Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Classification Algorithms”

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 559 records · Page 31Linked to original sources
PubMed2005

On the use of multi-objective evolutionary algorithms for the induction of fuzzy classification rule systems.

Extracting comprehensible and general classifiers from data in the form of rule systems is an important task in many problem domains. This study investigates the utility of a multi-objective evolutionary algorithm (MOEA) for this task. Multi-objective evolutionary algorithms are capable of finding several trade-off solutions between different objectives in a single run. In the context of the present study, the objectives to be optimised are the complexity of the rule systems, and their fit to the data. Complex rule systems are required to fit the data well. However, overly complex rule systems often generalise poorly on new data. In addition they tend to be incomprehensible. It is, therefore, important to obtain trade-off solutions that achieve the best possible fit to the data with the lowest possible complexity. The rule systems produced by the proposed multi-objective evolutionary algorithm are compared with those produced by several other existing approaches for a number of benchmark datasets. It is shown that the algorithm produces less complex classifiers that perform well on unseen data.

Algorithms↗
PubMed2002

A Bayesian method for classification of images from electron micrographs.

Particle classification is an important component of multivariate statistical analysis methods that has been used extensively to extract information from electron micrographs of single particles. Here we describe a new Bayesian Gibbs sampling algorithm for the classification of such images. This algorithm, which is applied after dimension reduction by correspondence analysis or by principal components analysis, dynamically learns the parameters of the multivariate Gaussian distributions that characterize each class. These distributions describe tilted ellipsoidal clusters that adaptively adjust shape to capture differences in the variances of factors and the correlations of factors within classes. A novel Bayesian procedure to objectively select factors for inclusion in the classification models is a component of this procedure. A comparison of this algorithm with hierarchical ascendant classification of simulated data sets shows improved classification over a broad range of signal-to-noise ratios.

Algorithms↗
PubMed2004

Classification of single-trial electroencephalogram during finger movement.

We present an algorithm to discriminate between the single-trial electroencephalograms (EEG) of two different finger movement tasks. The method uses a spatio-temporal analysis to classify the EEG recorded during voluntary left versus right finger movement tasks. This algorithm produced a classification accuracy of 92.1% on the data from five subjects, without requiring subject training or data selection. This technique can be employed in an EEG-based brain-computer interface due to its high recognition rate, insensitivity to noise, and simplicity in computation.

Adult↗
PubMed2005

The URMS-RMS hybrid algorithm for fast and sensitive local protein structure alignment.

We present an efficient and sensitive hybrid algorithm for local structure alignment of a pair of 3D protein structures. The hybrid algorithm employs both the URMS (unit-vector root mean squared) metric and the RMS metric. Our algorithm searches efficiently the transformation space using a fast screening protocol; initial transformations (rotations) are identified using the URMS algorithm. These rotations are then clustered and an RMS-based dynamic programming algorithm is invoked to find the maximal local similarities for representative rotations of the clusters. Statistical significance of the alignments is estimated using a model that accounts for both the score of the match and the RMS. We tested our algorithm over the SCOP classification of protein domains. Our algorithm performs very well; its main advantages are that (1) it combines the advantages of the RMS and the URMS metrics, (2) it searches extensively the transformation space, (3) it detects complex similarities and structural repeats, and (4) its results are symmetric. The software is available for download at biozon.org/ftp/software/urms/.

Algorithms↗
PubMed2001

Scene illuminant classification: brighter is better.

Knowledge of the scene illuminant spectral power distribution is useful for many imaging applications, such as color image reproduction and automatic algorithms for image database applications. In many applications accurate spectral characterization of the illuminant is impossible because the input device acquires only three spectral samples. In such applications it is sensible to set a more limited objective of classifying the illuminant as belonging to one of several likely types. We describe a data set of natural images with measured illuminants for testing illuminant classification algorithms. One simple type of algorithm is described and evaluated by using the new data set. The empirical measurements show that illuminant information is more reliable in bright regions than in dark regions. Theoretical predictions of the algorithm's classification performance with respect to scene illuminant blackbody color temperature are tested and confirmed by using the natural-image data set.

