Deep learning to detect anterior cruciate ligament tear on knee MRI: multi-continental external validation.

OBJECTIVES: To develop a deep-learning algorithm for anterior cruciate ligament (ACL) tear detection and to compare its accuracy using two external datasets. METHODS: A database of 19,765 knee MRI scans (17,738 patients) issued from different manufacturers and magnetic fields was used to build a deep learning-based ACL tear detector. Fifteen percent showed partial or complete ACL rupture. Coronal and sagittal fat-suppressed proton density or T2-weighted sequences were used. A Natural Language Processing algorithm was used to automatically label reports associated with each MRI exam. We compared the accuracy of our model on two publicly available external datasets: MRNet, Bien et al, USA (PLoS Med 15:e1002699, 2018); and KneeMRI, Stajduhar et al, Croatia (Comput Methods Prog Biomed 140:151-164, 2017). Receptor operating characteristics (ROC) curves, area under the curve (AUC), sensitivity, specificity, and accuracy were used to evaluate our model. RESULTS: Our neural networks achieved an AUC value of 0.939 for detection of ACL tears, with a sensitivity of 87% (0.875) and a specificity of 91% (0.908). After retraining our model on Bien dataset and Stajduhar dataset, our algorithm achieved AUC of 0.962 (95% CI 0.930-0.988) and 0.922 (95% CI 0.875, 0.962) respectively. Sensitivity, specificity, and accuracy were respectively 85% (95% CI 75-94%, 0.852), 89% (95% CI 82-97%, 0.894), 0.875 (95% CI 0.817-0.933) for Bien dataset, and 68% (95% CI 54-81%, 0.681), 93% (95% CI 89-97%, 0.934), and 0.870 (95% CI 0.821-0.913) for Stajduhar dataset. CONCLUSION: Our algorithm showed high performance in the detection of ACL tears with AUC on two external datasets, demonstrating its generalizability on different manufacturers and populations. This study shows the performance of an algorithm for detecting anterior cruciate ligament tears with an external validation on populations from countries and continents different from the study population. KEY POINTS: • An algorithm for detecting anterior cruciate ligament ruptures was built from a large dataset of nearly 20,000 MRI with AUC values of 0.939, sensitivity of 87%, and specificity of 91%. • This algorithm was tested on two external populations from different other countries: a dataset from an American population and a dataset from a Croatian population. Performance remains high on these two external validation populations (AUC of 0.962 and 0.922 respectively).

Applications of artificial intelligence and machine learning for the hip and knee surgeon: current state and implications for the future.

BACKGROUND: Artificial Intelligence (AI)/Machine Learning (ML) applications have been proven efficient to improve diagnosis, to stratify risk, and to predict outcomes in many respective medical specialties, including in orthopaedics. CHALLENGES AND DISCUSSION: Regarding hip and knee reconstruction surgery, AI/ML have not made it yet to clinical practice. In this review, we present sound AI/ML applications in the field of hip and knee degenerative disease and reconstruction. From osteoarthritis (OA) diagnosis and prediction of its advancement, clinical decision-making, identification of hip and knee implants to prediction of clinical outcome and complications following a reconstruction procedure of these joints, we report how AI/ML systems could facilitate data-driven personalized care for our patients.

Pre-surgical and Post-surgical Aortic Aneurysm Maximum Diameter Measurement: Full Automation by Artificial Intelligence.

OBJECTIVE: The aim of this study was to evaluate an automatic, deep learning based method (Augmented Radiology for Vascular Aneurysm [ARVA]), to detect and assess maximum aortic diameter, providing cross sectional outer to outer aortic wall measurements. METHODS: Accurate external aortic wall diameter measurement is performed along the entire aorta, from the ascending aorta to the iliac bifurcations, on both pre- and post-operative contrast enhanced computed tomography angiography (CTA) scans. A training database of 489 CTAs was used to train a pipeline of neural networks for automatic external aortic wall measurements. Another database of 62 CTAs, including controls, aneurysmal aortas, and aortic dissections scanned before and/or after endovascular or open repair, was used for validation. The measurements of maximum external aortic wall diameter made by ARVA were compared with those of seven clinicians on this validation dataset. RESULTS: The median absolute difference with respect to expert's measurements ranged from 1 mm to 2 mm among all annotators, while ARVA reported a median absolute difference of 1.2 mm. CONCLUSION: The performance of the automatic maximum aortic diameter method falls within the interannotator variability, making it a potentially reliable solution for assisting clinical practice.

