SEARCH · PubMed Health

Results for “Classification Algorithms”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

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

At least 1,351 records · Page 75Linked to original sources

PubMed2004

MRI-based individual 3D region-of-interest atlases of the human brain: a new method for analyzing functional data.

OBJECTIVES: Introduction of a new atlas-based method for analyzing functional data which takes into account the variability of individual human brains and the partial volume, effects of functional emission computed tomography, images in complex anatomical 3D regions, as well as, describing the underlying multi-modal image processing, principles. METHODS: 3D atlas extraction is done directly by automated segmentation of individual magnetic resonance images of the patient's head. This is done in two steps: voxel-based classification of T1-weighted images for tissue differentiation (low-level processing) is followed by knowledge-based analysis of the classified images for extraction of 3D anatomical regions (high-level processing). For atlas-based quantification of co-registered functional images, 3D anatomical regions can be convoluted with an idealized point spread function of the emission computed tomography system, after which a partial volume-dependent threshold can be determined. RESULTS: Quantitative evaluation studies, based on 50 realistic software head phantoms and 24 image data sets obtained from healthy subjects and patients, show low misclassification rates and stable results for the neural network-based classification approach (mean +/- SD 3.587 +/- 0.466%, range 2.726-4.927%) as well as for the adjustable parameters of the knowledge-based approach. Computation time is <5 min for classification, <1 min for most of the extraction algorithms. The influence of the partial volume-dependent threshold is shown for an activation study. CONCLUSIONS: This new method allows 3D atlas generation without the need to warp individual image data to an anatomical or statistical brain atlas. Going beyond the purely tissue-oriented approach, partial volume effects of emission computed tomography images can be analyzed in complex anatomical 3D regions.

Algorithms↗

PubMed1996

[How reliably does electromyography differentiate myopathies and neurogenic diseases?].

Electromyography is an essential part of neuromuscular diagnosis. Due to practicability, it mostly relies on qualitative judgement. Therefore, it is examiner dependent, and critical assessment of the EMG report is crucial. This may be furthered by (1) acquaintance with the genesis of the EMG, which has recently been rendered more transparent by computer simulation, and (2) experience with examiner-independent, quantitative diagnostic algorithms. By both approaches, it is intended to familiarize the reader with modern quality standards of EMG diagnosis. The diagnostic algorithm used to this end is discriminant MUP classification, which separated neurogenic and myopathic conditions without recourse to examiner judgements. This success shows that quantitative EMG, which has become available for routine practice, will add weight to EMG differential diagnosis.

Adult↗

PubMed1997

The alpha and omega of G-protein coupled receptors: a novel method for classification. Part 2. Bin classification.

A novel way of classification of G-protein coupled receptors is presented that is only based on receptor sequence information by counting of amino acid residues. It involves the number of amino acid residues between the Asn residue in TM1 and the residue Cys in the loop between TM4 and TM5, the number of residues between the latter Cys residue and Pro residue in TM6, and the number of residues between the latter Pro and the last amino acid residue (called omega) in the sequence. The classification of 131 sequences, covering biogenic amine, opioid and somatostatin receptors, is visualized by means of a diagram which is referred to as a bin map. Each bin in the diagram encloses all the sequences that belong to one and only one receptor type or subtype. This so-called bin classification was obtained by means of the genetic algorithm methodology, which offers new opportunities for classifying proteins.

Animals↗

PubMed2006

Nuclear envelope dystrophies show a transcriptional fingerprint suggesting disruption of Rb-MyoD pathways in muscle regeneration.