Journal Article↗
PubMed2001

A new classification of palatal fracture and an algorithm to establish a treatment plan.

Palatal fractures have previously been classified according to the anatomic location of the fracture line, which is helpful for understanding the types of palatal fracture, but which is insufficient for helping the surgeon to decide which fracture to open and how to do so. The purpose of this study was to aid in the establishment of a precise treatment plan by determining the surgical approach and the types of stabilization that should be used for different types of palatal fracture. In a retrospective review of 136 consecutive Le Fort maxillary fractures over 6 years, 18 patients (13.2 percent) with palatal fractures were analyzed. The principle of open reduction and internal fixation was applied to all the patients. In six patients (33 percent), exploration and fixation was done in the palatal surface. Eight patients (44 percent) needed an extended period of immobilization (4 to 6 weeks). No major complications were observed during the follow-up period. An algorithm was devised to help establish a proper treatment plan, and palatal fractures were classified into four types: closed reduction, anterior treatment, anterior and palatal treatment, and combined. The key elements considered in deciding the treatment principle and the classification of a palatal fracture were the possibility of closed reduction, surgical exposure, site of rigid fixation, and stability of fractured segments after rigid fixation. The outcome of reconstruction and the postoperative course differed depending on the type of palatal fracture. This classification scheme provided an easy and simple way to establish a treatment plan and was helpful in learning the treatment principles of palatal fracture.

Adult↗
PubMed1998

Identification of lung regions in chest radiographs using Markov random field modeling.

The authors present an algorithm utilizing Markov random field modeling for identifying lung regions in a digitized chest radiograph (DCR). Let x represent the classifications of each pixel in a DCR as either lung or nonlung. We model x as a realization of a spatially varying Markov random field. This model is developed utilizing spatial and textural information extracted from samples of lung and nonlung region-types in a training set of DCRs. With this model, the technique of Iterated Conditional Modes is used to determine the optimal classification of each pixel in a DCR. The algorithm's ability to identify lung regions is evaluated on a testing set of DCRs. The algorithm performs well yielding a sensitivity of 90.7% +/- 4.4%, a specificity of 97.2% +/- 2.0%, and an accuracy of 94.8% +/- 1.6%. In an attempt to gain insight into the meaning and level of the algorithm's performance numbers, the results are compared to those of some easily implemented classification algorithms.

Algorithms↗
PubMed2000

Detection of atrial arrhythmia for cardiac rhythm management by implantable devices.

Implantable atrial defibrillators (IAD) should provide pacing therapy whenever appropriate (ie, typical atrial flutter) to minimize shock-related patient discomfort. Additionally, IADs should provide diagnostics regarding atrial arrhythmia type and frequency of occurrence to enable improved physician management of atrial arrhythmia. To achieve this, IADs should accurately classify atrial arrhythmia such as atrial fibrillation (AF) and atrial flutter (AFL) This article evaluates the performance of an algorithm, atrial rhythm classification (ARC), designed to classify AF and AFL. The ARC algorithm uses maximum rate, standard deviation, and range of the 12 most recent atrial cycle lengths to plot a point in a three-dimensional space. A decision boundary divides the space into 2 regions--faster/unstable atrial cycle lengths (AF) or slower/stable cycle lengths (AFL). Classifications are made on a sliding window of 12 consecutive cycles until the end of the episode is reached. In this way, continuous episode feedback is provided that can be used to help guide device therapy, measure arrhythmia type and frequency of occurrence. Bipolar (1-cm) electrogram episodes of AF (n = 16) and AFL (n = 7) were acquired from 20 patients and retrospectively analyzed using the ARC algorithm. The sensitivity and specificity in this study was 0.993 and 0.982, respectively. The ARC algorithm would have appropriately guided atrial therapy and minimized discomfort associated with defibrillation shocks in this small patient data set warranting further studies. The ARC algorithm may also be beneficial as a diagnostic tool to assist physician management of atrial arrhythmia.

