ESGAR 2023 Book of Abstracts

Purpose: The SAVE randomised trial compared single CTC versus three biennial faecal immunochemical test (FIT) rounds for population screening of colo-rectal cancer (CRC). We evaluated herein the incidence of advanced adenoma and CRC in subjects of the CTC and FIT groups after the completion of the trial. Material and Methods: From 2017 to 2020, 1219 subjects in the CTC group and 5841 in the FIT group with a negative initial screening test were invited to two biennial FIT. Subjects positive at one of the subsequent FITs were referred to optical colonoscopy. The outcome was the number of advanced adenoma and CRC over the number of subjects with negative initial screening test. Results: The average time interval between the initial screening test and the last subsequent FIT was 6.1 years in the CTC group and 3.5 years in the FIT group. Participation to subsequent FIT in the CTC group was 80% (976/1219) at first and 54% (661/1219) at second round, whereas in the FIT group, it was 72% (4204/5841) at first and 45% (2651/5841) at second round. Overall in the two subsequent FIT rounds, 4 advanced adenomas and no CRC were detected in the CTC group, whereas 39 advanced adenomas and 8 cancers were found in the FIT group (p=0.06). Conclusion: We observed a trend in reduced incidence of advanced adenoma and CRC in subjects screened with single CTC as compared to those screened with three FIT rounds.

Intrascanner reproducibility of MRI radiomics features for pancreas cancer F. Ichinohe, K. Oyama, T. Aonuma, T. Nonaka, T. Fukuzawa, Y. Fujinaga; Matsumoto / JP Purpose: Although radiomics is a promising analysis for MRI, intrascanner reproducibility has not been fully evaluated because multiple MRIs of the same lesion are not usually acquired.The purpose of this retrospective study is to evaluate intrascanner reproducibility of MRI radiomic features for pancreatic cancer.Material and Methods: Twenty patients with pathologically proven pancreatic cancer between April 2018 and November 2022 were enrolled in this study.They underwent two contrast-enhanced MRIs within one month: with extracellular agents for evaluation of the pancreatic cancer using a 3-T scanner and with hepatobiliary agents for detection of liver metastases using the other type of 3-T scanner.Two radiologists set regions of interest by consensus on the lesions on unenhanced fat-suppressed T1-weighted images using the Dixon method (T1WI) and T2-weighted images (T2WI).We calculated 18 first-order features and 75 second-order features of the lesions with syngo.viaFrontier Radiomics (Siemens Healthcare, Erlangen, Germany).We evaluated the intrascanner reproducibility of them using intraclass correlation coefficients (ICC).Results: Out of 18 first-order features, 4 features (22%) in T1WI and 7 features (39%) in T2WI showed good reproducibility (ICC > 0.75).Out of 75 secondfeatures, 12 features (16%) in T1WI and 18 features (24%) in T2WI showed good reproducibility (ICC > 0.75).Conclusion: Intrascanner reproducibility of MRI radiomics in pancreatic cancer is insufficient, and this would be an obstacle to the clinical application of radiomics studies.

SS 3.8
Repeatability of MR elastography of the pancreas in healthy volunteers and pancreatic ductal adenocarcinoma patients N.P. Wassenaar, A.-S. van Schelt, E.M. Schrauben, H.W. van Laarhoven, J. Stoker, A.J. Nederveen, J.H. Runge; Amsterdam / NL Purpose: MR elastography (MRE) can measure visco-elastic properties of the pancreas.The ability to differentiate between healthy and diseased tissue and/ or assess treatment response, falls or stands with robust MRE.Our aims were to assess (I) MRE intra-and intersession repeatability in pancreatic ductal adenocarcinoma (PDAC) patients and healthy volunteers (HV) and (II) the ability to distinguish between healthy and malignant pancreatic tissue.Material and Methods: In total, 8 pathological-proven PDAC patients (independent of staging or prior treatment) and 8 age-and gender-matched HV underwent two back-to-back consecutive MRE scans (MRE-A and B) following MRE-C after repositioning.The shear-wave-speed (SWS) and phase-angle ( ) were calculated for healthy pancreas and tumor.One-way repeated measures ANOVA was used to determine intra-and intersession repeatability.Intraand intersession coefficient of variation (CV) were also calculated.Unpaired t test was used to test for differences between HV and patients.Results: SWS and did not significantly differ between the three MRE scans for patients and HV (Patients[SWS/ ]: F(2,14)=[0.63/1.64],p=[0.55/0.23];HV[SWS/ ]: F(2,14)=[2.42/1.90],p=[0.13/0.19]).The intra-and intersession CV for SWS/ were 10.4%/8.0%and 12.9%/10.5% for patients and 5.1%/2.8%and 8.1%/3.7%for HV.A significant difference between patients and HV was found for SWS at 1.89 versus 1.10m/s (p<0.001) and at 1.17 versus 0.81 radian (p<0.001).Conclusion: No significant differences were found between repeated scans for both visco-elastic parameters and stiffness values were comparable to the literature.The intrasession CVs should be taken into account when using pancreatic MRE for treatment response assessment in PDAC.

