mean AEF of each groups were compared with t-test. In addition, AEF of hepatic central zone (AEF-cz) and hepatic peripheral zone (AEF-pz) were also compared within each group. Results: Average AEF in each groups were as follows; CH-C -57.0±11.4%, LC-C PS(-) -61.0±10.9%, LC-C PS(+) -44.9±18.8%, NL -44.7±10.3%. There were significance differences between CH-C and NL (P=0.02), LC-C PS (-) and NL (P=0.01). However, there were no significant differences between AEF-cz and AEF-pz within each group. Conclusion: Quantitative AEF map generated with DECT might be useful as objective index to assess the hemodynamic differences of HCV-related chronic liver diseases.

SS 9.08
Gd Material and Methods: 75 patients underwent Gd-EOB-DTPA-enhanced MRI on a 3T system. Patients were classified into two groups: MELD-Score < 9 (n=49) and MELD-Score ≥ 9 (n=26). Two TurboFLASH sequences (TI = 400 ms, 1000 ms) were acquired before and 20 min after Gd-EOB-DTPA administration to obtain T1 maps. T1 relaxation times were determined indicating Gd-EOB-DTPA liver-uptake and correlated to the MELD-Score. Results: Significant changes between T1 relaxation times of non-enhanced MRI (778 ± 133 ms) and Gd-EOB-DTPA-enhanced MRI (339 ± 120 ms) were observed (p<0.001). T1 relaxation time for non-enhanced MRI showed no significant differences (p=0.526) between the group with MELD-Score < 9 (777 ± 150 ms) and the group with MELD-Score ≥ 9 (780 ± 97 ms). After administration of Gd-EOB-DTPA T1 relaxation time of patients with MELD-Score < 9 (299 ± 103 ms) and patients with MELD-Score ≥ 9 (416 ± 112 ms) shows a significant difference (p<0.001). Conclusion: Patients with advanced liver disease showed significantly lower changes in T1 mapping. Therefore, evaluation of changes in T1 mapping of the liver parenchyma may serve as a useful method to determine whole liver function, to improve the estimation of segmental liver function and finally to define the grade of liver disease. Report included the following information: tumor location, longitudinal extent and maximum thickness parietal involvement; lumen stenosis; distance from the distal margin of the lesion and puborectal muscles; extent of extramural invasion; lesion morphology (infiltrative or nodular); distance from the mesorectal fascia; relationship with elevator muscles and extramural vessels; lymph node involvement. Diagnostic accuracy and mean reporting time was assessed. Results: Diagnostic accuracy resulted superimposable in structured and nonstructured reporting (75% and 77%); mean reporting time resulted in 15 and 20 min, respectively.

Conclusion:
The main advantages of a well-structured report include reduction of reporting time, the possibility of creating an easily accessible database and clearer description of imaging findings to clinicians, allowing for detailed comparison with previous MRI examinations. years, mean 62 years) with a total of 44 lesions (37 sized between 6 and 9 mm, and 7 between 10 and 30 mm) were selected from a screening population for colorectal cancer (500 patients). Patients underwent CTC on a 64-row CT scanner following a 3-day low-residue diet plus a low dose of Macrogol 4000 and bisacodyl after lunch, and immediately before CTC they were rectally administered 50 mL of diatrizoate dimeglumine (Gastrografin ® , Bayer Schering) diluted in 300 mL warm tap water, followed by automatic insufflation of 3 L carbon dioxide. Patients were asked to turn themselves on the CT table to ensure homogeneous luminal enhancement. CTC findings were compared with optical colonoscopy, and per-segment image quality was assessed visually using a semiquantitative score (1=poor, 2=adequate, 3=excellent). Results: Sensitivity and specificity of CTC in lesion detection were 94.6% (CI 95% 80.5±99.1%) and 98.8% (CI 95% 95.3±99.8%) for lesions between 6 and 9 mm, and 100% (CI 95% 56.1±100%) and 99.4% (CI 95% 96.2±100%) for lesions between 10 and 30 mm, respectively. Mean image quality was adequate to excellent in all colonic segments (from 2.3 in the cecum to 3 in the sigmoid colon). Conclusion: CTC performed with rectal tagging showed high sensitivity and specificity for all lesion sizes with overall good image quality.