Algorithms↗
PubMed2000

Generating concise and accurate classification rules for breast cancer diagnosis.

In our previous work, we have presented an algorithm that extracts classification rules from trained neural networks and discussed its application to breast cancer diagnosis. In this paper, we describe how the accuracy of the networks and the accuracy of the rules extracted from them can be improved by a simple pre-processing of the data. Data pre-processing involves selecting the relevant input attributes and removing those samples with missing attribute values. The rules generated by our neural network rule extraction algorithm are more concise and accurate than those generated by other rule generating methods reported in the literature.

Algorithms↗
PubMed2002

Rule-based algorithm for automated King-type classification of idiopathic scoliosis.

Recent studies have demonstrated poor reliability of the King et al. classification of idiopathic scoliosis. The purpose of the present work was to determine whether the reliability of the King classification would be improved by employing unambiguous rules for classification encoded in a computer program. Thus the only possible source of variability in classifying a given radiograph would be variable landmark identification on the radiograph. Coordinates of the four comers of each thoracic and lumbar vertebra were obtained by digitizing radiographs, using the central sacral line to define the y-axis. A computer algorithm located curve apices as the most laterally deviated vertebra in a range of two above to two below a change in the sign of vertebral tilt. End vertebrae of each curve were located as those with the greatest tilt of the superior (proximal end vertebra) or inferior (distal end vertebra). In a lumbar curve, the apical vertebra was defined as 'crossing the midline' if all four corners laid on one side of the vertical central sacral line. Curves were defined as in Table I in King et al., except that curve flexibility data were not used. The algorithm was verified on the five examples in King et al. It then classified the six examples in Lenke et al. and Cummings et al. which were classified unreliably by human observers. The algorithm was also tested on 33 radiographs of 17 patients with AIS, 8 radiographs were repeat marked by two observers. The algorithm overcame accuracy and reliability problems, except in rare cases when it was borderline whether or not a lumbar curve crossed the midline, when the apex level was ambiguous, or when a Cobb angle was close to 10 degrees.

Algorithms↗
PubMed2000

Classifying incomplete spinal cord injury syndromes: algorithms based on the International Standards for Neurological and Functional Classification of Spinal Cord Injury Patients.

OBJECTIVE: To develop an objective and uniform means for classifying patients with incomplete spinal cord injury (SCI) according to SCI syndromes. DESIGN: Criteria for assigning the syndromes (defined by the International Standards for Neurological and Functional Classification of SCI Patients) were operationalized by means of sensory and motor scores and were incorporated into a set of six independent algorithms and two composite algorithms. SETTING: A regional SCI rehabilitation center in Canada. PATIENTS: SCI patients (n = 56) with incomplete injuries (American Spinal Injury Association classes B, C, D) and stable neurologic deficits. RESULTS: Individual algorithms allowed the highest classification rate but with some patients meeting the criteria for more than one syndrome. A composite, differential allocation algorithm, with selected thresholds at decision nodes, yielded a classification rate approximating that of the individual algorithms but without double classifications. CONCLUSIONS: The composite algorithm provided an objective and standardized means of assigning patients to syndromes based on clinically measurable sensory and motor scores. The thresholds used to implement criteria and the order of decision nodes greatly influenced the outcomes and may be adjusted to suit the needs of the classification, that is, embracing liberal or stringent criteria. Controversy remains about the interpretation of some syndromes, and many patients remain unclassifiable because of mixed clinical presentation.

Adult↗
PubMed2006

A modified SPIHT algorithm for image coding with a joint MSE and classification distortion measure.

