Computed tomographic prognostic factors for predicting local complications in patients with pancreatic necrosis

Deep Learning Models for Severity Prediction of Acute Pancreatitis in the Early Phase From Abdominal Nonenhanced Computed Tomography Images

OBJECTIVES

To develop and validate deep learning (DL) models for predicting the severity of acute pancreatitis (AP) by using abdominal nonenhanced computed tomography (CT) images.

METHODS

The study included 978 AP patients admitted within 72 hours after onset and performed abdominal CT on admission. The image DL model was built by the convolutional neural networks. The combined model was developed by integrating CT images and clinical markers. The performance of the models was evaluated by using the area under the receiver operating characteristic curve.

RESULTS

The clinical, Image DL, and the combined DL models were developed in 783 AP patients and validated in 195 AP patients. The combined models possessed the predictive accuracy of 90.0%, 32.4%, and 74.2% for mild, moderately severe, and severe AP. The combined DL model outperformed clinical and image DL models with 0.820 (95% confidence interval, 0.759-0.871), the sensitivity of 84.76% and the specificity of 66.67% for predicting mild AP and the area under the receiver operating characteristic curve of 0.920 (95% confidence interval, 0.873-0.954), the sensitivity of 90.32%, and the specificity of 82.93% for predicting severe AP.

CONCLUSIONS

The DL technology allows nonenhanced CT images as a novel tool for predicting the severity of AP.

Reporting of acute pancreatitis by radiologists-time for a systematic change with structured reporting template

Acute pancreatitis has a wide array of imaging presentations. Various classifications have been used in the past to standardize the terminology and reduce confusing and redundant terms. We aim to review the historical and current classifications of acute pancreatitis and propose a new reporting template which can improve communication between various medical teams by use of appropriate terminology and structured radiology template. The standardized reporting template not only conveys the most important imaging findings in a simplified yet comprehensive way but also allows structured data collection for future research and teaching purposes.

CT angiographic changes in dogs with acute pancreatitis: A prospective longitudinal study

Computed tomographic angiography (CTA) has recently been shown to be a useful tool in the diagnosis of acute canine pancreatitis, the identification of pancreatic necrosis, and the detection of sequelae. Evidence of pancreatic necrosis on CTA has been shown to be correlated with a poorer outcome in both humans and dogs and early diagnosis and intervention may improve outcomes. In humans, pancreatic necrosis is typically evident on CTA within 48 h of clinical signs, thus, repeat CTA examinations are often performed to identify pancreatic necrosis that may not have been evident on CTA examinations performed early in the course of disease. Published information investigating the timing of CTA examinations and the use of serial CTA in dogs with acute pancreatitis is lacking. In this prospective, longitudinal study, CTA examinations were performed at the time of hospitalization and repeated 3-5 days later in 11 dogs suffering from acute canine pancreatitis to determine if pancreatic necrosis or sequelae are under diagnosed on examinations performed at the time of hospitalization. Computed tomographic angiography studies were evaluated for changes in pancreatic size, pancreatic contrast enhancement, and peri-pancreatic tissues and vessels. The only statistically significant difference between the initial and repeat CTA examinations was the improvement of fat stranding on the repeat CTA examinations (P < .045). Based on these results, CTA performed at the time of admission is likely adequate in the diagnosis and evaluation of dogs with acute pancreatitis. Repeat CTA examinations are unlikely to add additional information in the absence of worsening clinical signs.

New Risk Factors for Infected Pancreatic Necrosis Secondary to Severe Acute Pancreatitis: The Role of Initial Contrast-Enhanced Computed Tomography

BACKGROUND AND AIMS

Pancreatic necrosis is a risk factor for poor prognosis of acute pancreatitis (AP). However, the associations between the findings on initial contrast-enhanced computed tomography (CT) of the pancreas and infected pancreatic necrosis (IPN) are unclear.

METHODS

This was a retrospective cohort study. Patients with severe AP (SAP) from January 2014 to December 2016 at the First Affiliated Hospital of Nanchang University were enrolled and assigned to an IPN group and a non-IPN group. Univariate and multivariate logistic regression analyses were sequentially performed to assess the associations between the variables and IPN development. A receiver operating characteristic (ROC) curve was generated for the qualified independent risk factor.