SS 10.02
Technical feasibility of elastosonography in the assessment of bowel wall in inflammatory bowel disease and inter-observer variability A. Colleoni 1 , L. Romanini 1 , M. Ravanelli 1 , V. Cantisani 2 , L. Grazioli 1 , P. Ricci 2 , R. Maroldi 1 ; 1 Brescia/IT, 2 Rome/IT Purpose: To investigate the feasibility of US-elastosonography to evaluate the abnormal bowel wall and the inter-observer variability as well.

Material and Methods:
In this prospective study, we examined 35 consecutive patients who underwent US examination of the bowel and showed at least one segment with thickened wall underwent an additional elastosonography assessment. Overall, 39 abnormal thickened bowel segments were evaluated. GE E9 equipment was used, with a linear 6-15 MHz probe. In 39 segments, the elastosonography was acquired with 7.5 MHz in 30/39, with 5 MHz in 33/39, and in 24/39 with both frequencies. All studies were performed by the same operator. The images acquired were subsequently evaluated by two other radiologists, separately. Inter-observer variability was calculated by Cohen's test (between two operators) and Randolph's test (comparing three operators). Results: All but 4 segments (35/39; 89.7%) were assessable by elastosonography. The elastosonography was considered adequate when the wall layers distinguished on standard ultrasound were equally detected in the color map (grading wall stiffness). The average thickness of the bowel segments was 6 mm. The Randolph test demonstrated a moderate agreement among the three operators (63%), while Cohen test provided a lesser agreement (k=0.39).

Conclusion:
The preliminary results indicate the feasibility of USelastosonography assessment of abnormal bowel segments in inflammatory bowel disease. This could offer the possibility to differentiate fibrotic from inflammatory thickening. However, only moderate correlation was observed. Material and Methods: Twenty-three patients submitted to LSG, complaining upper GI symptoms and/or weight regain and scheduled for a sleeve surgical revision were investigated. All patients underwent MDCT scan, upper GI barium swallow study and endoscopy. MDCT was compared to barium and endoscopy features as concern: esophageal dilatation, neo-fundus development and volume, hiatal hernia, sleeve size in toto and atrum dilatation. All patients underwent laparoscopic sleeve revision. Surgical findings were considered "as gold standard". Results: A total of 21 patients with hiatal hernia, neo-fundus or sleeve dilatation underwent surgical correction. All findings identified at MDCT were confirmed by intraoperative findings. The presence of hiatal hernia was significantly underestimated by both barium and endoscopy with a sensitivity of 57.1% and 50%, respectively (P=0.04, P=0.02). Conclusion: MDCT is an accurate method for the detection of hiatal hernias and quantification of gastric volumes and can be considered as non-invasive method to guide surgery.

SS 10.05
MDCT evaluation of bowel obstruction: can the radiologist's experience make a difference? R. Basilico, V. Calamita, A.R. Ferri, E. Rodolfino, N. Civitareale, A. Lella, A.R. Cotroneo; Chieti/IT Purpose: To evaluate the diagnostic performance of MDCT in the detection of site and cause of bowel obstruction and in the diagnosis of bowel wall ischemia or infarction, by assessing the added value of the radiologist's experience.
Material and Methods: The MDCT reports of 110 patients with surgical and istological diagnosis of bowel obstruction were retrospectively analyzed. We calculated the diagnostic accuracy of MDCT reports in the evaluation of the cause and the site of obstruction and the sensitivity, specificity, diagnostic accuracy values in the detection of bowel wall ischemia or infarction. The same MDCT examinations were then reviewed by an experienced abdominal radiologist, to compare her results with those of the original MDCT reports.

