A deep learning model enables accurate prediction and quantification of pulmonary edema from chest X-rays.

BACKGROUND: A quantitative assessment of pulmonary edema is important because the clinical severity can range from mild impairment to life threatening. A quantitative surrogate measure, although invasive, for pulmonary edema is the extravascular lung water index (EVLWI) extracted from the transpulmonary thermodilution (TPTD). Severity of edema from chest X-rays, to date is based on the subjective classification of radiologists. In this work, we use machine learning to quantitatively predict the severity of pulmonary edema from chest radiography. METHODS: We retrospectively included 471 X-rays from 431 patients who underwent chest radiography and TPTD measurement within 24 h at our intensive care unit. The EVLWI extracted from the TPTD was used as a quantitative measure for pulmonary edema. We used a deep learning approach and binned the data into two, three, four and five classes increasing the resolution of the EVLWI prediction from the X-rays. RESULTS: The accuracy, area under the receiver operating characteristic curve (AUROC) and Mathews correlation coefficient (MCC) in the binary classification models (EVLWI < 15, ≥ 15) were 0.93 (accuracy), 0.98 (AUROC) and 0.86(MCC). In the three multiclass models, the accuracy ranged between 0.90 and 0.95, the AUROC between 0.97 and 0.99 and the MCC between 0.86 and 0.92. CONCLUSION: Deep learning can quantify pulmonary edema as measured by EVLWI with high accuracy.

Identification of risk factors for upper gastrointestinal bleeding in intensive care unit patients (GIBICU study).

BACKGROUND AND GOALS: Risk stratification for the need for therapeutic endoscopy and prediction of mortality in patients with upper gastrointestinal bleeding (UGIB) can be assessed by several scores. However, current scores are not validated for variceal bleeding and Intensive Care Unit (ICU) patients. The aim of this study was to evaluate potential parameters for the prediction of UGIB and patient outcomes. PATIENTS AND STUDY METHODS: In this monocenter retrospective observational study, data from all esophagogastroduodenoscopies (EGD) between November 2014 and February 2020 with suspected hemorrhage in our ICU were evaluated. RESULTS: Out of 345 included EGD, 42.3% of UGIB was diagnosed. 51.9% needed endoscopic intervention. Overall, 52.3% of included patients with UGIB died. Logistic regression showed that preceding variceal or non-variceal UGIB (p < .001), serum lactate (p = .001), heart rate (HR) (p = .005), and blood transfusions (p = .001) were significant predictors of UGIB. Previous UGIB (p < .001), male sex (p = .015), known varices (p < .001), serum albumin (p = .19) and use of catecholamines (p = .040) were significant predictors for the need of endoscopic intervention. Higher mortality was significantly associated with the usage of steroids (p < .001), malignant preconditions (p = .021), serum albumin (p = .020) and prolonged PTT (partial thromboplastin time) (p = .001). CONCLUSIONS: We were able to identify additional parameters that had previously not been included in existing scores to predict the risk of UGIB, the need for therapeutic endoscopy and mortality in ICU patients. Therefore, an extension of these scores is necessary. Further validation of identified parameters in multicenter trials is needed to improve risk scores for ICU patients.

Extracorporeal multiorgan support including CO2-removal with the ADVanced Organ Support (ADVOS) system for COVID-19: A case report.