Mutations of lamin A/C (LMNA) cause a wide range of human disorders, including progeria, lipodystrophy, neuropathies and autosomal dominant Emery-Dreifuss muscular dystrophy (EDMD). EDMD is also caused by X-linked recessive loss-of-function mutations of emerin, another component of the inner nuclear lamina that directly interacts with LMNA. One model for disease pathogenesis of LMNA and emerin mutations is cell-specific perturbations of the mRNA transcriptome in terminally differentiated cells. To test this model, we studied 125 human muscle biopsies from 13 diagnostic groups (125 U133A, 125 U133B microarrays), including EDMD patients with LMNA and emerin mutations. A Visual and Statistical Data Analyzer (VISDA) algorithm was used to statistically model cluster hierarchy, resulting in a tree of phenotypic classifications. Validations of the diagnostic tree included permutations of U133A and U133B arrays, and use of two probe set algorithms (MAS5.0 and MBEI). This showed that the two nuclear envelope defects (EDMD LMNA, EDMD emerin) were highly related disorders and were also related to fascioscapulohumeral muscular dystrophy (FSHD). FSHD has recently been hypothesized to involve abnormal interactions of chromatin with the nuclear envelope. To identify disease-specific transcripts for EDMD, we applied a leave-one-out (LOO) cross-validation approach using LMNA patient muscle as a test data set, with reverse transcription-polymerase chain reaction (RT-PCR) validations in both LMNA and emerin patient muscle. A high proportion of top-ranked and validated transcripts were components of the same transcriptional regulatory pathway involving Rb1 and MyoD during muscle regeneration (CRI-1, CREBBP, Nap1L1, ECREBBP/p300), where each was specifically upregulated in EDMD. Using a muscle regeneration time series (27 time points) we develop a transcriptional model for downstream consequences of LMNA and emerin mutations. We propose that key interactions between the nuclear envelope and Rb and MyoD fail in EDMD at the point of myoblast exit from the cell cycle, leading to poorly coordinated phosphorylation and acetylation steps. Our data is consistent with mutations of nuclear lamina components leading to destabilization of the transcriptome in differentiated cells.

Biopsy↗

PubMed2004

BCI Competition 2003--Data sets Ib and IIb: feature extraction from event-related brain potentials with the continuous wavelet transform and the t-value scalogram.

The t-CWT, a novel method for feature extraction from biological signals, is introduced. It is based on the continuous wavelet transform (CWT) and Student's t-statistic. Applied to event-related brain potential (ERP) data in brain-computer interface (BCI) paradigms, the method provides fully automated detection and quantification of the ERP components that best discriminate between two samples of EEG signals and are, therefore, particularly suitable for classification of single-trial ERPs. A simple and fast CWT computation algorithm is proposed for the transformation of large data sets and single trials. The method was validated in the BCI Competition 2003, where it was a winner (provided best classification) on two data sets acquired in two different BCI paradigms, P300 speller and slow cortical potential (SCP) self-regulation. These results are presented here.

Algorithms↗

PubMed2000

Classification of ancient Roman glazed ceramics using the neural network of Self-Organizing Maps.

Artificial neural networks with unsupervised learning strategy known as Self-Organizing Maps were applied to classify ancient Roman glazed ceramics. Their clay ceramic bodies were analyzed by Inductively Coupled Plasma-Atomic Emission Spectroscopy and the chemical composition obtained was processed by this neural algorithm. The results obtained provide two types of information: firstly, classification of ceramic samples with identification of several groups and secondly, differentiation between the elemental chemical information. It was found that there are certain chemical elements which can be considered as principal and which can serve to differentiate between ceramics, whereas other elements give redundant information and do not contribute to sample differentiation. Seven chemical elements were considered principal and provide the necessary information. Two types of element were identified: 1- a group formed by common elements, such as: Ca, Fe, Mg, Mn and 2- another formed by optional elements: K or Na and Ba or Sr and Al or Ti.

Journal Article↗

PubMed2005

Surgical design and algorithm for correction of earlobe ptosis and pseudoptosis deformity.