Non-invasive measurement of liver iron concentration using 3-Tesla magnetic resonance imaging: validation against biopsy.

OBJECTIVES: To evaluate the performance and limitations of the R2* and signal intensity ratio (SIR) methods for quantifying liver iron concentration (LIC) at 3 T. METHODS: A total of 105 patients who underwent a liver biopsy with biochemical LIC (LICb) were included prospectively. All patients underwent a 3-T MRI scan with a breath-hold multiple-echo gradient-echo sequence (mGRE). LIC calculated by 3-T SIR algorithm (LICSIR) and by R2* (LICR2*) were correlated with LICb. Sensitivity and specificity were calculated. The comparison of methods was analysed for successive classes. RESULTS: LICb was strongly correlated with R2* (r = 0.95, p < 0.001) and LICSIR (r = 0.92, p < 0.001). In comparison to LICb, LICR2* and LICSIR detect liver iron overload with a sensitivity/specificity of 0.96/0.93 and 0.92/0.95, respectively, and a bias ± SD of 7.6 ± 73.4 and 14.8 ± 37.6 µmol/g, respectively. LICR2* presented the lowest differences for patients with LICb values under 130 µmol/g. Above this value, LICSIR has the lowest differences. CONCLUSIONS: At 3 T, R2* provides precise LIC quantification for lower overload but the SIR method is recommended to overcome R2* limitations in higher overload. Our software, available at www.mrquantif.org , uses both methods jointly and selects the best one. KEY POINTS: • Liver iron can be accurately quantified by MRI at 3 T • At 3 T, R2* provides precise quantification of slight liver iron overload • At 3 T, SIR method is recommended in case of high iron overload • Slight liver iron overload present in metabolic syndrome can be depicted • Treatment can be monitored with great confidence.

Hepatic Fibrosis, Inflammation, and Steatosis: Influence on the MR Viscoelastic and Diffusion Parameters in Patients with Chronic Liver Disease.

Purpose To determine the relationship of liver fibrosis, inflammation, and steatosis with the magnetic resonance (MR) viscoelastic and diffusion parameters in patients with chronic liver disease and to compare the diagnostic accuracy of the imaging parameters in staging liver fibrosis. Materials and Methods Consecutive patients with chronic liver disease scheduled for liver biopsy were prospectively recruited from November 2010 to October 2012 for this institutional review board-approved study after they provided written informed consent. Sixty-eight patients underwent three-dimensional MR elastography and intravoxel incoherent motion diffusion-weighted MR imaging with a 1.5-T MR system. Fibrosis, inflammation, and steatosis were assessed with the METAVIR and steatosis, activity, and fibrosis (or SAF) scoring systems. Spearman correlation and multiple regression analyses were performed to determine the relationship between liver fibrosis, inflammation, steatosis, and alanine aminotransferase (ALT) levels and viscoelastic and diffusion parameters. The accuracy of three-dimensional MR elastography and diffusion-weighted MR imaging in the determination of fibrosis stage was assessed with Obuchowski measures. Results At multiple regression analysis, fibrosis was the only variable associated with viscoelastic parameters (ß = 0.6, P < .001, R2 = 0.33 for shear modulus; ß = 0.6, P < .001, R2 = 0.32 for elasticity). Fibrosis had a weaker independent association with the apparent diffusion coefficient (ß = -0.3, P = .02, R2 = 0.33) than did steatosis (ß = -0.5, P < .001, R2 = 0.33). Steatosis was the only factor independently associated with the pure diffusion coefficient (ß = -0.4, P = .002, R2 = 0.22). Inflammation and ALT level were not associated with the viscoelastic or diffusion parameters. The diagnostic accuracy of fibrosis staging was significantly higher when measuring the shear modulus rather than the apparent diffusion coefficient (Obuchowski measures, 0.82 ± 0.04 vs 0.30 ± 0.06; P < .001). Conclusion Fibrosis is independently associated with the MR viscoelastic parameters and is less associated with the diffusion parameters than is steatosis. These results and those of diagnostic accuracy suggest that MR elastography should be preferred over diffusion-weighted MR imaging in the staging of liver fibrosis. © RSNA, 2016.

CT and MR imaging of multilocular acinar cell cystadenoma: comparison with branch duct intraductal papillary mucinous neoplasia (IPMNs).

OBJECTIVES: To describe CT and MR imaging findings of acinar cell cystadenoma (ACC) of the pancreas and to compare them with those of branch duct intraductal papillary mucinous neoplasia (BD-IPMN) to identify distinctive elements. METHODS: Five patients with ACC and the 20 consecutive patients with histologically proven BD-IPMN were retrospectively included. Clinical and biological information was collected and histological data reviewed. CT and MR findings were analysed blinded to pathological diagnosis in order to identify imaging diagnostic criteria of ACC. RESULTS: Patients with ACC were symptomatic in all but one case and were younger than those with BD-IPMN (p = 0.006). Four radiological criteria allowed for differentiating ACC from IPMN: five or more cysts, clustered peripheral small cysts, presence of cyst calcifications and absence of communication with the main pancreatic duct (p < 0.05). Presence of at least two or three of these imaging criteria had a strong diagnostic value for ACC with a sensitivity of 100% and 80% and a specificity of 85% and 100%, respectively. CONCLUSIONS: Preoperative differential diagnosis between ACC and BD-IPMN can be achieved using a combination of four CT and/or MR imaging criteria. Recognition of ACC patients could change patient management and lead to more conservative treatment. KEY POINTS: Four imaging findings are associated with acinar cell cystadenoma (ACC). Imaging could achieve differential diagnosis between ACC and BD-IPMN. Diagnosis on imaging would change patient management and avoid surgical resection.

High sensitivity of diffusion-weighted MR imaging for the detection of liver metastases from neuroendocrine tumors: comparison with T2-weighted and dynamic gadolinium-enhanced MR imaging.

PURPOSE: To compare the sensitivity and specificity of diffusion-weighted (DW) magnetic resonance (MR) imaging for identifying liver metastases from neuroendocrine tumor (NET) to those of T2-weighted fast spin-echo (FSE) and three-dimensional dynamic gadolinium-enhanced MR imaging, with surgical and histopathologic findings as the reference standard. MATERIALS AND METHODS: This retrospective study was approved by institutional review board, and informed consent was waived. Fifty-nine patients with NETs (41 patients with 162 liver metastases, and 18 control subjects with no liver metastases) underwent MR imaging that included DW, T2-weighted FSE, and dynamic gadolinium-enhanced MR sequences. Images were retrospectively reviewed by two abdominal radiologists, independently, for the detection and characterization of liver metastases. MR findings were compared with histopathologic and intraoperative ultrasonography findings for metastasis on a lesion-by-lesion basis to determine the sensitivity of each MR sequence alone and combined. Specificity was calculated by using the control population. Interreader agreement for each MR sequence and McNemar test were also calculated. RESULTS: There was excellent agreement between observers 1 and 2 for characterizing liver metastases at per-lesion analysis (κ coefficient: 0.86-1.00). DW MR was more sensitive (observer 1: sensitivity, 71.6% [116 of 162], 95% confidence interval [CI]: 64.2%, 78.0%; observer 2: sensitivity, 71.0% [115 of 162], 95% CI: 63.6%, 77.4%) than T2-weighted FSE (observer 1: sensitivity, 55.6% [90 of 162], 95% CI: 47.9%, 63.0%; observer 2: sensitivity, 55.6% [90 of 162], 95% CI: 47.9%, 63.0%) and dynamic gadolinium-enhanced MR (observer 1: sensitivity, 47.5% [77 of 162], 95% CI: 34.0%, 55.2%; observer 2: sensitivity, 48.1% [78 of 162], 95% CI: 40.6%, 55.8%) (P < .001 for both, McNemar test). The specificity of these sequences ranged from 88.9% to 100% (DW MR vs T2-weighted FSE MR: P > .99, DW MR vs dynamic gadolinium-enhanced MR: P = .61, and T2-weighted FSE MR vs dynamic gadolinium-enhanced MR: P = .61, McNemar test). CONCLUSION: DW MR imaging was more sensitive for the detection and characterization of liver metastases from NETs than T2-weighted FSE and dynamic gadolinium-enhanced MR imaging and should be systematically performed.

Fat deposition decreases diffusion parameters at MRI: a study in phantoms and patients with liver steatosis.

PURPOSE: Assess the effect of fat deposition on the MRI diffusion coefficients in lipid emulsion-based phantoms and patients with proven isolated liver steatosis. MATERIALS AND METHODS: Diffusion-weighted MRI with 11 b values from 0-500 s/mm(2) was performed in phantoms (fat fractions 0-18 %) with and without fat suppression and in 19 patients with normal liver (n = 14) or isolated liver steatosis (n = 5) proven by histopathology. The apparent, pure and perfusion-related diffusion coefficients and the perfusion fraction were measured. Spearman correlation coefficient and Mann-Whitney U test were used for comparisons. RESULTS: A strong correlation between the apparent and pure diffusion coefficients and fat fractions was seen in phantoms. The pure diffusion coefficient decreased significantly in patients with liver steatosis (0.96 ± 0.16 × 10(-3) mm(2)/s versus 1.18 ± 0.09 × 10(-3) mm(2)/s in normal liver, P = 0.005), whereas the decrease in apparent diffusion coefficient did not reach statistical significance (1.26 ± 0.25 × 10(-3) mm(2)/s versus 1.41 ± 0.14 × 10(-3) mm(2)/s in normal liver, P = 0.298). CONCLUSIONS: Fat deposition decreases the apparent and pure diffusion coefficients in lipid emulsion-based phantoms and patients with isolated liver steatosis proven by histopathology.

Obliterative portal venopathy: findings at CT imaging.

PURPOSE: To retrospectively analyze the computed tomographic (CT) findings in a single-center series of adult patients with biopsy-proved obliterative portal venopathy (OPV) and to compare them with those observed in patients with cirrhosis. MATERIALS AND METHODS: The requirement for informed consent was waived. This institutional review board-approved study included 42 consecutive patients with a histologically proved diagnosis of OPV who underwent CT at diagnosis. The clinical characteristics at diagnosis were recorded, and CT examination results were reviewed. Two radiologists evaluated portal vein patency and intrahepatic portal branches, the morphologic changes in the liver, the presence of hepatic nodules, and signs of portal hypertension in consensus. The control group consisted of 42 patients who had histologically proved cirrhosis. CT findings were compared between the OPV patient group and the cirrhotic group and also among the conditions associated with patients with OPV. The Fisher exact test was used. P values of .05 or less were considered to indicate significant differences. RESULTS: The following CT findings were observed significantly more frequently in OPV than in cirrhosis: extrahepatic portal vein thrombosis (18 [43%] of 42 vs five [12%] of 42); intrahepatic portal abnormalities (18 [58%] of 31 vs one [2%] of 42) such as reduced caliber, occlusive thrombosis, and lack of visibility; focal nodular hyperplasia-like nodules (six [14%] of 42 vs 0 [0%] of 42); and perfusion disorders (15 [36%] of 42 vs six [14%] of 42). Conversely, the combination of hypertrophy of the caudate lobe and atrophy of segment IV (27 [64%] of 42 vs 10 [24%] of 42) and nodular surface (37 [88%] of 42 vs seven [17%] of 42) were seen significantly more often in cirrhosis. CONCLUSION: Characteristic CT findings in patients with OPV that differ from those in patients with cirrhosis were shown, the most common being the presence of intra- or extrahepatic portal abnormalities.

Adhesion-related small bowel obstruction: deep learning for automatic transition-zone detection by CT.

BACKGROUND: To train a machine-learning model to locate the transition zone (TZ) of adhesion-related small bowel obstruction (SBO) on CT scans. MATERIALS AND METHODS: We used 562 CTs performed in 2005-2018 in 404 patients with adhesion-related SBO. Annotation of the TZs was performed by experienced radiologists and trained residents using bounding boxes. Preprocessing involved using a pretrained model to extract the abdominopelvic region. We modeled TZ localization as a binary classification problem by splitting the abdominopelvic region into 125 patches. We then trained a neural network model to classify each patch as containing or not containing a TZ. We coupled this with a trained probabilistic estimation of presence of a TZ in each patch. The models were first evaluated by computing the area under the receiver operating characteristics curve (AUROC). Then, to assess the clinical benefit, we measured the proportion of total abdominopelvic volume classified as containing a TZ for several different false-negative rates. RESULTS: The probability of containing a TZ was highest for the hypogastric region (56.9%). The coupled classification network and probability mapping produced an AUROC of 0.93. For a 15% proportion of volume classified as containing TZs, the probability of highlighted patches containing a TZ was 92%. CONCLUSION: Modeling TZ localization by coupling convolutional neural network classification and probabilistic localization estimation shows the way to a possible automatic TZ detection, a complex radiological task with a major clinical impact.

Simultaneous assessment of liver volume and whole liver fat content: a step towards one-stop shop preoperative MRI protocol.

OBJECTIVE: The purpose of this study was to evaluate the ability of a whole liver volume (WLV) segmentation algorithm to measure fat fraction (FF). METHODS: Twenty consecutive patients with histologically proven fatty liver disease underwent dual-echo in-phase/out-of-phase MRI and magnetic resonance spectroscopy (MRS) at 1.5 T. Two readers independently performed semiautomatic 3D liver segmentation on the out-of-phase sequences using an active contour model. FF was calculated for voxels, segments and WLV. Segmentation inter-observer reproducibility was assessed by intra-class correlation coefficient (ICC) for WLV and FF. Fat fraction correlation and agreement as determined by histology, MRS and MRI were determined. RESULTS: ICC was 0.999 (95% CI: 0.999-1, P < 0.001) for WLV FF calculation and 0.996 (95% CI: 0.990-0.998, P < 0.001) for whole liver volume calculations. Strong correlations were found between FF measured by histology, MRS and WLV-MRI. A Bland-Altman analysis showed a good agreement between FF measured by MRS and WLV-MRI. No systematic variations of FF was found between segments when analyzed by ANOVA (F = 1.78, P = 0.096). CONCLUSION: This study shows that a reproducible whole liver volume segmentation method to measure fat fraction can be performed. This strategy may be integrated to a "one-stop shop" protocol in liver surgery planning.

Necro-inflammatory activity grading in chronic viral hepatitis with three-dimensional multifrequency MR elastography.

The purpose of this study was to assess the diagnostic value of multifrequency MR elastography for grading necro-inflammation in the liver. Fifty participants with chronic hepatitis B or C were recruited for this institutional review board-approved study. Their liver was examined with multifrequency MR elastography. The storage, shear and loss moduli, and the damping ratio were measured at 56 Hz. The multifrequency wave dispersion coefficient of the shear modulus was calculated. The measurements were compared to reference markers of necro-inflammation and fibrosis with Spearman correlations and multiple regression analysis. Diagnostic accuracy was assessed. At multiple regression analysis, necro-inflammation was the only determinant of the multifrequency dispersion coefficient, whereas fibrosis was the only determinant of the storage, loss and shear moduli. The multifrequency dispersion coefficient had the largest AUC for necro-inflammatory activity A ≥ 2 [0.84 (0.71-0.93) vs. storage modulus AUC: 0.65 (0.50-0.79), p = 0.03], whereas the storage modulus had the largest AUC for fibrosis F ≥ 2 [AUC (95% confidence intervals) 0.91 (0.79-0.98)] and cirrhosis F4 [0.97 (0.88-1.00)]. The measurement of the multifrequency dispersion coefficient at three-dimensional MR elastography has the potential to grade liver necro-inflammation in patients with chronic vial hepatitis.

Pancreatic endocrine tumors: tumor blood flow assessed with perfusion CT reflects angiogenesis and correlates with prognostic factors.

PURPOSE: To prospectively correlate multidetector computed tomographic (CT) perfusion measurement of pancreatic endocrine tumors with tumor microvascular density (MVD) assessed by using histologic techniques and to determine whether perfusion CT parameters differ between tumor grades. MATERIALS AND METHODS: Institutional review board approval and informed consent were obtained. Thirty-six patients (15 men, 21 women; mean age, 53 years; range, 18-78 years) with resectable pancreatic endocrine tumors underwent presurgical dynamic perfusion CT. Twenty-eight (78%) of 36 patients were included in the study group; eight were excluded because of artifacts that were not compatible with perfusion postprocessing. Multidetector CT perfusion data were analyzed to calculate tumor and normal pancreatic blood flow, blood volume, mean transit time, and permeability-surface area product. Multidetector CT perfusion parameters were compared with intratumoral MVD by using the Spearman correlation coefficient and with World Health Organization (WHO) classification, tumor size, tumor proliferation index, hormonal profile, and presence of metastases by using Mann-Whitney tests. RESULTS: High correlation (r = 0.620, P < .001) was observed between tumor blood flow and intratumoral MVD. Blood flow was significantly higher (P = .02) in the group of benign tumors (WHO 1) than in the groups of tumors of indeterminate prognosis (WHO 2) or well-differentiated carcinomas (WHO 3). Blood flow was significantly higher in tumors with a proliferation index of 2% or less (P = .005) and in those without histologic signs of microscopic vascular involvement (P = .008). Mean transit time was longer in tumors with lymph node (P = .02) or liver (P = .05) metastasis. CONCLUSION: Perfusion CT is feasible in patients with pancreatic endocrine tumors and reflects MVD. Perfusion CT measurements are correlated with histoprognostic factors, such as proliferation index and WHO classification.

Noninvasive quantitation of human liver steatosis using magnetic resonance and bioassay methods.

The purpose was to evaluate the ability of three magnetic resonance (MR) techniques to detect liver steatosis and to determine which noninvasive technique (MR, bioassays) or combination of techniques is optimal for the quantification of hepatic fat using histopathology as a reference. Twenty patients with histopathologically proven steatosis and 24 control subjects underwent single-voxel proton MR spectroscopy (MRS; 3 voxels), dual-echo in phase/out of phase MR imaging (DEI) and diffusion-weighted MR imaging (DWI) examinations of the liver. Blood or urine bioassays were also performed for steatosis patients. Both MRS and DEI data allowed to detect steatosis with a high sensitivity (0.95 for MRS; 1 for DEI) and specificity (1 for MRS; 0.875 for DEI) but not DWI. Strong correlations were found between fat fraction (FF) measured by MRS, DEI and histopathology segmentation as well as with low density lipoprotein (LDL) and cholesterol concentrations. A Bland-Altman analysis showed a good agreement between the FF measured by MRS and DEI. Partial correlation analyses failed to improve the correlation with segmentation FF when MRS or DEI data were combined with bioassay results. Therefore, FF from MRS or DEI appear to be the best parameters to both detect steatosis and accurately quantify fat liver noninvasively.

MRI for the measurement of liver iron content, and for the diagnosis and follow-up of iron overload disorders.

MRI is now the reference method for detecting and quantifying hepatic and extrahepatic iron overload, regardless of its cause. The decrease of the hepatic signal is proportional to the amount of iron in the tissues. It is more pronounced with T2*-weighted gradient echo sequences. It increases proportionally with the strength of the magnetic field. Thus a 3-T MRI is be more sensitive and probably more accurate to detect a slight iron overload, as seen in dysmetabolic hepatosiderosis. Conversely, a 1.5-T MRI better estimates a high overload. Quantification can be done with the calculation of T2* (or R2*) or by using the liver to muscle signal intensity ratio (SIR). Today with a single multi-echo gradient-echo sequence, obtained in a unique apnea, the two methods can be used simultaneously. An associated quantification of steatosis is also obtained. This same type of sequence is proposed for quantification of iron in other tissues and in particular for the myocardium.

Neuroendocrine liver metastases: Vascular patterns on triple-phase MDCT are indicative of primary tumour location.

PURPOSE: To re-evaluate and compare CT features of neuroendocrine liver metastases (NLM) from pancreatic (p) and enteric (e) gastroenteropancreatic (GEP) tumours. MATERIAL AND METHODS: From 2006-2013, all patients with proven GEP-neuroendocrine tumours (NETs) with at least one NLM, no previous treatment were included. On unenhanced, arterial and portal phases, NLMs were characterized as hypo-, iso- or hyperattenuating in consensus by 2 radiologists blinded to clinical data. Enhancement patterns (EP) corresponded to the combination of arterial/portal CT attenuation. RESULTS: 78 patients (43 men, 55%, mean 56±13 yo) and 559NLMs were analyzed. pNLMs were more frequently hypoattenuating on unenhanced CT than eNLMs (72% vs. 57%, p<0.001). 70% of the lesions were hypervascular with no significant difference between pNLMs and eNLMs (p=0.32). eNLMs were more frequently hypoattenuating on portal phase than pNLMs (88% vs. 56%, p<0.001). eNLMs were more frequently hyper/hypo than pNLMs (56% vs. 28%, p<0.001). pNLMs were more frequently hyper/iso than eNLMs (33% vs. 8%, p<0.001). Other NLMs showed various patterns, including hypo/hypo in 12%. CONCLUSION: Most NLMs of GEP tumours are hypervascular but the enhancement pattern on multiphasic CT depends on the primary tumour. These differences are helpful when the primary tumour has not been diagnosed.

