Correct Boundaries Research Articles

An automatic variational level set segmentation framework for computer aided dental X-rays analysis in clinical environments

An automatic variational level set segmentation framework for Computer Aided Dental X-rays Analysis (CADXA) in clinical environments is proposed. Designed for clinical environments, the segmentation contains two stages: a training stage and a segmentation stage. During the training stage, first, manually chosen representative images are segmented using hierarchical level set region detection. Then the window based feature extraction followed by principal component analysis (PCA) is applied and results are used to train a support vector machine (SVM) classifier. During the segmentation stage, dental X-rays are classified first by the trained SVM. The classifier provides initial contours which are close to correct boundaries for three coupled level sets driven by a proposed pathologically variational modeling which greatly accelerates the level set segmentation. Based on the segmentation results and uncertainty maps that are built based on a proposed uncertainty measurement, a computer aided analysis scheme is applied. The experimental results show that the proposed method is able to provide an automatic pathological segmentation which naturally segments those problem areas. Based on the segmentation results, the analysis scheme is able to provide indications of possible problem areas of bone loss and decay to the dentists. As well, the experimental results show that the proposed segmentation framework is able to speed up the level set segmentation in clinical environments.

Local density approach for modeling fluids with density-dependent interactions.

In a recent paper [Phys. Rev. Lett. 86, 2038 (2001)] a simple fluid with a particular density-dependent pair potential was shown to exhibit, together with the vapor-liquid transition, a liquid-liquid phase separation and it was evidenced that, in order to adequately define the correct boundaries of stability, a simulation procedure based on the use of local densities had to be devised. It was found that for certain thermodynamic states the potential drives the system toward a phase separation that is otherwise frustrated by the change in the interactions induced by density fluctuations. Therefore, when integral equations or global density simulations are used, the critical points estimated from the thermodynamics are not associated with divergent correlations and vice versa. Here, we will explore in depth this fluid and introduce a detailed account of the proposed local density simulation technique. The results presented bear general significance for density-dependent potentials, like those of liquid metals or charge-stabilized colloids.

Database for mobile group II introns.

Group II introns are self-splicing RNAs and retroelements found in bacteria and lower eukaryotic organelles. During the past several years, they have been uncovered in surprising numbers in bacteria due to the genome sequencing projects; however, most of the newly sequenced introns are not correctly identified. We have initiated an ongoing web site database for mobile group II introns in order to provide correct information on the introns, particularly in bacteria. Information in the web site includes: (1) introductory information on group II introns; (2) detailed information on subfamilies of intron RNA structures and intron-encoded proteins; (3) a listing of identified introns with correct boundaries, RNA secondary structures and other detailed information; and (4) phylogenetic and evolutionary information. The comparative data should facilitate study of the function, spread and evolution of group II introns. The database can be accessed at http://www.fp.ucalgary.ca/group2introns/.

NOISY SPEECH SEGMENTATION/ENHANCEMENT WITH MULTIBAND ANALYSIS AND NEURAL FUZZY NETWORKS

This paper addresses the problem of speech segmentation and enhancement in the presence of noise. We first propose a new word boundary detection algorithm by using a neural fuzzy network (called ATF-based SONFIN algorithm) for identifying islands of word signals in fixed noise-level environment. We further propose a new RTF-based RSONFIN algorithm where the background noise level varies during the procedure of recording. The adaptive time-frequency (ATF) and refined time-frequency (RTF) parameters extend the TF parameter from single band to multiband spectrum analysis, and help to make the distinction of speech and noise signals clear. The ATF and RTF parameters can extract useful frequency information by adaptively choosing proper bands of the mel-scale frequency bank. Due to the self-learning ability of SONFIN and RSONFIN, the proposed algorithms avoid the need of empirically determining thresholds and ambiguous rules. The RTF-based RSONFIN algorithm can also find the variation of the background noise level and detect correct word boundaries in the condition of variable background noise level by processing the temporal relations. Our experimental results show that both in the fixed and variable noise-level environment, the algorithms that we proposed achieved higher recognition rate than several commonly used word boundary detection algorithms and reduced the recognition error rate due to endpoint detection.