A previously described classification system for earlobe ptosis and criterion for earlobe pseudoptosis deformity was based on height measurements of the two earlobe components: the free caudal segment and the attached cephalic segment. The "ideal" ear lobule free caudal segment was found to be between 1 and 5 mm (grade I ptosis), and the "ideal" attached cephalic segment was 15 mm or less. Earlobe pseudoptosis was defined by an attached cephalic segment measuring greater than 15 mm. Previous studies revealed an association between the elongated free caudal segment and increasing patient age and between the elongated attached cephalic segment and rhytidectomy. Sixteen fresh cadaver earlobes were used to design surgical patterns that would differentially reduce the free caudal segment, the attached cephalic segment, or both. A horizontal, medially based triangular excision pattern was designed. Triangular excisions limited to the attached cephalic segment resulted in 98 +/- 5 percent reduction of excision height from the attached cephalic segment but also resulted in an unexpected 32 +/- 2 percent augmentation of the excision height in the free caudal segment. Triangular excisions limited to the free caudal segment resulted in 88 +/- 4 percent reduction of the excision height from the free caudal segment and negligible reduction of 4 +/- 4 percent of excision height in the cephalic attached segment. An algorithm for correction of earlobe ptosis and pseudoptosis was subsequently derived and implemented in a clinical case. The authors propose that surgical treatment of patients with pseudoptosis be dependent on the ptosis grade. If the ptosis is grade I (1 to 5 mm), then excision of only the attached cephalic segment is recommended. If the ptosis is grade II or higher (more than 5 mm), then a combined attached cephalic and free caudal segment excision is recommended. In cases of isolated ptosis grade II or higher without pseudoptosis, then excision location of only the free caudal segment is recommended. The above simple algorithm and surgical designs will enable plastic surgeons to differentially correct earlobe ptosis and pseudoptosis.

Algorithms↗

PubMedDate not supplied

Raman spectroscopy in combination with background near-infrared autofluorescence enhances the in vivo assessment of malignant tissues.

The diagnostic ability of optical spectroscopy techniques, including near-infrared (NIR) Raman spectroscopy, NIR autofluorescence spectroscopy and the composite Raman and NIR autofluorescence spectroscopy, for in vivo detection of malignant tumors was evaluated in this study. A murine tumor model, in which BALB/c mice were implanted with Meth-A fibrosarcoma cells into the subcutaneous region of the lower back, was used for this purpose. A rapid-acquisition dispersive-type NIR Raman system was employed for tissue Raman and NIR autofluorescence spectroscopic measurements at 785-nm laser excitation. High-quality in vivo NIR Raman spectra associated with an autofluorescence background from mouse skin and tumor tissue were acquired in 5 s. Multivariate statistical techniques, including principal component analysis (PCA) and linear discriminant analysis (LDA), were used to develop diagnostic algorithms for differentiating tumors from normal tissue based on their spectral features. Spectral classification of tumor tissue was tested using a leave-one-out, cross-validation method, and the receiver operating characteristic (ROC) curves were used to further evaluate the performance of diagnostic algorithms derived. Thirty-two in vivo Raman, NIR fluorescence and composite Raman and NIR fluorescence spectra were analyzed (16 normal, 16 tumors). Classification results obtained from cross-validation of the LDA model based on the three spectral data sets showed diagnostic sensitivities of 81.3%, 93.8% and 93.8%; specificities of 100%, 87.5% and 100%; and overall diagnostic accuracies of 90.6%, 90.6% and 96.9% respectively, for tumor identification. ROC curves showed that the most effective diagnostic algorithms were from the composite Raman and NIR autofluorescence techniques.

Animals↗

PubMed2004

Support vector machines for separation of mixed plant-pathogen EST collections based on codon usage.