A substantial part of COVID-19-patients suffers from multi-organ failure (MOF). We report on an 80-year old patient with pulmonary, renal, circulatory, and hepatic failure. We decided against the use of extracorporeal membrane oxygenation (ECMO) due to old age and a SOFA-score of 13. However, the patient was continuously treated with the extracorporeal multi-organ- "ADVanced Organ Support" (ADVOS) device (ADVITOS GmbH, Munich, Germany). During eight 24h-treatment-sessions blood flow (100-300 mL/min), dialysate flow (160-320 mL/min) and dialysate pH (7.6-9.0) were adapted to optimize arterial PaCO2 and pH. Effective CO2 removal and correction of acidosis could be demonstrated by mean arterial- versus post-dialyzer values of pCO2 (68.7 ± 13.8 vs. 26.9 ± 11.6 mmHg; p < 0.001). The CO2-elimination rate was 48 ± 23mL/min. The initial vasopressor requirement could be reduced in parallel to pH-normalization. Interruptions of ADVOS-treatment repeatedly resulted in reversible deteriorations of paCO2 and pH. After 95 h of continuous extracorporeal decarboxylating therapy the patient had markedly improved circulatory parameters compared to baseline. In the context of secondary pulmonary infection and progressive liver failure, the patient had a sudden cardiac arrest. In accordance with the presumed patient will, we decided against mechanical resuscitation. Irrespective of the outcome we conclude that extracorporeal CO2 removal and multiorgan-support were feasible in this COVID-19-patient. Combined and less invasive approaches such as ADVOS might be considered in old-age-COVID-19 patients with MOF.

Aspergillus fumigatus cholangitis in a patient with cholangiocarcinoma: case report and review of the literature.

Aspergillus spp. cholangitis is an uncommon presentation of invasive aspergillosis. Only few cases are described in the literature affecting severely immunocompromised patients or patients following biliary surgery. Especially, invasive aspergillosis in non-haematological patients is associated with high mortality rates, caused by atypical presentations, which is associated with a delay in diagnosis and therapy. We report a 72-year-old man with primary diagnosis of cholangiocarcinoma and stent implantation by endoscopic retrograde cholangiopancreatography (ERCP) for biliary decompression who developed severe cholangitis with invasive aspergillosis. The patient had no history of prior hospitalisation, no immunosuppressive therapy and no preceding biliary surgery. Furthermore, in this exceptional case of extrapulmonary aspergillosis, there were no signs of pulmonary involvement. From the literature review, only three cases of Aspergillus cholangitis could be identified. Clinical manifestations of invasive aspergillosis can be variable and classical risk factors such as immunosuppression are not mandatorily present. Clinical awareness of these rare cases is of vital importance for initiation of correct therapy.

Colon diverticulosis is not a risk factor for the detection of adenomatous polyps - results of a prospective study.

BACKGROUND AND AIMS: Colon diverticula (CD) and adenomatous polyps are frequently found during colonoscopy. Data from the literature contains inconsistent information about whether patients with CD have a higher risk for colon adenomas. A positive correlation might influence the current guidelines for screening colonoscopies. The aim of this study was to examine whether presence of CD is associated with endoscopic adenoma detection. MATERIALS AND METHODS: This was a prospective study at 2 centers in Germany. Patients with an indication for colonoscopy were included. The number and localization of diverticula were recorded. Detected polyps were resected, and histopathological results were captured. Logistic regression models were fitted to the data to evaluate the association between CD and adenoma detection. RESULTS: A total of 938 colonoscopies was included. CD occurred in 49.1 % of the colonoscopies. The polyp and adenoma detection rates (PDR, ADR) were 50.3 % and 32.3 %. In 37.5 % of the patients with diverticula, at least 1 adenoma was detected, whereas this was the case in 27.3 % in the absence of diverticula. The presence of diverticula was positively correlated with the detection of adenomas in univariate analysis (p = 0.001), but no significant association could be found in multivariable analysis (p = 0.775). Increasing age (p < 0.001), male sex (p = 0.005), and longer withdrawal time (p < 0.001) were significant predictors for adenoma detection in the multivariable analysis. Similar results were also observed for both the distal and the proximal colon. DISCUSSION: Diverticula and adenomas are frequently found during colonoscopies. However, diverticula disease was not significantly associated with adenoma detection after adjustment for relevant prognostic factors. Older age, male sex, and duration of withdrawal time are predictors for the detection of adenomatous polyps.

Endocuff vision-assisted vs. standard polyp resection in the colorectum (the EVASTA study): a prospective randomized study.

