Interactions between pre-processing and classification methods for event-related-potential classification: best-practice guidelines for brain-computer interfacing.

Detecting event related potentials (ERPs) from single trials is critical to the operation of many stimulus-driven brain computer interface (BCI) systems. The low strength of the ERP signal compared to the noise (due to artifacts and BCI irrelevant brain processes) makes this a challenging signal detection problem. Previous work has tended to focus on how best to detect a single ERP type (such as the visual oddball response). However, the underlying ERP detection problem is essentially the same regardless of stimulus modality (e.g., visual or tactile), ERP component (e.g., P300 oddball response, or the error-potential), measurement system or electrode layout. To investigate whether a single ERP detection method might work for a wider range of ERP BCIs we compare detection performance over a large corpus of more than 50 ERP BCI datasets whilst systematically varying the electrode montage, spectral filter, spatial filter and classifier training methods. We identify an interesting interaction between spatial whitening and regularised classification which made detection performance independent of the choice of spectral filter low-pass frequency. Our results show that pipeline consisting of spectral filtering, spatial whitening, and regularised classification gives near maximal performance in all cases. Importantly, this pipeline is simple to implement and completely automatic with no expert feature selection or parameter tuning required. Thus, we recommend this combination as a "best-practice" method for ERP detection problems.

An online brain-computer interface based on shifting attention to concurrent streams of auditory stimuli.

We report on the development and online testing of an electroencephalogram-based brain-computer interface (BCI) that aims to be usable by completely paralysed users-for whom visual or motor-system-based BCIs may not be suitable, and among whom reports of successful BCI use have so far been very rare. The current approach exploits covert shifts of attention to auditory stimuli in a dichotic-listening stimulus design. To compare the efficacy of event-related potentials (ERPs) and steady-state auditory evoked potentials (SSAEPs), the stimuli were designed such that they elicited both ERPs and SSAEPs simultaneously. Trial-by-trial feedback was provided online, based on subjects' modulation of N1 and P3 ERP components measured during single 5 s stimulation intervals. All 13 healthy subjects were able to use the BCI, with performance in a binary left/right choice task ranging from 75% to 96% correct across subjects (mean 85%). BCI classification was based on the contrast between stimuli in the attended stream and stimuli in the unattended stream, making use of every stimulus, rather than contrasting frequent standard and rare 'oddball' stimuli. SSAEPs were assessed offline: for all subjects, spectral components at the two exactly known modulation frequencies allowed discrimination of pre-stimulus from stimulus intervals, and of left-only stimuli from right-only stimuli when one side of the dichotic stimulus pair was muted. However, attention modulation of SSAEPs was not sufficient for single-trial BCI communication, even when the subject's attention was clearly focused well enough to allow classification of the same trials via ERPs. ERPs clearly provided a superior basis for BCI. The ERP results are a promising step towards the development of a simple-to-use, reliable yes/no communication system for users in the most severely paralysed states, as well as potential attention-monitoring and -training applications outside the context of assistive technology.

A graphical model framework for decoding in the visual ERP-based BCI speller.

We present a graphical model framework for decoding in the visual ERP-based speller system. The proposed framework allows researchers to build generative models from which the decoding rules are obtained in a straightforward manner. We suggest two models for generating brain signals conditioned on the stimulus events. Both models incorporate letter frequency information but assume different dependencies between brain signals and stimulus events. For both models, we derive decoding rules and perform a discriminative training. We show on real visual speller data how decoding performance improves by incorporating letter frequency information and using a more realistic graphical model for the dependencies between the brain signals and the stimulus events. Furthermore, we discuss how the standard approach to decoding can be seen as a special case of the graphical model framework. The letter also gives more insight into the discriminative approach for decoding in the visual speller system.

Overlap and refractory effects in a brain-computer interface speller based on the visual P300 event-related potential.

We reveal the presence of refractory and overlap effects in the event-related potentials in visual P300 speller datasets, and we show their negative impact on the performance of the system. This finding has important implications for how to encode the letters that can be selected for communication. However, we show that such effects are dependent on stimulus parameters: an alternative stimulus type based on apparent motion suffers less from the refractory effects and leads to an improved letter prediction performance.

The target cell response to cytokines governs the autoreactive T cell repertoire in the pancreas of NOD mice.