Conclusion:
In this retrospective analysis of US features in patients with WD, steatosis was highly prevalent.Splenomegaly was observed in one third of the patients, also in the absence of cirrhosis.Abnormal thickness of periportal echogenicity is frequent and may increase the suspicion of Wilson´s disease.Material and Methods: 129 patients underwent US, UDFF and MRI-PDFF on the same day.A radiologist performed US and 10 UDFF measurements in the right hepatic lobe; a second radiologist drew 10 regions of interest in the right hepatic lobe on the MRI-PDFF map.Agreement and relationships between UDFF and MRI-PDFF measurements were tested with the Bland-Altman method and the Spearman's correlation; sensitivity, specificity, positive and negative predictive values, and accuracy of B-mode and UDFF to detect liver steatosis (MRI-PDFF ≥5.5%) were calculated.Results: Ten patients were excluded owing to MR (N=4) or US (N=6) artifacts; 119 patients were included (63 males, 56 females; median age 60 years).Thirty-one patients (26.1%) had steatosis.There were no significant differences between the overall UDFF and MRI-PDFF measurements as well as between those performed in segments 7-8 (p=0.092 and 0.447; mean bias, −0.43% and −0.23%, respectively).Measurements in segments 5-6 were significantly different between the two methods (p=0.019,mean bias −0.68%).There was a significant positive correlation between UDFF and MRI-PDFF (rho=0.669,p<0.001).Sensitivity, specificity, positive and negative predictive values, and accuracy of B-mode and UDFF for detection of liver steatosis were 58.1/87.1%,71.6/75%, 41.9/55.1%,82.9/94.3% and 68.1/78.2%,respectively.Conclusion: UDFF has agreement with MRI-PDFF and has higher diagnostic value than B-mode US for diagnosis of liver steatosis; UDFF measurements in segments 7-8 are more reliable than those in segments 5-6.Material and Methods: This prospective study included adult patients with NAFLD who underwent US with UGAP and liver biopsy within two months.UGAP measurements were performed by two independent and blinded radiologists.A total of 12 consecutive measurements were acquired in each patient, which were repeated twice by the first radiologist.The median values were considered for the analysis.Hepatic steatosis was graded at liver biopsy as (0)<5%; (1)5-33%; (2)33-66%; (3)>66%.Area under the receiver operating characteristic (ROC) curve (AUC) with Youden index was calculated to determine the diagnostic performance with optimal cutoff.The intraclass correlation coefficient (ICC) was used to assess the inter-and intra-observer reproducibility.

Scientific Sessions
SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 14, 2023 SS 4.6 Quantitative analysis of liver proton density fat fraction with deep learning on low-dose CT H.-J. Chung 1 , K. Jong-Min 1 , J. Lee 2 , J. Yoon 2 , S. Park 1 ; 1 Medical IP / KR, 2 Seoul / KR Purpose: The purpose of this study is to evaluate a deep learning-based method for estimating liver proton density fat fraction (PDFF) maps from lowdose CT (LDCT) images.Material and Methods: The dataset, which our PDFF estimator trained with, only contains small CT dataset (train: 18 cases, validation 2 cases) acquired from conventional CT Scanner (Brilliance, Philips).For low-dose simulation, the Poisson noise was inserted into projection of CT images to create a noisecorrupted sinogram at a quarter-dose level.To reconstruct LDCT images, the additional denoiser was trained using the American Association of Physicists in Medicine (AAPM) low-dose CT challenge dataset.145 adult clinical patients who had both CT (IQon, Philips Healthcare) and MRI (Skyra 3.0T, Siemens Healthineers) scan were retrospectively included for external validation.The performance of our estimation method was evaluated using R-squared metric.For measuring liver PDFF, the liver mask from vendor-provided solution (Siemens MedCom, Siemens Healthineers), and the commercial body composition analyzer (DeepCatch v1.2, Medical IP) was used for MRI and CT, respectively.The proposed deep neural network was implemented inDeep-Catch (v1.x,Medical IP, South Korea).Results: The R-squared value between MRI-and equation-based liver PDFF from original LDCT, deep learning-based liver PDFF from simulated LDCT without and with denoiser, and original standard dose CT were 0.6922, 0.7044, 0.8651, and 0.8664, respectively.

