Search

Bei der Therapie des Endometriumkarzinoms steht heute der operative Weg im Vordergrund und ermoglicht den stadiengerechten Einsatz zusatzlicher adjuvanter Masnahmen. Die Aufgabe der Diagnostik besteht darin, das operative Vorgehen moglichst individuell festzulegen. Der transvaginale Ultraschall ist eine kostengunstige, zumeist schmerz- und komplikationslose Methode in der Diagnostik des Endometriumkarzinoms. Sonographisch beurteilt werden Endometriumdicke, Echostruktur und Endometriumbegrenzung sowie – falls vorhanden – eine intrakavitare Flussigkeitsansammlung. Neben der konventionellen transvaginalen Sonographie (TVS) besteht die Moglichkeit einer Dignitatsbeurteilung mittels Farb- und Pulsdopplersonographie, z. B unter Verwendung von Pulsatility-Index und Resistance-Index. Bei der praoperativen Stadieneinteilung stellt die MRT bezuglich der Myometriuminfiltration und des zervikalen Befalls im Vergleich zu TVS und CT die praziseste Methode dar.

Learning representations that disentangle hidden explanatory factors in data has proven beneficial for effective pattern classification. Slow feature analysis (SFA) is a nonlinear dimensionality reduction technique that provides a useful representation for classification if the training data is sequential and transitions between classes are rare. The pattern discrimination ability of SFA has been attributed to the equivalence of linear SFA and linear discriminant analysis (LDA) under certain conditions. LDA, however, is often outperformed by locality preserving projections (LPP) when the data lies on or near a low-dimensional manifold. Here, we take a unified manifold learning perspective on LPP, LDA and SFA. We suggest that the discrimination ability of SFA is better explained by its relation to LPP than to LDA, and give an example of a situation where linear SFA outperforms LDA. We then propose a novel supervised manifold learning architecture that combines hierarchical nonlinear expansions, as commonly used for SFA, with supervised LPP. It learns a nonlinear parametric data representation that explicitly takes both the class labels and the manifold structure of the data into account. As an experimental validation, we show that this approach outperforms previously proposed models on the NORB object recognition dataset.

Genital findings in sexually abused prepubertal girls. Childhood sexual abuse is defined as the involvement of dependent, developmentally immature children and adolescents in sexual activities that they do not fully comprehend, to which they are unable to give informed consent or that violate the social taboos of family roles. Essential to the diagnosis of sexual abuse is an awareness of the problem and acknowledgement of its manifestations. The evaluation of the sexually abused girl usually is performed in a pediatric and adolescent gynecology outpatient department. Thus, the gynecologist will be part of a multidisciplinary approach to the problem and will need to be competent in the basic skills of history taking, physical examination, selection of laboratory tests und differential diagnosis. Findings secondary to sexual abuse are often subtle. Acute tears or bruisings are rare, because force is seldom part of sexual acts committed against a child. A vaginal opening of greater than 5 mm is not common and may indicate vaginal penetration. An intact hymen not necessarily exclude vaginal intercourse. Lack of physical evidence never rules out abuse because sexual acts may leave no physical findings. Language: de

Hierarchical temporal memory (HTM) is a biologically inspired framework that can be used to learn invariant representations of patterns in a wide range of applications. Classical HTM learning is mainly unsupervised, and once training is completed, the network structure is frozen, thus making further training (i.e., incremental learning) quite critical. In this letter, we develop a novel technique for HTM (incremental) supervised learning based on gradient descent error minimization. We prove that error backpropagation can be naturally and elegantly implemented through native HTM message passing based on belief propagation. Our experimental results demonstrate that a two-stage training approach composed of unsupervised pretraining and supervised refinement is very effective (both accurate and efficient). This is in line with recent findings on other deep architectures.