AIMS/HYPOTHESIS: The pancreatic beta cell response to cytokines is crucial for the development of type 1 diabetes in the NOD mouse. For example, beta cell production of suppressor of cytokine signalling-1 (SOCS-1) protects against diabetes. This finding and other recent studies indicated that cytokine-stressed beta cells might contribute to disease progression by affecting the pancreatic lymphocyte infiltrate. The aim of this study was to provide insight into how the beta cell influences the pancreas-infiltrating T cell repertoire. METHODS: Lymphocytes isolated from Socs1-transgenic (tg) and non-tg NOD mice were analysed by flow cytometry. mRNA and protein levels in pancreatic islets were measured by real-time PCR and immunofluorescence analysis, respectively. RESULTS: The percentages of regulatory T cells, total counts and ratios between infiltrating CD8+ and CD4+ T cells, and the expression of killer cell lectin-like receptor subfamily K, member 1 (NKG2D) on CD8+ T cells did not differ in pancreases from prediabetic Socs1-tg and non-tg NOD mice. However, a striking difference in the percentages of CD8+ T cells specific for glucose 6-phosphatase catalytic subunit-related protein 206-214 was found, showing that SOCS-1 prevents the accumulation of high percentages of self-reactive CD8+ T cells in the pancreas. It was also found that protection from diabetes in Socs1-tg NOD mice correlated with a reduced expression of Cxcl10 mRNA in IFN-gamma treated islets. CONCLUSIONS/INTERPRETATION: This study highlights an important role for the beta cell in the local regulation of the diabetogenic process. By responding to the pro-inflammatory pancreas milieu it strongly influences the islet-reactive T cell repertoire in the pancreas.

Do free-ranging common brushtail possums (Trichosurus vulpecula) play a role in the transmission of Toxoplasma gondii within a zoo environment?

To investigate the possible role of common brushtail possums (Trichosurus vulpecula) in the transmission of Toxoplasma gondii within a zoo environment, a serological survey of a free-ranging population resident within Taronga Zoo, Sydney, Australia was undertaken using the modified agglutination test (MAT). For comparison, the seroprevalence of T. gondii antibodies was also assessed in a possum population inhabiting a felid-free, non-urban woodland habitat. Six of 126 possums (4.8%) from the zoo population had antibodies to T. gondii with a MAT titre of 25 or higher, while in contrast, all of the 17 possums from woodland were seronegative. These observations suggest that possums were at a higher risk of exposure to the parasite as a consequence of co-existing with domestic, stray and captive felids associated with urbanisation. Screening of captive felids at the zoo indicated 16 of 23 individuals (67%) and all 6 species were seropositive for T. gondii, implicating them as a possible source of the parasite within the zoo setting. In addition captive, non-felid carnivores including the chimpanzee (Pan troglodytes), saltwater crocodile (Crocodylus porosus), dingo (Canis lupis) and leopard seal (Hydrurga leptonyx) were tested for the presence of T. gondii antibodies as these species predate and are a leading cause of death amongst zoo possums. In total, 5 of 23 individuals (22%) were seropositive, representing 2 of the 4 carnivorous species; the dingo and chimpanzee. These data suggest that carnivory was not a highly efficient pathway for the transmission of T. gondii and the free-ranging possum population posed minimal threat to the health of zoo animals.

Isolation of the mite Myocoptes musculinus Koch from the Spinifex Hopping mouse (Notomys alexis).

This paper reports on the isolation and identification of the fur-clasping mite, Myocoptes musculinus, from the faeces of the Spinifex Hopping mouse (Notomys alexis). This investigation adds to the sparse records of ectoparasites collected from native Australian murids.

Masking effect produced by Mach bands on the detection of narrow bars of random polarity.

Difficulties arise in measuring masking by Mach bands because very-low-contrast signals distort the bands. [J. Opt. Soc. Am. A 17, 1147 (2000).] Adding narrow luminance increments (bright bars) in the dark Mach band widens the dark band; adding decrements (dark bars) narrows the dark band, and conversely in the bright bands. Randomizing signal polarity prevents observers from using the distortion of the Mach bands as a cue to the presence of the signal. We measured (two-alternative-forced-choice) Mach bands' masking of randomly selected bright (incremental) or dark (decremental) bars. Detection was worse in both dark and bright Mach bands than on the neighboring plateaus. Separate analysis of trials containing only one polarity signal revealed 9-cycle/deg oscillations in performance as a function of location. Oscillations in the two polarities were approximately 180 degrees out of phase.

Statistical modeling of interlocus interactions in a complex disease: rejection of the multiplicative model of epistasis in type 1 diabetes.

