Robust Multimodal Person Identification With Limited Training Data

This paper presents a novel method of audio-visual feature-level fusion for person identification where both the speech and facial modalities may be corrupted, and there is a lack of prior knowledge about the corruption. Furthermore, we assume there are limited amount of training data for each modality (e.g., a short training speech segment and a single training facial image for each person). A new multimodal feature representation and a modified cosine similarity are introduced to combine and compare bimodal features with limited training data, as well as vastly differing data rates and feature sizes. Optimal feature selection and multicondition training are used to reduce the mismatch between training and testing, thereby making the system robust to unknown bimodal corruption. Experiments have been carried out on a bimodal dataset created from the SPIDRE speaker recognition database and AR face recognition database with variable noise corruption of speech and occlusion in the face images. The system's speaker identification performance on the SPIDRE database, and facial identification performance on the AR database, is comparable with the literature. Combining both modalities using the new method of multimodal fusion leads to significantly improved accuracy over the unimodal systems, even when both modalities have been corrupted. The new method also shows improved identification accuracy compared with the bimodal systems based on multicondition model training or missing-feature decoding alone.

Deficiencies of double-strand break repair factors and effects on mutagenesis in directly gamma-irradiated and medium-mediated bystander human lymphoblastoid cells

Using RNA interference techniques to knock down key proteins in two major double-strand break (DSB) repair pathways (DNA-PKcs for nonhomologous end joining, NHEJ, and Rad54 for homologous recombination, HR), we investigated the influence of DSB repair factors on radiation mutagenesis at the autosomal thymidine kinase (TK) locus both in directly irradiated cells and in unirradiated bystander cells. We also examined the role of p53 (TP53) in these processes by using cells of three human lymphoblastoid cell lines from the same donor but with differing p53 status (TK6 is p53 wild-type, NH32 is p53 null, and WTK1 is p53 mutant). Our results indicated that p53 status did not affect either the production of radiation bystander mutagenic signals or the response to these signals. In directly irradiated cells, knockdown of DNA-PKcs led to an increased mutant fraction in WTK1 cells and decreased mutant fractions in TK6 and NH32 cells. In contrast, knockdown of DNA-PKcs led to increased mutagenesis in bystander cells regardless of p53 status. In directly irradiated cells, knockdown of Rad54 led to increased induced mutant fractions in WTK1 and NH32 cells, but the knockdown did not affect mutagenesis in p53 wild-type TK6 cells. In all cell lines, Rad54 knockdown had no effect on the magnitude of bystander mutagenesis. Studies with extracellular catalase confirmed the involvement of H2O2 in bystander signaling. Our results demonstrate that DSB repair factors have different roles in mediating mutagenesis in irradiated and bystander cells. (C) 2008 by Radiation Research Society.

Context, institution, or accountability? Exploring the factors that shape the performance of national human rights and equality bodies

The past two decades witnessed a global proliferation of national human rights and equality bodies. Yet the research literature remains critical of their performance, positing a series of explanations for the gap between the expectations of civil society and the contribution they make. Through a comparative analysis of six statutory human rights and equality bodies in the United Kingdom and Ireland, this article explores the range of factors that shape their performance.

Combining environmental and medical datasets to explore potential associations between environmental factors and health: Policy implications for human health risk assessments

This review paper discusses the use of Tellus and Tellus Border soil and stream geochemistry data to investigate the relationship between medical data and naturally occurring background levels of potentially toxic elements (PTEs) such as heavy metals in soils and water. The research hypothesis is that long-term low level oral exposure of PTEs via soil and water may result in cumulative exposures that may act as risk factors for progressive diseases including cancer and chronic kidney disease. A number of public policy implications for regional human health risk assessments, public health policy and education are also explored alongside the argument for better integration of multiple data sets to enhance ongoing medical and social research. This work presents a partnership between the School of Geography, Archaeology and Palaeoecology, Northern Ireland Cancer Registry, Queen’s University Belfast, and the nephrology (kidney medicine) research group.

Human Papillomavirus and Oral Cancer: the International Agency for Research on Cancer Multicenter Study

