Despite various systems and safeguards available, unintentionally retained surgically placed foreign bodies remain difficult to eliminate completely. Developing a standardized approach to the request, "intraoperative film, rule out foreign body," is essential to reduce the adverse outcomes associated with this problem.
The purpose of this pilot study is to determine whether a deep convolutional neural network can be trained with limited image data to detect high-grade small bowel obstruction patterns on supine abdominal radiographs. Grayscale images from 3663 clinical supine abdominal radiographs were categorized into obstructive and non-obstructive categories independently by three abdominal radiologists, and the majority classification was used as ground truth; 74 images were found to be consistent with small bowel obstruction. Images were rescaled and randomized, with 2210 images constituting the training set (39 with small bowel obstruction) and 1453 images constituting the test set (35 with small bowel obstruction). Weight parameters for the final classification layer of the Inception v3 convolutional neural network, previously trained on the 2014 Large Scale Visual Recognition Challenge dataset, were retrained on the training set. After training, the neural network achieved an AUC of 0.84 on the test set (95% CI 0.78-0.89). At the maximum Youden index (sensitivity + specificity-1), the sensitivity of the system for small bowel obstruction is 83.8%, with a specificity of 68.1%. The results demonstrate that transfer learning with convolutional neural networks, even with limited training data, may be used to train a detector for high-grade small bowel obstruction gas patterns on supine radiographs.
Background.
End-stage liver disease (ESLD) patients requiring intensive care unit (ICU) care before liver transplantation (LT) often experience significant muscle mass loss, which has been associated with mortality. In this exploratory study, we primarily aimed to assess the feasibility of serial ultrasound (US) rectus femoris muscle area (RFMA) measurements for the evaluation of progressive muscle loss in ICU-bound potential LT candidates and describe the rate of muscle loss as assessed by sequential US RFMA measurements. Secondarily, we sought to identify patient characteristics associated with muscle loss and determine how muscle loss is associated with survival.
Methods.
We prospectively enrolled 50 ESLD adults (≥18 y old) undergoing evaluation for LT candidacy in the ICU. A baseline computed tomography measurement of psoas muscle area (PMA) and serial bedside US measurements of RFMA were obtained. The associations between patient characteristics, PMA, RFMA, ICU stay, and survival were analyzed.
Results.
Rapid decline in muscle mass by RFMA measurements was ubiquitously present and correlated to baseline PMA and length of ICU stay. RFMA normalized by body surface area decreased by 0.013 cm2/m2 (95% confidence interval, 0.010-0.016; P < 0.001) for each day in the ICU. Decreased RFMA normalized by body surface area was associated with poor overall survival (adjusted hazard ratio, 0.42; 95% confidence interval, 0.18-0.99; P = 0.047).
Conclusions.
In this exploratory, prospective study, serial US RFMA measurements in ESLD patients in the ICU are feasible, demonstrate progressive time-dependent muscle loss, and are associated with mortality. Further large-scale assessment of this modality compared with static PMA or performance-based dynamic assessments should be performed.
Hepatic infarction is infrequent due to the dual blood supply of the liver and the compensatory relationship between the hepatic artery and portal vein. Most cases occur in liver transplants due to vascular complications. Grayscale sonography combined with color and spectral wave Doppler can assess for vessel patency and parenchymal abnormalities. Liver infarctions appear as hypoechoic nonvascular regions on conventional and Doppler sonography. Here, we describe a grayscale ultrasound feature within liver infarctions in 2 liver transplants and in 1 native liver due to iatrogenic complication. This feature is similar to those described recently in the literature within splenic infarcts.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.