In general, common diseases do not follow a Mendelian inheritance pattern. To identify disease mechanisms and etiology, their genetic dissection may be assisted by evaluation of linkage in mouse models of human disease. Statistical modeling of multiple-locus linkage data from the nonobese diabetic (NOD) mouse model of type 1 diabetes has previously provided evidence for epistasis between alleles of several Idd (insulin-dependent diabetes) loci. The construction of NOD congenic strains containing selected segments of the diabetes-resistant strain genome allows analysis of the joint effects of alleles of different loci in isolation, without the complication of other segregating Idd loci. In this article, we analyze data from congenic strains carrying two chromosome intervals (a double congenic strain) for two pairs of loci: Idd3 and Idd10 and Idd3 and Idd5. The joint action of both pairs is consistent with models of additivity on either the log odds of the penetrance, or the liability scale, rather than with the previously proposed multiplicative model of epistasis. For Idd3 and Idd5 we would also not reject a model of additivity on the penetrance scale, which might indicate a disease model mediated by more than one pathway leading to beta-cell destruction and development of diabetes. However, there has been confusion between different definitions of interaction or epistasis as used in the biological, statistical, epidemiological, and quantitative and human genetics fields. The degree to which statistical analyses can elucidate underlying biologic mechanisms may be limited and may require prior knowledge of the underlying etiology.

The psychometric function: I. Fitting, sampling, and goodness of fit.

The psychometric function relates an observer's performance to an independent variable, usually some physical quantity of a stimulus in a psychophysical task. This paper, together with its companion paper (Wichmann & Hill, 2001), describes an integrated approach to (1) fitting psychometric functions, (2) assessing the goodness of fit, and (3) providing confidence intervals for the function's parameters and other estimates derived from them, for the purposes of hypothesis testing. The present paper deals with the first two topics, describing a constrained maximum-likelihood method of parameter estimation and developing several goodness-of-fit tests. Using Monte Carlo simulations, we deal with two specific difficulties that arise when fitting functions to psychophysical data. First, we note that human observers are prone to stimulus-independent errors (or lapses). We show that failure to account for this can lead to serious biases in estimates of the psychometric function's parameters and illustrate how the problem may be overcome. Second, we note that psychophysical data sets are usually rather small by the standards required by most of the commonly applied statistical tests. We demonstrate the potential errors of applying traditional chi2 methods to psychophysical data and advocate use of Monte Carlo resampling techniques that do not rely on asymptotic theory. We have made available the software to implement our methods.

The psychometric function: II. Bootstrap-based confidence intervals and sampling.

The psychometric function relates an observer's performance to an independent variable, usually a physical quantity of an experimental stimulus. Even if a model is successfully fit to the data and its goodness of fit is acceptable, experimenters require an estimate of the variability of the parameters to assess whether differences across conditions are significant. Accurate estimates of variability are difficult to obtain, however, given the typically small size of psychophysical data sets: Traditional statistical techniques are only asymptotically correct and can be shown to be unreliable in some common situations. Here and in our companion paper (Wichmann & Hill, 2001), we suggest alternative statistical techniques based on Monte Carlo resampling methods. The present paper's principal topic is the estimation of the variability of fitted parameters and derived quantities, such as thresholds and slopes. First, we outline the basic bootstrap procedure and argue in favor of the parametric, as opposed to the nonparametric, bootstrap. Second, we describe how the bootstrap bridging assumption, on which the validity of the procedure depends, can be tested. Third, we show how one's choice of sampling scheme (the placement of sample points on the stimulus axis) strongly affects the reliability of bootstrap confidence intervals, and we make recommendations on how to sample the psychometric function efficiently. Fourth, we show that, under certain circumstances, the (arbitrary) choice of the distribution function can exert an unwanted influence on the size of the bootstrap confidence intervals obtained, and we make recommendations on how to avoid this influence. Finally, we introduce improved confidence intervals (bias corrected and accelerated) that improve on the parametric and percentile-based bootstrap confidence intervals previously used. Software implementing our methods is available.

Wild-type and met-65-->Leu variants of human cystatin A are functionally and structurally identical.