BACKGROUND: Cap-assisted colonoscopy is frequently used to facilitate adenoma detection during endoscopy. However, data on how cap assistance influences polyp resection are scarce. We aimed to evaluate the impact of cap assistance with the Endocuff vision device (EVD) on the resection time for colorectal polyps in patients undergoing colonoscopy. METHODS : A randomized, prospective study was performed in a university hospital in Germany. A total of 250 patients were randomly assigned 1:1 to undergo either colonoscopy with the EVD (EVD arm) or standard colonoscopy without the use of a cap (standard arm). The primary outcome was the average duration of polypectomy. Secondary outcomes included adenoma detection rate, cecal and ileal intubation times, and propofol dosage. RESULTS: The use of EVD led to a significant reduction in the median polypectomy time in the EVD vs. standard arm (54 vs. 80 seconds, respectively; P = 0.02). This effect was strongest for polyps ≥ 6 mm. Compared with the standard group, Endocuff assistance also resulted in a shorter cecal intubation time (6 vs. 8 minutes; P = 0.03) and overall colonoscopy time (23 vs. 27 minutes; P = 0.02). In contrast, no difference in withdrawal time was observed. The polyp and adenoma detection rates did not differ significantly between the two groups. CONCLUSION: Endocuff-assisted colonoscopy reduces the duration of polypectomy, which may be due to a more stable scope position during resection. Further studies are needed to investigate whether comparable effects will be seen for other interventions, such as clipping or biopsy sampling.

Near focus NBI endoscopy plus acetic acid for optical polyp characterization in the colorectum - A proof of principle study.

Background and study aim: Optical polyp characterization (OPC) in the colorectum is an upcoming challenge for endoscopists. Narrow band imanging (NBI) has been proposed to be helpful for OPC. However, data from clinical studies have shown that quality of OPC differs markedly between endoscopists. The aim of this study was to test the value of a combined NBI plus acetic acid (NBI + AA) approach for OPC in the colorectum. Patients and methods: This was a prospective, single-arm study at a tertiary referral center in Germany. The study was designed as a proof of principle study. Initially polyps were characterized using High-definition white light (HDWL) only. Additionally, the same polyps were investigated using NBI + AA (1.5% solution) in order to predict polyp pathology in a real time setting. The near focus function was used for both HDWL and NBI + AA assessment. The primary endpoint was accuracy of colorectal polyp prediction when using NBI + AA. Results: A total of 63 polyps were detected in 55 patients. NBI + AA based accuracy of real-time predictions was 85.5% compared to 80.6% using HDWL (p = .450). Accuracy was 90.2% in the high confidence setting for both NBI + AA and HDWL predictions. A higher share of polyps were assessed with high confidence when using NBI + AA compared to HDWL (p = .006). The use of NBI + AA led to a better identification of polyp margins (p < .001) compared to HDWL. Conclusions: The use of acetic acid led to a high level of accuracy and confidence in the prediction of polyp histology. These data justify further investigation in a randomized controlled study.

Automated polyp detection in the colorectum: a prospective study (with videos).

BACKGROUND AND AIMS: Adenoma detection is a highly personalized task that differs markedly among endoscopists. Technical advances are therefore desirable for the improvement of the adenoma detection rate (ADR). An automated computer-driven technology would offer the chance to objectively assess the presence of colorectal polyps during colonoscopy. We present here the application of a real-time automated polyp detection software (APDS) under routine colonoscopy conditions. METHODS: This was a prospective study at a university hospital in Germany. A prototype of a novel APDS ("KoloPol," Fraunhofer IIS, Erlangen, Germany) was used for automated image-based polyp detection. The software functions by highlighting structures of possible polyp lesions in a color-coded manner during real-time colonoscopy procedures. Testing the feasibility of APDS deployment under real-time conditions was the primary goal of the study. APDS polyp detection rates (PDRs) were defined as secondary endpoints provided that endoscopists' detection served as criterion standard. RESULTS: The APDS was applied in 55 routine colonoscopies without the occurrence of any clinically relevant adverse events. Endoscopists' PDRs and ADRs were 56.4% and 30.9%, respectively. The PDRs and ADRs of the APDS were 50.9% and 29.1%, respectively. The APDS detected 55 of 73 polyps (75.3%). Smaller polyp size and flat polyp morphology were correlated with insufficient polyp detection by the APDS. CONCLUSION: Computer-assisted automated low-delay polyp detection is feasible during real-time colonoscopy. Efforts should be undertaken to improve the APDS with respect to smaller and flat shaped polyps. (Clinical trial registration number: NCT02838888.).

Optical classification of neoplastic colorectal polyps - a computer-assisted approach (the COACH study).

BACKGROUND AND AIMS: Clinical data suggest that the quality of optical diagnoses of colorectal polyps differs markedly among endoscopists. The aim of this study was to develop a computer program that was able to differentiate neoplastic from non-neoplastic polyps using unmagnified endoscopic pictures. METHODS: During colonoscopy procedures polyp photographies were performed using the unmagnified high-definition white light and narrow band image mode. All detected polyps (n = 275) were resected and sent to pathology. Histopathological diagnoses served as the ground truth. Machine learning was used in order to generate a computer-assisted optical biopsy (CAOB) approach. In the test phase pictures were presented to CAOB in order to obtain optical diagnoses. Altogether 788 pictures were available (602 for training the machine learning algorithm and 186 for CAOB testing). All test pictures were also presented to two experts in optical polyp characterization. The primary endpoint of the study was the accuracy of CAOB diagnoses in the test phase. RESULTS: A total of 100 polyps (of these 52% neoplastic) were used in the CAOB test phase. The mean size of test polyps was 4 mm. Accuracy of the CAOB approach was 78.0%. Sensitivity and negative predictive value were 92.3% and 88.2%, respectively. Accuracy obtained by two expert endoscopists was 84.0% and 77.0%. Regarding accuracy of optical diagnoses CAOB predictions did not differ significantly compared to experts (p = .307 and p = 1.000, respectively). CONCLUSIONS: CAOB showed good accuracy on the basis of unmagnified endoscopic pictures. Performance of CAOB predictions did not differ significantly from experts' decisions. The concept of computer assistance for colorectal polyp characterization needs to evolve towards a real-time application prior of being used in a broader set-up.

Cap Assisted Upper Endoscopy for Examination of the Major Duodenal Papilla: A Randomized, Blinded, Controlled Crossover Study (CAPPA Study).

OBJECTIVE: Examination of major duodenal papilla (MDP) by standard forward-viewing esophagogastroduodenoscopy (S-EGD) is limited. Cap assisted esophagogastroduodenoscopy (CA-EGD) utilizes a cap fitted to the tip of the endoscope that can depress the mucosal folds and thus might improve visualization of MDP. The aim of this study was to compare CA-EGD to S-EGD for complete examination of the MDP. METHODS: Prospective, randomized, blinded, controlled crossover study. Subjects scheduled for elective EGD were randomized to undergo S-EGD (group A) or CA-EGD (group B) before undergoing a second examination by the alternate method. Images of the MDP were evaluated by three blinded multicenter-experts. Our primary outcome measure was complete examination of the papilla. Secondary outcome measures were duration and overall diagnostic yield. RESULTS: A total of 101 patients were randomized and completed the study. Complete examination of MDP was achieved in 98 patients using CA-EGD compared to 24 patients using S-EGD (97 vs. 24%, P<0.001). Median duration from intubation of the esophagus until localization of the MDP was shorter with CA-EGD (46. vs. 96 s., P<0.001). In group A, 11 extra lesions and 12 additional incidental findings were detected by secondary CA-EGD, whereas neither were detected by secondary S-EGD in group B (22 vs. 0% and 24 vs. 0%, P<0.001 and P<0.001). CONCLUSION: CA-EGD enabled complete examination of MDP in almost all cases compared to a low success rate of S-EGD. CA-EGD detected a significant amount of lesions and incidental findings when added to S-EGD. CA-EGD is a safe and effective method for examination of MDP.