Results:
The diagnostic accuracy in determining the cause of bowel obstruction and in defining the site of obstruction were respectively 89% and 91% for MDCT reports and 92% and 94% for the experienced reader. The sensitivity, specificity and diagnostic accuracy of MDCT reports in identifying bowel wall ischemia or infarction were respectively 73%, 96% and 91%. The sensitivity value for bowel wall ischemia or infarction significantly increased to 83% for the abdominal radiologist. Material and Methods: 104 patients with primary gastric cancer (mean age 68.67 years) who consecutively underwent MDCT scan followed by radical surgical treatment were prospectively evaluated. Regional lymph nodes were considered involved when the short-axis diameter was >5mm for the lymph nodes of group 1 and >8mm for the lymph nodes of other group according to the Japanese Classification of Gastric Carcinoma. All patients underwent a radical lymph node dissection (D2-D3) according to Japanese Research Society for Gastric Cancer (JRSGC) guidelines. The removal of nodal stations was always preceded by Indian-ink injection in the lesser and greater curvature of the stomach; after operation, single lymph nodes were retrieved on the fresh specimen by the surgeon, and classified in JRSGC nodal stations for pathological examination. Results: Lymph node invasion was found in 85 cases (81.73%) with a MDCT sensitivity and specificity of 89% and 85%, respectively. The rate of understaging was higher (15%) than that of overstaging (8%). Lymph node status of early forms was correctly staged in all cases. Furthermore, all N3 cases were correctly staged. Conclusion: MDCT is a useful technique in the preoperative assessment of lymphatic cancer spread and could have a positive impact in clinical decision making in the era of neoadjuvant treatment. To quantitative analysis, a radiologist measured signal intensities of the lesion, the liver and the spleen, and calculated liver-spleen contrast, lesion-to-liver contrast-to-noise ratios and SAR. To qualitative analysis, three reviewers independently reviewed the image sequences using a five-point rating scale, focused on hypointense lesion detection and the image qualities. Results: The high FA sequence had significantly higher liver-spleen contrast and lesion-to-liver CNR compared to those of low FA (p<0.05, respectively). The per-lesion and per-person sensitivities of high FA were higher than those of low FA in all three reviewers (p<0.05, respectively). There was statistical differences for the detection of HCCs smaller than 1.5 cm in two of three reviewers (p<0.05, respectively). Interobserver agreements were more than moderate degree. Conclusion: Increasing the FA in T1-weighted hepatobiliary phase liver MRI with gadoxetatic acid improves lesion detection and conspicuity for smaller lesions and helps the diagnosis of HCC.

