Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Feature selection”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 847 records · Page 47Linked to original sources
PubMed1997

T cell receptor biochemistry, repertoire selection and general features of TCR and Ig structure.

T cell recognition is a central event in the development of most immune responses, whether appropriate or inappropriate (i.e. autoimmune). We are interested in reducing T cell recognition to its most elemental components and relating this to biological outcome. In a model system involving a cytochrome c-specific I-Ek restricted T cell receptor (TCR) derived from the 2B4 hybridoma, we have studied the interaction of soluble TCR and soluble peptide-MHC complexes using surface plasmon resonance. We find a striking continuum in which biological activity correlates best with the dissociation rate of the TCR from the peptide-MHC complex. In particular, we have found that weak agonists have significantly faster off-rates than strong agonists and that antagonists have even faster off-rates. This suggests that the stability of TCR binding to a given ligand is critically important with respect to whether the T cell is stimulated, inhibited or remains indifferent. It also suggests that the phenomenon of peptide antagonists might be explained purely by kinetic models and that conformation, either inter- or intramolecular, may not be a factor. We have also studied TCR repertoire selection during the establishment of a cytochrome c response, initially using an anti-TCR antibody strategy, but more recently using peptide-MHC tetramers as antigen-specific staining reagents. These tetramers work well with either class I or class II MHC-specific TCRs and have many possible applications. Lastly, we have also tried to correlate the structural and genetic features of TCRs with their function. Recent data on TCR structure as well as previous findings with antibodies suggest that both molecules are highly dependent on CDR3 length and sequence variation to form specific contacts with antigens. This suggests a general "logic' behind TCR and Ig genetics as it relates to structure and function that helps to explain certain anomalous findings and makes a number of clear predictions.

Animals↗
PubMed2001

Selection of valid and reliable EEG features for predicting auditory and visual alertness levels.

A selection procedure with three rules, high efficiency, low individual variability, and low redundancy, was developed to screen electroencephalogram (EEG) features for predicting behavioral alertness levels. A total of 24 EEG features were derived from temporal, frequency spectral, and statistical analyses. Behavioral alertness levels were quantified by correct rates of performance on an auditory and a visual vigilance task, separately. In the auditory task study, a subset of three EEG features, the relative spectral amplitudes in the alpha (alpha%, 8-13 Hz) and theta (theta%, 4-8 Hz) bands, and the mean frequency of the EEG spectrum (MF), was found to be the best combination for predicting the auditory alertness level. In the visual task study, the mean frequency of the beta band (Fbeta, 13-32 Hz) was the only EEG feature selected. The application of an averaging subwindow procedure within a moving time window to EEG analysis increased the predictive power of EEG features and decreased the disturbing effect of movement artifacts on the EEG data.

Acoustic Stimulation↗
PubMed1992

Interclinician agreement on the recognition of selected gross morphologic features of pigmented lesions. Studies of melanocytic nevi V.

BACKGROUND: Since the late 1970s clinical criteria for dysplastic melanocytic nevi (DMN) have been proposed and discussed. However, to our knowledge, no rigorous quantitative evaluation of the ability of examiners to agree on the gross morphologic features ascribed to DMN or atypical melanocytic lesions in general has been conducted. OBJECTIVE: The purpose of the study was to determine rates of interclinician agreement for recognizing seven clinical features associated with melanocytic lesions. METHODS: The gross morphologic features of 156 pigmented lesions, judged to be the clinically most atypical, from 156 consecutively examined patients with cutaneous melanoma were analyzed. Independent of the other clinicians' examinations, four physicians (two medical oncologists, one internist-epidemiologist, and one dermatologist-dermatopathologist) recorded seven gross morphologic features of the most atypical pigmented lesion on each patient. For up to 122 patients, the rates of interobserver agreement on recognizing these clinical features were measured by intraclass correlation. RESULTS: Among the clinical features assessed, the examiners noted macular components in 85.7% to 96.3% of lesions, asymmetry in 47.2% to 79.5%, irregular borders in 55.1% to 80.2%, ill-defined borders in 40.1% to 83.9%, and haphazard color in 40.8% to 73.9% of lesions. Despite this range of variance among examiners, there were statistically significant rates of interclinician agreement for the recognition of the latter features. There was less agreement in assessing individual colors within lesions. CONCLUSION: These findings substantiate that certain gross morphologic features routinely used in the clinical evaluation of melanocytic lesions can be recognized with a significant degree of reliability.

Adult↗
PubMed1995

Structural features mediating fibrin selectivity of vampire bat plasminogen activators.