The set partitioning in hierarchical trees (SPIHT) algorithm is an efficient wavelet-based progressive image-compression technique, designed to minimize the mean-squared error (MSE) between the original and decoded imagery. However, the MSE-based distortion measure is not in general well correlated with image-recognition quality, especially at low bit rates. Specifically, low-amplitude wavelet coefficients that may be important for classification are given low priority by conventional SPIHT. In this paper, we use the kernel matching pursuits (KMP) method to autonomously estimate the importance of each wavelet subband for distinguishing between different textures, with textural segmentation first performed via a hidden Markov tree. Based on subband importance determined via KMP, we scale the wavelet coefficients prior to SPIHT coding, with the goal of minimizing a Lagrangian distortion based jointly on the MSE and classification error. For comparison we consider Bayes tree-structured vector quantization (B-TSVQ), also designed to obtain a tradeoff between MSE and classification error. The performances of the original SPIHT, the modified SPIHT, and B-TSVQ are compared.

Algorithms↗
PubMed2003

Functional classification of proteins using a nearest neighbour algorithm.

With the large volume of genomic data being analysed nowadays it becomes extremely important to provide automated ways of protein classification to give the scientist a good overview of the analysed data. The system described here is very flexible and can be used with any given protein classification scheme. Before using the system, it has to be trained with a set of already classified proteins. Afterwards, other proteins can be classified by the system automatically. Several tests have been performed to assess the quality of this classification; they have shown the usefulness of the system. The system will be available as part of a commercial software package.

Algorithms↗
PubMed2003

Molecular classification of cancer types from microarray data using the combination of genetic algorithms and support vector machines.

Simultaneous multiclass classification of tumor types is essential for future clinical implementations of microarray-based cancer diagnosis. In this study, we have combined genetic algorithms (GAs) and all paired support vector machines (SVMs) for multiclass cancer identification. The predictive features have been selected through iterative SVMs/GAs, and recursive feature elimination post-processing steps, leading to a very compact cancer-related predictive gene set. Leave-one-out cross-validations yielded accuracies of 87.93% for the eight-class and 85.19% for the fourteen-class cancer classifications, outperforming the results derived from previously published methods.

Algorithms↗
PubMed2004

[Diagnosis and classification of calcaneal fractures in computed tomography].

PURPOSE: To compare image quality of single-slice spiral CT (SSCT) and multislice spiral CT (MSCT) in the diagnosis and classification of calcaneal fractures and to present a rapid and precise algorithm for the classification system of Stuermer. MATERIALS AND METHODS: In 102 patients with 124 calcaneal fractures, spiral CT was performed, in 82 cases as SSCT with a slice thickness (SD) of 3 mm, a table speed (TS) of 3 - 4 mm/rot and an increment of 1.5 mm. In 42 cases, patients were scanned using MSCT (SD of 1.25 mm, increment 0.8 mm). For these examinations, 2 different scan protocols were used, with a TS of 3.75 mm/rot in one group (n = 21) and a TS of 7.5 mm/rot in the other group (n = 21). The image quality of axial sections and reconstructed images was assessed on a scale from 1 to 5 (1 = very good; 5 = insufficient). The fractures were evaluated using a classification system according to Stuermer, which assigns three main groups (A/B/C) and three subtypes (1/2/3). RESULTS: MSCT had substantial advantages over SSCT with respect to scan time and image quality, especially for multiplanar reformatting (median 1.5 versus 4.0). TS showed no significant influence on the image quality. The standardized evaluation of the images enabled a classification of fractures within 5 minutes. All fractures could be assigned to the different types and subtypes. This classification system takes into account the severity of the fracture and the therapeutic approach. The most common type (90 of 124 fractures) were "joint depression" fractures (type C), which were treated by surgery in 92 %. Type A fractures were treated conservatively in 72 %. CONCLUSION: Spiral CT, especially MSCT, allows rapid diagnosis and precise classification of calcaneal fractures, achieved with high quality multiplanar reformatting. The presented classification in different fracture types and subtypes allows an adequate planning of therapy.

Adolescent↗
PubMed2005

Self-organizing and self-correcting classifications of biological data.