Conclusion:
In this study, we proposed the deep learning method for generating liver PDFF maps from LDCT images through estimator and denoiser.The proposed method may be a promising tool for measuring liver fat content without additional MRI acquisition.Material and Methods: Thirty patients (nine females) with multiple metabolic risk factors were prospectively enrolled between October and December 2022.Each patient underwent 1.5T upper abdomen MRI examination; acquisition protocol included axial proton density fat fraction (PDFF), and axial T1-weighted dual-echo breath hold sequences targeted to the evaluation of lumbar paraspinal muscles at the third lumbar vertebra (L3) level.Two expert radiologists performed quantitative image analysis on a dedicated workstation.Steatosis estimation expressed as percentage was collected and grading of steatosis assigned.The presence of sarcopenia was assessed by segmenting the L3 dual-echo images with ImageJ pixel analysis software and calculating the SMI.Correlations among parameters were assessed using Spearman's coefficient of rank with a dedicated software and a p value<0.05 was considered significant.

Conclusion:
Patients with multiple metabolic risk factors and hepatic steatosis quantified at MRI are correlated with lower SMI values and sarcopenia and the correlation is stronger in female patients.This method might be a noninvasive, radiation-free and repeatable method for a comprehensive metabolic patients' assessment at diagnosis and follow-up.

SS 4.8
Functional liver imaging score: an intraindividual comparison between gadoxetate disodium and gadobenate dimeglumine M. Milazzo, R. Cannella, F. Agnello, G. Rancatore, G. Brancatelli; Palermo / IT Purpose: The functional liver imaging score (FLIS) is a semiquantitative tool developed to estimate liver function based on the three hepatobiliary phase (HBP) features on gadoxetate disodium (Gd-EOB-DTPA) MRI (EOB-MRI).The aim of this study is to perform an intraindividual comparison between the FLIS obtained with EOB-MRI and gadobenate dimeglumine (Gd-BOPTA) MRI.Material and Methods: This study included cirrhotic patients with available laboratory and clinical data, who underwent EOB-MRI and BOPTA-MRI within 6 months acquired with an HBP at 20 minutes and 2 hours, respectively.Three readers with different experience levels independently reviewed the HBP images to evaluate the FLIS based on the following features on a 0-2 scale: hepatic enhancement, biliary excretion, and portal vein signal intensity.Intraindividual differences in FLIS were assessed with the Wilcoxon signed rank-sum test and inter-reader agreement with the intraclass correlation coefficient (ICC).

SS 4.9
Contribution of functional liver imaging score obtained from gadoxetic acid-enhanced MRI to predict liver function in patients with chronic liver disease and liver cirrhosis N. Inan Gurcan 1 , I. Sever 1 , A. Atasever 1 , S. Yazici 1 , T. Sahin 1 , S. Server 1 , O. Ulusoy 1 , Y. Yuzer 1 ; 1 Istanbul / TR Purpose: To evaluate liver function in patients with liver cirrhosis and chronic liver disease using functional liver imaging scores (FLIS) obtained from gadoxetic acid-enhanced MRI and its association with clinical-laboratory parameters including indirect bilirubin, INR, albumin-bilirubin (ALBI) grade, Child-Pugh (CP), and model for end-stage liver disease (MELD) score.Material and Methods: We retrospectively analyzed 173 patients (130 men, 43 women) between January 2017 and December 2022 with the following inclusion criteria: patients diagnosed with liver cirrhosis or chronic liver disease (CLD) who underwent gadoxetic acid-enhanced MRI.Patients with mechanical cholangiectasis on MRI were excluded.Three parameters on hepatobiliary phase images were evaluated for FLIS: liver parenchymal enhancement, biliary excretion, and signal intensity of the portal vein and all of which were scored on the 0-2 ordinal scale.We assessed the correlation between clinical-laboratory liver function parameters (indirect bilirubin, INR, ALBI grade, CP, and MELD score) and FLIS using Spearman rank correlation.Receiver operating characteristic (ROC) curve analysis was performed to demonstrate the ability of FLIS for the prediction of liver function.Results: FLIS parameters showed strong to very strong correlation with liver function parameters.ROC curve analysis showed that FLIS ≥ 5 was the optimal cutoff for the prediction of CLD (sensitivity, 81%; specificity, 87%).Conclusion: FLIS showed a strong correlation with liver function parameters; hence, it can be useful for the prediction of liver function in clinical practice.

Scientific Sessions
SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 14, 2023 SS 4.10 MELIF score: an automated, non-invasive measurement of liver function using MRI that outperforms standard liver function scores C. Rio Bartulos 1 , K. Senk 2 , R. Bade 3 , M. Schumacher 3 , J. Platz 3 , N. Kaiser 3 , J. Woetzel 3 , P. Wiggermann 1 ; 1 Braunschweig / DE, 2 Regensburg / DE, 3 Bremen / DE Purpose: Diffuse and chronic liver disease along with HCC are a global health burden.These diseases need to be monitored during their progression, which requires tools to assess liver function.Here we present a new liver function score, which is based on T1 relaxometry and calculated fully automatically by AI-based software.Material and Methods: An image-based software system was developed that includes automatic image processing with elastic spatial registration, AI liver segmentation, and incorporation of patient weight and height into liver function modeling, resulting in the MELIF score.Gadolinium ethoxybenzyldiethylenetriaminepentaacetic acid-enhanced MR images are used to calculate the MELIF score, and data from up to 195 patients were analyzed.To analyze its diagnostic potential, the MELIF score is retrospectively compared with classical liver function parameters such as MELD, Child-Pugh (CP), and ALBI scores, as well as with the T1 reduction rate (rrT1), a classical imagebased liver function parameter on which the MELIF score is based.For this purpose, we use Pearson correlation and receiver operating characteristic area-under-the-curve analysis.