SS 11.09
Imaging Material and Methods: This retrospective study includes 10 patients (10 males, mean age 65 years) who had either an hepatectomy (n=5) for HCC and CC, either a liver transplantation (LT) (n=5) for HCC. All the patients were evaluated with presurgical imaging study by US (n=10), MRI (n=6) and CT (n=10). All patients had a clinical history of chronic liver disease and 4/10 were treated with presurgical Transarterial Chemoembolization (TACE). After histologic and immunohistochemical stain confirmation, two radiologists analyzed the imaging features of cHCC-CC: size, shape, margins, signal intensity, components of tumor, enhancement pattern, satellite tumors, metastasis, lymphadenopathy, vascular invasion, biliary dilatation and findings of liver disease. Results: The cHCC-CC was a single mass (8/10) with a mean size of 4 cm (range 1-10 cm), heterogeneous signal (6/10), hyperintensity on T2-weighted and hypointensity on T1-weighted images (6/10). On post-contrast imaging, cHCC-CC were divided in 3 enhanced patterns: HCC-like (early enhancement with wash-out on later phases, 4/10), CC-like (early ring-enhancement with progressive enhancement in central portions, 4/10) and hypovascular pattern (2/10). Other findings included chronic hepatopathy (10/10), late capsule enhancement (2/10), lymphadenopathies (2/10), portal venous invasion (1/10) and biliary dilatation (1/10). Conclusion: In our study, cHCC-CC presents in male patient most commonly as a single mass, heterogeneous, hyperintensity on T2-weighted images with 2 predominant enhanced patterns. Purpose: Patient compliance in ingestion of oral contrast can be crucial in obtaining a technically adequate study. We assessed the palatability and preference of four iodinated oral contrast agents commonly used in abdominopelvic CT and CT colonography (CTC). Material and Methods: 80 volunteers sampled four common contrast agents, Omnipaque, Telebrix, Gastromiro and Gastrografin in a computer-generated random order. Each agent was anonymised and diluted to a standard concentration of 30 mg iodine/ml, the dilution used for faecal tagging in our institution. A 100 mm visual analogue scale and a 7-point Likert scale were used to rate palatability. Willingness to drink 3 aliquots of 200 ml of each of the contrast agents over a 48 h period was also assessed, as per our institution's current bowel preparation protocol for CTC. Results: Gastrografin was rated significantly less palatable than the remaining agents (p<0.005). Omnipaque and Telebrix were significantly more palatable than Gastromiro. No difference existed between Omnipaque and Telebrix. 39% of participants would refuse to consume the quantities of Gastrografin required for a CTC examination compared to Telebrix (7%) and Omnipaque (9%) (p<0.05). Conclusion: Omnipaque and Telebrix are significantly more palatable than both Gastromiro and Gastrografin with participants more willing to ingest them in larger quantities, thereby reiterating the influence of palatability on patient compliance. We feel that this will directly influence the technical adequacy of CTC studies.

SS 14.06
CT colonography: clinical evaluation of novel software to automatically co-register polyps between follow-up surveillance studies E. Helbren 1 , H. Roth 1 , T. Hampshire 1 , P.J. Pickhardt 2 , D. Hawkes 1 , S. Halligan 1 ; 1 London/UK, 2 Madison, WI/US Purpose: To evaluate the software that aims to automatically identify and then register polyp location across sequential CTC acquisitions performed for polyp surveillance.

Material and Methods:
We developed a registration algorithm to match endoluminal colonic surfaces between prone and supine CTC acquisitions. Haustral folds were automatically matched between scans to initialize a registration method that establishes correspondence over the entire colonic surface via non-rigid registration. Initial and follow-up CTC from 11 patients (14 polyps) undergoing surveillance were selected and the algorithm was tested via two methods: "Direct" (the algorithm used polyp co-ordinates from the initial prone and supine acquisitions to automatically predict polyp location on follow-up CTC) and "loop" (polyp co-ordinates from initial supine acquisition were used to predict polyp location on the initial prone acquisition, then followup prone, follow-up supine and back to initial supine, respectively). Registration accuracy was assessed by two observers via the Euclidean distance between true and predicted polyp locations. Results: Successful polyp registration was achieved in all 11 cases for all 14 polyps using both loop and direct methods. Mean Euclidean registration error for individual polyps was 2.0 cm (range 0.1 to 9 cm) for "direct" method and 3.1 cm (range 0.3 to 8.5 cm) for "loop" method. To determine how healthcare professionals balance the potential benefits of detecting extracolonic cancer at screening CT colonography (CTC) versus risks of unnecessary investigations due to false-positives. Material and Methods: 30 healthcare professionals underwent a discrete choice experiment. Participants chose between two hypothetical tests for colorectal cancer screening. One was unable to diagnose extracolonic cancer but generated no false positives (FP), whereas the other diagnosed a curable extracolonic cancer in 1 in 600 screenees, but generated FP. The FP rate was varied systematically to determine the "tipping point" at which the rate was deemed sufficient to outweigh the benefit of detecting cancer. FP leading to follow-up imaging and those precipitating invasive tests were presented separately. Results: 14/30 (47%) participants were prepared to tolerate any rate of followup imaging to diagnose 1 extracolonic cancer in 600 screenees. Across all participants, the mean rate of further imaging deemed acceptable was 53.5%. Invasive tests were less acceptable: The mean "tipping point" was an invasive test rate of 9.7% to diagnose 1 in 600 screenees with a curable extracolonic cancer. Conclusion: Healthcare professionals will tolerate numerous extracolonic FP by CTC if the consequence is simply further imaging. Tolerance for invasive tests is lower. The specificity of CTC for clinically significant extracolonic findings is likely to be highly acceptable to healthcare professionals, including surgeons, gastroenterologists and radiologists.

