Supervised and semi-supervised 3D organ localisation in CT images combining reinforcement learning with imitation learning.

Computer aided diagnostics often requires analysis of a region of interest (ROI) within a radiology scan, and the ROI may be an organ or a suborgan. Although deep learning algorithms have the ability to outperform other methods, they rely on the availability of a large amount of annotated data. Motivated by the need to address this limitation, an approach to localisation and detection of multiple organs based on supervised and semi-supervised learning is presented here. It draws upon previous work by the authors on localising the thoracic and lumbar spine region in CT images. The method generates six bounding boxes of organs of interest, which are then fused to a single bounding box. The results of experiments on localisation of the Spleen, Left and Right Kidneys in CT Images using supervised and semi supervised learning (SSL) demonstrate the ability to address data limitations with a much smaller data set and fewer annotations, compared to other state-of-the-art methods. The SSL performance was evaluated using three different mixes of labelled and unlabelled data (i.e. 30:70,35:65,40:60) for each of lumbar spine, spleen left and right kidneys respectively. The results indicate that SSL provides a workable alternative especially in medical imaging where it is difficult to obtain annotated data.

Performance of the cobas® HBV RNA automated investigational assay for the detection and quantification of circulating HBV RNA in chronic HBV patients.

BACKGROUND: The amount of HBV RNA in peripheral blood may reflect HBV covalently closed circular DNA (cccDNA) transcriptional activity within infected hepatocytes. Quantification of circulating HBV RNA (cirB-RNA) is thus a promising biomarker for monitoring antiviral treatment. OBJECTIVES: We evaluated the performance of an automated, prototype quantitative HBV RNA assay for use on the Roche cobas® 6800/8800 systems. STUDY DESIGN: The sensitivity, specificity, linearity, and potential interference by HBV DNA of the cobas® HBV RNA assay were assessed using synthetic HBV armored RNA and clinical specimens. RESULTS: cobas® HBV RNA results were linear between 10 and 107 copies/mL in clinical samples of several HBV genotypes, and up to 109 copies/mL with synthetic RNA. Precision and reproducibility were excellent, with standard deviation below 0.15 log10 copies/mL and coefficients of variation below 5% throughout the linear range. The presence of HBV DNA had minimal (<0.3 log10 copies/mL) impact on HBV RNA quantification at DNA:RNA ratios of up to approximately one million. In a panel of 36 untreated patient samples, cirB-RNA concentrations were approximately 200-fold lower than HBV DNA. cirB-RNA was detected in all 13 HBeAg-positive patients (mean 6.0 log10 copies/mL), and in 20 of 23 HBeAg-negative patients (mean of quantifiable samples 2.2 log10 copies/mL). Finally, cirB-RNA was detected in 12 of 20 nucleoside analog-treated patients (mean of quantifiable samples 3.4 log10 copies/mL). CONCLUSIONS: The cobas® 6800/8800 investigational HBV RNA assay is a high throughput, sensitive and inclusive assay to evaluate the clinical relevance of cirB-RNA quantification in patients with chronic hepatitis B.

A Highly Verified Assay for KRAS Mutation Detection in Tissue and Plasma of Lung, Colorectal, and Pancreatic Cancer.

CONTEXT.­: KRAS Mutation Test v2 is used for the qualitative detection and identification of 28 mutations in exons 2, 3, and 4 of the human KRAS gene. OBJECTIVE.­: To verify the performance of KRAS Mutation Test v2 and to evaluate its accuracy by correlation with a next-generation sequencing method on Illumina MiSeq. DESIGN.­: In this study, we used formalin-fixed, paraffin-embedded tissue and plasma specimens from non-small cell lung cancer, colorectal cancer, and pancreatic cancer patients. Results of specificity, precision, analytical sensitivity, and accuracy as compared with a MiSeq method are reported. RESULTS.­: The KRAS Mutation Test v2 demonstrated exquisite sensitivity and specificity and broad coverage of KRAS mutations. Precision was 100% (108 of 108) across all samples, operators, and instruments for formalin-fixed, paraffin-embedded tissue and 99.8% (615 of 616) for plasma. Analytical sensitivity was high with detection of 1% mutant sequence in formalin-fixed, paraffin-embedded tissue samples and as low as 25 mutant sequence copies/mL for plasma samples. The test also showed high overall concordance for formalin-fixed, paraffin-embedded tumor tissue as well as for plasma specimens when compared with MiSeq sequencing results. CONCLUSIONS.­: The KRAS Mutation Test v2 is a highly robust, reproducible, and sensitive test for the qualitative detection of 28 mutations in exons 2, 3, and 4 of the KRAS gene in both solid (tissue) and liquid (plasma) biopsies from colorectal cancer, non-small cell lung cancer, and pancreatic cancer, and is a convenient option for KRAS mutation testing.