Use of covariance analysis for the prediction of structural domain boundaries from multiple protein sequence alignments.

Current methods for identification of domains within protein sequences require either structural information or the identification of homologous domain sequences in different sequence contexts. Knowledge of structural domain boundaries is important for fold recognition experiments and structural determination by X-ray crystallography or nuclear magnetic resonance spectroscopy using the divide-and-conquer approach. Here, a new and conceptually simple method for the identification of structural domain boundaries in multiple protein sequence alignments is presented. Analysis of covariance at positions within the alignment is first used to predict 3D contacts. By the nature of the domain as an independent folding unit, inter-domain predicted contacts are fewer than intra-domain predicted contacts. By analysing all possible domain boundaries and constructing a smoothed profile of predicted contact density (PCD), true structural domain boundaries are predicted as local profile minima associated with low PCD. A training data set is constructed from 52 non-homologous two-domain protein sequences of known 3D structure and used to determine optimal parameters for the profile analysis. The alignments in the training data set contained 48 +/- 17 (mean +/- SD) sequences and lengths of 257 +/- 121 residues. Of the 47 alignments yielding predictions, 35% of true domain boundaries are predicted to within 15 amino acids by the local profile minimum with the lowest profile value. Including predictions from the second- and third-lowest local minima increases the correct domain boundary coverage to 60%, whereas the lowest five local minima cover 79% of correct domain boundaries. Through further profile analysis, criteria are presented which reliably identify subsets of more accurate predictions. Retrospective analysis of CASP3 targets shows predictions of sufficient accuracy to enable dramatically improved fold recognition results. Finally, a prediction is made for geminivirus AL1 protein which is in full agreement with biochemical data, yielding a plausible, novel threading result.

CROSS-REFERENCE MAXIMUM LIKELIHOOD ESTIMATE RECONSTRUCTION FOR POSITRON EMISSION TOMOGRAPHY

Maximum likelihood estimate (MLE) is a widely used approach for PET image reconstruction. However, it has been shown that reconstructing emission tomography based on MLE without regularization would result in noise and edge artifacts. In the attempt to regularize the maximum likelihood estimate, we propose a new and efficient method in this paper to incorporate the correlated but possibly incomplete structure information which may be derived from expertise, PET systems or other imaging modalities. A mean estimate smoothing the MLE locally within each region of interest is derived according to the boundaries provided by the structure information. Since the boundaries may not be correct, a penalized MLE using the mean estimate is sought. The resulting reconstruction is called a cross-reference maximum likelihood estimate (CRMLE). The CRMLE can be obtained through a modified EM algorithm, which is computation and storage efficient. By borrowing the strength from the correct portion of boundary information, the CRMLE is able to extract the useful information to improve reconstruction for different kinds of incomplete and incorrect boundaries in Monte Carlo studies. The proposed CRMLE algorithm not only reduces the estimation errors, but also preserves the correct boundaries. The penalty parameters can be selected through human interactions or automatically data-driven methods, such as the generalized cross validation method.

Soft image segmentation by weighted linked pyramid

This paper introduces the notion of soft segmentation and the use of a weighted linked pyramid algorithm to compute it. A fundamental problem of region segmentation is that for many images no unique, correct region boundaries can be determined. In this situation, the normal desire for a definitive crisp segmentation may be unreasonable. It is proposed that soft segmentation is a more natural way to segment digital image data than crisp segmentation. One method of deriving a soft segmentation from a weighted linked pyramid is shown. We introduce methods for comparing a soft segmentation with a crisp one and show that in some ways the soft segmentation is better.

Excess Enthalpies (HE) for 2-Butoxyethanol/Water and Determination of Liquid-Liquid Equilibrium Phase Boundaries from Fits to HE of Redlich-Kister, Pade, and Critical-Scaling Equations.

