Protection against microbial infection by the induction of inflammation is a key function of the IL-1 superfamily, including both classical IL-1 and the new IL-36 cytokine families. is a frequent human fungal pathogen causing mucosal infections. Although the initiators and effectors important in protective host responses to are well described, the key players in driving these responses remain poorly defined. Recent work has identified a central role played by IL-1 in inducing innate Type-17 immune responses to clear infections. Despite this, lack of IL-1 signaling does not result in complete loss of immunity, indicating that there are other factors involved in mediating protection to this fungus. In this study, we identify IL-36 cytokines as a new player in these responses. We show that infection of the oral mucosa induces the production of IL-36. As with IL-1α/β, induction of epithelial IL-36 depends on the hypha-associated peptide toxin Candidalysin. Epithelial IL-36 gene expression requires p38-MAPK/c-Fos, NF-κB, and PI3K signaling and is regulated by the MAPK phosphatase MKP1. Oral candidiasis in IL-36R mice shows increased fungal burdens and reduced IL-23 gene expression, indicating a key role played by IL-36 and IL-23 in innate protective responses to this fungus. Strikingly, we observed no impact on gene expression of IL-17 or IL-17-dependent genes, indicating that this protection occurs via an alternative pathway to IL-1-driven immunity. Thus, IL-1 and IL-36 represent parallel epithelial cell-driven protective pathways in immunity to oral infection.

Imaging tests are central to the diagnosis and staging of pancreatic adenocarcinoma. We performed a systematic review and meta-analysis of the pertinent evidence on 5 imaging tests (computed tomography (CT), magnetic resonance imaging, CT angiography, endoscopic ultrasound with fine-needle aspiration, and combined positron emission tomography with CT). Searches of several databases up to March 1, 2014, yielded 9776 articles, and 24 provided comparative effectiveness of 2 or more imaging tests. Multiple reviewers applied study inclusion criteria, extracted data from each study, rated the risk of bias, and graded the strength of evidence. Data included accuracy of diagnosis and resectability in primary untreated pancreatic adenocarcinoma, including tumor stage, nodal stage, metastases, and vascular involvement. Where possible, study results were combined using bivariate meta-analysis. Studies were at low or moderate risk of bias. Most comparisons between imaging tests were insufficient to permit conclusions, due to imprecision or inconsistency among study results. However, moderate-grade evidence revealed that CT and magnetic resonance imaging had similar sensitivities and specificities for both diagnosis and vascular involvement. Other conclusions were based on low-grade evidence. In general, more direct evidence is needed to inform decisions about imaging tests for pancreatic adenocarcinoma.

A significant volume of medical data remains unstructured. Natural language processing (NLP) and machine learning (ML) techniques have shown to successfully extract insights from radiology reports. However, the codependent effects of NLP and ML in this context have not been well-studied. Between April 1, 2015 and November 1, 2016, 9418 cross-sectional abdomen/pelvis CT and MR examinations containing our internal structured reporting element for cancer were separated into four categories: Progression, Stable Disease, Improvement, or No Cancer. We combined each of three NLP techniques with five ML algorithms to predict the assigned label using the unstructured report text and compared the performance of each combination. The three NLP algorithms included term frequency-inverse document frequency (TF-IDF), term frequency weighting (TF), and 16-bit feature hashing. The ML algorithms included logistic regression (LR), random decision forest (RDF), one-vs-all support vector machine (SVM), one-vs-all Bayes point machine (BPM), and fully connected neural network (NN). The best-performing NLP model consisted of tokenized unigrams and bigrams with TF-IDF. Increasing N-gram length yielded little to no added benefit for most ML algorithms. With all parameters optimized, SVM had the best performance on the test dataset, with 90.6 average accuracy and F score of 0.813. The interplay between ML and NLP algorithms and their effect on interpretation accuracy is complex. The best accuracy is achieved when both algorithms are optimized concurrently.

No abstract

No abstract

No abstract

This cohort study assesses the frequency of ovarian cancer and evaluates the diagnostic performance of the American College of Radiology Ovarian-Adnexal Reporting and Data System ultrasound risk score among 913 women with suspected or known adnexal lesions.

Radiologists have higher levels of professional satisfaction than do other physicians; however, as with physicians overall, their satisfaction has decreased over time.