RESULTS

Forty-two patients with IPN were compared with 100 patients without IPN. Contrast-enhanced CT was performed 7 (range 3-10) days after AP onset. Multivariate stepwise logistic regression analyses showed that the number of acute peripancreatic fluid collections (APFCs) (OR 1.328, P = 0.006), presence of peripancreatic and pancreatic parenchymal necrosis (OR 4.001, P = 0.001), and gastrointestinal wall thickening (OR 3.353, P = 0.006) were independent risk factors for IPN secondary to SAP. The area under an ROC curve for the number of APFCs was 0.714, the sensitivity was 78.60%, and the specificity was 57.30% at a cutoff value of 4.5.

CONCLUSIONS

The number of APFCs, presence of peripancreatic and pancreatic parenchymal necrosis, and gastrointestinal wall thickening were independent risk factors associated with IPN. As initial contrast-enhanced CT (about 7 days from AP onset) plays an important role in predicting IPN, it is important for clinicians to consider initial imaging of the pancreas.

Asian consensus statements on endoscopic management of walled-off necrosis Part 1: Epidemiology, diagnosis, and treatment

Walled-off necrosis (WON) is a relatively new term for encapsulated necrotic tissue after severe acute pancreatitis. Various terminologies such as pseudocyst, necroma, pancreatic abscess, and infected necrosis were previously used in the literature, resulting in confusion. The current and past terminologies must be reconciled to meaningfully interpret past data. Recently, endoscopic necrosectomy was introduced as a treatment option and is now preferred over surgical necrosectomy when the expertise is available. However, high-quality evidence is still lacking, and there is no standard management strategy for WON. The consensus meeting aimed to clarify the diagnostic criteria for WON and the role of endoscopic interventions in its management. In the Consensus Conference, 27 experts from eight Asian countries took an active role and examined key clinical aspects of WON diagnosis and endoscopic management. Statements were crafted based on literature review and expert opinion, employing the modified Delphi method. All statements were substantiated by the level of evidence and the strength of the recommendation. We created 27 consensus statements for WON diagnosis and management, including details of endoscopic procedures. When there was not enough solid evidence to support the statements, this was clearly acknowledged to facilitate future research. Proposed management strategies were formulated and are illustrated using flow charts. These recommendations, which are based on the best current scientific evidence and expert opinion, will be useful for guiding endoscopic management of WON. Part 1 of this statement focused on the epidemiology, diagnosis, and timing of intervention.

Pancreatic fluid collections: What is the ideal imaging technique?

Pancreatic fluid collections (PFCs) are seen in up to 50% of cases of acute pancreatitis. The Revised Atlanta classification categorized these collections on the basis of duration of disease and contents, whether liquid alone or a mixture of fluid and necrotic debris. Management of these different types of collections differs because of the variable quantity of debris; while patients with pseudocysts can be drained by straight-forward stent placement, walled-off necrosis requires multi-disciplinary approach. Differentiating these collections on the basis of clinical severity alone is not reliable, so imaging is primarily performed. Contrast-enhanced computed tomography is the commonly used modality for the diagnosis and assessment of proportion of solid contents in PFCs; however with certain limitations such as use of iodinated contrast material especially in renal failure patients and radiation exposure. Magnetic resonance imaging (MRI) performs better than computed tomography (CT) in characterization of pancreatic/peripancreatic fluid collections especially for quantification of solid debris and fat necrosis (seen as fat density globules), and is an alternative in those situations where CT is contraindicated. Also magnetic resonance cholangiopancreatography is highly sensitive for detecting pancreatic duct disruption and choledocholithiasis. Endoscopic ultrasound is an evolving technique with higher reproducibility for fluid-to-debris component estimation with the added advantage of being a single stage procedure for both diagnosis (solid debris delineation) and management (drainage of collection) in the same sitting. Recently role of diffusion weighted MRI and positron emission tomography/CT with ¹⁸F-FDG labeled autologous leukocytes is also emerging for detection of infection noninvasively. Comparative studies between these imaging modalities are still limited. However we look forward to a time when this gap in literature will be fulfilled.

Comparison of scoring systems in predicting the severity of acute pancreatitis

AIM: To investigate the prognostic usefulness of several existing scoring systems in predicting the severity of acute pancreatitis (AP).

METHODS: We retrospectively analyzed the prospectively collected clinical database from consecutive patients with AP in our institution between January 2011 and December 2012. Ranson, Acute Physiology and Chronic Health Evaluation (APACHE)-II, and bedside index for severity in acute pancreatitis (BISAP) scores, and computed tomography severity index (CTSI) of all patients were calculated. Serum C-reactive protein (CRP) levels were measured at admission (CRPi) and after 24 h (CRP₂₄). Severe AP was defined as persistent organ failure for more than 48 h. The predictive accuracy of each scoring system was measured by the area under the receiver-operating curve (AUC).