Excess enthalpies (HE) for the binary n-butanol/water measured by isothermal flow calorimetry at 30 and 50°C, and nonionic amphiphile 2-butoxyethanol/water at 10 different temperatures (from 48.5 to 70°C) reported in the literature were analyzed for the determination of phase boundaries. HE exhibited S-shape behavior in the former system and U-shape behavior in the latter system. When the HE data in the single-phase were fitted by semi-empirical polynomials and critical-scaling equations, plots of specific HE vs. weight fraction provided more accurate fitting with fewer parameters than conventionally drawn molar HE vs. mole fraction plots. This was due to the enhanced symmetry of specific HE vs. weight fraction plots. Liquid-liquid equilibrium phase boundaries between the single- and two-phase regions were determined from HE. The phase boundary points were obtained as the intersections of the curve and the straight line, which describe the composition dependence of HE for the single- and two-phase regions, respectively. When Redlich-Kister (RK) and Pade polynomials were employed, the phase boundary points could be determined for the n-butanol/water but not for the 2-butoxyethanol/water. These results imply that correct phase boundaries may not be obtained with the semi-empirical polynomials when HE behavior is of the U-type. However, when the critical-scaling equations for HE of binary mixtures were used, the phase boundaries were obtained accurately, irrespective of the type of HE data.

A recurrent neural fuzzy network for word boundary detection in variable noise-level environments

This paper discusses the problem of automatic word boundary detection in the presence of variable-level background noise. Commonly used robust word boundary detection algorithms always assume that the background noise level is fixed. In fact, the background noise level may vary during the procedure of recording. This is the major reason that most robust word boundary detection algorithms cannot work well in the condition of variable background noise level. In order to solve this problem, we first propose a refined time-frequency (RTF) parameter for extracting both the time and frequency features of noisy speech signals. The RTF parameter extends the (time-frequency) TF parameter proposed by Junqua et al. from single band to multiband spectrum analysis, where the frequency bands help to make the distinction between speech signal and noise clear. The RTF parameter can extract useful frequency information. Based on this RTF parameter, we further propose a new word boundary detection algorithm by using a recurrent self-organizing neural fuzzy inference network (RSONFIN). Since RSONPIN can process the temporal relations, the proposed RTF-based RSONFIN algorithm can find the variation of the background noise level and detect correct word boundaries in the condition of variable background noise level. As compared to normal neural networks, the RSONFIN can always find itself an economic network size with high-learning speed. Due to the self-learning ability of RSONFIN, this RTF-based RSONFIN algorithm avoids the need for empirically determining ambiguous decision rules in normal word boundary detection algorithms. Experimental results show that this new algorithm achieves higher recognition rate than the TF-based algorithm which has been shown to outperform several commonly used word boundary detection algorithms by about 12% in variable background noise level condition, It also reduces the recognition error rate due to endpoint detection to about 23%, compared to an average of 47% obtained by the TF-based algorithm in the same condition.

Numerical validation of a Eulerian hydrochemical code using a 1D multisolute mass transport system involving heterogeneous kinetically controlled reactions