MOTIVATION: Discovery of host and pathogen genes expressed at the plant-pathogen interface often requires the construction of mixed libraries that contain sequences from both genomes. Sequence identification requires high-throughput and reliable classification of genome origin. When using single-pass cDNA sequences difficulties arise from the short sequence length, the lack of sufficient taxonomically relevant sequence data in public databases and ambiguous sequence homology between plant and pathogen genes. RESULTS: A novel method is described, which is independent of the availability of homologous genes and relies on subtle differences in codon usage between plant and fungal genes. We used support vector machines (SVMs) to identify the probable origin of sequences. SVMs were compared to several other machine learning techniques and to a probabilistic algorithm (PF-IND) for expressed sequence tag (EST) classification also based on codon bias differences. Our software (Eclat) has achieved a classification accuracy of 93.1% on a test set of 3217 EST sequences from Hordeum vulgare and Blumeria graminis, which is a significant improvement compared to PF-IND (prediction accuracy of 81.2% on the same test set). EST sequences with at least 50 nt of coding sequence can be classified using Eclat with high confidence. Eclat allows training of classifiers for any host-pathogen combination for which there are sufficient classified training sequences. AVAILABILITY: Eclat is freely available on the Internet (http://mips.gsf.de/proj/est) or on request as a standalone version. CONTACT: friedel@informatik.uni-muenchen.de.

Algorithms↗

PubMedDate not supplied

Support vector machine for optical diagnosis of cancer.

We report the application of a support vector machine (SVM) for the development of diagnostic algorithms for optical diagnosis of cancer. Both linear and nonlinear SVMs have been investigated for this purpose. We develop a methodology that makes use of SVM for both feature extraction and classification jointly by integrating the newly developed recursive feature elimination (RFE) in the framework of SVM. This leads to significantly improved classification results compared to those obtained when an independent feature extractor such as principal component analysis (PCA) is used. The integrated SVM-RFE approach is also found to outperform the classification results yielded by traditional Fisher's linear discriminant (FLD)-based algorithms. All the algorithms are developed using spectral data acquired in a clinical in vivo laser-induced fluorescence (LIF) spectroscopic study conducted on patients being screened for cancer of the oral cavity and normal volunteers. The best sensitivity and specificity values provided by the nonlinear SVM-RFE algorithm over the data sets investigated are 95 and 96% toward cancer for the training set data based on leave-one-out cross validation and 93 and 97% toward cancer for the independent validation set data. When tested on the spectral data of the uninvolved oral cavity sites from the patients it yielded a specificity of 85%.

Algorithms↗

PubMed2006

Molecular classification of human cancers using a 92-gene real-time quantitative polymerase chain reaction assay.

CONTEXT: Correct diagnosis of the tissue origin of a metastatic cancer is the first step in disease management, but it is frequently difficult using standard pathologic methods. Microarray-based gene expression profiling has shown great promise as a new tool to address this challenge. OBJECTIVE: Adoption of microarray technologies in the clinic remains limited. We aimed to bridge this technological gap by developing a real-time quantitative polymerase chain reaction (RT-PCR) assay. DESIGN: We constructed a microarray database of 466 frozen and 112 formalin-fixed, paraffin-embedded (FFPE) samples of both primary and metastatic tumors, measuring expression of 22,000 genes. From the microarray database, we used a genetic algorithm to search for gene combinations optimal for multitumor classification. A 92-gene RT-PCR assay was then designed and used to generate a database for 481 frozen and 119 FFPE tumor samples. RESULTS: The microarray-based K-nearest neighbor classifier demonstrated 84% accuracy in classifying 39 tumor types via cross-validation and 82% accuracy in predicting 112 independent FFPE samples. We successfully translated the microarray database to the RT-PCR platform, which allowed an overall success rate of 87% in classifying 32 different tumor classes in the validation set of 119 FFPE tumor samples. CONCLUSIONS: The RT-PCR-based expression assay involving 92 genes represents a powerful tool for accurately and objectively identifying the site of origin for metastatic tumors, especially in the cases of cancer of unknown primary. The assay uses RT-PCR and routine FFPE samples, making it suitable for rapid clinical adoption.

Algorithms↗

PubMed2006

The linear separability problem: some testing methods.

The notion of linear separability is used widely in machine learning research. Learning algorithms that use this concept to learn include neural networks (single layer perceptron and recursive deterministic perceptron), and kernel machines (support vector machines). This paper presents an overview of several of the methods for testing linear separability between two classes. The methods are divided into four groups: Those based on linear programming, those based on computational geometry, one based on neural networks, and one based on quadratic programming. The Fisher linear discriminant method is also presented. A section on the quantification of the complexity of classification problems is included.