Results:
The MELIF score correlates significantly stronger with MELD and CP score than ALBI or rrT1.Furthermore, we can show that the MELIF score is better able to distinguish between good and impaired liver function (AUC 0.8), defined by the MELD score, and between patients with and without cirrhosis (0.83) than the ALBI score (0.77 and 0.79).

Conclusion:
In the future, this developed software will enable fully automated routine determination of liver function.The MELIF score can compete with common liver function parameters and perform better than the ALBI score or rrT1.Material and Methods: PD patients suspended motility-influencing drugs 24 hrs before their scan (REC #: 11/LO/1634).They sipped 800 mL 2% mannitol over 45 mins before successive 2D dynamic sequences were acquired in the supine position.Images were processed using a commercially available non-rigid registration algorithm (GIQuant, Motilent, UK) to quantify small bowel motility.Mean ± S.D motility (arbitrary units, a.u) was compared using a t test against healthy volunteers (HV) scanned using a similar protocol-1L of 1.7% mannitol in the prone position.

Conclusion:
The abbreviated MRE protocol was feasible in most PD patients.Small bowel motility may be reduced compared to HVs.Future work will focus on minimising acquisition time and oral contrast load and comparing age-matched controls to PD patients at different stages of disease to assess whether differences are a function of age or disease progression.Purpose: Steatohepatitic HCC (sh-HCC) is characterized by >50% area of steatohepatitic changes.However, fat (+/-inflammation) can be found in nonotherwise specified HCC (NOS-HCC).We compared the imaging features and outcomes of sh-HCC, fatty NOS-HCC, and NOS-HCC with a steatohepatitic component <50% (sh-NOS-HCC).

Authors' Index
Material and Methods: Patients undergoing hepatic resection for an HCC with fat content on pathology between 2012 and 2021 were included.Imaging features were assessed per liver imaging reporting and data system (LI-RADS) v2018 (features and categories).Fat quantification was performed on chemical-shift MRI.Tumor groups were compared, and recurrence-free (RFS) and overall survival (OS) were estimated.

Conclusion:
Imaging and prognostic characterization of fat containing HCC subtypes largely overlap.A steatohepatitic HCC may be suspected in homogenous tumors with fat-in-mass and developed on NASH.Purpose: To assess the CT changes associated with chemotherapy-associated liver injury (CALI) and their influence on severe postoperative complications in patients undergoing pancreaticoduodenectomy (PD) after neoadjuvant oxaliplatin-based chemotherapy for pancreatic ductal adenocarcinoma (PDAC).Material and Methods: Patients undergoing PD after oxaliplatin-based chemotherapy for PDAC (2017-2020) were included.Baseline and preoperative CT were reviewed to extract qualitative and quantitative features.Severe complications were defined as Clavien-Dindo ≥3.Features associated with CALI and severe complications were identified by paired univariate analysis.
More granular analysis showed that CT features were mostly associated with ascites and sepsis, but not with clinically significant pancreatic fistula.

Conclusion:
Neoadjuvant oxaliplatin-based chemotherapy for PDAC leads to CT changes, including splenic volume increase, heterogeneous hepatic enhancement, and portosystemic shunts.In addition to clinical and surgical characteristics, lower precontrast hepatic parenchyma attenuation and heterogeneous enhancement were associated with severe liver-related complications after PD.

SS 8.4
CT evaluation of liver steatosis after neoadjuvant chemotherapy as a predictive factor for surgical complications in patients with pancreatic adenocarcinoma E. Boffa 1 , A. Spezia 1 , G.A. Zamboni 1 , G. Mansueto 2 ; 1 Verona / IT, 2 Verona / IT Purpose: To evaluate the liver density changes on CT after neoadjuvant treatment (NAT) for pancreatic adenocarcinoma to quantitatively assess druginduced hepatic damage and evaluate the role of moderate/severe steatosis as a predictive factor for post-operative complications.
Material and Methods: Informed consent for the utilization of clinical and radiologic data was provided by all patients (PAD-R registry, n1101CESC).IRB approval was not required for this retrospective study.Two readers reviewed the pre-and post-chemotherapy CT scans performed between April 2019 and February 2022 on 76 patients with pancreatic adenocarcinoma who received NAT and went to surgery.Liver steatosis was evaluated by drawing multiple ROIs on liver and spleen parenchyma (6 on liver, 4 on spleen) on the venous phase, obtaining the mean difference in density between liver and spleen.The population was divided into 2 groups based on steatosis degree after NAT (group 1: <30%; group 2: >30%), then post-surgical complications were compared across groups applying the chi-squared test.
Results: Twenty-eight patients (37%) developed or worsened their steatosis degree after neoadjuvant chemotherapy.42 patients (55%) developed postsurgical complications (POPF, liver failure, bile leak, PPAP).Moderate/severe steatosis had a positive trend of correlation with post-surgical PPAP, although not significant, and in the non-PPAP patients was significantly correlated with post-surgical complications frequency (Clavien-Dindo index after 90 days >3