Human parainfluenza type 3 virus impairs the efficacy of glucocorticoids to limit allergy-induced pulmonary inflammation in guinea-pigs.

Viral exacerbations of allergen-induced pulmonary inflammation in pre-clinical models reportedly reduce the efficacy of glucocorticoids to limit pulmonary inflammation and airways hyper-responsiveness to inhaled spasmogens. However, exacerbations of airway obstruction induced by allergen challenge have not yet been studied. hPIV-3 (human parainfluenza type 3 virus) inoculation of guinea-pigs increased inflammatory cell counts in BAL (bronchoalveolar lavage) fluid and caused hyper-responsiveness to inhaled histamine. Both responses were abolished by treatment with either dexamethasone (20 mg/kg of body weight, subcutaneous, once a day) or fluticasone propionate (a 0.5 mg/ml solution aerosolized and inhaled over 15 min, twice a day). In ovalbumin-sensitized guinea-pigs, allergen (ovalbumin) challenge caused two phases of airway obstruction [measured as changes in sGaw (specific airways conductance) using whole body plethysmography]: an immediate phase lasting between 4 and 6 h and a late phase at about 7 h. The late phase, airway hyper-responsiveness to histamine and inflammatory cell counts in BAL were all significantly reduced by either glucocorticoid. Inoculation of guinea-pigs sensitized to ovalbumin with hPIV-3 transformed the allergen-induced airway obstruction from two transient phases into a single sustained response lasting up to 12 h. This exacerbated airway obstruction and airway hyper-responsiveness to histamine were unaffected by treatment with either glucocorticoid whereas inflammatory cell counts in BAL were only partially inhibited. Virus- or allergen-induced pulmonary inflammation, individually, are glucocorticoid-sensitive, but in combination generate a phenotype where glucocorticoid efficacy is impaired. This suggests that during respiratory virus infection, glucocorticoids might be less effective in limiting pulmonary inflammation associated with asthma.

The Environmental Determinants of Diabetes in the Young (TEDDY): genetic criteria and international diabetes risk screening of 421 000 infants.

AIMS: The Environmental Determinants of Diabetes in the Young (TEDDY) study seeks to identify environmental factors influencing the development of type 1 diabetes (T1D) using intensive follow-up of children at elevated genetic risk. This study requires a cost-effective yet accurate screening strategy to identify the high-risk cohort. METHODS: The TEDDY cohort was identified through newborn screening using human leukocyte antigen (HLA) class II genes based on criteria established with pre-TEDDY data. HLA typing was completed at six international centers using different genotyping methods that can achieve >98% accuracy. RESULTS: TEDDY developed separate inclusion criteria for the general population (GP) and first-degree relatives (FDRs) of T1D patients. The FDR eligibility includes nine haplogenotypes (DR3/4, DR4/4, DR4/8, DR3/3, DR4/4b, DR4/1, DR4/13, DR4/9, and DR3/9) for broad HLA diversity, whereas the GP eligibility includes only the first four haplogenotypes with DRB1*0403 as an exclusion allele. TEDDY has screened 414 714 GP infants, of which 19 906 (4.8%) were eligible, whereas 1415 of the 6333 screened FDR infants (22.2%) were eligible. High-resolution confirmation testing of the eligible subjects indicated that the low-cost and low-resolution genotyping techniques employed at the screening centers yielded an accuracy of 99%. There were considerable variations in eligibility rates among the centers for GP (3.5-7.4%) and FDR (19-32%) subjects. The eligibility rates among US ethnic groups were 0.9, 1.3, 5.0, and 6.9% for Asians, Black, Caucasians, and Hispanics, respectively. CONCLUSIONS: Different low-cost and low-resolution genotyping methods are useful for the efficient and accurate identification of a high-risk cohort for follow-up based on the TEDDY HLA inclusion criteria.

Bradykinin-induced lung inflammation and bronchoconstriction: role in parainfluenze-3 virus-induced inflammation and airway hyperreactivity.