Algorithms↗

PubMed2005

Motif extraction and protein classification.

We present a novel unsupervised method for extracting meaningful motifs from biological sequence data. This de novo motif extraction (MEX) algorithm is data driven, finding motifs that are not necessarily over-represented in the data. Applying MEX to the oxidoreductases class of enzymes, containing approximately 7000 enzyme sequences, a relatively small set of motifs is obtained. This set spans a motif-space that is used for functional classification of the enzymes by an SVM classifier. The classification based on MEX motifs surpasses that of two other SVM based methods: SVMProt, a method based on the analysis of physical-chemical properties of a protein generated from its sequence of amino acids, and SVM applied to a Smith-Waterman distances matrix. Our findings demonstrate that the MEX algorithm extracts relevant motifs, supporting a successful sequence-to-function classification.

Algorithms↗

PubMed2000

An alternative view of protein fold space.

Comparing and subsequently classifying protein structures information has received significant attention concurrent with the increase in the number of experimentally derived 3-dimensional structures. Classification schemes have focused on biological function found within protein domains and on structure classification based on topology. Here an alternative view is presented that groups substructures. Substructures are long (50-150 residue) highly repetitive near-contiguous pieces of polypeptide chain that occur frequently in a set of proteins from the PDB defined as structurally non-redundant over the complete polypeptide chain. The substructure classification is based on a previously reported Combinatorial Extension (CE) algorithm that provides a significantly different set of structure alignments than those previously described, having, for example, only a 40% overlap with FSSP. Qualitatively the algorithm provides longer contiguous aligned segments at the price of a slightly higher root-mean-square deviation (rmsd). Clustering these alignments gives a discreet and highly repetitive set of substructures not detectable by sequence similarity alone. In some cases different substructures represent all or different parts of well known folds indicative of the Russian doll effect--the continuity of protein fold space. In other cases they fall into different structure and functional classifications. It is too early to determine whether these newly classified substructures represent new insights into the evolution of a structural framework important to many proteins. What is apparent from on-going work is that these substructures have the potential to be useful probes in finding remote sequence homology and in structure prediction studies. The characteristics of the complete all-by-all comparison of the polypeptide chains present in the PDB and details of the filtering procedure by pair-wise structure alignment that led to the emergent substructure gallery are discussed. Substructure classification, alignments, and tools to analyze them are available at http://cl.sdsc.edu/ce.html.

Algorithms↗

PubMed2005

Systematic identification of statistically significant network measures.

We present a graph embedding space (i.e., a set of measures on graphs) for performing statistical analyses of networks. Key improvements over existing approaches include discovery of "motif hubs" (multiple overlapping significant subgraphs), computational efficiency relative to subgraph census, and flexibility (the method is easily generalizable to weighted and signed graphs). The embedding space is based on scalars, functionals of the adjacency matrix representing the network. Scalars are global, involving all nodes; although they can be related to subgraph enumeration, there is not a one-to-one mapping between scalars and subgraphs. Improvements in network randomization and significance testing--we learn the distribution rather than assuming Gaussianity--are also presented. The resulting algorithm establishes a systematic approach to the identification of the most significant scalars and suggests machine-learning techniques for network classification.

Algorithms↗

PubMed2005

On the quality of tree-based protein classification.