Material and Methods:
We performed a retrospective review of pancreas transplant patients in our centre from January 2009 to January 2021.We considered acute venous thromboses those occurring during the first 30 days after transplantation.Our protocol includes at least one CDUS at 24-48h post-transplant and another at 7 days.CEUS was performed when CDUS detected neither flow nor thrombus in the splenic or mesenteric veins of the pancreatic graft.
Results: During this period, 247 pancreas transplants were performed.The CDUS detected 35 venous thromboses.CEUS was performed in 28 other cases where CDUS did not identify either flow or thrombus.In 15 of these patients, CEUS demonstrated that the splenic and mesenteric veins were permeable, but with low flow.In 11 cases, thrombosis was confirmed by angio-CT or arteriography.In one case, angio-CT showed misdiagnosis of thrombosis due to filiform vessels.The remaining case corresponded to peripheral thrombosis and no further tests were performed.There were only two cases of acute venous thrombosis in which the pancreas could not be assessed ultrasonographically and CT had to be performed.Material and Methods: A dataset of 170 liver MRI was used to train a convolutional neural network for liver parenchyma segmentation by excluding the liver vessels on non-contrast T1 vibe Dixon in-phase acquisitions.21 prospectively acquired liver MRI of healthy volunteers unseen by the neural network with T1 vibe Dixon and shMOLLI T1 mapping sequences were assessed by measuring parenchymal T1 relaxation time based on a manual ROI of the whole liver by excluding the liver vessels.The previously unseen T1 vibe Dixon in-phase acquisitions were then automatically segmented with the AI model and registered to the T1-maps using affine transformations.Subsequently, the segmentation was eroded along parenchymal and vessel borders to exclude partial-volume effects.Finally, T1-relaxation times with standard deviation and ROI voxel size were determined using the AI segmentation mask and compared with the manual annotations as the ground truth.
Results: The automated AI-based parenchyma ROI calculation achieved a mean voxel intensity of 862+/−72ms, compared to the manually annotated ROI with 846+/−72ms.The Spearman correlation between the automated AI and the manual calculation was 0.94 (p<0.001test were used for statistical analysis.ROC curve was used in the diagnostic power evaluation and p < 0.05 was considered significant. Results: There was no statistically significant difference between the group responders and non-responders, in terms of SFA, VFA, total fat area (TFA), tumor volume, SMA, sarcopenia, and sarcopenic obesity.On the other hand, MFV was significantly less in the non-responder group.Ryan score decreased as MFV increased (p=0.02).In addition, there was weak positive correlation between MFV and SFA (p<0.01),VFA (p<0.01) and TFA (p<0.01).

Conclusion:
CT-based mesorectal fat volume measurements, which are noninvasive and easy to apply, are important parameters that can be used practically in the prediction of neoadjuvant treatment response in rectal cancer patients.Future studies are needed for validation of this parameter.

SS 10.6
Association between dynamic contrast-enhanced MRI parameters and prognostic factors in patients with primary rectal cancer S. Kim, H. Kim, K. Nam; Busan / KR Purpose: We aimed to evaluate the association between perfusion parameters derived from dynamic contrast-enhanced MRI (DCE-MRI) with prognostic factors in patients with primary rectal cancer.Material and Methods: A total of 51 patients (31 men, 20 women; mean age, 69 years; range, 45-89 years) who had pathologically proven rectal adenocarcinoma and were treated via surgery were retrospectively enrolled.All the patients underwent preoperative DCE-MRI.Two blinded radiologists determined the tumor border after radiologic-pathologic correlation in each patient and drew regions of interest along the tumor border on consecutive slices bearing tumor to cover the whole tumor volume.The four perfusion parameters, including the volume transfer constant (Ktrans), were calculated under the extended Toft model.Tumor stage, lymph node stage, extramural venous invasion, Kirstenras mutation, carcinoembryonic antigen, circumferential resection margin status, tumor size and tumor differentiation were included as prognostic factors.
The association was assessed via correlation or t test.In the case of significant prognostic factors, receiver operating characteristic (ROC) curve analyses were performed to estimate the diagnostic predictive values.Results: Ktrans only showed a significant difference according to tumor differentiation, specifically, between the well-differentiated (n = 6) and moderately differentiated (n = 45) groups (0.127 ± 0.032, 0.084 ± 0.036, P = 0.036).The AUC was 0.838 (95% CI, 0.702-0.929),and the estimated accuracy, sensitivity, and specificity were 87, 90, and 60%, respectively.However, none of the other perfusion parameters showed significant differences in any of the prognostic factors.Conclusion: Ktrans showed a significant difference according to tumor differentiation.