Inhaled bradykinin causes bronchoconstriction in asthmatic subjects but not nonasthmatics. To date, animal studies with inhaled bradykinin have been performed only in anesthetized guinea pigs and rats, where it causes bronchoconstriction through sensory nerve pathways. In the present study, airway function was recorded in conscious guinea pigs by whole-body plethysmography. Inhaled bradykinin (1 mM, 20 s) caused bronchoconstriction and influx of inflammatory cells to the lungs, but only when the enzymatic breakdown of bradykinin by angiotensin-converting enzyme and neutral endopeptidase was inhibited by captopril (1 mg/kg i.p.) and phosphoramidon (10 mM, 20-min inhalation), respectively. The bronchoconstriction and cell influx were antagonized by the B(2) kinin receptor antagonist 4-(S)-amino-5-(4-{4-[2,4-dichloro-3-(2,4-dimethyl-8-quinolyloxymethyl)phenylsulfonamido]-tetrahydro-2H-4-pyranylcarbonyl}piperazino)-5-oxopentyl](trimethyl)ammonium chloride hydrochloride (MEN16132) when given by inhalation (1 and 10 µM, 20 min) and are therefore mediated via B(2) kinin receptors. However, neither intraperitioneal MEN16132 nor the peptide B(2) antagonist icatibant, by inhalation, antagonized these bradykinin responses. Sensitization of guinea pigs with ovalbumin was not sufficient to induce airway hyperreactivity (AHR) to the bronchoconstriction by inhaled bradykinin. However, ovalbumin challenge of sensitized guinea pigs caused AHR to bradykinin and histamine. Infection of guinea pigs by nasal instillation of parainfluenza-3 virus produced AHR to inhaled histamine and lung influx of inflammatory cells. These responses were attenuated by the bradykinin B(2) receptor antagonist MEN16132 and H-(4-chloro)DPhe-2'(1-naphthylalanine)-(3-aminopropyl)guanidine (VA999024), an inhibitor of tissue kallikrein, the enzyme responsible for lung synthesis of bradykinin. These results suggest that bradykinin is involved in virus-induced inflammatory cell influx and AHR.

Do non-HLA genes influence development of persistent islet autoimmunity and type 1 diabetes in children with high-risk HLA-DR,DQ genotypes?

OBJECTIVE: Specific alleles of non-HLA genes INS, CTLA-4, and PTPN22 have been associated with type 1 diabetes. We examined whether some of these alleles influence development of islet autoimmunity or progression from persistent islet autoimmunity to type 1 diabetes in children with high-risk HLA-DR,DQ genotypes. RESEARCH DESIGN AND METHODS: Since 1993, the Diabetes Autoimmunity Study in the Young (DAISY) has followed 2,449 young children carrying HLA-DR,DQ genotypes associated with type 1 diabetes. Of those, 112 have developed islet autoimmunity (persistent autoantibodies to insulin, GAD65, and/or IA-2), and 47 of these have progressed to type 1 diabetes. The influence of polymorphisms of INS(-23Hph1), CTLA-4(T17A), and PTPN22(R620W) on development of persistent islet autoimmunity and progression to type 1 diabetes was evaluated by parametric models and by survival analyses. RESULTS: PTPN22(R620W) allele T was associated with development of persistent islet autoimmunity (hazard ratio 1.83 [95% CI 1.27-2.63]) controlling for ethnicity, presence of HLA-DR3/4,DQB1*0302, and having a first-degree relative with type 1 diabetes. Survival analyses showed a significantly (P = 0.002) higher risk of persistent islet autoimmunity by age 10 years for the TT genotype (27.3%) than for the CT or CC genotype (7.9 and 5.3%, respectively). Cumulative risk of persistent islet autoimmunity was slightly higher (P = 0.02) for the INS(-23Hph1) AA genotype (7.8%) than for the AT or TT genotype (4.2 and 6.4% risk by age 10 years, respectively). CONCLUSIONS: Whereas the HLA-DR3/4,DQB1*0302 genotype had a dramatic influence on both development of islet autoimmunity and progression to type 1 diabetes, the PTPN22(R620W) T allele significantly influences progression to persistent islet autoimmunity in the DAISY cohort.

An 'overwhelming illness': women's experiences of learning to live with chronic fatigue syndrome/myalgic encephalomyelitis.