MOTIVATION: Phylogenetic analysis of protein sequences is widely used in protein function classification and delineation of subfamilies within larger families. In addition, the recent increase in the number of protein sequence entries with controlled vocabulary terms describing function (e.g. the Gene Ontology) suggests that it may be possible to overlay these terms onto phylogenetic trees to automatically locate functional divergence events in protein family evolution. Phylogenetic analysis of large datasets requires fast algorithms; and even 'fast', approximate distance matrix-based phylogenetic algorithms are slow on large datasets since they involve calculating maximum likelihood estimates of pairwise evolutionary distances. There have been many attempts to classify protein sequences on the family and subfamily level without reconstructing phylogenetic trees, but using hierarchical clustering with simpler distance measures, which also produce trees or dendrograms. How can these trees be compared in their ability to accurately classify protein sequences? RESULTS: Given a 'reference classification' or 'group membership labels' for a set of related protein sequences as well as a tree describing their relationships (e.g. a phylogenetic tree), we propose a method for dividing the tree into monophyletic or paraphyletic groups so as to optimize the correspondence between the reference groups and the tree-derived groups. We call the achieved optimal correspondence the 'accuracy of a tree-based classification (TBC)', which measures the ability of a tree to separate proteins of similar function into monophyletic or paraphyletic groups. We apply this measure to compare classical NJ and UPGMA phylogenetic trees with the trees obtained from hierarchical clustering using different protein similarity measures. Our preliminary analysis on a set of expert-curated protein families and alignments suggests that there is no uniformly superior algorithm, and that simple protein similarity measures combined with hierarchical clustering produce trees with reasonable and often the most accurate TBC. We used our measure to help us to design TIPS, a tree-building algorithm, based on agglomerative clustering with a similarity measure derived from profile scoring. TIPS is comparable with phylogenetic algorithms in terms of classification accuracy and is much faster on large protein families. Due to its time scalability and acceptable accuracy, TIPS is being used in the large-scale PANTHER protein classification project. The trees produced by different algorithms for different protein families can be viewed at http://panther.appliedbiosystems.com/pub/tree_quality/trees.jsp. For every tree and every level of classification granularity we provide the optimal TBC along with the reference classification. AVAILABILITY: The script that evaluates the accuracy of TBC is available at http://panther.appliedbiosystems.com/pub/tree_quality/index.jsp

Algorithms↗

PubMed2001

Molecular classification of multiple tumor types.

Using gene expression data to classify tumor types is a very promising tool in cancer diagnosis. Previous works show several pairs of tumor types can be successfully distinguished by their gene expression patterns (Golub et al. 1999, Ben-Dor et al. 2000, Alizadeh et al. 2000). However, the simultaneous classification across a heterogeneous set of tumor types has not been well studied yet. We obtained 190 samples from 14 tumor classes and generated a combined expression dataset containing 16063 genes for each of those samples. We performed multi-class classification by combining the outputs of binary classifiers. Three binary classifiers (k-nearest neighbors, weighted voting, and support vector machines) were applied in conjunction with three combination scenarios (one-vs-all, all-pairs, hierarchical partitioning). We achieved the best cross validation error rate of 18.75% and the best test error rate of 21.74% by using the one-vs-all support vector machine algorithm. The results demonstrate the feasibility of performing clinically useful classification from samples of multiple tumor types.

Algorithms↗

PubMed2006

Automated image processing method for the diagnosis and classification of malaria on thin blood smears.

Malaria is a serious global health problem, and rapid, accurate diagnosis is required to control the disease. An image processing algorithm to automate the diagnosis of malaria on thin blood smears is developed. The image classification system is designed to positively identify malaria parasites present in thin blood smears, and differentiate the species of malaria. Images are acquired using a charge-coupled device camera connected to a light microscope. Morphological and novel threshold selection techniques are used to identify erythrocytes (red blood cells) and possible parasites present on microscopic slides. Image features based on colour, texture and the geometry of the cells and parasites are generated, as well as features that make use of a priori knowledge of the classification problem and mimic features used by human technicians. A two-stage tree classifier using backpropogation feedforward neural networks distinguishes between true and false positives, and then diagnoses the species (Plasmodium falciparum, P. vivax, P. ovale or P. malariae) of the infection. Malaria samples obtained from the Department of Clinical Microbiology and Infectious Diseases at the University of the Witwatersrand Medical School are used for training and testing of the system. Infected erythrocytes are positively identified with a sensitivity of 85% and a positive predictive value (PPV) of 81%, which makes the method highly sensitive at diagnosing a complete sample provided many views are analysed. Species were correctly determined for 11 out of 15 samples.

Algorithms↗