ESGAR 2016 Book of Abstracts

REVIEWING PANEL S562 D. Akata (Ankara/TR) O. Akhan (Ankara/TR)

Authors' Index S604-S609 Purpose: Majority of patients with pancreatic cancer present with metastatic and locally advanced disease at the time of initial diagnosis. The action to control cardiovascular risk in diabetes trial demonstrated that the survival in pancreatic adenocarcinoma treated with neoadjuvant FOLFIRINOX (fluorouracil, leucovorin, oxaliplatin, and irinotecan) is improved when compared with gemcitabine-based regime. We evaluated CT examinations of patients with adenocarcinoma before and after FOLFIRINOX to assess neoplasm modifications; we also compared imaging findings with histopathological results given by surgery or biopsy. Material and methods: Twenty-two patients with pancreatic adenocarcinoma treated with FOLFIRINOX were enrolled; we evaluated: longest dimension and volume of the neoplasm, density of the lesion, vessels involvement, metastatic lymphnodes, perineural invasion and resectability (according to Americas hepato-pancreato-biliary association guidelines). After FOLFIRINOX every patient underwent exploratory laparotomy with multiple biopsies of peritumoral tissue.
Results: After FOLFIRINOX treatment we observed significant reduction of tumor volume and longest dimension (p<0,01); increase of density and vascularisation; reduction of cases with major vessels involvement. In 6/22 patients (27.3%) CT showed vessels encasement but biopsy results were only fibrosis. CT sensitivity and specificity to assess perineural invasion were, respectively, 22% and 86%; CT sensitivity and specificity to evaluate resectability were 92% and 60%. CT did not recognize metastatic lymphnodes.
Conclusion: FOLFIRINOX is a highly effective treatment for advanced pancreatic adenocarcinoma; CT is indicated to evaluate tumor response, but it might underestimate the percentage of successfully resectable patients. SS 1.3 Correlation between appearance of the retroportal fat plane at preoperative CT and pathology findings in resected adenocarcinoma of the pancreatic head F. Lombardo Material and methods: We included 41 patients with resected PDAC of the pancreatic head (25M, 16F, mean age 65 years). All patients had a multiphasic preoperative MDCT. All cases were re-evaluated at pathology for the state of the retroportal lamina, lymphnode and perineural invasion. CT images were reviewed in consensus by two radiologists for assessment of the fat plane between the pancreatic head and the mesenteric artery and vein: this was graded in 3 categories (clear, effaced, infiltrated). Fisher's test was used to assess the correlation between CT and pathology findings.
Results: A clear fat plane between the pancreatic head and the mesenteric vessels was significantly associated with a negative retroportal lamina at pathology (p=0.0266). This was also observed for the plane between the head and the superior mesenteric artery (p=0.0011) and the superior mesenteric vein (p=0.0327). No different results were observed between effaced and clearly infiltrated fat. No association was observed between the appearance of the fat planes at CT and the presence of lymphnode or perineural invasion (p=n.s.).
Conclusion: A clear fat plane between the pancreatic head and the mesenteric vessels is significantly associated with negative retroportal lamina at pathology. CT is not accurate in predicting perineural or nodal invasion.

SS 2.4
Noninvasive biomarker for prediction of treatment response to concurrent chemo-radiotherapy in patients with locally advanced HCC Y.E. Chung Purpose: To investigate noninvasive biomarker for prediction of treatment response in patients with locally advanced HCC. Material and methods: Thirty patients (55.5±10.2 years old, M:F=24:6) who underwent concurrent chemo-radiotherapy (CCRTx) due to advanced HCC were prospectively enrolled. Contrast-enhanced US (CEUS) and perfusion MRI were obtained before and immediately after CCRTx. The third CEUS was obtained 1 month after the end of CCRTx. Response assessment was performed 3 months after the end of CCRTx based on RECIST criteria. Quantitative biomarkers measured by CEUS including peak enhancement (PE), time to peak enhancement (TTP), and mean transit time (MTT) and by perfusion MR including volume transfer constant (Ktrans), reflex constant (Kep), and extravascularextracellular volume fraction (Ve) were compared with the Mann-Whitney U test. A cutoff value of PE was calculated with ROC analysis and overall survival (OS) was compared by Breslow method. Results: PE of the first CEUS before CCRTx was significantly lower in responder (median, 18.6%; IQR, 20.9%) than in non-responder (59.1%, 13.5%; P=0.002). There was no significant difference between two groups in other quantitative biomarkers. In receiver operating characteristic analysis, responder could be diagnosed with sensitivity of 90.9% and specificity of 100% with a cutoff value of 42.6% of PE. OS was also significantly longer in patients with PE<42.6% than the others (P=0.014). Conclusion: Early treatment response and OS could be predicted by PE on CEUS before CCRTx in patients with locally advanced HCC.

SS 2.6
Added value of hepatobiliary-phase and T2-weighted images to the liver imgaing reporting and data system for the diagnosis of HCC: preliminary results G. Rosiak, A. Grodzicka, E. Rosiak, B. Górnicka, O. Rowinski, A. Cieszanowski; Warsaw/PL Purpose: To retrospectively assess the added value of T2-weighted and hepatobiliary-phase (HP) images in the liver imaging reporting and data system (LI-RADS) for the diagnosis of HCC in patients with chronic liver disease. Material and methods: 22 patients (18 men, 5 women) in the mean age of 65 years (range, 34-87) with chronic liver disease and 35 histopathologically confirmed HCC (mean diameter of 33 mm, range: 7-138 mm) underwent magnetic resonance imaging (MRI) with hepatobiliary contrast agent (Gd-BOPTA in 15, Gd-EOB-DTPA in 7 patients). Lesions were solitary in 10 patients and multiple in 12 patients. Two abdominal radiologists evaluated dynamic contrast-enhanced images, T2-weighted images and HP images. In case of disagreement between readers, a consensus interpretation was formed. The diagnostic performance of LI-RADS was assessed and compared to T2-weighted and HP images.
Results: Nineteen of 35 HCCs (54.3%) were classified as LI-RADS category 5 lesions, 12 (34.3%) as LI-RADS category 4 lesions and 4 (11.4%) as LI-RADS 3 category lesions. Combining LI-RADS category 4 lesions with hypointensity on HP images and hyperintensity on T2-weighted images for the diagnosis of HCC would improve the sensitivity from 54.3% (for LI-RADS category 5 lesions only) to 80%. .5 years ± 10; 26-84) underwent subsequent contrast-enhanced CT (n= 56) and Gd-EOB-DTPA-enhanced MRI (n=108). US reports were correlated with CT and MR images. Nodules were classified as true positive (nodules detected by US and confirmed as HCC), false positive (nodules detected by US but not confirmed as HCC) and false negative (HCCs not detected by US). Reference standard for HCC diagnosis was AASLD 2010 criteria and pathology. US sensitivity for HCC detection and agreement among US and CT/MRI were calculated.

SS 2.9
MDCT perfusion imaging biomarkers as prognostic indicators of survival in hepatocarcinoma A. Alberich-Bayarri, A. Torregrosa Andrés, R. Nombela, J. Tomás Cucarella, L. Martí-Bonmatí; Valencia/ES Purpose: To analyze the relationship of perfusion CT imaging biomarkers in treated hepatocarcinoma with overall survival (OS) to add insights into prognosis evaluation of the disease. Material and methods: A total of 28 patients with hepatocarcinoma were included in the study (mean age 68±10). Perfusion MDCT acquisitions were performed in a scanner with 256 detectors (Brilliance iCT, Philips Healthcare, The Netherlands). The acquisition had a voxel size of 0.25x0.25x5mm. A mean volumen of 41±5 ml of iodated contrast was administered (IOMERON 400©). Images were pre-processed by a gaussian filtering and a non-linear registration. The parametric maps of arterial perfusion (AP), perfusion index (PI), arterial time-to-peak (TTP) and permeability (P) were calculated. The pre-processing and parametric maps calculation was performed at the IntelliSpace Portal© platform of Philips. Finally, regions of interest (ROI) were placed in the lesions to register the mean values obtained. OS was calculated using the date of the first MR post-treatment. OS was higher in the group with an HP>=0.6 (354 days) vs HP<0.6 (226 days).
No differences were observed for the other parameters. Conclusion: Perfusion CT and especially the hepatic perfusion index is an imaging biomarker with growing applications in HCC prognosis evaluation, as derived from the preliminary evaluation in this study. Purpose: To assess the accuracy of unenhanced MRI immediately after the percutaneous ablation of liver malignancy in predicting the treatment efficacy at CT follow-up. Material and methods: Percutaneous ablation was prospectively performed in 17 liver malignancies (14 HCCs, 3 METs). After the procedure, on the same day all the lesions were studied with unenhanced MRI. The MRI protocol was essential (T2, FST2, T1, FST1 and DWI) to be performed in short time. The best sequence to detect the coagulative necrosis was established. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and accuracy of MRI in predicting the correct centring and the complete treatment of the lesion were calculated with respect to the standard one month follow-up CT.
Results: Coagulative necrosis was hyperintense in T1 sequences in 13/17 (76%). FST1 was the sequence in which the best conspicuity of the ablated area was depicted. With respect to follow-up CT, MRI predicted the correct centring of the lesions in 13/17 (76%) lesions with sensitivity, specificity, PPV, NPV and accuracy of 100%, 75%, 93%, 100% and 94%, respectively. MRI predicted the complete treatment of the lesions in 12/17 lesions with sensitivity, specificity, PPV, NPV and accuracy of 100%. Conclusion: MRI with the single T1FS weighted sequence was highly accurate in predicting the treatment efficacy of percutaneous ablation of liver malignancies as shown at CT follow-up. Unnecessary CT in case of incomplete treatment can be, therefore, easily avoided. Percutaneous electrochemotherapy of malignant main portal veins thrombosis: a prospective case series L. Tarantino 1 , G. Busto 1 , A. Nasto 1 , R. Fristachi 1 , L. Cacace 1 , M. Talamo 1 , P. Ambrosino 2 , P. Gallo 3 , P. Tarantino 2 , C. Accardo 1 ; 1 Pagani/IT, 2 Naples/IT, 3 Rome/IT Purpose: We report our experience in thrombosis of portal vein and/or firstorder branches (Vp3-4PVTT) from HCC treated with percutaneous electrochemotherapy (ECT). Material and methods: Six cirrhotics (5 males; 61-85 years, 4 in Child-Pugh A, 2 in Child-Pugh B class) prevously studied with three-phase CT, contrastenhanced ultrasound (CEUS) and US-guided percutaneous biopsy of the thrombus, underwent ECT (Cliniporator Vitae -IGEA, Carpi, Italy) of Vp3-4PVTT and the associated HCC nodule in the same session, under general anesthesia.
Results: 24-hour posttreatment CEUS showed complete absence of enhancement of the treated trhombus in all cases. Post-treatment biopsy showed apoptosis and necrosis of tumor cells in all cases. The follow-up ranges from 3 to 12 months (median: 8 months). In 1 patient, the 3-6-9-12 months CEUS and CT demonstrate complete recanalization of the treated PVTT; in another patient, 3-6-9 months CEUS and CT show a partially patent portal vein; other 3 patients (8, 4 and 3 month follow-up, respectively) show avascular complete thrombosis at CEUS and CT. In all 5 patients, CEUS and CT show absence of abnormal intravascular or perivascular enhancement consistent with residual tumor or local recurrence. One patient, was lost to follow-up because of death from gastrointestinal hemorrhage 5 weeks after ECT. Conclusion: ECT seems effective and safe for curative treatment of Vp3-Vp4 PVTT from HCC.