The processes through which people learn to live with CFS/ME are poorly understood and have not been rigorously explored within the literature. Semi-structured interviews were conducted with eight women and analysed using interpretative phenomenological analysis. Participants initially described being 'overwhelmed' by CFS/ME. Attempts at seeking help were unsatisfactory and participants described feeling let down and disbelieved. Participants reacted to this by identifying types of 'self-help' and assertively taking more responsibility for their illness and its treatment. Acquiring social support and greater knowledge were key mediating factors in the emergence of control and acceptance. The relevance of the themes to existing research and the implications for clinical practice are considered.

Association of non-HLA genes with type 1 diabetes autoimmunity.

Approximately 50% of the genetic risk for type 1 diabetes is attributable to the HLA region. We evaluated associations between candidate genes outside the HLA region-INS, cytotoxic T-lymphocyte-associated antigen (CTLA)-4, interleukin (IL)-4, IL-4R, and IL-13 and islet autoimmunity among children participating in the Diabetes Autoimmunity Study in the Young (DAISY). Children with persistent islet autoantibody positivity (n = 102, 38 of whom have already developed diabetes) and control subjects (n = 198) were genotyped for single nucleotide polymorphisms (SNPs) in the candidate genes. The INS-23Hph1 polymorphism was significantly associated with both type 1 diabetes (OR = 0.30; 95% CI 0.13-0.69) and persistent islet autoimmunity but in the latter, only in children with the HLA-DR3/4 genotype (0.40; 0.18-0.89). CTLA-4 promoter SNP was significantly associated with type 1 diabetes (3.52; 1.22-10.17) but not with persistent islet autoimmunity. Several SNPs in the IL-4 regulatory pathway appeared to have a predisposing effect for type 1 diabetes. Associations were found between both IL-4R haplotypes and IL-4-IL-13 haplotypes and persistent islet autoimmunity and type 1 diabetes. This study confirms the association between the INS and CTLA-4 loci and type 1 diabetes. Genes involved in the IL-4 regulatory pathway (IL-4, IL-4R, IL-13) may confer susceptibility or protection to type 1 diabetes depending on individual SNPs or specific haplotypes.

Incremental training of first order recurrent neural networks to predict a context-sensitive language.

In recent years it has been shown that first order recurrent neural networks trained by gradient-descent can learn not only regular but also simple context-free and context-sensitive languages. However, the success rate was generally low and severe instability issues were encountered. The present study examines the hypothesis that a combination of evolutionary hill climbing with incremental learning and a well-balanced training set enables first order recurrent networks to reliably learn context-free and mildly context-sensitive languages. In particular, we trained the networks to predict symbols in string sequences of the context-sensitive language [a(n)b(n)c(n); n > or = 1. Comparative experiments with and without incremental learning indicated that incremental learning can accelerate and facilitate training. Furthermore, incrementally trained networks generally resulted in monotonic trajectories in hidden unit activation space, while the trajectories of non-incrementally trained networks were oscillating. The non-incrementally trained networks were more likely to generalise.

Phylogenetic analyses of the homologous transmembrane channel-forming proteins of the F0F1-ATPases of bacteria, chloroplasts and mitochondria.

Sequences of the three integral membrane subunits (subunits a, b and c) of the F0 sector of the proton-translocating F-type (F0F1-) ATPases of bacteria, chloroplasts and mitochondria have been analysed. All homologous-sequenced proteins of these subunits, comprising three distinct families, have been identified by database searches, and the homologous protein sequences have been aligned and analysed for phylogenetic relatedness. The results serve to define the relationships of the members of each of these three families of proteins, to identify regions of relative conservation, and to define relative rates of evolutionary divergence. Of these three subunits, c-subunits exhibited the slowest rate of evolutionary divergence, b-subunits exhibited the most rapid rate of evolutionary divergence, and a-subunits exhibited an intermediate rate of evolutionary divergence. The results allow definition of the relative times of occurrence of specific events during evolutionary history, such as the intragenic duplication event that gave rise to large c-subunits in eukaryotic vacuolar-type ATPases after eukaryotes diverged from archaea, and the extragenic duplication of F-type ATPase b-subunits that occurred in blue-green bacteria before the advent of chloroplasts. The results generally show that the three F0 subunits evolved as a unit from a primordial set of genes without appreciable horizontal transmission of the encoding genetic information although a few possible exceptions were noted.