It is demonstrated that at steady state, the 1D thermo-kinetic hydrochemical Eulerian mass balance equations in pure advective mode are indeed identical to the governing mass balance equations of a single reaction path (or geochemical) code in open system mode. Thus, both calculated reaction paths should be theoretically identical whatever the chemical complexity of the water–rock system (i.e., multicomponent, multireaction zones kinetically and equilibrium-controlled). We propose to use this property to numerically test the thermo-kinetic hydrochemical Eulerian codes and we employ it to verify the algorithm of the 1D finite difference code KIRMAT. Compared to the other methods to perform such numerical tests (i.e., comparisons with analytical, semi-analytical solutions, between two Eulerian hydrochemical codes), the advantage of this new method is the absence of constraints on the chemical complexity of the modelled water–rock systems. Moreover, the same thermo-kinetic databases and geochemical functions can be easily and mechanically used in both calculations, when the numerical reference comes from the Eulerian code with no transport terms ( u and D=0) and modify to be consistent with the definition of the open system mode in geochemical modelling. The ability of KIRMAT to treat multicomponent pure advective transport, subjected to several kinetically equilibrium-controlled dissolution and precipitation reactions, and to track their boundaries has been successfully verified with the property of interest. The required numerical validation of the reference calculations is bypassed in developing the Eulerian code from an already checked single reaction path code. A forward time-upstream weighting scheme (a mixing cell scheme) is used in this study. An appropriate choice of grid spacing allows to calculate within the grid size uncertainty the correct mineral reaction zone boundaries, despite the presence of numerical dispersion. Its correction enables us to improve the convergence and to extend the numerical test to mixed advective–dispersive mass transport. However, the skewness factor involves numerical oscillations that prevent to compute different grid spacing. The use of a different chemically controlled time step constraint in both calculations induces some inconsistencies into the validation tests. This numerical validation method may be applied as well as to check a thermo-kinetic hydrochemical finite element based code, from a 1D heterogeneous systems, and 2D–3D systems provided that they are designed so as to be 1D equivalent. A one-step algorithm and the use of a numerical reference coming from the Eulerian code to be tested ensure the potential success (accuracy) of the numerical validation method.

Selectivity, sharing and competitive interactions in the regulation of Hoxb genes.

The clustered organisation of Hox complexes is highly conserved in vertebrates and the reasons for this are believed to be linked with the regulatory mechanisms governing their expression. In analysis of the Hoxb4-Hoxb6 region of the HoxB complex we identified enhancers which lie in the intergenic region between Hoxb4 and Hoxb5, and which are capable of mediating the correct boundaries of neural and mesodermal expression for Hoxb5. We examined their regulatory properties in the context of the local genomic region spanning the two genes by transgenic analysis, in which each promoter was independently marked with a different reporter, to monitor simultaneously the relative transcriptional read-outs from each gene. Our analysis revealed that within this intergenic region: (i) a limb and a neural enhancer selectively activate Hoxb4 as opposed to Hoxb5; (ii) a separate neural enhancer is able to activate both genes, but expression is dependent upon competition between the two promoters for the enhancer and is influenced by the local genomic context; (iii) mesodermal enhancer activities can be shared between the genes. We found similar types of regulatory interactions between Hoxb5 and Hoxb6. Together these results provide evidence for three separate general mechanisms: selectivity, competition and sharing, that control the balance of cis-regulatory interactions necessary for generating the proper spatial and temporal patterns of Hox gene expression. We suggest that these mechanisms are part of a regulatory basis for maintenance of Hox organisation.

Regulation of POU genes by castor and hunchback establishes layered compartments in the Drosophila CNS.

POU transcription factors participate in cell-identity decisions during nervous system development, yet little is known about the regulatory networks controlling their expression. We report all known Drosophila POU genes require castor (cas) for correct CNS expression. drifter and I-POU depend on cas for full expression, whereas pdm-1 and pdm-2 are negatively regulated. cas encodes a zinc finger protein that shares DNA-binding specificity with another pdm repressor: the gap segmentation gene regulator Hunchback (Hb). Our studies reveal that the embryonic CNS contains sequentially generated neuroblast sublineages that can be distinguished by their expression of either Hb, Pdm-1, or Cas. Hb and Cas may directly silence pdm expression in early and late developing sublineages, given that pdm-1 cis-regulatory DNA contains >=32 Hb/Cas-binding sites and its enhancer(s) are ectopically activated in cas- neuroblasts. In addition, the targeted misexpression of Cas in all neuroblast lineages reduces Pdm-1 expression without altering Hb expression. By ensuring correct POU gene expression boundaries, hb and cas maintain temporal subdivisions in the cell-identity circuitry controlling CNS development.