SS 3.2
Microwave ablation of large HCCs by simultaneous multiple antennae insertion: long-term follow-up L. Tarantino 1 , P. Ambrosino 2 , P. Gallo 3 , P. Tarantino 2 , A. Nasto 1 ; 1 Pagani/IT, 2 Naples/IT, 3 Rome/IT Purpose: We report long-term results of microwawe (MW) ablation with simultaneous insertion of multiple antennae in the treatment of large HCC. Material and methods: Between 2008 and 2013, 36 cirrhotics with a single HCC nodule >3 cm (3.2-7.0 cm; mean: 4.4 cm) underwent MW ablation in a single session by simultaneous insertion of multiple 13-gauge-MW-antennae (Viva-Wave, Covidien, USA). After intraoperative contrast-enhanced ultrasound (CEUS), residual viable tumor was treated in the same session by reinsertion of 2-3 MW antennae.
Results: According to the tumor size, 10 and 18 patients were treated with a single insertion of 2-3 synchronous antennae, respectively. 8 patients were treated with 2 insertions of 3 antennae in the same session. Intraoperative CEUS showed residual tumor in 12 patients and these patients underwent an additional insertion of 2 or 3 antennae. Intraoperative CEUS at the end of the procedure showed complete necrosis in all patients. 1-month CT showed complete necrosis in 33/36 patients. Only a severe hemoperitoneum occurred in one patient after treatment. Local recurrence occurred in 7 patients. Recurrences in other liver segments occurred in 35/36 patients within 6 to 24 months (mean: 15 months). 20 patients were alive at 18-78 months follow-up (mean: 42 months). Conclusion: Aggressive ablation of large HCC by simultaneous insertion of multiple MW antennae is safe and seems to result in patients' survival comparing with surgery and small HCC ablation.

Verona/IT
Purpose: To determine the frequency of tumor seeding after percutaneous US-guided FNA of solid pancreatic neoplasms. Material and methods: Follow-up examinations (CT or MR) of 124 patients with solid pancreatic neoplasms who underwent US-FNAs were retrospectively evaluated. Two radiologists evaluated the presence of imaging features suggesting tumor seeding as peritoneal/subcutaneous solid nodules or ascites.
Results: Median follow-up length was 197 days. Subcutaneous nodules along the needle track were found in 1/124 cases (0.8%). Ascites without local or systemic disease progression was found in 7/124 patients (5.6%), after a median time interval of 210 days after FNA. Conclusion: Tumor seeding is uncommon after percutaneous US-FNA of solid pancreatic neoplasms. Ascites without any sign of local/systemic disease progression can develop after US-FNA and may suggest microscopic peritoneal tumor seeding; nevertheless, as the time interval to the development of ascites is comparable to the disease-specific time to progression, ascites may also be the consequence of disease progression unrelated to the procedure.

SS 3.4
Microwave ablation of large HCCs using a new device: a case series L. Tarantino 1 , P. Ambrosino 2 , P. Gallo 3 , P. Tarantino 2 , A. Nasto 1 ; 1 Pagani/IT, 2 Naples/IT, 3 Rome/IT Purpose: We evaluated a device designed to achieve large volumes of necrosis in HCC by synchronous activation of multiple microwave (MW) antennae. Material and methods: 10 consecutive patients with a single large HCC nodule (3.5-6.5 cm; mean: 4.6 cm) underwent US-guided percutaneous MW ablation by synchronous insertion of multiple MW antennae (SynchroWave 915 MHz antennas -MicroThermX® microwave ablation system, Terumo, Belgium, Europe). A single insertion of 2 antennae (3 cases), and 3 antennae (5 cases) was performed. 2 insertions of 3 antennae in the same session were performed in 2 cases. Treatment efficacy was assessed by CT and bimonthly US followup.
Results: Post-treatment CT showed complete necrosis in 8/10 HCC nodules (80%). 2 patients with incomplete ablation underwent an additional MW ablation session. CT showed complete necrosis in both of them. Several major complications (recovered with medical treatment) occurred: anaerobic infection of the treated necrotic area in 2 cases, severe right pleural effusion in one case, jaundice from transient liver failure. All patients are still alive (follow-up 12-20 months). In 6/10 (60%) cases, intrahepatic recurrence occurred within 6-14 months (mean 10 months) and it could be successfully treated with ablation in 3 cases. The other 3 patients underwent chemotherapy with sorafenib and/or best supportive care. Conclusion: The MicroThermX microwave ablation system seems an effective and relatively safe device for treatment of large HCC. Results: The rate of biochemical pancreatitis, which was resolved in 3 days after stent insertion, was significantly higher in group A. The average primary stent patency was 5.9 and 5.7 months in groups A and B, respectively; 3-month and 6-month stent failure was 0% and 6.1% in group A and 6.3% and 25% in group B. The median survival from the initial drainage was 12.3 (6.7-20.1) and 12.8 (5.7-14.7) months in groups A and B, respectively. Conclusion: The effect of an endoluminal ablation on patients survival was not proven in the prospective randomised clinical study. However, in the group of patients undergoing ablation there is a tendency of a lower rate of early stent failure. The intervention should be associated with very mild biochemical pancreatitis. Purpose: To identify the most common pathogens isolated from biliary cultures in patients undergoing percutaneous transhepatic intervention and to assess antibiotic sensitivity. To establish the optimal prophylactic antibiotic regime.
Material and methods: All percutaneous transhepatic interventions performed over a two-year period were reviewed retrospectively. Those where no biliary culture was obtained were excluded. Analysis of the culture results including pathogens grown and antibiotic sensitivity was performed.
Results: A total of 58 patients were included in the analysis. No pathogens were grown in 27.6% of cultures (n=16). Of those with positive cultures Enterococci and Pseudomonas were the most common pathogens grown in 59.5% of cases (n=25). Vancomycin was the most effective antimicrobial demonstrating sensitivity in 31% (n=16) of positive cultures followed by ciprofloxacin showing sensitivity in 26.2% (n=11). Gentamycin was the fifth most effective antimicrobial demonstrating sensitivity in only 14% (n=6). Conclusion: Effective antibiotic prophylaxis requires knowledge of likely pathogens and procedure-specific infection risks, which may vary between hospitals and patient cohorts. In our institution, gentamycin is traditionally administered prophylactically prior to percutaneous biliary intervention; however, this study has demonstrated this is comparably ineffective, necessitating a change in protocol. The choice of antimicrobial requires continued review due to emerging antibiotic resistance.

SS 3.9
Irreversible electroporation of locally advanced pancreatic cancer: case study of a next day postinterventional fluorodeoxyglucose positron emission tomography in one patient L. Lambert Material and methods: 34 cases of the RCHA were evaluated retrospectively using MDCT. We categorized them into the three subtypes according to pancreatic penetration and the passing routes. The distance between the orifice of the superior mesenteric artery (SMA) and RCHA bifurcation (DSMA-RCHA) was measured using advanced 3D imaging software. Analysis of variance was used to evaluate the difference in DSMA-RCHA according to the RCHA subtype.
Results: Type A (n=17, 50%) referred to RCHA penetrating the pancreatic parenchyma, all crossing the dorsal aspect of the SMV. Among them, three cases were accompanied by the circumportal pancreas. Type B (n=10, 29%) referred to RCHA without penetration of the pancreatic parenchyma and crossing of the dorsal aspect of the MPV or SMV. Type C (n=7, 21%) referred to RCHA without penetration of the pancreas parenchyma and crossing of the ventral aspect of the MPV or SMV. The mean DSMA-RCHA of each subtype was as follows: type A, 3.13 cm (95% CI, 2.70-3.57); type B, 2.04 cm (95% CI, 1.40-2.68); and type C, 2.14 cm (95% CI, 2.23-2.92). The DSMA-RCHA of the penetrating pancreatic parenchyma of the RCHA was significantly longer than that of the non-penetrating pancreatic parenchyma (P=0.007).
Conclusion: Half of RCHA shows penetrating the pancreatic parenchyma and this type A takes off from the SMA more distally than RCHA without intrapancreatic penetration.

SS 4.10
CT angiography protocol for pancreatic cancer essential for surgical resectability J. Grubor-Pilipovic, T. Kokovic, U. Milosevic, S. Stojanovic; Novi Sad/RS Purpose: Early tumor detection and accurate radiological staging in patients with pancreatic carcinoma are crucial. The purpose was to review the technical aspects of the CT angiography (CTA) pancreas protocol and the findings relevant to diagnosis and staging of pancreatic carcinoma. Material and methods: CTA pancreas protocol included: IV spasmolytic and oral contrast agent prior to the study, bolus tracking at the level of the celiac axis using an enhancement threshold of 150HU, triple phase acquisition. Major vessels running within 1cm from the tumor margin were evaluated. CT appearance was graded on a 0-4 scale (0: none, 1: <24%, 2: 25-49%, 3: 50-74%, 4: 75-100%) by circumferential contiguity of tumor to vessels. Results: In forty-eight patients, CTA had the highest accuracy in assessing extent of primary tumor (73%), locoregional extension (74%), vascular invasion (83%), distant metastases (88%), tumor TNM stage (46%), and tumor resectability (83%). Surgical correlation of CT findings was available in 89 veins and 83 arteries, and both surgical and histologic correlations were available for 42 veins and 29 arteries. At surgical observation, 29 of 35 veins (82.9%) evaluated as CT grade 3 or 4 were found to be involved, whereas only 18 of 30 arteries (60%) evaluated as CTA grade 3 or 4 were proved to be involved.

Conclusion:
The key to management in pancreatic carcinoma evaluation is determining resectability. It is, therefore, important that radiologists describe in detail the findings that are relevant for staging of pancreatic carcinoma and have a clear understanding of the implications of these findings. Material and methods: 65 patients, diagnosed with locally advanced rectal cancer were prospectively enrolled in the study. All patients underwent MRI on a 3-Tesla before, during and after chemoradiotherapy (CRT). All patients underwent total mesorectal excision (TME). MR-TRG was evaluated on T2weighted fast spin-echo (FSE) multi-planar imaging. The MR-TRG was determined by the fibrosis/tumour ratio and was divided into 4 grades based on the percentage of fibrosis (<25%, <50%, <75%, 100%). Measurements were performed on all axial images including the tumour. MR-TRG evaluated on the second examination (during therapy) was correlated to the pathological finding after surgery, defined as partial response or complete response.
Results: A complete pathological response was observed only in patients with MR-TRG 4 (100% fibrosis) with a negative predictive value of 100%. In lower MR-TRG groups (1, 2 and 3), a partial response was observed. Conclusion: MR-TRG 4 is an accurate predictor of complete response after CRT. When a lower MR-TRG is observed the persistence of disease should be suspected. This method, applied during therapy, may reduce the time to surgery.

SS 5.6
Long-term follow-up features on rectal MRI during "wait-and-see Purpose: Non-operative treatment with stringent follow-up ("wait-and-see") is emerging as an alternative to surgery in clinical complete responders after chemoradiotherapy for rectal cancer. MRI is one of the main follow-up tools.
The aim was to describe the long-term evolution in the morphology of the rectal wall during long-term follow-up of these patients. Material and methods: 68 patients with a sustained complete response during "wait-and-see" follow-up were analysed. Patients underwent MRI 3 monthly (first year) and 6 monthly (second to fifth year). Two readers in consensus analysed the rectal wall morphology on the initial post-CRT scan and the evolution in morphology on the various sequential follow-up MRIs.
Results: Median follow-up was 30 months (range 6-98). 512 MRIs were analysed (median 7, range 3-15/patient). In 7% of patients, the rectal wall completely normalised post-CRT. The other 93% showed a fibrotic remnant (60% minimal fibrosis limited to the bowel wall; 21% thick/mass-like fibrosis and 12% irregular/spicular fibrosis). In 94% the rectal wall morphology remained unchanged during long-term follow-up, in 2% initial fibrosis later developed into a normalised wall, in 3% the fibrosis slightly thickened (without evidence of recurrence). Conclusion: In the majority of patients with a complete response residual fibrosis is present post-CRT, which remains unchanged during long-term follow-up in almost all patients. A completely normalised wall is observed in 1 in 10-20 patients.