SS 10.7
A systematic review of prognostic models which incorporate imaging data that predict outcomes in rectal cancer R. Mitchell-Hay, H. O'Brien, A. Murray, D. Mclernon; Aberdeen / GB Purpose: To review the methodological quality and predictive performance of prognostic models that predict overall survival or response to chemoradiotherapy in rectal cancer patients.Material and Methods: Medline, PubMed, The Cochrane Library and Web of Science were searched for papers concerning the development and validation of predictive and/or prognostic models for adult patients with rectal cancer that included both clinical and imaging data.Two authors independently screened titles, then subsequently abstracts and full texts.Any uncertainty was resolved by consensus.Data were extracted from the included papers using the CHARMS checklist with risk of bias assessed using the PROBAST tool.
Results: 5288 records were initially identified with 1766 duplicates removed prior to screening.3523 records were screened with 95 full texts reviewed.25 papers were included in the final review.28% of models were externally validated on an independent data set.Following internal validation, AUC where it was reported ranged from 0.61 to 0.97.The overall risk of bias was recorded as high in 15 papers with unclear risk of bias in the remaining 10. High risk of bias was predominantly found in the analysis section of the PROBAST tool and primarily related to issues such as low event rate and univariate screening being used for predictor selection.

Conclusion:
All models within rectal cancer that incorporate imaging data showed high or unclear risk of bias.Methodological improvements such as employing larger data sets and avoiding poor model building approaches could address these issues.Material and Methods: We retrospectively included patients (age> 18 years) with colorectal liver metastases, who underwent a DECT with a 3rd generation Dual-Source scanner (Somatom Force, Siemens, post-contrast DECT: 80-100/150Sn kV, modulated mA, iopamidol 370 mg I/mL, 1.3 ml/kg) between January 2018 and January 2019.The DECT venous dataset was processed with a 3-material decomposition algorithm (Liver VNC, Siemens).Two readers in consensus placed a 10 mm diameter circular region of interest (ROI) within 15 mm from liver metastases avoiding vascular structures, and an identical control ROI in a contralateral lesion-free liver segment.Statistical analysis was performed with non-parametric tests.
Results: 53 patients (39 M/14 F) were included, with a median age of 64 years, for a total of 102 nodules (median diameter 23 mm).The distribution of steatosis showed a median fat fraction of 4% in the perilesional site and 7% in the contralateral lesion-free parenchyma (Wilcoxon p = 0.046).

Conclusion:
The quantification of liver steatosis from DECT datasets found subtle, but statistically significant differences between perilesional and lesionfree liver parenchyma.
, M. Zerunian, T. Polidori, A. Del Gaudio, D. Caruso, A. Laghi; Rome / IT Purpose: To analyse the possible correlation between hepatic steatosis and sarcopenia evaluated with skeletal muscle index (SMI) at unenhanced MRI of the abdomen in patients with multiple metabolic risk factors.
To evaluate an automated T1 mapping quantification of the liver with deep-learning-based 3D T1 vibe Dixon liver parenchyma segmentation and affine transformation with voxel-wise co-registration.
, A. Agostini, M. Marchegiani, E. Montecchiesi, L. Ottaviani, R. Rossi, M. Vivarelli, A. Giovagnoni; Ancona / IT Purpose: To evaluate the role of dual-energy CT (DECT) in the assessment of the distribution of liver steatosis in patients with colorectal liver metastases.

2 A preliminary study on the prediction of HCC in patients with chronic hepatitis B using heterogeneous data
J.-M.Kim 1 , H.-J. Chung 1 , D. Lee 2 , J.-H. Lee 2 , S. Park 1 ; 1 Medical IP., Co., Ltd / KR, 2 Seoul National University College of Medicine / KR The validation dataset including 42 HCC and 233 non-HCC patients was studied.The AUROC of model with only 10 baseline parameters was 0.6232 and those of model with 10 baseline parameters and volume of liver and spleen was 0.7073.
Conclusion:In this study, we developed and evaluated two HCC prediction models with and without using CT-derived quantitative factors.According to the results, the model with liver and spleen volumes has significantly better predictive performance than those with only 10 baseline parameters.SS 5.3Prospective randomized controlled trial of US versus US and biomarkers for early detection of HCC H. Farhang Zangeneh, O. Cerocchi, K. Khalili, L.A. El-Karim, H. Janssen, B. Hansen, J. Feld; Toronto, ON / CA Purpose: To determine the impact of triple blood biomarkers in US screening for early HCC.Material and Methods: Adults with cirrhosis or high-risk HBV infection followed at one institution were randomized to HCC screening with US alone (Group A) or US+BM (Group B) with measurement of alpha-fetoprotein (AFP), lectin-reactive fraction of AFP (AFP-L3) and des-gamma-carboxy prothrombin (DCP).Elevated BM levels and/or findings on US triggered CT/MRI for confirmation of HCC diagnosis.The primary outcome was the proportion of HCCs diagnosed at a curable stage (BCLC 0/A) within Milan criteria.