To overcome the limitations of the state-of-the-art influenza surveillance systems in Europe, we established in 2008 a European-wide consortium aimed at introducing an innovative information and communication technology approach for a web-based surveillance system across different European countries, called Influenzanet. The system, based on earlier efforts in The Netherlands and Portugal, works with the participation of the population in each country to collect real-time information on the distribution of influenza-like illness cases through web surveys administered to volunteers reporting their symptoms (or lack of symptoms) every week during the influenza season. Such a large European-wide web-based monitoring infrastructure is intended to rapidly identify public health emergencies, contribute to understanding global trends, inform data-driven forecast models to assess the impact on the population, optimize the allocation of resources, and help in devising mitigation and containment measures. In this article, we describe the scientific and technological issues faced during the development and deployment of a flexible and readily deployable web tool capable of coping with the requirements of different countries for data collection, during either a public health emergency or an ordinary influenza season. Even though the system is based on previous successful experience, the implementation in each new country represented a separate scientific challenge. Only after more than 5 years of development are the existing platforms based on a plug-and-play tool that can be promptly deployed in any country wishing to be part of the Influenzanet network, now composed of The Netherlands, Belgium, Portugal, Italy, the UK, France, Sweden, Spain, Ireland, and Denmark.

Trisomy 22 is a common trisomy in spontaneous abortions. In contrast, live-born trisomy 22 is rarely seen due to severe organ malformations associated with this condition. Here, we report on a male infant with complete, non-mosaic trisomy 22 born at 35 + 5 weeks via caesarean section. Peripheral blood lymphocytes and fibroblasts showed an additional chromosome 22 in all metaphases analyzed (47,XY,+22). In addition, array CGH confirmed complete trisomy 22. The patient’s clinical features included dolichocephalus, hypertelorism, flattened nasal bridge, dysplastic ears with preauricular sinuses and tags, medial cleft palate, anal atresia, and coronary hypospadias with scrotum bipartitum. Essential treatment was implemented in close coordination with the parents. The child died 29 days after birth due to respiratory insufficiency and deterioration of renal function. Our patient’s history complements other reports illustrating that children with complete trisomy 22 may survive until birth and beyond.

From 1982 to 1993 67 diabetic women attended the University Women's Hospital obstetric clinic at Wurzburg. These women were separated into two groups: group 1 (inpatient group) delivered between 1/82 and 7/87, group 2 (outpatient group) delivered between 8/87 and 4/93. Between 1982 and 1987 we used extensive inpatient care. But did not so after 1987. The fetus was monitored by ultrasound scans, cardiotocography, oxytocin stress test and doppler measurements. Nearly all patients used home blood glucose monitoring. We compared blood glucose levels, mode of delivery and fetal outcome for the two groups. There was no significant difference in the average blood glucose levels between the two groups. Cesarean section rates (37%) and gestational age at delivery (37 weeks) were similar in both groups. Fetal macrosomia (birth weight > 4000 g) occured more frequently in the outpatient group (group l: 4%, group 2: 25%,P

Hierarchical Temporal Memory is a biologically-inspired framework that can be used to learn invariant representations of patterns. Classical HTM learning is mainly unsupervised and once training is completed the network structure is frozen, thus making further training quite critical. In this paper we develop a novel technique for HTM (incremental) supervised learning based on error minimization. We prove that error backpropagation can be naturally and elegantly implemented through native HTM message passing based on Belief Propagation. Our experimental results show that a two stage training composed by unsupervised pre-training + supervised refinement is very effective. This is in line with recent findings on other deep architectures.

Excitonics is a rapidly expanding field of nanophotonics in which the harvesting of photons, ensuing creation and transport of excitons via Forster resonant energy transfer (FRET), and subsequent charge separation or photon emission has led to the demonstration of excitonic wires, switches, Boolean logic and light harvesting antennas for many applications. FRET funnels excitons down an energy gradient resulting in energy loss with each step along the pathway. Conversely, excitonic energy up conversion via up conversion nanoparticles (UCNPs), although currently inefficient, serves as an energy ratchet to boost the exciton energy. Although FRET-based up conversion has been demonstrated, it suffers from low FRET efficiency and lacks the ability to modulate the FRET. We have engineered an up conversion FRET-based switch by combining lanthanide-doped UCNPs and fluorophores that demonstrates excitonic energy up conversion by nearly a factor of 2, an excited state donor to acceptor FRET efficiency of nearly 25%, and an acceptor fluorophore quantum efficiency that is close to unity. These findings offer a promising path for energy up conversion in nanophotonic applications including artificial light harvesting, excitonic circuits, photovoltaics, nanomedicine, and optoelectronics.