SS 5.10
Comparing MRI with fluoroscopy in defaecating proctography: the patient's perspective R. Prasad, C. White, R. Wiles; Liverpool/GB Purpose: Flouroscopy and MRI are currently used to investigate defaecation and pelvic floor problems. Fluoroscopy has advantages of imaging in a physiological sitting position and may better demonstrate intussusception. MRI has advantages of imaging all pelvic compartments and avoiding ionising radiation. Anecdotally it is suspected that MRI, allowing more privacy, may be better tolerated by patients. The authors aimed to evaluate patient experience of both techniques to potentially guide future modality choice. Material and methods: This prospective study was conducted during June-December 2015 in a large teaching hospital. Patients completed a post-procedure questionnaire rating out of 5 (1=strongly disagree and 5=strongly agree) their satisfaction of each test including staff communication, patient facilities, replication of symptoms and whether it was comfortable, dignified and pain free.
Results: There were 19 and 8 patients in the fluoroscopy and MRI cohorts, respectively. Every question for both modalities scored median 5.0 and mean 4.4 or more, except for ease of replicating symptoms, for which MRI scored less than fluoroscopy at median 4.5 and mean 3.8 (vs 5 and 4.6, respectively). Conclusion: Overall satisfaction for both tests was very high. Despite the anecdotal suspicion that MRI may be better tolerated by patients, this study did not find this. For ease of replicating symptoms MRI scored less than fluoroscopy. This may be due to the non-physiological positioning, potentially indicating that fluoroscopy may better demonstrate pathology compared to MRI. To compare the performance of pulsed-gradient-spin-echo (PGSE) and oscillating-gradient-spin-echo (OGSE) diffusion-weighted imaging for characterization of hepatocellular nodules in liver cirrhosis. Material and methods: Twenty-four Wistar rats were included. Cirrhosis was induced by weekly intra-peritoneal injection of diethylnitrosamine (50mg/kg) during 16 weeks. After sacrifice, a cylindrical liver sample was resected and imaged in a 7T MR scanner. The protocol included T1-weighted/T2-weighted/ PGSE/OGSE images (b=0/150/300/500s.mm-2). For precise radio-pathological correlations, only 80 nodules identified on T1/T2-weighted images and on pathological examination were analysed. Apparent diffusion coefficient (ADC) was calculated with a monoexponential fit. Two pathologists classified the nodules in regenerative (RN), low (LGDN) or high (HGDN) grade dysplastic nodules, early or progressed HCC. ADC were compared in group 1 (RN+LGDN), group 2 (HGDN+HCCearly) and group 3 (HCCprogressed) with Kruskal-Wallis test and areas under the receiver operating characteristic curves (AUROCs).

SS 6.2
Comparison of gadoxetic acid-enhanced MRI and non-specific extracellular gadolinium contrastenhanced MRI for the assessment of HCC response to loco-regional treatment J. Rimola 1 , M. Davenport 2 , P.S. Liu 2 , T. Conclusion: Gd-MRI is more accurate than EOB-MRI for the determination of viable HCC following LRT, and therefore the preferred contrast agent to be used for its assessment in patients with cirrhosis. Specificity for Gd-MRI and EOB-MRI is similar.

SS 6.3
Comparison of the efficiency of tenth and twentieth minute delayed gadoxetate disodium (Gd-EOB-DTPA) enhanced MR images for diagnosis of small HCC (less than 2 cm) A. Gocmez, M.G. Kartal, B. Bakir, B. Acunas; Istanbul/TR Purpose: To find out whether 10-minute delayed images obtained after the injection of Gd-EOB-DTPA are sufficient to characterize small HCC (<2cm) in comparison to the images obtained at 20-minute delayed images, the recommended optimal image acquisition time.
Material and methods: 63 lesions in 48 patients diagnosed with small-sized (<2 cm) HCC on gadoxetate-enhanced MRI performed at Istanbul Medicine Faculty were retrospectively evaluated. MRI images were divided into 2 sets. 10-minute set contained 10-minute images in addition to T1, T2, DWI, and postcontrast phases and 20-minute set contained 20-minute hepatobiliary image in addition to T1, T2, DWI, and postcontrast phases for qualitative analysis. Two independent observers scored focal lesion signal intensities at 10and 20-minute images. For the quantitative analysis, signal intensities were measured by placing ROI and calculated as was the contrast ratio.
Results: The quantitative analyses showed significant positive correlation between the 10-minute and 20-minute contrast ratio of the lesions. In qualitative analyses (rs=0.914,p<0.001), there was an excellent concordance between both observers for 10-minute and 20-minute images (Kappa for 10-minute images 0.907±0.053, for 20-minute images 0.930±0.049, p<0.001). There was moderate concordance between 10-and 20-minute images for each observer (Kappa for the first observer 0.523±0.095, for the second observer 0.490±0.095, p<0.001).
Conclusion: Compared to 10-minute images, 20-minute images did not reveal significant differences between HCC-liver signal ratio and diagnostic efficency. Therefore, especially for HCC evaluation in Child A group cirrhotic patients, obtaining only 10-minute images for delayed phase images depending on routine clinical needs will shorten image acquisition time but not affect the accuracy of the technique to a significant degree.

SS 6.4
Computed Material and methods: 19 patients (mean age 69±6) undergoing P-CT before and directly after TACE due to multifocal HCC were prospectively included in this dual center study. Two readers determined arterial-liver-perfusion (ALP, in mL/min/100mL), portal-venous-perfusion (PLP, in mL/min/100mL) and hepatic-perfusion-index (HPI, in%) placing circular regions-of-interest covering the maximum diameter of each lesion (N=19) before and after TACE. Imaging follow-up with contrast-enhanced CT or MRI was used to distinguish responders (complete response/partial response) from non-responders (stable disease/ progressive disease) following EASL criteria. Diagnostic performance of percentual changes in perfusion parameters before and after treatment (Δ) for early assessment of treatment response was determined using receiver operating characteristics.
Results: Mean ALP, PLP and HPI were 36.5, 16.6 and 75.9 before and 11.9, 23.3 and 46.8 after TACE, resulting in a ΔALP, ΔPLP and ΔHPI of -53%, 17% and -25%. Interreader agreement was fair to excellent for all perfusion parameters (ICC, 0.758-0.978). Before TACE, no significant correlations were found between perfusion parameters and treatment response (all p>0.05). After TACE, PLP was the only parameter to significantly correlate with treatment response (p<0.003) showing high accuracy for identification of TACE non-responders (AUC 0.943, 95% confidence interval 0.808-1.000, p<0.01). Conclusion: P-CT is useful for monitoring the effects of TACE in HCC patients. Decreased PLP after TACE is a potential biomarker for tumor progression. Material and methods: The total of 40 patients treated in the University Hospital Brno with HCC were included in this study. All of them underwent TACE and their tumor range was evaluated on the CT/MR input and output. The volumetric analysis was performed semiautomatically. The survival of the patients was evaluated since the date of the first chemoembolization. The overall survival was evaluated by the Kaplan-Meier method and the differences in survival by the log-rank test.
Results: The strongest correlation has been proven between the length of survival and determination of the viable part of a tumor using volumetric analysis, and between the length of survival and the ratio viable/nonviable parts of the tumor. The median of survival since the first performed TACE is 15.0 months. RECIST and mRECIST have not been proven as a statistically significant factor of correlation with the overall survival. Conclusion: Volumetric analysis was statistically proven the strongest factor of the correlation with the length of the survival of the patients, contrary to RECIST and mRECIST. It is a convenient way to evaluate the response of HCC to treatment, particularly in complex tumors after TACE. Quantitative studies have shown that pathological assessment of disease stage is compromised in inadequate biopsies (<25mm in length, fewer than 11 portal tracts, or tissue area less than 22mm2). This study aimed to determine if the liver biopsy sample affected pathological analysis between side-notch and end-cutting needles. Material and methods: Retrospective analysis of a total of 105 non-targetted US liver biopsies was carried out using either end-cutting and side-notch needles within a 42-month period (2011)(2012)(2013)(2014). Blinded pathological analysis of the samples was performed by a consultant pathologist for adequacy, length and fragmentation. Furthermore, each biopsy was reviewed for associated complications, gauge of needle and number of throws.
Results: 49 biopsies using end-cutting needle and 56 with side-notch needle were analysed. The samples were deemed inadequate from 0% of the endcutting group vs 24.5% of the side-cutting group. Length yield was higher in the end-cutting group (mean of 26.7mm vs 20.1mm), respectively, with less fragmentation in the end-cutting group (12% v 50%). Total number of portal tracts was also higher in the end-cutting group (mean 11.25 vs 7.3). The difference in length and number of portal tracts was statistically significant with p values of 0.0009 and 0.0263, respectively. Conclusion: End-cutting needle performed better with regard to sample adequacy, specifically the number of portal tracts, length yield and degree of fragmentation.

SS 7.2
Evaluation of 3D VIBE-DIXON imaging sequence for quantification of hepatic iron overload at 3T A. Kiani, E. Bannier, G. D'Assignies, G. Gambarota, H. Saint-Jalmes, Y. Gandon; Rennes/FR Purpose: To assess at 3T the ability of 3D gradient echo (GRE) imaging to quantify liver iron concentration (LIC). Material and methods: After IRB approval and written consent from all participants, 202 patients suspected of hepatic iron overload were included. Examinations were performed on a 3T MRI (MAGNETOM Verio, Siemens). Our reference 2D 11 echos multi-echo (ME)-GRE sequence (body coil), previously validated versus biopsy, was compared to a prototype 3D ME GRE (VIBE) sequence (surface coil). R2* mapping was computed and compared to R2* of reference sequence. VIBE R2* was converted to LIC according to Wood's formula adapted at 3T and compared to LIC provided by R2* and signal intensity ratio method of reference sequence. The shortest TE was 1.23 ms for both sequences.
Results: According, respectively, to 2D GRE and 3D ME VIBE results, mean R2* were 144±63Hz and 108±44Hz; mean LIC were 58±65µmol/g and 24±10µmol/g (p<0.001). Over all patients, correlation coefficients were 0.33 and 0.11 for R2* and LIC. Excluding patients with high LIC (>120 µmol/g), coefficients were 0.74 and 0.66. LIC correlation curve slope was 0.24. Conclusion: 3D ME VIBE results are well correlated to our reference method for patients with low/medium overloads. Evaluation of major overloads is limited at 3T using only R2* if the first TE is not short enough. 3T extrapolated Wood's formula seems to significantly underestimate LIC with the use of VIBE sequence.

Porto/PT, 2 Valencia/ES
Purpose: Multiecho GRE MR imaging is being used to determine liver R2* as an imaging biomarker of iron overload. Our purpose was to determine the R2* of liver, pancreas, spleen and bone marrow in patients with diffuse liver diseases and to evaluate their relationship.
Material and methods: The series included 100 consecutive patients with diffuse liver disorders, liver biopsy and abdominal MR examination (3T, single breath-hold, 12 echoes GRE sequence). Parametric iron R2* quantification was performed with a dedicated software selecting ROIs in liver, pancreas, spleen and vertebral bone marrow. Liver biopsy was used as gold standard for liver iron deposits grading (0-4).
Conclusion: R2* measurements showed a correlation between liver, pancreas, spleen and bone marrow iron deposits in patients with diffuse liver diseases. These results add insights into liver iron disorders and their relationship with other abdominal organs and tissues.

