Vaginal Tampon Colonization by Staphylococcus aureus in Healthy Women.

Tampons recovered from a cohort of 737 healthy women (median age, 32 years) were analyzed for the presence of Staphylococcus aureus A total of 198 tampons (27%) were colonized by S. aureus, 28 (4%) by a strain producing toxic shock syndrome toxin 1 (TSST-1). S. aureus was detected more frequently in tampons that did not require an applicator for their insertion (74/233 [32%] versus 90/381 [24%]; odds ratio [OR] = 1.51 [95% confidence interval, 1.04 to 2.17]) and in women who used an intrauterine device for contraception (53/155 [34%] versus 145/572 [27%]; OR = 1.53 [95% confidence interval, 1.05 to 2.24]). The S. aureus strains isolated from tampons belonged to 22 different clonal complexes (CCs). The most prevalent CC was CC398 agr1 (n = 57 [27%]), a clone that does not produce superantigenic toxins, followed by CC30 agr3 (n = 27, 13%), producing TSST-1 (24/27 [89%]), the principal clone of S. aureus involved in menstrual toxic shock syndrome (MTSS).IMPORTANCE Menstrual toxic shock syndrome (MTSS) is an uncommon severe acute disease that occurs in healthy menstruating women colonized by TSST-1-producing S. aureus who use intravaginal protection, such as tampons and menstrual cups. The catamenial product collected by the protection serves as a growth medium for S. aureus and allows TSST-1 production. Previous studies evaluated the prevalence of genital colonization by S. aureus by vaginal swabbing, but they did not examine tampon colonization. This study demonstrated a high prevalence of tampon colonization by S. aureus and the presence of the CC30 TSST-1 S. aureus clone responsible for MTSS in tampons from healthy women. The results support the vaginal carriage of this lineage in healthy women. In addition, the higher prevalence of S. aureus within tampons that do not require an applicator indicates a crucial role for handwashing before tampon handling to decrease the risk of tampon contamination.

Beryllium solubility in occupational airborne particles: Sequential extraction procedure and workplace application.

Modification of an existing sequential extraction procedure for inorganic beryllium species in the particulate matter of emissions and in working areas is described. The speciation protocol was adapted to carry out beryllium extraction in closed-face cassette sampler to take wall deposits into account. This four-step sequential extraction procedure aims to separate beryllium salts, metal, and oxides from airborne particles for individual quantification. Characterization of the beryllium species according to their solubility in air samples may provide information relative to toxicity, which is potentially related to the different beryllium chemical forms. Beryllium salts (BeF(2), BeSO(4)), metallic beryllium (Bemet), and beryllium oxide (BeO) were first individually tested, and then tested in mixtures. Cassettes were spiked with these species and recovery rates were calculated. Quantitative analyses with matched matrix were performed using inductively coupled plasma mass spectrometry (ICP-MS). Method Detection Limits (MDLs) were calculated for the four matrices used in the different extraction steps. In all cases, the MDL was below 4.2 ng/sample. This method is appropriate for assessing occupational exposure to beryllium as the lowest recommended threshold limit values are 0.01 µg.m(-3) in France([) (1) (]) and 0.05 µg.m(-3) in the USA.([ 2 ]) The protocol was then tested on samples from French factories where occupational beryllium exposure was suspected. Beryllium solubility was variable between factories and among the same workplace between different tasks.

Exploiting Negative Evidence for Deep Latent Structured Models.

The abundance of image-level labels and the lack of large scale detailed annotations (e.g. bounding boxes, segmentation masks) promotes the development of weakly supervised learning (WSL) models. In this work, we propose a novel framework for WSL of deep convolutional neural networks dedicated to learn localized features from global image-level annotations. The core of the approach is a new latent structured output model equipped with a pooling function which explicitly models negative evidence, e.g. a cow detector should strongly penalize the prediction of the bedroom class. We show that our model can be trained end-to-end for different visual recognition tasks: multi-class and multi-label classification, and also structured average precision (AP) ranking. Extensive experiments highlight the relevance of the proposed method: our model outperforms state-of-the art results on six datasets. We also show that our framework can be used to improve the performance of state-of-the-art deep models for large scale image classification on ImageNet. Finally, we evaluate our model for weakly supervised tasks: in particular, a direct adaptation for weakly supervised segmentation provides a very competitive model.

SyMIL: MinMax Latent SVM for Weakly Labeled Data.

Designing powerful models able to handle weakly labeled data are a crucial problem in machine learning. In this paper, we propose a new multiple instance learning (MIL) framework. Examples are represented as bags of instances, but we depart from standard MIL assumptions by introducing a symmetric strategy (SyMIL) that seeks discriminative instances in positive and negative bags. The idea is to use the instance the most distant from the hyper-plan to classify the bag. We provide a theoretical analysis featuring the generalization properties of our model. We derive a large margin formulation of our problem, which is cast as a difference of convex functions, and optimized using concave-convex procedure. We provide a primal version optimizing with stochastic subgradient descent and a dual version optimizing with one-slack cutting-plane. Successful experimental results are reported on standard MIL and weakly supervised object detection data sets: SyMIL significantly outperforms competitive methods (mi/MI/Latent-SVM), and gives very competitive performance compared to state-of-the-art works. We also analyze the selected instances of symmetric and asymmetric approaches on weakly supervised object detection and text classification tasks. Finally, we show complementarity of SyMIL with recent works on learning with label proportions on standard MIL data sets.