SS 14.09
Accuracy of CT colonography in detecting colorectal cancer at our District General Hospital in North Wales S. Sinha, C. Corr; Wrexham/UK Purpose: To assess accuracy of CTC in diagnosing colorectal cancer (CRC) at our hospital with respect to nationally and internationally agreed standards.

Material and Methods:
A retrospective study analysing all CTC's from November 2006 to May 2011 was performed. Radiology reports were reviewed and correlated with data of all patients with CRC diagnosed up to November 2012, which was obtained from pathology and cancer services. The following were recorded from radiology reports: demographics; cancer diagnosis; equivocal findings and extra-colonic cancers. Results: 841 CTC's were performed of which 47 had confirmed CRC identified as cancer, suspicious for cancer, or suspicious polyp on CTC. Sensitivity and negative predictive value (NPV) were 100%. 40 cases were identified as definite CRC with no colonoscopy advised and 39 were proven to be cancer. Specificity was 97.5% for definitive CRC diagnosis. 16 cases were 'suspicious but uncertain' and colonoscopy advised -6 of these were cancer. 18 cases diagnosed as benign were confirmed as benign. Combined specificity with CTC and colonoscopy for CRC diagnosis was 96%. 30 unexpected extracolonic malignancies were identified where 3 were concurrent with CRC. Conclusion: CTC is very accurate for diagnosing CRC with high NPV and sensitivity. Specificity is also high and increases with the complementary use of colonoscopy in equivocal cases. CTC should be used as a first-line imaging investigation for diagnosis of CRC.

SS 14.10
Appendiceal length as an independent risk factor for acute appendicitis P.J. Pickhardt, J. Suhonen, B.D. Pooler; Madison, WI/US Purpose: To determine if appendiceal lengths differ among adults with proven acute appendicitis compared with normal adult controls. Material and Methods: Preoperative MDCT in 321 consecutive adults with surgically proven appendicitis was compared with MDCT in 321 consecutive adult controls undergoing CTC screening. Appendiceal length at MDCT was obtained using curved reformatted images along the appendiceal long axis. For validation, MDCT length was compared with gross pathology length for the appendicitis group. Results: Appendiceal length at MDCT correlated well with gross pathology (mean length, 6.76 vs 6.56 cm). Mean appendiceal length and standard deviation of length at MDCT were both significantly greater in controls compared with the acute appendicitis group (7.87±3.49 cm vs 6.76±1.94 cm; p<0.001). Appendicitis cases outnumbered controls at every 0.5-cm interval between 4.0 cm and 9.5 cm. The odds ratio (OR) for acute appendicitis within the 4.0-9.5 cm interval was 5.4 compared with controls, and increased to 37.9 for the 2.25-13.75 cm interval. 88.5% (284/321) of appendicitis cases fell within the 4.0-9.5 cm range, compared with only 56.1% (180/321) of controls (p<0.001). Conclusion: Appendiceal length strongly correlates with the likelihood of acute appendicitis. Specifically, "intermediate" appendiceal lengths (4.0-9.5 cm) are more frequently complicated by acute appendicitis, whereas both "long" (>9.5 cm) and "short" (<4.0 cm) appendiceal lengths are more frequently seen in adult controls and may be more resistant to developing appendicitis.