Automatic detection of accentual phrase boundaries using prosodic features and phoneme boundaries

There are two approaches for constructing an appropriate fundamental frequency (F0) control method for speech synthesis: statistical and rule-based. The statistical approach has the advantage of automatic training, but it requires a large corpora of speech that is annotated with prosodic boundaries. Recently, a method is proposed for high-accuracy detection of these boundaries [Ostendorf and Ross (1996)], given a set of prosodic boundary candidates in which almost all the correct boundaries are included. This paper proposes a detection method to generate these boundary candidates, specifically for accentual phrases which represent one of the smallest prosodic units. The detection algorithm uses local maximums and minimums of the F0 contour and low-energy regions of the speech waveform for finding candidate regions that correspond to accentual phrases and pauses in speech. The candidate phrase boundaries are then aligned to the nearest phoneme boundaries, which are detected automatically using forced alignment with a speaker-independent speech recognition system given a phoneme transcription. This method was applied to 250 read Japanese sentences. High-detection accuracy (97%) was obtained, with almost all the missed detections having valid candidates within ±3 phonemes. The insertion error rate was less than double the number of correct boundaries.

A variable threshold edge-detector for improved quantitation of gated tomographic imaging of the left ventricular blood pool

The accurate measurement of left ventricular volume from tomographic MUGA studies is difficult due to the limited resolving power of the gamma camera, which causes errors in the detection of the correct ventricular boundaries. Therefore, the use of fixed threshold or second-derivative edge-detectors results in overestimates at small volumes. A variable threshold edge-detection technique was developed to overcome this. Computer-simulated short-axis slices through the heart over a range of left ventricular dimensions were convolved by the Point Spread Response Function of the system to model the acquired image. The maximum pixel value and the threshold value required to detect the true ventricular edge from each simulation were then combined into a look-up table for the calculation of the required threshold value. As the dimension of the ventricle decreased, the threshold value chosen to detect the ventricular edge increased. Left ventricular volumes and ejection fraction measurements were calculated for seven patients using cine-MRI as the gold-standard technique for validation of the proposed method. The single photon emission tomographic studies were analysed using both the standard second-derivative edge-detection software and the proposed variable threshold technique. The variable threshold technique was shown to increase significantly the accuracy of ventricular volume measurements and ejection fraction calculations. The average error in the measurement of volumes was reduced from 41.4 +/- 45.1% to 18.5 +/- 14.6% and the accuracy of ejection fraction measurement was increased from 29.7 +/- 4.6% to 11.3 +/- 6.9%.

Diabetes susceptibility at IDDM2 cannot be positively mapped to the VNTR locus of the insulin gene.

An inconsistency has come to light between the conclusion of Lucassen et al. that IDDM2 (11p15.5) must lie within a 4.1 kilobase (kb) segment at the insulin (INS) locus and their own data showing statistically significant associations between insulin-dependent diabetes mellitus (IDDM) and markers beyond the boundaries of that segment. We present data from an independent study of 201 IDDM patients and 107 non-diabetic control subjects that also show significant association with a marker 5' of the INS locus. Patients and control subjects were genotyped at INS/+ 1140 A/C (a surrogate for the variable number tandem repeat (VNTR) polymorphism in the regulatory part of the INS gene) and a marker 5' of the tyrosine hydroxylase (TH) gene, TH/pINS500-RsaI, making it 10 kb 5' of the VNTR. Homozygotes for INS/ + 1140 allele '+' were significantly more frequent among IDDM patients than among control subjects (73 vs 45%, p < 0.001) giving an odds ratio of 3.3 (95% confidence interval (CI): 2.0-5.3). A very similar association was found for homozygotes for the TH/RsaI allele '+' (53 vs 31%, p < 0.001) giving an odds ratio of 2.6 (95%CI 1.6-4.2). By multilocus analysis, the TH/RsaI allele '+' identified a subset of INS/ + 1140 alleles '+' haplotypes that are more specifically associated with IDDM (odds ratio = 5.4, 95%CI 2.9-10.4) than allele + 1140 '+' as a whole. In conclusion, the segment of chromosome 11 that is associated with IDDM spans, at least, the INS and TH loci. No legitimate claim can be made that IDDM2 corresponds to the VNTR polymorphism at the INS locus until the correct boundaries for IDDM2 have been defined and other loci within them have been excluded as determinants of IDDM.