RESULTS: Of 161 patients, 21 (13%) were classified as severe AP, and 3 (1.9%) died. Statistically significant cutoff values for prediction of severe AP were Ranson ≥ 3, BISAP ≥ 2, APACHE-II ≥ 8, CTSI ≥ 3, and CRP₂₄≥ 21.4. AUCs for Ranson, BISAP, APACHE-II, CTSI, and CRP₂₄ in predicting severe AP were 0.69 (95%CI: 0.62-0.76), 0.74 (95%CI: 0.66-0.80), 0.78 (95%CI: 0.70-0.84), 0.69 (95%CI: 0.61-0.76), and 0.68 (95%CI: 0.57-0.78), respectively. APACHE-II demonstrated the highest accuracy for prediction of severe AP, however, no statistically significant pairwise differences were observed between APACHE-II and the other scoring systems, including CRP₂₄.

CONCLUSION: Various scoring systems showed similar predictive accuracy for severity of AP. Unique models are needed in order to achieve further improvement of prognostic accuracy.

Percutaneous drainage of abdominal and pelvic abscesses in children

It has only been in the last several decades that abscesses within deep compartments, particularly within the abdomen and pelvis, have become safely accessible with imaging guidance. Since that time, percutaneous abscess drainage has become the standard of care in children. We review the clinical features, diagnosis, and image-guided management of abdominal and pelvic abscesses in children.

Global survey of controversies in classifying the severity of acute pancreatitis

BACKGROUND

The need to accurately classify the severity of patients with acute pancreatitis is widely acknowledged but some questions pertinent to this remain debatable. The aim of this study was to benchmark opinions of pancreatologists worldwide with regard to the issues related to classifying the severity of acute pancreatitis.

METHODS

An online survey was conducted using an independent commercial service. The corresponding authors of all articles pertinent to clinical aspects of acute pancreatitis published over the last 5 years were invited to participate.

RESULTS

A total of 528 invitations were sent and 240 (45%) responses from 49 countries, representing all the inhabited continents, were received. The Atlanta approach to classifying the severity of acute pancreatitis was considered adequate for modern clinical practice and clinical research by 40 (17%) of the respondents. The determinants-based approach to classifying the severity of acute pancreatitis was considered adequate for modern clinical practice and clinical research by 188 (78%) and 191 (80%) of the respondents, respectively. The definitions of local and systemic determinants of severity were also clarified.

CONCLUSION

Classifying the severity of acute pancreatitis on the basis of Atlanta approach was considered inadequate by the overwhelming majority of respondents. An international consensus on the new classification of severity of acute pancreatitis has to take into account the results of this survey.

Comparative evaluation of blood markers and extrapancreatic inflammatim on CT score in the early prediction of the severity of acute panereatitis

AIM: To compare the value of blood markers and extrapancreatic inflammation on CT score (EPIC) in early prediction of the severity of acute pancreatitis (AP).

METHODS: The clinical, laboratory and CT data obtained on admission (within 24 h of hospitalization) for 96 patients with AP who were hospitalized from September 2010 to September 2011 were analyzed. Severe AP (SAP) was defined as the presence of one or more of the following signs: mortality, persistent organ failure and/or admission ICU, and/or operation. Blood markers and EPIC were compared between SAP group and mild AP (MAP) using the t test. The correlation between blood markers, EPIC and severity of AP was analyzed. The value of blood markers and EPIC in predicting the severity of AP was assessed using receiver operation curve analysis. The sensitivity, positive predictive value and accuracy were also studied.

RESULTS: There were 20 patients with SAP and 76 patients with MAP. The levels of blood markers and EPIC in the SAP group were significantly higher than those in the MAP group [white blood cell (WBC): (15.16 ± 5.06) × 10⁹/L vs (11.05 ± 1.76) × 10⁹/L, neutrophil-lymphocyte ratio (NLR): 18.95 ± 12.13 vs 6.63 ± 3.44, high-sensitivity C-reactive protein (hs-CRP): 58.35 mg/L ± 20.47 mg/L vs 28.59 mg/L ± 12.92 mg/L, D-dimer (DD): 1596.95 μg/L ± 1409.05 μg/L vs 412.52 μg/L ± 316.66 μg/L, EPIC: 3.30 ± 0.86 vs 1.50 ± 0.96, all P = 0.000]. The Spearman correlation coefficients (rs) between severity of AP and WBC, NLR, hs-CRP, DD and EPIC were 0.419, 0.571, 0.568, 0.434 and 0.613, respectively (all P = 0.000). The area under the curve (AUC) of WBC, NLR, hs-CRP, DD and EPIC in predicting the severity of AP were 0.798 (0.670-0.925), 0.906 (0.830-0.981), 0.904 (0.838-0.970), 0.808 (0.638-0.938) and 0.917 (0.851-0.983), respectively. The predictive sensitivities were 70.00%, 85.00%, 85.00%, 75.00% and 85.00%; the positive predictive values were 58.33%, 73.91%, 51.52%, 48.39% and 72.00%; and the accuracies were 83.33%, 90.63%, 80.21%, 78.13% and 90.63%, respectively.