M. Ubeda; San Sebastián/ES
Purpose: To systematically review the T2* relaxometry methods used to quantify hepatic iron concentration. Material and methods: A literature search was conducted in MEDLINE and EMBASE, using the search terms "MRI", "Liver" and "Iron", combined with appropriate boolean operators. Were considered relevant those articles that: (1) used T2* relaxometry to determine hepatic iron concentration and (2) included acquisition parameter specifications. Four radiologists agreed on the parameters to be considered in each article and designed a data collection sheet.
Results: From all retrieved papers (357) only 61 were judged to be relevant for our purpose. Reference T2* values used to differentiate between normal and iron overload (n=16) varied from 6.3 to 21 ms. First echo time (TE) varied from 0.8 to 4.7 ms. Only 14 articles compared the MRI results with direct biopsy measurement of hepatic iron. The threshold to differentiate normal from iron overload varied from 4.6 to 14.5 ms. Eighteen articles gave not only T2* values, but also its transformation into mg/ Fe/gr. Eight of them used the same mathematical formula but with different acquisition parameters.
Conclusion: There is no standard method. There is no standardised acquisition protocol. Results are not reproducible. Thus, T2* relaxometry is not yet a valid method for standardised assessment of hepatic iron concentration. Results: Enhanced CT imaging showed hepatomegaly, massive ascites, multiple parenchymal undefined areas and absence of clear space-occupying lesions. CEUS showed large and completely avascular, irregular areas in the liver, crossed from patent portal, arterial and hepatic vessels. These aspects were highly consistent with massive occlusion of sinusoids without any vascular damage to portal spaces. The patient died because of liver failure 4 days after admission. At autopsy microscopic examination showed markedly dilated sinusoids filled with thrombi and fibrin and hepatic venules lumen obstruction.
Conclusion: Contrast material for CEUS, as blood-pool agent, allows a clear and well-defined detection of thrombosed sinusoids. CT and MR contrast materials could miss these aspects due to their interstitial diffusion.

SS 7.10
Hepatic enhancement of gadolinium-ethoxybenzyldiethylenetriamine pentaacetic acid-enhanced 3-T MRI predicts the severity of liver cirrhosis S. Lee; Seoul/KR Purpose: To evaluate the effectiveness of gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced MR in the assessment for the severity of liver cirrhosis and quantitative liver function. Material and methods: This retrospective study consists of 120 patients who underwent Gd-EOB-DTPA-enhanced 3-T MR (normal liver, n=30; Child-Pugh class A, n=30; B, n=30; C, n=30), using matching method by underlying disease, age (±5 years), gender, and creatinine (±0.05 mg/dL). Contrast enhancement index (CEI) was calculated and compared between normal liver and each cirrhotic group. We analyzed the correlation between hepatic function parameters and CEI on 20-min hepatobiliary phase (HP). The diagnostic performance of CEI on HP for the severity of cirrhosis was evaluated by area under curve analysis.
Results: The degree and time course of hepatic enhancement differed significantly between normal and each cirrhotic group (P<0.001). The mean CEI on HP constantly and significantly decreased as severity of liver cirrhosis progressed (P<0.001). Total bilirubin (P=0.022), albumin (P<0.001), platelet count (P=0.04) and model for end stage liver disease (MELD) score (P=0.01) were independent predictors of hepatic enhancement on HP. The CEI on HP showed good discriminatory ability in the severity of cirrhosis (AUC ≥ 0.94).
Conclusion: The degree of hepatic enhancement with Gd-EOB-DTPA indicates the severity of liver cirrhosis and correlates with hepatic function parameters. Material and methods: 8 patients with known pancreatic neuroendocrine tumor (4 patients -well-differentiated tumors (G1) and 4 patients -moderately differentiated tumors (G2)) underwent whole pancreas perfusion by a 256-slice CT (Brilliance iCT; Philips). 80-kVp/100mAs (low-dose) image data were reconstructed with iDose5 iterative reconstruction. Perfusion parameters were calculated with maximum-slope and dual-input one-compartment model methods. The parameters generated included the blood flow (BF, ml/min/100 ml tissue), blood volume (BV, ml/100 ml tissue), arterial blood flow (AF, ml/min/100 ml tissue), portal blood flow (PF, ml/min/100 ml tissue) and perfusion index (    Material and methods: In our department, twenty-five patients (mean age 33,12 ± 7,32) with CD underwent MR enterography and in the same time a real-time USE from July 2014 to October 2015. ADC values were calculated in the mesentery of pathological ileum (study group) and of normal ileum (control group) and were compared with the USE colour images in the same location. These results were statistically analysed. Results: In the study group, the USE colour-scale coding showed a colour change from blue to red in the fibrotic change of mesentery, and blue-green in the oedematous change, 10 and 15 patients, respectively. Moreover, there was a significant (p<0,05) restriction of the diffusion in 13 patients with CD in the active phase (mean ADC values for the fibrotic mesentery: 2,83 ± 0,21 x 10-3, mean ADC values for oedematous mesentery: 2,17 ± 0,33 x 10-3). However, there was a significant difference between the control and the study group.
Conclusion: Evaluation of CD through USE and DWI is a more and more growing field, and many tools are available. USE also gives a confirmation of the nonsolid nature of the mesenteric mass because the tumor appeared almost entirely green (soft) on hardness colorimetric scale. The aim of this retrospective study was to introduce and validate a CT-scoring system to discriminate between infected and non-infected postoperative abdominal fluid collections. Material and methods: Between May and November 2015, all patients with portal-venous CT within 24 hours before CT-guided intervention were included. CTs were independently reviewed by two radiologists. Imaging signs (Hounsfield units -HU, wall enhancement, entrapped gas) and C-reactive protein ≤ 24h before CT were retrospectively correlated with results of microbiology (CT-guided intervention).
Results: 50 patients were included. On binary logistic regression analysis, the four parameters were associated with the incidence of infected abdominal fluid collections. A scoring system consisting of nominal categorization of the four variables was selected to develop a clinical scoring from 0 to 11 (C-reactive protein ≥ 150 mg/l: 4 points; HU ≥ 20: 2 points; wall enhancement: 2 points; entrapped gas: 3 points). The model was well calibrated (Hosmer-Lemeshow test). A cut-off of 5 points showed a 83% positive predictive value for the presence of infected fluid and a 74% negative predictive value. The sensitivity was 73% and the specificity was 83%.
Conclusion: This study provides a validation of a newly developed scoring system and cut-off values for the discrimination between infected and noninfected postoperative abdominal fluid collections. These findings might help to prevent unnecessary interventions. were reviewed. Two abdominal radiologists reviewed medical records and CT images of all patients. A total 155 patients took abdominal CT scan and we excluded 124 patients who had only lymphadenopathy or organomegaly due to leukemic involvement, or didn't have any abdominal lesion, who hadn't undergone any chemotherapy. Consequently, 31 leukemic patients who underwent CT examination for their abdominal symptoms were included in this study and they were divided into two groups. The first group was the fungal ball group consisting of 8 patients who had poorly enhancing ball-shaped lesion involving bowel or liver and the second group consisted of other patients. We compared ANC, duration between chemotherapy induction and CT scan, and clinical diagnosis for two groups.
Results: The result showed ANC was significantly low and the duration was significantly short for the fungal ball group. Clinical diagnosis of invasive fungal infection was significantly high for the fungal ball group. Conclusion: Poorly enhancing ball-shaped lesion involving the liver or intestine of immunocompromised patients may suggest invasive fungal infection. The CT findings are not pathognomonic but are helpful in narrowing the differential diagnosis and early treatment in an immunocompromised patient. Material and methods: Over 14 months, patients from our emergency department who underwent CT for suspected AA were retrospectively included. The CT protocol included both unenhanced (UCT) and IV-enhanced CT (IVCT) scans. Four readers (two residents, one senior abdominal radiologist, one senior non abdominal radiologist (SNAR)) read each patient scans: UCT first then IVCT. At each reading, they were asked to confirm or rule out AA, to record their level of confidence and to propose an alternative diagnosis. The reference standard was pathologic specimen when available or a composite reference standard with other tests and follow-up when the patient did not undergo surgery. The main goal of this prospective study was to assess the added value of DWI in patients with inflammatory appendicial mass, who were decided a conservative treatment rather than immediate surgery. We also aimed to reveal alternative diagnoses such as tumors of cecum and appendix. This paper hypothesizes that DWI has the ability show the morphological and inflammatory changes of IAM to monitor the treatment response and show alternative diagnoses. Material and methods: A total of 19 consecutive patients (mean age, 37+-13.1; range, 19-69; F/M: 9/10) with a clinical diagnosis of IAM followed-up with conservative treatment with or without interval appendectomy, were enrolled in this prospective study during a period of 19 months. All the diagnoses of IAM were made with contrast enhanced CT. After the CT diagnosis of IAM, only those patients who were decided a follow-up period of conservative treatment were included to our study. For follow-up period, DWI was chosen as the modality of imaging. Results: It provided statistical confirmation that ADC values increase as CRP and WBC drop towards their normal level. Additionally, we assessed the size of IAM that the decrement also corresponded to increase of ADC value.

Conclusion:
The study revealed that DWI has the ability to show the morphological and inflammatory changes of IAM to monitor the treatment response and show alternative diagnoses.

SS 9.10
Reliability of Alvarado score in acute appendicitis and establishment of a compound sonographic score M. Mannil, C. Polysopoulos, D. Weishaupt, A. Hansmann; Zurich/CH Purpose: The purpose of this clinical retrospective study is to improve the diagnosis of acute appendicitis using the established Alvardo score in its modified version [1.2] combined with sonographic criteria. Material and methods: 132 patients with clinical suspicion of acute appendicitis received an abdominal US between 2012 and 2014. Sonographic criteria included appendix found, appendix diameter, appendix stone, compressibility, free fluid, mesenteric lymph nodes, inflamed mesenteric fatty tissue, perforation, abscess formation, small/large bowel affection and gynecological/urological pathologies. 100 patients had a surgically and pathologically confirmed acute appendicitis. The remaining 32 patients served as controls. Two models were computed in case the appendix was found (n=93) and when it was not (n=39) using logistic regression.
Results: The modified Alvarado score shows already excellent correlation with acute appendicitis with a cutoff value of 8. However, the diameter of the appendix adds significant information to this already strong predictor (p = 0.003).
Adding the appendix diameter to the Alvarado score improves its sensitivity (~100%) and specificity (91.4%), resulting in the formula: modified Alvarado score + appendix diameter [mm] ≥13 is highly suggestive of an acute appendicitis, while 1 point/per mm is subtracted in case the appendix shows a diameter of <8 mm and 1 point/per mm is added if the diameter is >8 mm.