Liver steatosis assessed by preoperative MRI: An independent risk factor for severe complications after major hepatic resection.

BACKGROUND: Steatosis assessed by histology is commonly considered to be a significant risk factor for liver surgery. MRI is considered as the new gold standard for noninvasive liver fat quantification. The purpose was to assess whether liver steatosis determined by preoperative MRI is an independent risk factor of complications after major liver resection. METHODS: All patients who underwent liver MRI before major liver resection in our institution between January 2001 and December 2011 were included in this retrospective study. The liver fat fraction (LFF) was assessed on in- and opposed-phase T1-weighted dual echo gradient echo MRI and steatosis was defined as a MRI LFF of ≥ 5%. The association between steatosis and postoperative complications (Clavien-Dindo classification, ascites > 500 mL at day 5, 50-50 criteria, fistula/collection, blood liver test alterations, pulmonary complications, nonpulmonary complications, >1 complication, duration of stay in the intensive care unit, duration of hospital stay, and death) was assessed by multivariate analysis using the appropriate model. RESULTS: A MRI LFF of ≥ 5% was associated with severe postoperative complications (Clavien-Dindo score ≥ IIIa; P = .04), more pulmonary complications (P = .02), and longer duration of hospital stay (P = .02) on the multivariate model adjusted for confounding factors. The postoperative aminotransferase levels were higher in patients with a MRI LFF of ≥ 5%, than in other patients (P = .0008). CONCLUSION: Liver steatosis assessed by routine preoperative MRI is shown to be an independent risk factor of severe postoperative complications after major liver resection.

Obesity and Fatty Pancreatic Infiltration Are Risk Factors for Pancreatic Precancerous Lesions (PanIN).

PURPOSE: The roles of intravisceral and subcutaneous fat are unknown, and the prevalence of precancerous lesions in obese patients was never evaluated. This study aims to assess the frequency and severity of pancreatic intraepithelial neoplasia (PanIN) and to correlate pathologic findings with metabolic abnormalities, type of fat, and fatty pancreatic infiltration. EXPERIMENTAL DESIGN: Normal pancreatic tissue from surgical specimens was analyzed. Fatty infiltration and fibrosis in intra- and extralobular locations and PanIN lesions were assessed. General characteristics were collected: body mass index (BMI), diabetes, and tobacco intake. Liver steatosis and subcutaneous and intravisceral fat were assessed by CT scan (ImageJ software). RESULTS: Of note, 110 patients were included [median age, 53.8 (17-85) years]. Arterial hypertension, diabetes, and tobacco intake were found in 19%, 9%, and 23%, respectively. Median BMI was 24 (16-37; BMI < 25: 45%, 25 ≤ 30: 24%, ≥30: 11%). Overall, PanIN lesions were found in 65% (type I, II, and III PanIN in 62%, 38%, and 1%, respectively). Fibrosis and fatty pancreas (intra- and extralobular locations) were found in 1% and 24% and in 30% and 51%, respectively. A correlation was observed between PanIN lesions and fatty pancreas [extralobular (0.01) and intralobular (<0.0001)], intralobular fibrosis (0.003), high BMI (P = 0.02), and subcutaneous (P = 0.02) and intravisceral fat (P = 0.02). The number of PanIN lesions was correlated with intravisceral fat (r = 0.22, P = 0.04), but not with subcutaneous fat (r = 0.14, P = 0.22). In multivariate analysis, PanIN lesions were associated with intralobular fibrosis [OR, 5.61; 95% confidence interval (CI), 1.18-42.99] and intralobular fat (OR, 17.86; 95% CI, 4.935-88.12). CONCLUSIONS: Obesity (especially android obesity) and pancreatic fatty infiltration are risk factors for pancreatic precancerous lesions.