A simple scoring system using MR enterography based on the most inflamed segment to evaluate disease activity in Crohn's disease
N. Seo, J. Lim, S.-S.Kim, H. Bae, J. Cheon; Seoul / KR Presence or absence of residual small bowel abnormality does not in itself predict future relapse.However, in those with residual disease, bowel wall thickness is greater in those who relapse and may be a simple tool to guide the decision to discontinue biologic medications.
Conclusion:The maximal segmental MARIAs based on the most inflamed bowel segment can be a rapid and practical MRE-based index to represent overall disease activity and to predict endoscopic remission in CD.SS 6.5MRI factors associated with relapse following discontinuation of biologic medication in Crohn's disease J. Pakpoor, H. Fitzke, M. Hameed, T. Glover, J. Holmes, W. Blad, L. Whitley, S. Taylor; London / GB Purpose: The decision to stop biologic medication in Crohn's disease (CD) in apparent clinical remission is challenging.We investigated MR enterography (MRE) observations associated with future relapse.Material and Methods: 49 CD patients discontinuing biologics due to clinical remission and undergoing MRE within 12 months prior, or 1 month after, were identified.Two blinded radiologists assessed small bowel disease presence/ activity and determined length-of-disease, wall-thickness, fat-wrapping, and s-MARIA.Relapse status was assigned based on clinical records including patient symptoms, biochemistry, imaging and medications.Groups were compared with Chi-square and Wilson-Cox statistics.Results: Patient mean age was 30.1 (SD: 13.0, 19 female).Median follow-up was 55 months.Twenty-five (51%) had disease relapse (DR)
Purpose: Conclusion: Replacing IV Gad with MI did not affect the diagnostic accuracy or inter-reader agreement for the detection of SBCD activity or most severe segment.SS 6.

7 Motility in small bowel strictures in Crohn's disease measured with cine-MRI
Strictures (wall thickening >3mm and >50% luminal reduction) and pre-stenotic dilations (luminal diameter >3cm) were delineated on a reference image and motility was quantified within these regions of interest (ROI) on a motility map, producing a single, numerical motility score (arbitrary units=AU).Stricture and pre-stenotic dilation motility scores are presented in medians[IQR].Correlations were tested between stricture motility, disease duration and HBI by means of spearman's rank correlation test.
Purpose: Conclusion:In this proof-of-concept study, motility is potentially reduced in the ascending and transverse colon in patients with left-sided UC and symptoms of UCAC compared to controls, justifying study in larger cohorts.Hypomotility may be a contributing factor in UCAC suggesting stimulants/prokinetics may be effective.SS 6.

9 Quantified small bowel motility as a biomarker of Parkinson's disease: a feasibility study using an abbreviated MR enterography protocol
J. Holmes, H. Fitzke, J. Pakpoor, A. Menys, D. Atkinson, G. Bhatnagar, S. Taylor; London / GB Purpose: GI dysmotility is an early feature of Parkinson's disease (PD), preceding diagnosis by >10 years (Schapira et al. 2017).We assessed the feasibility of an abbreviated MR enterography protocol (MRE) with reduced oral contrast volume to quantify global small bowel motility in PD and compared to healthy volunteers.

3 Prognostic value of MRI features in patients with single large HCC after curative resection
Conclusion:The pattern of imaging features of HCC appears different according to AFP and PIVKA-II serum levels.An increase in PIVKA-II seems more representative of local tumor aggressiveness, while an increase in AFP may be more informative about distant or extrahepatic tumor spread.Both dosages are complementary and should be done.SS 7.K.Gu, J. Min; Seoul / KR