The distinguishing characteristic of vampire bat (Desmodus rotundus) salivary plasminogen activators (DSPAs) is their strict requirement for fibrin as a cofactor. DSPAs consist of structural modules known from urokinase (u-PA) and tissue-type plasminogen activator (t-PA) such as finger (F), epidermal growth factor (E), kringle (K), and protease (P), combining to four genetically and biochemically distinct isoenzymes, exhibiting the formulas FEKP (DSPA alpha 1 and alpha 2) and EKP and KP (DSPA beta and DSPA gamma). Only DSPA alpha 1 and alpha 2 bind to fibrin. All DSPAs are single-chain molecules, displaying substantial amidolytic activity. In a plasminogen activation assay, all four DSPAs are almost inactive in the absence of fibrin but strongly stimulated by fibrin addition. The catalytic efficiency (kcat/Km) of DSPA alpha 1 increases 10(5)-fold, whereas the corresponding value of t-PA is only 550. The ratio of the bimolecular rate constants of plasminogen activation in the presence of fibrin versus fibrinogen (fibrin selectivity) of DSPA alpha 1, alpha 2, beta, gamma, and t-PA was found to be 13,000, 6500, 250, 90, and 72, respectively. Whereas all DSPAs are therefore more fibrin dependent and fibrin selective than t-PA, the extent depends on the respective presence of the various domains. The introduction of a plasmin-sensitive cleavage site in a position akin to the one in t-PA partially obliterates fibrin cofactor requirement. Fibrin dependence and fibrin selectivity of DSPAs are accordingly mediated by fibrin binding, which involves the F domain, as yet undefined determinants within the K and P domains, and by the absence of a plasmin-sensitive activation site. These findings transcend the current understanding of fibrin-mediated stimulation of plasminogen activation: in addition to fibrin binding, specific protein-protein interactions come into play, which stabilize the enzyme in its active conformation.

Aminocaproic Acid↗
PubMed1994

Computer-aided diagnosis in mammography: classification of mass and normal tissue by texture analysis.

Computer-aided diagnosis schemes are being developed to assist radiologists in mammographic interpretation. In this study, we investigated whether texture features could be used to distinguish between mass and non-mass regions in clinical mammograms. Forty-five regions of interest (ROIs) containing true masses with various degrees of visibility and 135 ROIs containing normal breast parenchyma were extracted manually from digitized mammograms as case samples. Spatial-grey-level-dependence (SGLD) matrices of each ROI were calculated and eight texture features were calculated from the SGLD matrices. The correlation and class-distance properties of extracted texture features were analysed. Selected texture features were input into a modified decision-tree classification scheme. The performance of the classifier was evaluated for different feature combinations and orders of features on the tree. A classification accuracy of about 89% sensitivity and 76% specificity was obtained for ordered features, sum average, correlation, and energy, during the training procedure. With a leave-one-out method, the test result was about 76% sensitivity and 64% specificity. The results of this preliminary study demonstrate the feasibility of using texture information for classification of mass and normal breast tissue, which will be likely to be useful for classifying true and false detections in computer-aided diagnosis programmes.

Algorithms↗
PubMed1994

TBP-DNA interactions in the minor groove discriminate between A:T and T:A base pairs.

In this report, we test the hypothesis that TBP binds DNA promiscuously due to its manner of recognition of the DNA minor groove. The experiment performed was to select TBP-binding sequences from a pool of random double stranded oligonucleotides. Sixty two clones from this pool were sequenced. Surprisingly, the results show that TBP has a marked preference for stably binding one sequence (TATATAA) over all others, yet only four classes of TATA box were selected. The features of the selected sequences allow definition of a binding consensus for TBP. The DNA binding properties of TBP to the four TATA variants was examined, the results being in accord with the observed selection frequencies. However, the nature of TBP-DNA binding is strongly affected by ionic strength. We infer that recognition of DNA via the minor groove can be highly selective even where A:T and T:A discrimination is required. Models for how this might be accomplished are discussed.

Base Composition↗
PubMed2002

Useful clinical features for the selection of ideal patients with atrial fibrillation for mapping and catheter ablation.

OBJECTIVE: To identify useful clinical characteristics for selecting patients eligible for mapping and ablation of atrial fibrillation. METHODS: We studied 9 patients with atrial fibrillation, without structural heart disease, associated with: 1) antiarrhythmic drugs, 2) symptoms of low cardiac output, and 3) intention to treat. Seven patients had paroxysmal atrial fibrillation and 2 had recurrent atrial fibrillation. RESULTS: In the 6 patients who underwent mapping (all had paroxysmal atrial fibrillation), catheter ablation was successfully carried out in superior pulmonary veins in 5 patients (the first 3 in the left superior pulmonary vein and the last 2 in the right superior pulmonary vein). One patient experienced a recurrence of atrial fibrillation after 10 days. We observed that patients who had short episodes of atrial fibrillation on 24-hour Holter monitoring before the procedure were those in whom mapping the focus of tachycardia was possible. Tachycardia was successfully suppressed in 4 of 6 patients. The cause of failure was due to the impossibility of maintaining sinus rhythm long enough for efficient mapping. CONCLUSION: Patients experiencing short episodes of atrial fibrillation during 24-hour Holter monitoring were the most eligible for mapping and ablation, with a final success rate of 66%, versus the global success rate of 44%. Patients with persistent atrial fibrillation were not good candidates for focal ablation.

Adult↗
PubMed2006

Discriminating benign from malignant thyroid lesions using artificial intelligence and statistical selection of morphometric features.

The objective of this study was to perform a comparative investigation of the capability of various classifiers in discriminating benign from malignant thyroid lesions. Using May Grunvald-Giemsa-stained smears taken by fine needle aspiration (FNA) and a custom image analysis system, 25 nuclear features describing the size, shape and texture of the nuclei were measured in each case. A statistical pre-processing of features revealed that only 4 of the 25 features are important when discriminating benign from malignant thyroid lesions, which were transformed and fed to four classifiers for subsequent analysis. The cases were divided into one set used for the training of classifiers, a second set used as the test set, and the remaining cases with no clear classification formed an ambiguous test set. Classification was performed at the nuclear and patient level. The technique described in this study produced encouraging results and promises to be a helpful tool in the daily cytological laboratory routine.

Anthropometry↗
PubMed1992

[Prognostic value of selected hemodynamic parameters featuring left ventricular systolic and diastolic function in patients with dilated cardiomyopathy].

UNLABELLED: The goal of the study was the choice of hemodynamic parameters most useful in prognosing the survival of the patients with dilated cardiomyopathy (DCM). The investigated group comprised 40 patients, who underwent left and right cardiac catheterization with left ventricular (LV) quantitative angiography. The day of catheterization was the starting point of the observation, which was performed by means of follow-up regular examinations and information from questionnaires sent to and returned by the patients or their families. The follow-up spanned from 1 to 120 months (means = 43). The hemodynamic parameters of LV systole, ejection, isovolumic relaxation and filling were assessed. The patients were divided into two subgroups featuring survival of less and more than 3 years, respectively. In these two subgroups the following parameters were compared: LVEADP, T constant of relaxation, LVEDP, LVEDVI,+dp/dt max, EF, LV mass/volume ratio, MCSD (midwall circumferential LV stress in early diastole), -dp/dt min. They were also correlated with survival and analyzed in Cutler-Ederer survival tables. CONCLUSIONS: 1. 32% of patients survived more than 5 years. 2. The prognostically most valuable parameters seem to be: -dp/dt min, LVEADP, LVEDVI, EF, LVEDP, MCSD, +dp/dt max. 3. The prognosing of survival in patients with DCM should be based on a multifactorial analysis because of lack of domination of any single factor.

Adolescent↗
PubMed1994

Selection of morphometric features that identify responders and nonresponders in stage IV Wilms' tumors.

Wilms' tumor is the most common renal malignancy of childhood. The use of histologic grade and stage has divided those patients into two main groups, responders and nonresponders, based on the absence or presence of extreme cytologic atypia (anaplasia). Patients with Wilms' tumor who have favorable histology experience excellent cure rates with relatively conservative treatment regimens. However, some patients with a favorable histologic diagnosis die of the disease. Destained hematoxylin and eosin-stained tissue sections were restained stoichiometrically with Feulgen stain, and 100 tumor cells per case were measured on a CAS 200 running CMP software. Using multivariant stepwise discriminant analysis on 13 cases of stage IV, favorable-histology Wilms' tumor, we successfully classified 92% of individuals into their correct prognostic groupings using 10 features of the quantitative morphology of tumor cells, including six Markovian textures. This morphometric technique may identify patients who can benefit from reduced therapy and those who must be treated with the more aggressive, classic therapy regimens.

Cell Nucleus↗
PubMed2004

Computer-aided diagnosis of mammographic microcalcification clusters.

Computer-aided diagnosis techniques in medical imaging are developed for the automated differentiation between benign and malignant lesions and go beyond computer-aided detection by providing cancer likelihood for a detected lesion given image and/or patient characteristics. The goal of this study was the development and evaluation of a computer-aided detection and diagnosis algorithm for mammographic calcification clusters. The emphasis was on the diagnostic component, although the algorithm included automated detection, segmentation, and classification steps based on wavelet filters and artificial neural networks. Classification features were selected primarily from descriptors of the morphology of the individual calcifications and the distribution of the cluster. Thirteen such descriptors were selected and, combined with patient's age, were given as inputs to the network. The features were ranked and evaluated for the classification of 100 high-resolution, digitized mammograms containing biopsy-proven, benign and malignant calcification clusters. The classification performance of the algorithm reached a 100% sensitivity for a specificity of 85% (receiver operating characteristic area index Az = 0.98 +/- 0.01). Tests of the algorithm under various conditions showed that the selected features were robust morphological and distributional descriptors, relatively insensitive to segmentation and detection errors such as false positive signals. The algorithm could exceed the performance of a similar visual analysis system that was used as basis for development and, combined with a simple image standardization process, could be applied to images from different imaging systems and film digitizers with similar sensitivity and specificity rates.

Algorithms↗