CONCLUSION: WBC and DD have a moderate value in predicting the severity of AP, while NLR, hs-CRP, and EPIC have a much higher value.

A comparative evaluation of radiologic and clinical scoring systems in the early prediction of severity in acute pancreatitis

OBJECTIVES

The early identification of clinically severe acute pancreatitis (AP) is critical for the triage and treatment of patients. The aim of this study was to compare the accuracy of computed tomography (CT) and clinical scoring systems for predicting the severity of AP on admission.

METHODS

Demographic, clinical, and laboratory data of all consecutive patients with a primary diagnosis of AP during a two-and-half-year period was prospectively collected for this study. A retrospective analysis of the abdominal CT data was performed. Seven CT scoring systems (CT severity index (CTSI), modified CT severity index (MCTSI), pancreatic size index (PSI), extrapancreatic score (EP), ''extrapancreatic inflammation on CT'' score (EPIC), ''mesenteric oedema and peritoneal fluid'' score (MOP), and Balthazar grade) as well as two clinical scoring systems: Acute Physiology, Age, and Chronic Health Evaluation (APACHE)-II and Bedside Index for Severity in AP (BISAP) were comparatively evaluated with regard to their ability to predict the severity of AP on admission (first 24 h of hospitalization). Clinically severe AP was defined as one or more of the following: mortality, persistent organ failure and/or the presence of local pancreatic complications that require intervention. All CT scans were reviewed in consensus by two radiologists, each blinded to patient outcome. The accuracy of each imaging and clinical scoring system for predicting the severity of AP was assessed using receiver operating curve analysis.

RESULTS

Of 346 consecutive episodes of AP, there were 159 (46%) episodes in 150 patients (84 men, 66 women; mean age, 54 years; age range, 21-91 years) who were evaluated with a contrast-enhanced CT scan (n = 131 episodes) or an unenhanced CT scan (n = 28 episodes) on the first day of admission. Clinically severe AP was diagnosed in 29/159 (18%) episodes; 9 (6%) patients died. Overall, the Balthazar grading system (any CT technique) and CTSI (contrast-enhanced CT only) demonstrated the highest accuracy among the CT scoring systems for predicting severity, but this was not statistically significant. There were no statistically significant differences between the predictive accuracies of CT and clinical scoring systems.

CONCLUSIONS

The predictive accuracy of CT scoring systems for severity of AP is similar to clinical scoring systems. Hence, a CT on admission solely for severity assessment in AP is not recommended.

Severe acute pancreatitis requiring drainage therapy: findings on computed tomography as predictor of patient outcome

BACKGROUND/AIMS

To evaluate whether morphologic features on computed tomography (CT) correlate with outcome of patients with severe acute pancreatitis (SAP).

METHODS

80 patients with SAP requiring percutaneous drainage therapy were retrospectively analyzed. Twelve CT features beyond the CT severity index (CTSI) were studied. Endpoints for patient outcome were patient death, length of hospital and ICU stay. The twelve features and the CTSI score were correlated with mortality using Kaplan-Meier estimator and correlated with length of hospital and ICU stay using the χ(2) test. A p value ≤0.05 was considered statistically significant.

RESULTS

Two CT features exhibited a significant correlation with mortality: (1) the number of parts of pancreas (head, corpus, tail) that exhibited areas of necrosis and (2) the presence of distant fluid collections (posterior pararenal space and/or paracolic gutter). Mortality was 42% (21 of 50 patients) and 20% (6 of 30 patients) if two/all three parts or none/one part of the pancreas exhibited necrosis, respectively. Mortality was 46% (18 of 39 patients) and 22% (9 of 41 patients) if distant fluid collections were present or absent, respectively. All other imaging features including the CTSI showed no significant correlation with patient outcome.

CONCLUSION

We identified two morphologic features on CT that might be helpful to predict prognosis of patients suffering from SAP. and IAP.