5 MRI and CT findings after irreversible electroporation for locally advanced pancreatic cancer
This study is an ancillary study from a prospective single-center, open-label, non-randomized study called IRECAP (clinicaltrials.govidentifier: NCT03105921).All participants provided written informed consent.Patients underwent enhanced CT and MRI before IRE and within six days (D6), one month (D30), and three months (D90) after IRE.On CT and MRI, 4 patterns were evaluated, normal pancreas parenchyma, fat inflammation, tumor and necrosis.Only descriptive statistic data were performed.Results: 15 patients were included (mean age 61 years, range 37-77).After IRE, ablation zone was bigger in size than the tumor targeted, without clearly demarcated margins due to peripheral inflammation.Transient necrosis was observed in all patients.A tissular mass can be detected at one month after IRE inside the treated area.The nature of this mass, tumoral or fibrotic, remains unknown but pathological analysis is on-going and results will be presented for the congress.Vascular involvement seems unmodified after IRE.To determine the value of contrast-enhanced US (CEUS) in the diagnosis of acute venous pancreas graft thrombosis when colour Doppler US (CDUS) identifies neither flow nor thrombus.
). Conclusion: Moderate/severe liver steatosis after neoadjuvant treatment demonstrates a correlation with post-surgical complication frequency, thus reflecting an increased surgical risk in patients with chemotherapy-induced liver damage.SS 8.M. Vionnet, B. Belardy, J. Damion, J. Goujon, S. Velasco, A. Coulibaly, G. Herpe, D. Tougeron, J.-P.Tasu; Poitiers / FR Purpose: Locally advanced pancreatic cancer (LAPC) accounts for 30% of patients with pancreatic adenocarcinoma.In LAPC, irreversible electroporation (IRE) has been tested to improve survival and quality of life.However, imaging findings post-IRE remains described in only 2 previous reports.The aim of this study was, therefore, to describe pancreas post-IRE imaging findings.Material and Methods:Purpose:

Early intravoxel incoherent motion diffusion-weighted imaging for the prediction of post-pancreatectomy acute
Conclusion: Contrast-enhanced US (CEUS) allows the differentiation between venous low flow and thrombosis when no Doppler signal is found in the pancreas graft veins.Radiomic features were derived from preoperative CT scans from adult patients with an indication for a pancreatoduodenectomy at the Amsterdam University Medical Center to develop the preoperative-fistula risk score (pre-FRS).These radiomics features were analyzed with four machine learning classifiers.The model was then validated in an independent dataset from Verona University Hospital.The predictive performances of the models were evaluated using a range of metrics.The pre-FRS was successfully validated in an external cohort and found to have adequate performance for predicting POPF.This preoperative prediction of POPF has the potential to improve patient prognosis and could assist surgeons making tailored treatment decisions.

Usefulness of 2D shear wave elastography for evaluation of hepatic fibrosis and treatment response in patients with autoimmune hepatitis
E. Soh, Y. Lee, Y. Kim; Iksan / KR Patients diagnosed with AIH and underwent 2D SWE between June 2014 and June 2021 were enrolled in this retrospective study.We classified the patients into four groups according to the histologic stage of fibrosis (F1-F4).The baseline characteristics, laboratory test results, histologic results, and 2D SWE results were analyzed.We investigated the diagnostic performance of LS measurements in hepatic fibrosis staging and compared variables before and after steroid treatment for AIH.Results: A total of 69 patients were analyzed.The LS values differed according to the stage of liver fibrosis (P < 0.001).The area under the curve of the LS value was 0.903, 0.815, and 0.854 for F2, F3, and F4, respectively.The diagnostic performance of LS measurement was significantly greater than that of serum biomarkers, except for fibrosis index-4 for F4 (P < 0.05).There was a significant difference in both the LS value and DS in patients who received steroid therapy during the follow-up examination (P = 0.012 and 0.011, respectively).Conclusion: 2D SWE is a useful method for the assessment of hepatic fibrosis in patients with AIH.For follow-up examinations, LS and DS can be used as reliable parameters to evaluate treatment response to AIH.

based virtual portal vein pressure gradient: a noninvasive computational model for the diagnosis of portal hypertension
S. Wan 1 , L. Ren 2 , Y. Wei 3 , B. Song 4 ; 1 Cheng Du / CN, 2 Beijing / CN, 3 Chengdu / CN, 4 Cheng DU / CN We confirmed that the computational model could be used to noninvasively diagnose portal hypertension in cirrhosis and may be used as a substitute for invasive transjugular PVPG measurements.

Concomitant left gastric vein embolization during tips placement for acute variceal bleeding has no effect on its recurrence: results of an observational retrospective multicenter
Radiotherapy and/or chemotherapy is applied before the operation in medium-risk and high-risk rectal cancers.In recent years, interest in the relationship between body composition and prognosis in cancer patients has increased.In this study, we aim to investigate the importance of visceral fat quantification, sarcopenia and sarcopenic obesity before the neoadjuvant chemoradiotherapy (CRT) treatment of rectal cancer patients and evaluate the treatment response using those parameters.Material and Methods: 78 patients who were diagnosed with rectum cancer and received neoadjuvant CRT treatment between 2010 and 2022, with preoperative abdominopelvic CT and rectum MRI were included.Musclefat distribution, visceral fat area (VFA), subcutaneous fat area (SFA), skeletal muscle area (SMA) measurements on axial images through L3 vertebra and mesorectal fat volume (MFV) measurement were performed on CT scan.Rectal tumor volume was measured on MRI.Pathological treatment response was assessed with the Modified Ryan score.Mann-Whitney U test and Pearson 2 Conclusion:In this multicenter national real-life study, we did not observe any benefit of concomitant LGVE during TIPS placement for acute variceal bleeding on recurrence rate.Purpose: