Faculty Bibliography

Probe-based confocal laser endomicroscopy (pCLE) allows for real-time diagnosis of dysplasia and cancer in Barrett's esophagus (BE) but is limited by low sensitivity. Even the gold standard of histopathology is hindered by poor agreement between pathologists. We deployed deep-learning-based image and video analysis in order to improve diagnostic accuracy of pCLE videos and biopsy images. Blinded experts categorized biopsies and pCLE videos as squamous, non-dysplastic BE, or dysplasia/cancer, and deep learning models were trained to classify the data into these three categories. Biopsy classification was conducted using two distinct approaches-a patch-level model and a whole-slide-image-level model. Gradient-weighted class activation maps (Grad-CAMs) were extracted from pCLE and biopsy models in order to determine tissue structures deemed relevant by the models. 1970 pCLE videos, 897,931 biopsy patches, and 387 whole-slide images were used to train, test, and validate the models. In pCLE analysis, models achieved a high sensitivity for dysplasia (71%) and an overall accuracy of 90% for all classes. For biopsies at the patch level, the model achieved a sensitivity of 72% for dysplasia and an overall accuracy of 90%. The whole-slide-image-level model achieved a sensitivity of 90% for dysplasia and 94% overall accuracy. Grad-CAMs for all models showed activation in medically relevant tissue regions. Our deep learning models achieved high diagnostic accuracy for both pCLE-based and histopathologic diagnosis of esophageal dysplasia and its precursors, similar to human accuracy in prior studies. These machine learning approaches may improve accuracy and efficiency of current screening protocols.

One of the greatest obstacles in the adoption of deep neural networks for new medical applications is that training these models typically require a large amount of manually labeled training samples. In this body of work, we investigate the semi-supervised scenario where one has access to large amounts of unlabeled data and only a few labeled samples. We study the performance of MixMatch and FixMatch-two popular semi-supervised learning methods-on a histology dataset. More specifically, we study these models' impact under a highly noisy and imbalanced setting. The findings here motivate the development of semi-supervised methods to ameliorate problems commonly encountered in medical data applications.

INTRODUCTION:Excessive alcohol consumption has steadily risen to become the third leading cause of preventable death in the USA. One consequence of heavy alcohol use recently under considerable investigation is alcoholic hepatitis. Although many risk factors for developing alcoholic hepatitis have been documented, our aim in this study was to examine the potential association between sarcopenia and severity, mortality, 30 days readmission rate, complication, infections and length of hospital stay in alcoholic hepatitis patients. METHODS:A retrospective analysis was performed at a large, academic hospital in 194 alcoholic hepatitis patients aged 18-60 who had cross-sectional computed tomography imaging and met our clinical definition of alcoholic hepatitis. The fifth percentile of the psoas muscle index was used as a cutoff for sarcopenia. RESULTS:One hundred ninety-four patients met the criteria for alcoholic hepatitis and had cross-sectional imaging. Higher Model for End-Stage Liver disease score was found in the sarcopenia group when compared to the non-sarcopenia group (mean Model for End-Stage Liver disease 21.5 and 24.2, respectively, P = 0.03). Sarcopenia also correlated with significantly longer hospital stay; the average length of stay in the sarcopenia group was 17.2 days while the non-sarcopenia patients had an average of 12.4 days. We found higher risk of developing pneumonia, sepsis and hepatic encephalopathy in sarcopenic patients. CONCLUSION:Alcoholic hepatitis patients with sarcopenia have significantly worse outcomes when compared with the patients without sarcopenia, including a severe form of alcoholic hepatitis, longer hospital stays, higher risk of developing pneumonia, sepsis and hepatic encephalopathy.

BACKGROUND:Liver cirrhosis is the late stage of hepatic fibrosis and is characterized by portal hypertension that can clinically lead to decompensation in the form of ascites, esophageal/gastric varices or encephalopathy. The most common sequelae associated with liver cirrhosis are neurologic and neuropsychiatric impairments labeled as hepatic encephalopathy (HE). Well established triggers for HE include infection, gastrointestinal bleeding, constipation, and medications. Alterations to the gut microbiome is one of the leading ammonia producers in the body, and therefore may make patients more susceptible to HE. AIM/OBJECTIVE:To investigate the relationship between the use of proton pump inhibitors (PPIs) and HE in patients with cirrhosis. METHODS:non-users at admission to the hospital and throughout their hospital course. Secondary outcomes included rate of infection, gastrointestinal bleeding within the last 12 mo, mean ammonia level, and model for end-stage liver disease scores at admission. RESULTS:0.220). CONCLUSION/CONCLUSIONS:Chronic PPI use in cirrhotic patients is associated with significantly higher average West Haven Criteria for HE compared to patients that do not use PPIs.

INTRODUCTION/BACKGROUND:Mature peripancreatic fluid collection (MPFC) is a known and often challenging consequence of acute pancreatitis and often requires intervention. The most common method accepted is the "step-up approach," which consists of percutaneous drainage followed, if necessary, by minimally invasive retroperitoneal necrosectomy. Our paper aims to distinguish between plastic stents and lumen-apposing stents in the endoscopic management of MPFC in terms of morbidity, mortality, and haste of fluid collection resolution. METHODS:A retrospective analysis was performed at UMass Memorial Medical Center in patients with a diagnosis of MPFC. Utilizing medical records, clinical data, radiology, as well as endoscopic evidence, patients were differentiated by stent type used (plastic versus lumen-apposing) for the management of the MPFC. The primary outcome of the study was to assess the time to MPFC resolution following the placement of either plastic or lumen-apposing stents (on endoscopic ultrasound or computerized tomography scan) using a multivariate analysis with a logistic regression model. RESULTS:=0.01). Stent migration was seen in 5 patients (11%) who received a plastic pigtail stent compared to 0 (0%) patients who received a lumen-apposing stent. DISCUSSION/CONCLUSIONS:Our study demonstrates that lumen-apposing stents result in a significant reduction in the interval between stent placement and MPFC resolution as well as the time from stent placement to removal, when compared to plastic pigtail stents, the prior standard-of-care. Our study reached similar conclusions regarding the number of stents placed. However, we did not find a significant difference between the complication rates, specifically peri- and postprocedural bleeding or perforation, between the 2 study groups, as demonstrated in prior papers.

INTRODUCTION/BACKGROUND:The perioperative mortality is significantly higher in patients with cirrhosis undergoing certain surgical procedures. In this study, we examined the inpatient perioperative mortality and morbidities in cirrhotic people who underwent laparoscopic appendectomy. METHODS:value less than .05 were determined to define the statistical significance. RESULT/RESULTS: = .65). CONCLUSIONS:Appendectomy-related morbidity and mortality in cirrhotic patients are not different from noncirrhotic patients.