Conclusion:
The modified Alvarado score shows excellent correlation with acute appendicitis. However, the diameter of the appendix adds significant additional information. To determine if early changes in MRI-measured small bowel (SB) motility, occurring after initiation of anti-TNF treatment, are associated with longer-term response to therapy. Material and methods: 11 patients were recruited prospectively; 20 patients were identified retrospectively. All completed a standard SB-MRI protocol preand post-initiation of anti-TNF therapy. Patients ingested mannitol orally and underwent 8-12 breath-hold coronal TruFISP/BTFE imaging at a frame rate of 1/second, encompassing the entire SB. Two radiologists independently segmented diseased ileum and motility was measured using a validated opticflow algorithm. Subsequently, patients were followed-up for a mean of 10.3 months, with response to therapy being judged via a physician global assessment (PGA). Motility changes in responders vs. non-responders were compared via the Wilcoxon rank sum test. The sensitivity and specificity of improved motility for response to therapy was calculated. Results: Mean time between initiation of anti-TNF therapy and follow-up SB-MRI was 16 weeks (range = 8-28 weeks). Patients who were durable responders (i.e. sustained response to anti-TNF therapy at a mean of 10.3 months) had significantly greater motility changes (median=83.3% rise from baseline) than non-responders (median=22.7% reduction from baseline, p<0.001). Improved motility at mean 16 weeks had 93% Sn and 83% Sp for predicting longer term response.
Conclusion: Early changes in segmental SB motility measured by MRI may be able to predict longer-term durable response to anti-TNF therapy.

SS 10.2
Elevated fecal calprotectin (FCP) is associated with correlation between restricted diffusion (RD) and mucosal inflammation ( Purpose: The aim of the study is to assess the usability of the ADC as a biomarker of the inflammatory process activity in the CLC disease. Material and methods: We analyzed the MRI scans of 43 patients, performed on 1.5T scanner. A standard protocol of MR enterography was performed including the DWI sequence and the following values b (0, 100, 300, 500, 800 s/ mm2), the ADC maps were generated. The ADC value was obtained by way of a 5-fold measurement and marking a region of interest (ROI) for each of the affected bowel area. The results were averaged and compared with the value of CRP and colonoscopy results and histo-pathological examination.
Results: Based on the colonoscopy and histopathological examination, in 32 of 43 examined patients an active stage of CLC was ascertained, and in 11 people -a chronic stage. The average value of the ADC for the active stage of disease was ~1.411x10-3 mm2/s, and for the chronic stage ~1.750x10-3 mm2/s. 27 patients' CRP was accordant with the clinical stage of the disease activity and the ADC value, low in active stage and high in chronic stage. In 6 cases of the active process, the value of the ADC correlated with disease activity better than the CRP.  Material and methods: From July 2013 to July 2015, 427 patients with a history of adverse reactions to iodinated CM during CT (group A) and propensity score-matched control group (group B, n=427) without adverse reactions were studied. Age, sex, CM types, CM injection protocols including velocity, total amount and duration, the presence or absence of prophylaxis and accompanying allergic symptoms were compared between patients with angioedema and those without angioedema in group A. In addition, the incidence of bowel angioedema was compared between group A and B using a clustered logistic regression method.
Results: The incidences of CM-induced bowel angioedema in group A were 3.3 % (14/427) in per-patient analysis and 2.6 % (15/578) in per-exam analysis. Angioedema was found in 1 of 14 patients in consecutive CT examinations. None of the patient demographics or CM related factors were different between the patients with and without bowel angioedema (P > 0.05). The incidences of CM-induced bowel angioedema in group B were 1.7 % (8/458) and 1.9 % (8/427) in per-patient and per-exam analyses, and these rates were not significantly different between group A and B (P=0.35 and P=0.37, respectively).
Conclusion: The incidence of CM-induced bowel angioedema during CT was 1.7%-3.3% and none of the studied risk factors was associated with bowel angioedema. Results: Fifty-one cases of primary SBNET were detected and 29 (57%) were multifocal. Tumor densitometry showed a statistical non-significant difference between phases (p>0.05). DCBT did not allow a significant radiation dose reduction (p>0.05). Among all the 75 CT-E in patients with SBNET, 53% had liver metastasis. Desmoplastic reaction was present in 38 (51%) cases, but only in 9 (12%) figured out as "Sun Burst Sign". In the 57% (43 patients) of cases mesenteric lymph nodes were noticed and a mesenteric mass was present in 41 cases (33 cases presented vascular encasement). Conclusion: CT-E might be proposed as the main imaging technique in the diagnosis and planning the management of SBNETs.

SS 10.10
Contrast

Birmingham/GB
Purpose: The Endobarrier is a 60cm duodenal bypass liner placed endoscopically for up to 1-year, improving glucose control and weight in obesity-related diabetes. The aim of this study was to evaluate whether failure of response to Endobarrier relates to chyme leakage around the device's proximal anchor. Material and methods: Adults with suboptimally controlled type-2 diabetes and obesity who had undergone minimum 6 months' treatment were classified as responders (≥3% initial body weight loss; ≥1% HbA1c reduction) and nonresponders. They underwent limited CT post oral contrast. Leakage was evaluated (none, mild, moderate, severe) by an independent experienced radiologist, blinded to responder status, and correlated with metabolic response. Anteroposterior, craniocaudal dimensions of the anchor, wall thickness, liver HU were measured.  (MT), T2-weighted, and contrast-enhanced 3DT1-weighted imaging (with gadofosveset-trisodium). Sequences were processed using the Intellispace Discovery research platform (Philips Healthcare). One reader delineated whole-tumour volumes on b1000-DWI, which were subdivided into three equal subvolumes (high, intermediate and low DWI-signal). The following parameters were compared between these 3 DWI-signal groups: ADC, MT-ratio, signal intensity on T2W and contrast-enhanced MRI, and various DCE parameters (initial slope, wash-in/wash-out, initial signal excess, wash-in time to peak (TTP), final slope, TTP, max enhancement, mean transit time, AUC).
Results: Significant differences between the 3 DWI-signal groups were found for the DCE-parameters initial slope and initial signal excess (both higher with increasing DWI-signal; P=0.008/0.026). ADC, wash-in TTP, TTP and AUC60 were significantly different for the high vs. low DWI-signal groups with lower ADC and TTP, and higher wash-in TTP and AUC60 in the high DWI-signal group. Amsterdam/NL, 4 Roermond/NL, 5 Leeuwarden/NL Purpose: Wait-and-see policy can be offered to clinical complete responders (cCR) after neoadjuvant chemoradiation. Aim was to evaluate whether the timing of selection influences outcome, by comparing patients included for waitand-see at initial assessment with patients who were selected after a second assessment 3 months later. Material and methods: 114 eligible patients underwent initial assessment with endoscopy and MRI+DWI ±8 weeks post-CRT. 61 had a typical cCR and were selected for wait-and-see immediately(W&S-1). 24 had residual tumour and underwent TME. The other 29 patients had a near cCR (not meeting 1 or 2 criteria for cCR) and underwent a second assessment after 3 months, after which 24 were included for wait-and-see(W&S-2) and 5 for TME. 3-6 monthly follow-up (with MRI+DWI and endoscopy) was performed. Results: 2/24 patients that underwent TME after first assessment had a pathologic complete response(pCR). 1/5 that underwent TME after second assessment had pCR. In total 85/114 (75%) were in wait-and-see with 25 months median follow-up. 3-year LRFS was 88% for W&S-1 compared to 77% for W&S-2 (p=0.22), 3-year overall survival was 96% and 96%, respectively (p=0.95). All local regrowths could be easily salvaged with standard TME. Conclusion: A second response assessment after 3 months can offer waitand-see to more patients. However, there is a trend towards lower 3-year LRFS in patients included after second assessment, although overall survival is not influenced by the lower LRFS. Purpose: It is believed that chemoradiation for low rectal cancer increases sphincter preservation. Aim was to evaluate with MRI whether sphincter preservation is increased by chemoradiation and whether MRI can predict sphincter preservation after chemoradiation. Material and methods: A radiologist independently evaluated the T2-weighted MRIs (in 3 directions) in 47 patients before and after CRT with tumours <5 cm from the anorectal junction (ARJ) and measured distance of the lower tumour pole to the ARJ. Also, a confidence level score for feasibility of sphincter preservation was scored (CL=0 definitely no sphincter preservation, CL4=sphincter preservation definitely possible). Likelihood for sphincter preservation before and after CRT was compared and receiver operator characteristics(ROC) curves with area under the curve (AUC) were calculated. Results: Mean distance from ARJ increased significantly during CRT from 21±16mm pre-CRT to and 31±18mm post-CRT (P<0.001). In 42% sphincter preservation was deemed not feasible pre-CRT, which decreased to 23% after CRT. AUC for sphincter preservation based on confidence level score was 0.84(0.72-0.96), with sensitivity of 100% and specificity of 44%. Based on post-CRT height measurement AUC was 0.87(0.76-0.98), with optimal size cut-off at 26 mm (sens: 86%, spec: 71%).
Conclusion: This is the first study to show that CRT increases the distance to the ARJ and thus leads to a higher rate of sphincter preservation. MRI can accurately predict sphincter preservation after CRT.

SS 11.9
Artefacts on diffusion-  Purpose: Feasibility of sphincter-preserving surgery in low rectal tumours (<5 cm from anorectal junction) is based on the exact distance from the anorectal junction (ARJ), rigidity upon digital rectal examination and presence of sphincter invasion. Aim was to evaluate whether a radiologist and a surgeon can predict feasibility of sphincter-preserving surgery at T2W-MRI. Material and methods: 44 patients with a rectal tumour <5 cm from the ARJ that did not undergo CRT were included. T2W-MRI in 3 directions was evaluated by a specialized radiologist and a specialized surgeon for feasibility of sphincter-preservation. Distance from ARJ (mm) and confidence level scores for sphincter-preserving surgery and sphincter invasion were scored. Reference standard was type of surgery combined radicality of the resection. Results: Both the radiologist and surgeon could predict sphincter preservation accurately with AUC 0.81 (radiologist) and 0.82 (surgeon). Sensitivity and specificity are 75% and 83% (radiologist) and 50% and 87% (surgeon), respectively. Height was predictive of sphincter preservation for both readers with AUC of 0.90 (radiologist) and 0.88 (surgeon), with optimal cut-off at 18 and 26 mm, respectively. Interobserver reproducibility of height was ICC 0.70 (0.51-0.82) and of confidence level scores for sphincter preservation was k0.53 (0.31-0.76). Conclusion: Sphincter preservation can accurately be predicted with T2W-MRI by both specialized radiologists as well as specialized surgeons. Height is the most accurate predictor with a cut-off of approximately 18-26 mm. Purpose: Neoadjuvent chemo-radiotherapy for locally advanced rectal cancer (LARC) aims to downstage prior to definitive management. Radiological reassessment of the tumour post-therapy has implications for treatment. Various methods of assessing tumour response exist including TNM stage, RECIST and TRG-score. Material and methods: We assessed the inferior mesenteric vein diameter (IMV) and the IMV:common iliac vein ratio (IMV:CIV) pre and post-radiotherapy of 100 patients with LARC to ascertain if these are surrogate markers of tumour response. IMV measurements were performed by two radiologists and MRI response assessed by two radiologists blinded to CT measurements. Results: At baseline IMV diameter was significantly higher for cases with local lymphadenopathy -N0 = 5.2 mm vs N1/2 =6 mm (p=0.0059) -and lymphovascular invasion (LVI) --ve 5.4 mm vs +ve 6.4 mm (p=0.0001). Post radiotherapy there was a significant decrease in the IMV diameter and IMV:CIV in cases with treatment response compared to non-responders -percentage change IMV:CIV = -16.46% vs +1.18% (p=0.0002). These results were reproduced between TRG groups using ANOVA (p=0.0001). There was also a significant decrease in IMV when assessing lymph-node and LVI response versus non-responders (p=0.0001 and 0.0001 respectively). Conclusion: We confirm that IMV diameter and IMV:CIV are surrogate markers of lymph-node status and LVI at baseline. IMV diameter and IMV:CIV are also a markers of tumour , LN and LVI response to chemo-radiotherapy.

SS 12.3
Dynamic contrast-enhanced MRI: use in predicting pathological complete response to neoadjuvant chemoradiation in locally advanced rectal cancer T. Tong, Y. Sun, Y. Gu; Shanghai/CN Purpose: To determine the ability of dynamic contrast enhanced (DCE-MRI) to predict pathological complete response (pCR) before preoperative chemoradiotherapy (CRT), in locally advanced rectal cancer. Material and methods: In a prospective clinical trial, 38 enrolled patients underwent pre-and post-CRT DCE-MRI at 3.0T. The tumor length and the following perfusion parameters (Ktrans, kep, ve) were measured for the tumor and compared between the pCR group and the non-pCR group, as well as before and after CRT. For categorical variable comparison, the Kruskal-Wallis test was used. P<0.05 was considered significant. Results: No difference in tumor length was found between pCR and non-pCR group pre-and post-CRT (P=0.26, 0.35, respectively). Before CRT, the mean tumor Ktrans in the pCR group was significantly higher than in the non-pCR group (P=0.01). A Ktrans of 0.66 emerged as the best cut-off for distinguishing pCR from non-pCR. Regarding kep and ve, significant differences were also observed between pCR and non-pCR groups (P=0.02, 0.01, respectively). The mean Ktrans, kep and ve values post-CRT were not significantly different between pCR and non-pCR groups (P=0.10, 0.12, 0.08, respectively). Conclusion: Before neoadjuvant chemoradiotherapy in rectal cancer, DCE-MRI can distinguish between complete and incomplete response using a Ktrans threshold of 0.66 with a sensitivity of 100%.

SS 12.4
Standardized index of shape (DCE-MRI) and standardized uptake values (PET): two quantitative approaches to discriminate chemo-radiotherapy locally advanced rectal cancer responders under a functional profile. Two biological sides of the same coin? A. Petrillo, M. Petrillo, R. Fusco, A. Avallone, S. Lastoria; Naples/IT Purpose: We investigated the potential use of dynamic contrast-enhanced MRI (DCE-MRI) to discriminate responder from non-responder patients after neoadjuvant chemo-radiotherapy (CRT) for locally-advanced rectal cancer (LARC) in comparison with [18F]2-fluoro-2-deoxy-D-glucose positron emission tomography (FDG-PET/CT). Material and methods: 75 consecutive patients with LARC were enrolled in a prospective study. Each patient gave written informed consent to participate in the trial. Pathological TNM and tumor regression grade (TRG) were estimated. DCE-MRI analysis was performed by measuring SIS value before start and at the end of therapy as well as maximum SUV (SUVmax) by FDG-PET/CT. Nonparametric sample tests, ROC analysis and diagnostic performance were performed.
Results: Fifty-five patients (73.3%) with TRG 1-2 were classified as responders, while 20 subjects (26.7%) with TRG 3-4 were considered non responders. Assessment via ΔSIS reached a sensitivity of 92.7%, a specificity of 80.0% and an accuracy of 89.3% using a cut-off of 6.0% while ΔSUVmax reached a sensitivity of 67.3%, a specificity of 75.0% and an accuracy of 69.7% using a cut-off of 59.7%. Conclusion: Pre-surgery assessment of CRT response by ΔSIS showed a higher predictive ability than ΔSUVmax in LARC, increasing sensitivity and negative predictive value. SIS percentage change could play a relevant role in LARC management improving the ability to identify complete pathological response, allowing for conservative strategy or for a "wait and see" policy, reducing significant morbidity and functional complications related to total mesorectal excision.

SS 12.5
Diagnostic accuracy of magnetic resonance imaging in predicting response to neoadjuvant therapy in patients with locally advanced rectal cancer J. Munir, I. Niazi, R. Sayyed, S. Hasan Raza, M. Hussain, S. Khattak, A. Syed; Lahore/PK Purpose: Magnetic Resonance Imaging of pelvis is the imaging modality of choice for preoperative staging in rectal cancer. We aim to look at the accuracy of MRI in evaluating response to neoadjuvant therapy. Material and methods: Medical records of patients undergoing surgery for rectal cancer following neoadjuvant therapy at our institution in year 2014 were reviewed. Data was collected regarding T, N stage on MRI at presentation, after neoadjuvant therapy and on histopathology of resected specimen. Concordance of post-treatment MRI and histopathology was evaluated. Factors that might affect this concordance were assessed in univariate analysis.
Results: A total of 108 patients underwent surgery for rectal cancer after neoadjuvant therapy in 2013-14. Most of these had T3 or above tumor (97%) or nodal involvement (95%) on MRI at presentation. Post neoadjuvant therapy MRI scan was able to predict a negative circumferential resection margin, however it was not able to predict complete pathological response. The post treatment MRI scan couldn't accurately differentiate T3-T4 tumors from T1-T2 tumors or node positive from node negative tumors. No specific factor could be identified to be significantly associated with this concordance. The diagnostic accuracy of MRI for predicting a negative CRM was 84.54%. Conclusion: Although post treatment changes result in limitation of MRI to evaluate response to treatment, MRI shows acceptable diagnostic accuracy in predicting a negative circumferential resection margin. Purpose: To compare the diagnostic performance for evaluating hepatic lesions between gadoxetic acid-enhanced MRI (Gd-EOB-MRI) and MRI using extracellular contrast agent (ECA-MRI), and to determine whether the type of contrast agent affects early intrahepatic recurrence. Material and methods: Between January 2005 and December 2010, 418 colorectal cancer patients who underwent both preoperative CT and liver MRI were retrospectively reviewed. Image interpretation was based on initial radiologic reports and all focal hepatic lesions were confirmed by pathology or follow-up imaging. The diagnostic performance was assessed by the area under the receiver operating characteristic curve (AUROC) and the rate of indeterminate lesions on each MRI. For patients who underwent curative liver surgery, early intrahepatic recurrence rate within 6 months were evaluated. Results: Total 291 and 127 patients underwent Gd-EOB-MRI and ECA-MRI, respectively. In per-patient analysis, AUROCs of both MRI groups were not significantly different between Gd-EOB-MRI (0.990; 95 % confidence interval, 0.980-0.999) and ECA-MRI (0.985, 0.968-1.000) (P=0.836). The rates of indeterminate lesions on MRI were 6.98 % (30/430) for Gd-EOB-MRI and 4.88 % (10/205) for ECA-MRI, which did not reach significant difference (P=0.309). Early recurrence rate in ECA-MRI group (28.6 %) was significantly higher than that in Gd-EOB-MRI group (11. 6%) (P=0.031).
Conclusion: Gd-EOB-MRI and ECA-MRI showed comparable excellent diagnostic performance in colorectal cancer patients. Further study comparing survival outcome between Gd-EOB-MRI and ECA-MRI should be followed.

SS 13.2
Gd-EOB-DTPA-enhanced magnetic resonance imaging (MRI) for detecting colorectal hepatic metastases in patients who have not previously undergone treatment: a meta-analysis Y. Material and methods: Twenty-three patients underwent abdominal contrastenhanced MRI on a 3-T system and were included in this intra-individual comparison study. Each patient underwent both CAIPIRINHA-VIBE and conventional VIBE within 3 months showing stable disease. The sequences were reviewed by two blinded radiologists (A+B) using a 5-point rating scale for liver lesions and affected segments and with regard to overall image quality and sharpness of intrahepatic vessels. Cohen's analysis and Wilcoxon matched pairs tests were performed. Results: Superior interobserver agreement was observed for CAIPIRINHA-VIBE for both per-lesion and per-segment analysis (per-lesion, κ=0.18, p<0.04; per-segment κ=0.47, p<0.01) compared to conventional VIBE (per-lesion, κ=0.16, p<0.03; per-segment, κ=0.38; p<0.001). CAIPIRINHA-VIBE series received better scores (image quality: A=4.8, B=3.7; sharpness of intrahepatic vessels: A=4.2, B=3.7) compared to standard VIBE (image quality: A=4.0, B=3.0; sharpness of intrahepatic vessels A=3.5, B=3.0) from both reviewers (p<0.01).
Conclusion: CAIPIRINHA-VIBE sequences facilitate superior image quality and consequently improved inter-observer agreement regarding dignity and allocation of focal liver lesions compared to standard VIBE sequences.

SS 13.4
Liver metastases detection using sparsity-based learned dictionaries E. Klang 1 , A. Ben-Cohen 2 , I. Diamant 2 , E. Konen 1 , H. Greenspan 2 , M.-M. Amitai 1 ; 1 Ramat Gan/IL, 2 Tel Aviv/IL Purpose: Automatic liver lesions detection in CT examinations is an ongoing task for the computer vision community. Sparse classification has lately been proved to be powerful algorithms for data analysis. We evaluated sparse based learned dictionaries approach for the liver detection task.
Material and methods: This study included CT examinations of twenty patients with 68 metastases, segmented in 2D. The published SLIC algorithm was used to divide 2D liver images into 500 "super pixels". Each "super pixel" is a homogeneous region that includes neighboring pixels with similar Hounsfield units. Several features sets were extracted from each "super pixel" (grey level histogram, acutance and others), and using this features a dictionary of "super pixels" was created. Sparsisty classification was used to optimize the dictionary, limiting its number of words, creating "super pixels" groups of metastases, normal parenchyma, blood vessels and others, thus, effectively detecting the "super pixels" of metastases. The sensitivity and specificity of the sparse based approach were compared to that of the state of the art "random forests" classification algorithm.
Results: The sensitivity and specificity to detect liver metastases were 71% and 98% for the sparse based algorithm and 66% and 99% for the "random forests" algorithm. Conclusion: Sparse based approach to detect liver metastases showed promising results in comparison to "random forests" algorithm, with better sensitivity but slightly worse specificity. Purpose: To evaluate the diagnostic performance of diffusion-weighted MR imaging (DW-MRI) and Gd-EOB-DTPA-enhanced MRI at 3T-device in the preoperative staging of colorectal liver metastases in patients previously undergone chemotherapy. Material and methods: Fifty patients with colorectal cancer and focal liver lesions underwent MR imaging at 3T-device (GE DISCOVERY MR750;GE Healthcare) after preoperative chemotherapy. After preliminary acquisition of axial T1w (in/out of phase) and T2w (propeller and SS-FSE) images, DW-MRI was performed using an axial spin-echo echo-planar sequence with multiple b-values (150,500,1000,1500 sec/mm²) in all diffusion directions. Gd-EOB-DTPA-enhanced MRI was performed using a 3D breath-hold fat-suppressed T1w LAVA-flex sequence including both dynamic and hepato-biliary phase. MR images were reviewed by two observers in conference in order to detect and characterize (benign/malignant) focal liver lesions. MRI findings were correlated with surgery and histopathology, which was our gold standard. Only clear benign lesions at intraoperative ultrasound remained unresected. Statistical analysis was performed on a per-lesion basis.
Results: A total of 306 hepatic lesions were detected; of these, 220 were metastases (72%), whereas the remaining 86 (28%) were characterized as benign lesions (hemangiomas, cysts and nodular regenerative hyperplasia). The sensitivity, specificity, PPV, NPV and diagnostic accuracy of the reviewers for the detection and characterization of focal liver lesions were 98%,93%,97%,95% and 97%, respectively. Conclusion: The combination of DW-MRI with Gd-EOB-DTPA-enhanced MRI at 3T-device is particularly effective in preoperative staging of colorectal liver metastases after chemotherapy.
Conclusion: Though differences in mean ADC values of benign and malignant FLL were significant, confidence intervals indicate their large overlap, so ADC can't be the main differential diagnostic criterion. Analysis of DWI images can help in difficult diagnostic cases, but should only be used in conjunction with standard MRI, including contrast-enhanced MRI.