Individual recognition of song repertoires in two wood warblers

We examined recognition of individual neighbors in hooded warblers (Wilsonia citrina) and Kentucky warblers (Oporornis formosus) by comparing responses to playbacks of neighbors' songs from appropriate and inappropriate boundaries of the subjects' territories. Kentucky warblers, which sing one song type each, responded more strongly to the songs of neighbors broadcast from incorrect than from correct boundaries. Hooded warblers use their repertoires of three to nine song types in two distinct modes of singing: repeat mode involves repeated presentation of one song type; mixed mode involves presentation of two or more song types in irregular sequence. Playbacks of neighbors' mixed-mode songs from appropriate and inappropriate boundaries indicated capabilities of individual recognition similar to those reported previously for repeat-mode songs (Godard 1991) and to those of Kentucky warblers. Repertoires of moderate size, therefore, have no pronounced influence on eventual recognition of individual neighbors in hooded warblers.

Counterpropagation networks applied to the classification of alkanes through infrared spectra

Enhanced uni-flow counterpropagation networks are used as pattern recognition systems and applied to the identification of chemical structure from corresponding infrared spectra. It is shown that such networks are more suitable for this type of problem than backpropagation networks, both in terms of training times and network performance. The problem of optimum classification between highly similar infrared spectra is addressed, and factors such as training set size, sampling rate, data pre-processing, output data representation and the number of Kohonen layer nodes are considered in this context. It is shown that such networks may achieve rates of correct classification in excess of 90%, although the learning of correct decision boundaries is highly sensitive to the above parameters in cases where the non-informational content of training and test data varies considerably with respect to the informational content, and hence clustering of classes in pattern space is incomplete.

ERRATUM

"Low-Level Lead Exposure and Children's Cognitive Function in the Preschool Years" by Bellinger et al (Pediatrics. 1991:87:219-227) has an error in Fig 2 on page 225. The error bars for the mean change in score for children in the "low" lead group should cover an interval that is approximately 50% larger than that indicated. The correct boundaries for mean ± SE are 12.3 and 1.2 points.

Multiple spatially specific enhancers are required to reconstruct the pattern of Hox-2.6 gene expression.

Murine Hox genes are organized into four clusters that share many features with the homeotic clusters of Drosophila. This evolutionary conservation and the clear relationships between the position of a gene within a cluster and its expression pattern have led to the suggestion that the structure of the cluster is essential for proper regulation. Using a Hox-2.6-lacZ reporter gene in transgenic mice we have shown that the overall expression pattern of the endogenous Hox-2.6 gene can be reconstructed when it is isolated from the complex. The transgene was expressed in the proper tissues, with the correct spatial distribution and temporal pattern. Furthermore, direct comparison by in situ hybridization revealed that the levels of transgene expression are similar to those of the endogenous gene. This has allowed us to define three elements that regulate particular aspects of the Hox-2.6 pattern, two of which act as spatially specific enhancers. One enhancer, region A, directed expression only in the neural tube, whereas the other, region C, specified the majority of the Hox-2.6 pattern. Both were also capable of imposing the correct boundaries of expression on heterologous promoters. The definition of such elements will allow the characterization of the trans-acting factors that mediate spatial regulation in the mammalian embryo.

Predictability Experiments Using a High-Resolution Limited-Area Model

Abstract Recently reported results indicate that limited-area mesoscale models with prescribed lateral boundaries do not exhibit the same predictability error growth as observed in large-scale (global) models. These results have been reanalyzed in greater detail. New methods of limited-area initialization and spectral analysis have been used. The new analyses indicate that several model properties act to restrict the growth of perturbation. These include: the Projection of initial perturbations onto gravity waves which interact only weakly with other, more significant motions; the “sweeping out” of errors by correct or perfect lateral boundaries; and the reduction of differences by subgrid dissipation. This last property suggests that there is a strong dynamical forcing of small scales by much larger scales, so that this forcing is only weakly affected by typical, small perturbations in this model. New experiments suggest that some quasi-geostrophic components of the forecasts, away from the inflow bounda...