MOTIVATION: Rapid, automated means of organizing biological data are required if we hope to keep abreast of the flood of data emanating from sequencing, microarray and similar high-throughput analyses. Faced with the need to validate the annotation of thousands of sequences and to generate biologically meaningful classifications based on the sequence data, we turned to statistical methods in order to automate these processes. RESULTS: An algorithm for automated classification based on evolutionary distance data was written in S. The algorithm was tested on a dataset of 1436 small subunit ribosomal RNA sequences and was able to classify the sequences according to an extant scheme, use statistical measurements of group membership to detect sequences that were misclassified within this scheme and produce a new classification. In this study, the use of the algorithm to address problems in prokaryotic taxonomy is discussed. AVAILABILITY: S-Plus is available from Insightful, Inc. An S-Plus implementation of the algorithm and the associated data are available at http://taxoweb.mmg.msu.edu/datasets

Algorithms↗
PubMed2006

Eigengene-based linear discriminant model for tumor classification using gene expression microarray data.

MOTIVATION: The nearest shrunken centroids classifier has become a popular algorithm in tumor classification problems using gene expression microarray data. Feature selection is an embedded part of the method to select top-ranking genes based on a univariate distance statistic calculated for each gene individually. The univariate statistics summarize gene expression profiles outside of the gene co-regulation network context, leading to redundant information being included in the selection procedure. RESULTS: We propose an Eigengene-based Linear Discriminant Analysis (ELDA) to address gene selection in a multivariate framework. The algorithm uses a modified rotated Spectral Decomposition (SpD) technique to select 'hub' genes that associate with the most important eigenvectors. Using three benchmark cancer microarray datasets, we show that ELDA selects the most characteristic genes, leading to substantially smaller classifiers than the univariate feature selection based analogues. The resulting de-correlated expression profiles make the gene-wise independence assumption more realistic and applicable for the shrunken centroids classifier and other diagonal linear discriminant type of models. Our algorithm further incorporates a misclassification cost matrix, allowing differential penalization of one type of error over another. In the breast cancer data, we show false negative prognosis can be controlled via a cost-adjusted discriminant function. AVAILABILITY: R code for the ELDA algorithm is available from author upon request.

Artificial Intelligence↗
PubMed2002

Classification of renal proteinuria: a simple algorithm.

Total protein, albumin, alpha1-microglobulin, and immunoglobulin G (IgG) were analyzed in 1,622 urine samples without Bence-Jones proteinuria or gross hematuria. There was correlation with the histological picture obtained on renal biopsy in 61 patients. We established 24-h reference intervals for alpha1-microglobulin and IgG on 659 urine samples with total protein and albumin excretion rates below 100 mg/24 h and 30 mg/24 h, respectively, and creatinine clearance above 80 ml/min. The central 95% reference interval was found to be between 4 and 17 mg/24 h for alpha1-microglobulin and between 3 and 8.5 mg/24 h for IgG. In 80 urine samples with albumin excretion rate above 30 mg/24 h and alpha1-microglobulin and IgG within their reference intervals, we analyzed the 95% central interval of the distribution of the IgG/albumin ratios, and it was found to be within 0.01 and 0.20 (0.90 confidence interval: 0.17-0.24). Proteinuria was considered to be of the selective glomerular type if the albumin excretion rate was abnormal and the IgG/albumin ratio was under 0.20, even when the IgG excretion was within a pathological range. For the classification of proteinuria as predominantly tubular, we estimated the alpha1-microglobulin/albumin ratio in 173 urine samples with normal excretion rates of albumin and IgG and pathological excretion of alpha1-microglobulin. The discriminating value of 0.91 (0.90 confidence interval: 0.78-1.08) was accepted in order to define proteinuria of a tubular origin in the presence of a pathological albumin excretion rate. The association between albumin and IgG excretion rates and tubular reabsorption of the alpha1-microglobulin normally filtered by the glomerulus was studied in 33 urine samples from patients with no histologically significant tubulo-interstitial or vascular disease and a serum creatinine concentration below 141 pmol/l. The optimal curve-fitting function between albumin plus IgG and alpha1-microglobulin excretion rates was of the quadratic type (r = 0.927). Mixed proteinuria was considered when both, albumin and alpha1-microglobulin excretion rates were pathological and could not be included in the previously described groups.

Algorithms↗