SS 13.8
Therapy response assessment of colorectal liver metastases after preoperative chemotherapy: diagnostic performance of diffusion-weighted MR imaging at 3T device Material and methods: Our study group included thirty-two patients with colorectal liver metastases undergone MR imaging at 3T-device (GE DISCOV-ERY MR750; GE Healthcare) after preoperative chemotherapy. DW-MRI was performed using a spin-echo echo-planar sequence with multiple b-values (150,500,1000,1500 sec/mm²), obtaining an ADC map. Fitted ADC values were calculated by two observers in conference for each liver lesion (more than 1 cm of diameter) drawing a ROI around the entire tumor and another one at the tumor periphery. All MRI findings were correlated with histopathology after surgery. Hepatic metastases were pathologically classified into three groups on the basis of tumor regression grading (TRG): MHR (major histological response,TRG 1-2), PHR (partial histological response,TRG 3), and NHR (no histological response,TRG 4-5). Statistical analysis was performed on a perlesion basis.
Results: A total of 104 colorectal liver metastases were analyzed. MHR, PHR and NHR were observed in 18%, 38% and 44% of lesions, respectively. Periphery ADC value was significantly different in the three groups (p<0.01) and was significantly higher in MHR than in NHR (p<0.01). However, ADC value of the entire tumor was not significantly different in the three groups. Conclusion: DW-MRI, using ADC map and value, can be useful to assess the efficacy of preoperative chemotherapy in colorectal liver metastases. Oslo/NO, 2 London/GB Purpose: To determine the value of entropy, a CT texture analysis (CTTA) parameter, for preoperative prediction of recurrence after resection of colorectal liver metastases (CRLM). Material and methods: Forty-four patients who underwent resection of CRLM between 2007-2009 were retrospectively included and followed-up until December 2015. CTTA was performed of the largest CRLM on contrast-enhanced CT using a research software TexRAD. Entropy was evaluated with different spatial scale filters (ssf2-6) corresponding to fine to coarse textures. A chemonaïve group (n=41) and a chemo-exposed group (n=20) were evaluated. TexRAD identified optimal threshold-values for entropy to divide the groups into poor/good prognosis. ROC and Kaplan-Meier/multivariate Cox analyses were performed.
Results: In the chemo-naïve group, the overall time-to-recurrence was 37.3 months and the threshold-value 4.9 (filter ssf2). Below the threshold-value the overall time-to-recurrence was 46.7 months and above 21.7 months (p=0.008). Sensitivity and specificity for recurrence was 54.2% and 82.4%, respectively (p=0.447). In the chemo-exposed group, the overall time-to-recurrence was 19.2 months and the threshold-value was 4.7 (filter ssf3). Below the thresholdvalue the overall time-to-recurrence was 33.1 months and above 5.2 months (p=0.007). Sensitivity and specificity for recurrence was 62.5% and 100%, respectively (p=0.001).
Conclusion: Entropy derived from preoperative CTTA may have the potential to predict early recurrence, especially in patients treated with preoperative chemotherapy. In practice, this may allow improved selection for resection or focused postoperative follow-up in high-risk patients.

SS 13.10
Assessment of portal venous phase in dual-energy CT of the liver: initial experience with multiple monoenergetic levels and advanced monoenergetic reconstructions D. Material and methods: This prospective study included subjects who underwent PVP DECT for assessment of liver lesions. From this data, linear-blended (LB) poly-energetic equivalent datasets and multiple monoenergetic reconstructions from 40-190 keV were performed. Standard virtual monoenergetic images (VMI) and advanced image-based virtual monoenergetic images (VMI+) were performed. Quantitative and qualitative image quality were assessed. An assessment of diagnostic accuracy was performed against Magnetic Resonance, as a reference standard. Direct comparisons were assessed by using repeated measure of variance. Results: In 29 subjects with 49 liver lesions, the highest peak CNR for liver parenchyma and hyper-vascular liver lesions was achieved for VMI+, although for hypo-vascular liver lesions both VMI+ and VMI had lower CNR than LB. However, qualitative image quality assessments revealed a higher diagnostic quality and also reader preference for VMI+ over LB or VMI. There was also a trend toward higher diagnostic performance and confidence with VMI+ over the other two reconstructions: 83.7 % on LB, 75.5 % on the best VMI keV level (75 keV), and 95.9% on the best VMI+ keV level (50 keV) Conclusion: Low keV VMI+ may improve detection of liver lesions on PVP despite a lack of clear improvement in measured CNR for hypo-dense lesions. Purpose: To evaluate the usefulness of 3D magnetic resonance cholangiopancreatography (MRCP) with/without partial maximum intensity projection (MIP) in patients with autoimmune pancreatitis (AIP). Material and methods: 3D MRCP and endoscopic retrograde cholangiopancreatography (ERCP) images were retrospectively analyzed in 24 patients with AIP. Three types of pancreatic duct findings, length of the main pancreatic duct narrowing (NR-MPD), multiple skipped MPD narrowing (SK-MPD) and side branches arising from the narrowed portion of the MPD (SB-MPD) were evaluated on four sets of MRCP images, MIP with 5-7.5 mm (MIP5), 10-15 mm (MIP10), 72-90 mm thickness (full-MIP), and the three data sets combined (a-MIP) were scored using a three-or five-point scale. The scores in the four MRCP data sets were statistically analyzed, and the positive rate of each finding was compared in the MRCP data sets and ERCP.
Results: Scores on NR-MPD and SK-MPD did not differ significantly among the four MRCP data sets. However, full-MIP yielded a significantly lower score on SB-MPD than the other three data sets (P < 0.05). NR-MPD and SK-MPD positivity rates were higher in all MRCP data sets than in ERCP (P < 0.05), whereas the SB-MPD positivity rate was similar in three MRCP data sets, excluding full-MIP, and ERCP. To determine whether Gd-EOB-DTPA-enhanced 3T MR Cholangiography may provide additional information in the evaluation of biliary strictures after orthotopic liver transplantation. Material and methods: Fifty-two liver transplant patients with clinical-echographical suspicion of biliary strictures underwent MR imaging at 3T device (GE-DISCOVERY MR750; GE Healthcare). After acquisition of T1w/T2w images and conventional T2-weighted MR Cholangiography (image set 1), a 3D T1-weighted fat-suppressed LAVA sequence was performed before and 20-120 minutes after intravenous administration of 10 ml Gd-EOB-DTPA (Primov-ist®, Bayer HealthCare) (image set 2). The diagnostic value of Gd-EOB-DTPAenhanced MR Cholangiography in the assessment of biliary strictures was tested by separate analysis results of image set 1 alone and image set 1 and 2 together. MRI results were correlated with direct cholangiography, surgery and/or clinical-radiological follow-up.
Results: The level of confidence in the assessment of biliary anastomotic and non-anastomotic strictures was significantly increased by the administration of Gd-EOB-DTPA (p<0.05). Particularly, contrast-enhanced T1-weighted LAVA sequences tended to out-perform conventional T2-weighted MR Cholangiography in the visualization of the extra-hepatic biliary system and in the grading and extension of non-anastomotic strictures. Functional information provided by Gd-EOB-DTPA biliary excretion was especially helpful in patients previously undergoing biliary-enteric anastomosis with not or poorly dilated biliary system. Computer-aided detection for CT colonography: an evaluation of the optimal reader paradigm using evidence-based medicine methodology C.E. Redmond, D. Malone, S.J. Skehan; Dublin/IE Purpose: Computer-aided detection (CAD) software for CT colonography (CTC) interpretation, which can be used in various reader paradigms, is widely available. The aim of this project was to establish the optimal reader paradigm in terms of per-polyp sensitivity (≥ 6mm) and reporting time, based on the current best available evidence. Material and methods: Evidence-based medicine (EBM) methods were used to retrieve and appraise studies published on the use of CAD for CTC. Appraisal was performed using the Oxford Centre for Evidence Based Medicine (CEBM) tools for diagnostic studies.
Results: No studies were retrieved from the secondary literature. From the primary literature, 9 relevant studies were identified, analysed and ranked in a "hierarchy of evidence". When compared to unassisted interpretation, using CAD as a second reader was associated with a higher per-polyp sensitivity (77.1% vs. 81.5%, respectively, p=0.001). Using CAD as a concurrent reader has a statistically significant shorter reporting time (p <0.001) and equivalent per-polyp sensitivity when compared to CAD as a second reader. The use of CAD as a first reader is controversial, however it appears to have similar perpolyp sensitivity levels.
Conclusion: The best current evidence indicates that CAD for CTC as a second reader results in higher per-polyp sensitivity when compared to unassisted interpretation (grade B recommendation). Using CAD as a concurrent reader is more time efficient and has a similar sensitivity (grade C recommendation).

SS 15.2
Participation to sequential fecal immunochemical test of non-responders to CT colonography and colonoscopy in a randomized screening trial (SAVE) L. Sali  Material and methods: This pragmatic single-centre randomized trial recruited asymptomatic and symptomatic patients aged 45 years or older and referred by clinicians for CT colonography. Patients were randomly assigned in a 1:1 ratio, in block of ten, to receive a reduced bowel preparation with diet restriction or the same reduced bowel preparation without diet restriction. All investigators were masked to treatment allocation. The primary endpoint, analyzed by ITT, was a composite of indices describing the quality of large bowel cleansing.
Results: 100 patients were randomly allocated to treatments (50 with diet restriction; 50 without diet restriction) and were included in intention-to-treat analysis. Six patients withdrew consent, leaving for analysis 46 assigned to reduced bowel preparation with diet restriction and 48 assigned to reduced bowel preparation without diet restriction. For all patients with confirmed important pathology identified at contrast-enhanced CTC, the low-dose, pre-contrast, prone acquisitions were re-read by a resident, blind to the contrast-enhanced imaging, reports, outcomes and study purpose. Results were compared to the reference standard and descriptive statistics produced. Results: In total, 94 (12%) studies were classified C-RADS E4, for which follow-up data were available for 87 (median 24 months; range 14-48). One patient refused investigation and 6 had died. 24 were ultimately considered incorrectly classified (12 metastatic disease from colorectal primary; 12 pre-existing pathology). Of 46 potential new extracolonic cancers, 29 were benign; of the 17 confirmed malignancies, 100% were identified at unenhanced CT. Of 7 aortic aneurysms and 10 further important, non-neoplastic findings, all were conspicuous on the pre-contrast acquisition. Each report was categorised as positive or negative for 6mm+ polyps/cancer. The time of day at which the report was issued was extracted from the hospital Radiology Information System (RIS), as was the number of CTC reported by each radiologist on that particular day. Logistic regression models were constructed to assess the statistical significance of (a) time of day and (b) number of prior cases reported that day by a given radiologist on 6mm+ polyp/cancer detection.

