University of Iceland

About: University of Iceland is a education organization based out in Reykjavik, Suðurnes, Iceland. It is known for research contribution in the topics: Population & Genome-wide association study. The organization has 5423 authors who have published 16199 publications receiving 694762 citations. The organization is also known as: Háskóli Íslands.

The clinical value of somatic TP53 gene mutations in 1,794 patients with breast cancer.

Abstract: To investigate the clinical value of somatic TP53 mutations in breast cancer, we assembled clinical and molecular data on 1,794 women with primary breast cancer with long-term follow-up and whose tumor has been screened for mutation in exons 5 to 8 of TP53 by gene sequencing. TP53 mutations were more frequent in tumors of ductal and medullar types, aggressive phenotype (high grade, large size, node positive cases, and low hormone receptor content) and in women 10 years; P < 0.0001) compared with patients with no such mutation. The prognostic value of TP53 mutation was independent of tumor size, node status, and hormone receptor content, confirming and reconciling previous findings in smaller series. Moreover, an interaction between TP53 mutation and progesterone receptor (PR) status was revealed, TP53 mutation combined with the absence of progesterone receptor being associated with the worst prognosis. Whereas previous studies have emphasized the fact that missense mutations in the DNA-binding motifs have a worse prognosis than missense mutations outside these motifs, we show that non-missense mutations have prognostic value similar to missense mutations in DNA-binding motifs. Nonetheless, specific missense mutants (codon 179 and R248W) seem to be associated with an even worse prognosis. These results, obtained on the largest series analyzed thus far, show that TP53 mutations identified by gene sequencing have an independent prognostic value in breast cancer and could have potential uses in clinical practice.

The Nordic countries as a cohort for pharmacoepidemiological research.

Abstract: The Nordic countries have a long tradition of registry-based epidemiological research. Many population-based health registries were established in the 1960s, with use of unique personal identifiers facilitating linkage between registries. In recent years, each country has established a national database to track prescription drugs dispensed to individuals in ambulatory care. The objectives were to present an overview of the prescription databases established in the Nordic countries, as well as to elaborate on their unique potential for record linkage and cross-national comparison of drug utilization. Five Nordic countries collect drug exposure data based on drugs dispensed at pharmacies and have the potential to link these data to health outcomes. The databases together cover 25 million inhabitants (Denmark: 5.5 million; Finland: 5.3 million; Iceland: 0.3 million; Norway: 4.8 million; and Sweden: 9.2 million). In 2007, the registries encompassed 17 million prescription drug users (68% of the total population). We provide examples of how these databases have been used for descriptive drug utilization studies and analytical pharmacoepidemiological studies linking drug exposure to other health registries. Comparisons are facilitated by many similarities among the databases, including data source, content, coverage and methods used for drug utilization studies and record linkage. There are, however, some differences in coding systems and validity, as well as in some access and technical issues. To perform cross-national pharmacoepidemiological studies, resources, networks and time are needed, as well as methods for pooling data. Interpretation of results needs to account for inter-country heterogeneity and the possibility of spurious relationships. The Nordic countries have a unique potential for collaborative high-quality cross-national pharmacoepidemiological studies with large populations. This research may assist in resolving safety issues of international interest, thus minimizing the risk of either over-reacting on possible signals or underestimating drug safety issues.

Seismic structure of the Iceland mantle plume

Abstract: Oceanic hotspots are generally accepted to be the manifestations of plumes of hot, upwelling mantle material1,2, but the nature of such flows remains enigmatic. Iceland, for example, is one of the most thoroughly investigated hotspots, yet previous seismological3–5 and geodynamic6–12 studies have been unable to constrain the width or temperature of the plume. Here we report the results of a regional broadband seismic experiment undertaken to determine the three-dimensional velocity structure of the upper mantle beneath Iceland using relative travel times of body waves from teleseismic earthquakes. Inversion solutions of the data show a cylindrical zone of low P- and S-wave velocities that extends from 100km to at least 400km depth beneath central Iceland. The radius of the low-velocity anomaly is about 150km, and its magnitude is approximately 2% for P waves and 4% for S waves, indicating that Iceland is underlain by a hot, narrow plume of upwelling mantle.

Generative Adversarial Networks for Hyperspectral Image Classification

Abstract: A generative adversarial network (GAN) usually contains a generative network and a discriminative network in competition with each other. The GAN has shown its capability in a variety of applications. In this paper, the usefulness and effectiveness of GAN for classification of hyperspectral images (HSIs) are explored for the first time. In the proposed GAN, a convolutional neural network (CNN) is designed to discriminate the inputs and another CNN is used to generate so-called fake inputs. The aforementioned CNNs are trained together: the generative CNN tries to generate fake inputs that are as real as possible, and the discriminative CNN tries to classify the real and fake inputs. This kind of adversarial training improves the generalization capability of the discriminative CNN, which is really important when the training samples are limited. Specifically, we propose two schemes: 1) a well-designed 1D-GAN as a spectral classifier and 2) a robust 3D-GAN as a spectral–spatial classifier. Furthermore, the generated adversarial samples are used with real training samples to fine-tune the discriminative CNN, which improves the final classification performance. The proposed classifiers are carried out on three widely used hyperspectral data sets: Salinas, Indiana Pines, and Kennedy Space Center. The obtained results reveal that the proposed models provide competitive results compared to the state-of-the-art methods. In addition, the proposed GANs open new opportunities in the remote sensing community for the challenging task of HSI classification and also reveal the huge potential of GAN-based methods for the analysis of such complex and inherently nonlinear data.