The solution structure of an N-terminally truncated and mutant form (M65L(2-98)) of the human cysteine protease inhibitor cystatin A has been reported that reveals extensive structural differences when compared to the previously published structure of full-length wild-type (WT) cystatin A. On the basis of the M65L(2-98) structure, a model of the inhibitory mechanism of cystatin A was proposed wherein specific interactions between the N- and C-terminal regions of cystatin A are invoked as critical determinants of protease binding. To test this model and to account for the reported differences between the two structures, we undertook additional structural and mechanistic analyses of WT and mutant forms of human cystatin A. These show that modification at the C-terminus of cystatin A by the addition of nine amino acids has no effect upon the affinity of papain inhibition (K(D) = 0.18+/-0.02 pM) and the consequences of such modification are not propagated to other parts of the structure. These findings indicate that perturbation of the C-terminus can be achieved without any measurable effect on the N-terminus or the proteinase binding loops. In addition, introduction of the methionine-65 --> leucine substitution into cystatin A that retains the N-terminal methionine (M65L(1-98)) has no significant effect upon papain binding (K(D) = 0.34+/-0.02 pM). Analyses of the structures of WT and M65L(1-98) using (1)H NMR chemical shifts and residual dipolar couplings in a partially aligning medium do not reveal any evidence of significant differences between the two inhibitors. Many of the differences between the published structures correspond to major violations by M65L(2-98) of the WT constraints list, notably in relation to the position of the N-terminal region of the inhibitor, one of three structural motifs indicated by crystallographic studies to be involved in protease binding by cystatins. In the WT structure, and consistent with the crystallographic data, this region is positioned adjacent to another inhibitory motif (the first binding loop), whereas in M65L(2-98) there is no proximity of these two motifs. As the NMR data for both WT9C and M65L(1-98) are wholly consistent with the published structure of WT cystatin A and incompatible with that of M65L(2-98), we conclude that the former represents the most reliable structural model of this protease inhibitor.

NOD Idd5 locus controls insulitis and diabetes and overlaps the orthologous CTLA4/IDDM12 and NRAMP1 loci in humans.

A genome scan for B10-derived loci that reduce the frequency of diabetes and insulitis in NOD mice demonstrated a large region (34 cM) of linkage on the proximal end of chromosome 1. This locus was designated Idd5 and encompassed candidate genes including Il1r1, Il1r2, Stat1, Stat4, Nramp1, and Bcl2. In the current study, we have confirmed the existence of Idd5 by developing a series of congenic mouse strains that are resistant to diabetes and determined that Idd5 is actually two genes located within a 9.4-cM interval. Idd5.1 is in the proximal 1.5-cM portion of the interval and contains the candidates Casp8, Cflar (FLIP), Cd28, and Cd152 (CTLA4). Idd5.1 overlaps the orthologous CTLA4/IDDM12 locus in humans. Idd5.2 is in the distal 5.1-cM portion of the 9.4-cM interval and contains the candidates Nramp1, which has a functional polymorphism between NOD and B10, and Cmkar2 (CXCR2, interleukin [IL]-8 receptor alpha). Candidate genes eliminated by this analysis include Il1r1, Ilr2, Zap70, Orch5, Stat1, Stat4, Bcl2, Cmkar4 (CXCR4), and Il10. On its own, the Idd5 locus provides a significant amount of protection from diabetes (50% reduction from parental frequency) and when combined with another resistance locus (Idd3 on chromosome 3), provides nearly complete protection from diabetes and insulitis.

The NOD Idd9 genetic interval influences the pathogenicity of insulitis and contains molecular variants of Cd30, Tnfr2, and Cd137.

Previous analyses of NOD mice have shown that some genes control the development of both insulitis and diabetes, while other loci influence diabetes without reducing insulitis. Evidence for the existence of a gene only influencing diabetes, Idd9 on mouse chromosome 4, is provided here by the development of a novel congenic mouse strain, NOD.B10 Idd9. NOD.B10 Idd9 mice display profound resistance to diabetes even though nearly all develop insulitis. Subcongenic analysis has demonstrated that alleles of at least three B10 genes, Idd9.1, Idd9.2, and Idd9.3 are required to produce Idd9-mediated diabetes resistance. Candidate genes with amino acid differences between the NOD and B10 strains have been localized to the 5.6 cM Idd9.2 interval (Tnfr2, Cd30) and to the 2.0 cM Idd9.3 interval (Cd137).

Detection of incremental and decremental bars at different locations across Mach bands and related stimuli.

Two-alternative forced-choice procedures were used to measure the detectability of bright and dark bars at various locations across luminance patterns that produced Mach bands. Detection performance was significantly affected by both dark and bright Mach bands: poor detection performance was observed at locations near, but not in, the Mach bands; relatively good detection performance at locations within the Mach bands was caused by reliable changes in the width, depth, or symmetry of the bands produced by the signal bars. The changes were apparent with signals of lower luminance than that needed for detection in the plateau regions far from the bands, but, because the cues were not sufficiently reliable to allow errorless performance, unusually shaped psychometric functions were obtained.