Background: Human papillomavirus (HPV), the causal agent of cervical cancer, appears to be involved in the etiology of cancer of the oral cavity and oropharynx. To investigate these associations, we conducted a multicenter case-control study of cancer of the oral cavity and oropharynx in nine countries. Methods: We recruited 1670 case patients (1415 with cancer of the oral cavity and 255 with cancer of the oropharynx) and 1732 control subjects and obtained an interview, oral exfoliated cells, and blood from all participants and fresh biopsy specimens from case patients. HPV DNA was detected by polymerase chain reaction (PCR). Antibodies against HPV16 L1, E6, and E7 proteins in plasma were detected with enzyme-linked immunosorbent assays. Multivariable models were used for case-control and case-case comparisons. Results: HPV DNA was detected in biopsy specimens of 3.9% (95% confidence interval [CI]=2.5% to 5.3%) of 766 cancers of the oral cavity with valid PCR results and 18.3% (95% CI=12.0% to 24.7%) of 142 cancers of the oropharynx (oropharynx and tonsil combined) with valid PCR results. HPV DNA in cancer biopsy specimens was detected less frequently among tobacco smokers and paan chewers and more frequently among subjects who reported more than one sexual partner or who practiced oral sex. HPV16 DNA was found in 94.7% of HPV DNA-positive case patients. HPV DNA in exfoliated cells was not associated with cancer risk or with HPV DNA detection in biopsy specimens. Antibodies against HPV16 L1 were associated with risk for cancers of the oral cavity (odds ratio [OR]=1.5, 95% CI=1.1 to 2.1) and the oropharynx (OR=3.5, 95% CI=2.1 to 5.9). Antibodies against HPV16 E6 or E7 were also associated with risk for cancers of the oral cavity (OR=2.9, 95% CI=1.7 to 4.8) and the oropharynx (OR=9.2, 95% CI=4.8 to 17.7). Conclusions: HPV appears to play an etiologic role in many cancers of the oropharynx and possibly a small subgroup of cancers of the oral cavity. The most common HPV type in genital cancers (HPV16) was also the most common in these tumors. The mechanism of transmission of HPV to the oral cavity warrants further investigation.

Riboflavin, Flavin Mononucleotide and Flavin Adenine Dinucleotide in Human Plasma and Erythrocytes at Baseline and after Low-Dose Riboflavin Supplementation.

Background: Vitamin B2 exists in blood as riboflavin and its cofactors, flavin mononucleotide (FMN) and FAD. The erythrocyte glutathione reductase activation coefficient (EGRAC) has traditionally been used to assess vitamin B2 status in humans. We investigated the relationships of EGRAC and plasma and erythrocyte concentrations of riboflavin, FMN, and FAD in elderly volunteers and their responses to riboflavin administration. Methods: EGRAC and plasma and erythrocyte concentrations of riboflavin, FMN, and FAD were determined in 124 healthy individuals with a mean age of 69 years. The same measurements were made in a subgroup of 46 individuals with EGRAC 1.20 who participated in a randomized double-blind 12-week intervention study and received riboflavin (1.6 mg/day; n = 23) or placebo (n = 23). Results: Median plasma concentrations were 10.5 nmol/L for riboflavin, 6.6 nmol/L for FMN, and 74 nmol/L for FAD. In erythrocytes, there were only trace amounts of riboflavin, whereas median FMN and FAD concentrations were 44 and 469 nmol/L, respectively. Erythrocyte FMN and FAD correlated with each other and with EGRAC and plasma riboflavin (P

An Immunohistochemical Study of the Distribution of the Measles Virus Receptors, CD46 and SLAM, In Normal Human Tissues and Subacute Sclerosing Panencephalitis

We have compared the expression of the known measles virus (MV) receptors, membrane cofactor protein (CD46) and the signaling lymphocyte-activation molecule (SLAM), using immunohistochemistry, in a range of normal peripheral tissues (known to be infected by MV) as well as in normal and subacute sclerosing panencephalitis (SSPE) brain. To increase our understanding of how these receptors could be utilized by wild-type or vaccine strains in vivo, the results have been considered with regard to the known route of infection and systemic spread of MV. Strong staining for CD46 was observed in endothelial cells lining blood vessels and in epithelial cells and tissue macrophages in a wide range of peripheral tissues, as well as in Langerhans' and squamous cells in the skin. In lymphoid tissues and blood, subsets of cells were positive for SLAM, in comparison to CD46, which stained all nucleated cell types. Strong CD46 staining was observed on cerebral endothelium throughout the brain and also on ependymal cells lining the ventricles and choroid plexus. Comparatively weaker CD46 staining was observed on subsets of neurons and oligodendrocytes. In SSPE brain sections, the areas distant from lesion sites and negative for MV by immunocytochemistry showed the same distribution for CD46 as in normal brain. However, cells in lesions, positive for MV, were negative for CD46. Normal brain showed no staining for SLAM, and in SSPE brain only subsets of leukocytes in inflammatory infiltrates were positive. None of the cell types most commonly infected by MV show detectable expression of SLAM, whereas CD46 is much more widely expressed and could fulfill a receptor function for some wild-type strains. In the case of wild-type stains, which are unable to use CD46, a further as yet unknown receptor(s) would be necessary to fully explain the pathology of MV infection.