Results:
The detection rate of 6mm+ polyps/cancer dropped from 20.6% (for scans reported between 08:00am and 09:00am) to only 15.3% (for the 17:00pm to 18:00pm period), p=0.048. The detection rate also progressively declined over the course of a reporting session (from 22.2% to 15.3%, p=0.0123). Radiologists reporting their first CTC study on a given day had a detection rate of 22.5% whereas later scans had progressively lower detection, falling to 10.8% by their 5th scan of the day (p=0.0013). Conclusion: Detection rates at CTC progressively decline during the day (mirroring previous findings for colonoscopy), raising the possibility of clinically significant reporting fatigue for radiologists. Material and methods: Volumetric measurements of FLR using CT with 3D liver reconstruction were obtained in pre-and post-first stage of ALPPS. CT scans were performed with a 80-section Aquilion PRIME Toshiba MS scanner, with 1 mm slice thickness, reconstruction interval 0,5mm, pitch factor 1.8, voltage 120 kV, 100 mA, tube rotation time 0.35 s. As a CT-contrast agent we used omnipack 350 (Iohexol) 200 ml, injection speed 5 ml/s. We used standard liver multiphase bolus contrast enhancement program. The circumscribed areas were then automatically multiplied by the CT section thickness, yielding an approximate volume for each liver section, and the volumes of all sections were summed to give the selected liver volume. The critical minimum FLR has been estimated to be approximately 25-30% in normal liver, and 35-40% in liver after chemotherapy.
Results: Between January 2013 and September 2015 in 15 patients FLR ratio were < 26%, determining the indication for ALPPS. CT scans were performed on postoperative day 8, after liver transection and portal vein ligation. All patients have had significantly increases in FLR (195,1%) and extended right hepatectomies were successfully performed in all cases. Conclusion: CT-volumetry with 3D reconstructions allows to increase the safety of major liver resections for primary and metastatic liver cancer.

SS 16.4
Is a CT unique portal phase enough for causal diagnosis of non traumatic abdominopelvic emergencies? G. Herpe, M. Verdier, G. Vesselle, J.P. Tasu; Poitiers/FR Purpose: Computed tomography (CT) is widely used in the management of acute non traumatic abdominal emergencies (ANTAE). The aim of this study is to evaluate if a reduced CT protocol including a unique portal phase (Pp) is as accurate than a protocol including pre-contrast and post-contrast phases (Pc). Material and methods: A retrospective study included adults undergone precontrast and post-contrast abdominopelvic CT for ANTAE was conducted. Suspected haemorrhagic and/or ischemic disease were excluded. Two blind CT readings were carried out, one using only Pp, the second all phases of the Pc. In case of reading discrepancy, a third reading was performed, by an abdominal imaging expert. The final diagnosis obtained by consulting the clinical report of the patient was considered as the gold standard for the diagnosis. Total radiation dose was compared for each CT protocol. Results: 196 patients were included. There was no statistically significant difference in term of diagnosis between Pp and Pc (concordance ratio 98.5%; CI95% = 95.6% -99.7%). Three errors due to an inappropriated protocol were observed (1.5%), Two of them were related to biliary tract obstruction causes. The use of single portal phase led to a 61% decrease of the global radiation dose. Conclusion: Using a single portal phase in cases of ANTAE, CT accuracy remains the same and radiation dose decreases by 61%. Precontrast phase should be nevertheless added in suspicious biliary tract pathologies. Purpose: To evaluate the differences at CT histogram texture analysis between pancreatic neuroendocrine tumours (PNETs) and pancreatic metastasis of renal clear cell carcinoma (PMRCCCs). Material and methods: We selected 28 patients (15 PNETs and 13 PMRCC-Cs) with a CT with baseline and arterial phase scans, and histologically proven lesions. The DICOM files of each patient were loaded in a texture analysis freeware software (MaZda 4.6), then lesions were selected by manually drawing a Region of Interest (ROI). Since lesions were not always clear on unenhanced CT, ROIs were drawn on the arterial CT scan and then copied to the unenhanced CT. Mean, variance, skewness and kurtosis histogram values were extracted for each ROI in both unenhanced and enhanced CT images. Mean and standard deviation were calculated for each parameter and an unpaired t-test was used to compare PNETs and PMKCCCs. Results: No significant difference was observed between PNETs and PMRC-CCs for all the analyzed parameters, both for the unenhanced and arterial phases. Kurtosis and skewness extracted from the unenhanced CT phase showed a trend to significance (respectively p = 0.056 and p = 0.072). Conclusion: From our preliminary results, a differentiation between PNETs and PMRCCCs by means of CT histogram texture analysis parameters does not appear feasible. Purpose: To compare Iodixanol 320mgI/mL and Iomeprol 400mgI/mL in terms of liver enhancement in MDCT at 100kV and 120kV. Material and methods: 110 patients were prospectively randomized into three groups. Group A received 637.5mgI/kg LBW of Iodixanol 320 and images were acquired at 120kV. Group B received 637.5mgI/kg LBW of Iodixanol 320 and images were acquired at 100kV. Group C received 750mgI/kg LBW of Iomeprol 400 and images were acquired at 120kV. Attenuation values were measured on pre-contrast and portal-venous phase. Contrast Enhancement Indexes (CEI) were calculated subtracting basal densities from post-contrast acquisitions. Means were compared with paired T-test. A blinded independent reader evaluated image quality.
Results: Mean CEIs for groups A, B and C were respectively 49.37, 58.04 and 54.55 HU. Liver enhancement achieved injecting Iodixanol320 was significantly higher at 100kV compared to 120kV (P=0.0369). Liver enhancement achieved injecting Iodixanol320 at 100 and 120kV was not significantly different from that achieved injecting Iomeprol400 at 100 and 120kV, respectively (P=0.4183 and P=0.0526). No significant differences were observed in terms of image quality among the three groups. Conclusion: Similar liver enhancement values were observed injecting a lower amount of Iodixanol 320 compared to the ones achieved injecting Iomeprol 400. Values were even more similar when images obtained at 100kV with Iodixanol 320 were compared to the ones obtained at 120kV with Iomeprol 400 with no significant differences in terms of image quality.

SS 17.2
Iodine quantification in an ex-vivo calf liver model with simulated lesions using a single source dual energy CT (ssDECT) and three segmentation methods: inter-reader agreement and reproducibility. A. Agostini, U. Mahmood, D. Ryan, P. Sawan, L. Mannelli; New York, NY/US Purpose: To evaluate inter-reader agreement and reproducibility of iodine quantification using a single source dual energy CT (ssDECT) and three segmentation methods on an ex-vivo animal model. Material and methods: A radiologist manually injected different volumes of Iohexol 300 mgI/ml in ten chicken sausages and inserted them in two calf livers, recording injected volumes and positions. Livers were scanned 7 times with a GE Discovery CT750HD (80/140kVp, 260 mA, 0.984 pitch, 0.7s rotation time, and 2.5mm slice thickness). Three other radiologists, blinded to iodine injected and sausage positions, segmented the sausages with three methods (manual, semi-automatic segmentation, and fixed threshold) by using a GE AW VolumeShare 5; iodine concentrations were calculated on iodine (-water) images. The estimated and injected iodine quantities were compared with Bland-Altman plots. Inter-reader agreement and reproducibility were calculated with intra-class correlation coefficients. Results: 8 sausages were injected with a total of 1590mg of iodine; two sausages were excluded because of macroscopic iodine leakage. Mean errors of estimated iodine quantities were -2.76mg (-2.73%), -3.73mg (-3.27%), and -2.62mg (-2.67%) for manual, semi-automatic segmentation, and fixed threshold. Inter-reader agreements were respectively 0.9998, 0.9951, and 0.9883, while intraclass correlation coefficients among 7 scans were 0.9993, 0.9977, and 0.9999 respectively for the three segmentation methods.
Conclusion: ssDECT has high inter-reader agreement and reproducibility in iodine quantification.

SS 17.3
Contrast media protocol optimization in MDCT of the liver: advantages of using a bioimpedance device S. Picchia, M. Rengo, D. Caruso, D. Bellini, D. De Santis, A. Laghi; Latina/IT Purpose: To compare two different approaches to the quantification of CM volume for MDCT of the liver, in a population of patients with high (>30) BMI, one using nomograms and the second using a bioimpedance device. Material and methods: 9 patients were prospectively randomized into two groups. In Group A LBW was calculated using nomograms while in Group B using a bioimpedance devices. In both groups patients received 750mgI per Kg of LBW of iodinated contrast medium. Attenuation values were measured on pre-contrast and portal-venous phase. Contrast Enhancement Indexes (CEI) were calculated sottracting basal densities from post-contrast acquisitions. Means were compared with Paired T-test. Image quality was evaluated by a blinded independent reader. Results: The average amount of iodine administered in group A (41,9g) was significantly higher (p=0,04) than in group B (35,91g). Mean CEIs of the liver in portal-phase for groups A and B were respectively 66,09HU and 58,6HU. A significant greater portal-enhancement of the liver was observed in group A (p=0,03). No significant differences were observed in terms of image quality among the three groups. Conclusion: In patients with high BMI (>30) a significant higher liver enhancement was achieved using a bioimpedance device for the quantification of CM volume. The use of nomograms for the calculation of LBW in high BMI patients determine an underestimation of the adequate amount of iodine needed to obtain a proper liver enhancement.

SS 17.4
Incidence of acute transient dyspnea (ATD) after administration of gadoxetic acid (EOB) in liver magnetic resonance imaging L. Grazioli 1 , R. Faletti 2 , G. Battisti 3 , B. Frittoli 1 , P. Fonio 2 ; 1 Brescia/IT, 2 Turin/IT, 3 Spoleto/IT Purpose: The aim of the study was the evaluation of the incidence of ATD after administration of EOB and its correlation with motion artifacts during arterial phase. Material and methods: Two hundred fifty consecutives patients who underwent liver MRI with EOB were prospectively analyzed with multi-arterial CAIP-IRINHA algorithm. The incidence of ATD and the breath impairment were evaluated either after the saline flush and the MR scan procedure using a specific questionnaire and a breath hold registration (PACE system). Quality of the images was evaluated with a semiquantitative five-point scale by three different radiologists. Results: PACE system alterations were recognized in 16/250 patients (6%). Incomplete breath-hold at the end of arterial phase was reported on specific questionnaire in 11/16 patients. Acute transient dyspnea, in agreement with PACE system and questionnaire, was identified in 2/250 patients (0,8%). Motion artifacts are present in 51/250 (20%) and increase from first and second phase to the third in both different administration (saline flush and EOB) using CAIPIRINHA multi-arterial phase technique. Inter-observer concordance by three radiologists was good (k=0.78 with Choen test). Conclusion: Incidence of ATD after EOB administration was significantly lower than previously reported by literature. Motion-related artifacts incidence increased during third phase independently by saline or EOB administration.

SS 17.10
Early clinical experience in the management of postoperative lymphatic leakage using lipiodol lymphangiography and two additional glue embolization techniques S. Hur; Seoul/KR Purpose: To evaluate the safety and efficacy of lipiodol lymphangiography and two additional embolization techniques using glue (N-butyl cyanoacrylate) for the management of postoperative lymphatic leakage.
Material and methods: This study included patients with post-operative lymphatic leakage who were referred to the interventional radiology department of two tertiary referral centers for Lipiodol lymphangiography from August 2010 to December 2015. Lipiodol lymphangiography was performed for both diagnostic and therapeutic purposes. Additional embolization using glue was performed as needed using two different techniques: 1) lympho-pseudoaneurysm embolization and 2) lymph node embolization. The safety and efficacy of these techniques were determined. Results: All 18 patients underwent successful lymphangiography. Fifteen patients were observed for the therapeutic effect of lymphangiography and 7 (47%) recovered without further embolization. Nine patients, including 3 patients who underwent embolization immediately after lymphangiography and 6 patients whose conditions did not respond to lymphangiography, underwent 7 lympho-pseudoaneurysm and 4 lymph node embolizations. The clinical success rate of additional embolization was 89% (8/9) and the overall clinical success rate was 83% (15/18). The median time from initial lymphangiography to recovery was 5.5 days. No major complication related to these procedures were reported other than transient leg edema in 3 patients, all of which could be managed conservatively.