Congenic mapping of the type 1 diabetes locus, Idd3, to a 780-kb region of mouse chromosome 3: identification of a candidate segment of ancestral DNA by haplotype mapping.

Type 1 diabetes in the nonobese diabetic (NOD) mouse arises as a consequence of T cell-mediated destruction of the insulin-producing beta cells of the pancreas. Although little is known of the events that initiate and subsequently drive beta-cell destruction it is clear that the entire process is under complex genetic control. At present 19 loci have been mapped that influence the development of diabetes either at the level of initiation of insulitis or at the level of progression from insulitis to overt diabetes, or both. Previously, we have mapped one of these loci, Idd3, to a 0.35-cM interval on proximal mouse chromosome 3. In the present study we have narrowed the map position of this locus to an interval of 0.15 cM by a combination of novel congenic strains and an ancestral haplotype analysis approach. We have constructed a physical contig in bacterial artificial chromosome (BAC) clones across the minimal interval. Restriction mapping of the BAC contig placed the maximum size of the Idd3 interval at 780 kb between the markers D3Nds36 and D3Nds76. To refine further the Idd3 interval we developed a series of novel single nucleotide polymorphisms (SNPs) and carried out haplotype analysis on DNA from mouse strains known to carry either Idd3 susceptibility or protective alleles. This haplotype analysis identified a 145-kb segment of ancestral DNA between the microsatellite marker D3Nds6 and the SNP 81.3. One haplotype of this ancestral segment of DNA is found in mouse strains carrying an Idd3 susceptibility allele and another is found in mouse strains carrying an Idd3 protective allelle. Within the 780-kb congenically defined interval this 145-kb segment represents the most likely location for Idd3. The Il2 gene, which encodes the cytokine interleukin 2 (IL2), maps to this interval and is a strong candidate for Idd3. To investigate whether sequence variation exists in the promoter region of the Il2 gene, which might alter its expression, we sequenced the promoter region of the Il2 gene from mouse strains carrying either an Idd3 susceptibility or resistance allele. Two sequence variants were identified, neither of which fell in known regulatory elements within the Il2 promoter. In agreement with this observation steady-state Il2 mRNA levels showed no variation between susceptible and resistant mouse strains. These data suggest that the profound protection from diabetes seen in congenic mice carrying an Idd3 protective allele is unlikely to be due to differences in the level of expression of the Il2 gene. Instead, all of the current data support our hypothesis that Idd3 corresponds to amino acid variation at the amino terminus of Il2.

The role of Gly-4 of human cystatin A (stefin A) in the binding of target proteinases. Characterization by kinetic and equilibrium methods of the interactions of cystatin A Gly-4 mutants with papain, cathepsin B, and cathepsin L.

The importance of the evolutionarily conserved Gly-4 residue for the affinity and kinetics of interaction of cystatin A with several cysteine proteinases was assessed by site-directed mutagenesis. Even the smallest replacement, by Ala, resulted in approximately 1000-, approximately 10- and approximately 6000-fold decreased affinities for papain, cathepsin L, and cathepsin B, respectively. Substitution by Ser gave further 3-8-fold reductions in affinity, whereas the largest decreases, >10(5)-fold, were observed for mutations to Arg and Glu. The kinetics of inhibition of papain by the mutants with small side chains, Ala and Ser, were compatible with a one-step bimolecular reaction similar to that with wild-type cystatin A. The decreased affinities of these mutants for papain and cathepsin L were due exclusively to increased dissociation rate constants, but the reduced affinities for cathepsin B were due also to decreased association rate constants. The latter finding indicates that the intact N-terminal region serves as a guide directing cystatin A to the active site of cathepsin B, as has been proposed for cystatin C. The kinetics of binding of the mutants with charged side chains, Arg and Glu, to papain were consistent with a two-step binding mechanism, in which the mutant side chains are accommodated in the complex by a conformational change. The NMR solution structure of the Ala and Trp mutants showed only minor changes compared with wild-type cystatin A, indicating that the large reductions in affinity for proteinases are not due to altered structures of the mutants. Instead, a side chain larger than a hydrogen atom at position 4 affects the interaction with the proteinase most likely by interfering with the binding of the N-terminal region.