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 361 records · Page 20Linked to original sources

PubMed1991

Echographic differentiation of histological types of intraocular melanoma.

The aim of this study was to assess the differentiation of histologically different types of choroidal melanomas by means of clinical and quantitative acoustic/texture parameters of echograms. Clinical parameters were graded by morphological, kinetic and quantitative statements about the A- and B-mode echograms by a skilled diagnostician. Acoustic/texture parameters were obtained by processing and analysing radio frequency, AM-demodulated and FM-demodulated echograms. The data were preprocessed to remove influences induced by beam diffraction and focusing. The correlations between the clinical parameters were lower than 0.5 but significant in a few cases. The best set of four clinical echographic parameters enabled a retrospective classification of spindle type melanomas vs. mixed + epithelioid type melanomas with an accuracy of 77% (area under the ROC-curve of 86%). These figures do not enable, however, a prospective diagnosis. The correlations of the clinical parameters with the acoustic/texture parameters were investigated. The mutual correlations between the latter parameters were also assessed, the most significant (0.70) being between the speckle size and the slope of the linearized backscattering spectrum. The discriminant analysis performed with the best four acoustic/texture parameters (n = 30) yielded a sensitivity of 89%, a specificity of 92% and an area under the ROC-curve corresponding to a probability of correct classification of 96.6%. When using these data prospectively to classify tumours of unknown cell type (n = 21), this classification could be performed in 86% of cases.

Algorithms↗

PubMed1992

Echographic differentiation of intraocular melanomas by computer analysis.

The aim of this study was to assess the differentiation of histologically different types of choroidal melanomas by means of clinical and quantitative acoustic/texture parameters of echograms. Clinical parameters were graded by morphological, kinetic and quantitative statements about the A- and B-mode echograms by a skilled diagnostician. Acoustic/texture parameters were obtained by processing and analysing radio frequency, AM-demodulated and FM-demodulated echograms. The latter data were preprocessed to remove influence induced by beam diffraction and focusing. The best set of four clinical echographic parameters enabled a retrospective classification of spindle type melanomas vs. mixed+epithelioid type melanomas with an accuracy of 77% (area under the ROC-curve of 86%). The discriminant analysis performed with the best four acoustic/texture parameters (n = 30) yielded a sensitivity of 89%, a specificity of 92%, and an area under the ROC-curve which corresponds to a probability of correct classification of 96.6%. When using these data prospectively to classify tumours of unknown cell type (n = 21), this classification could be performed in 86% of cases.

Algorithms↗

PubMed1991

Flow phenomena, classification, and clinical aspects of magnetic resonance angiography.

Both time-of-flight and phase-contrast techniques are currently being employed in producing angiographic MR images of arteries and veins. Both two-dimensional and three-dimensional acquisition methods can be used to display projective vascular images with the maximum intensity projection algorithm, eliminating the need for image subtraction. MR angiography has expanded the armamentarium of MR imaging, especially in the head and neck, and should in some organ systems reduce the need for conventional x-ray catheter angiography.

Blood Flow Velocity↗

PubMed1989

Adapting best linear unbiased prediction (BLUP) for timely genetic evaluation: II. Progeny traits in multiple contemporary groups within a herd.

The second step of a procedure to partially circumvent the voluminous calculations for some BLUP (Best Linear Unbiased Prediction) computing algorithms for genetic evaluation is presented. In addition, the procedure allows timely evaluations of each contemporary group. This procedure is pertinent especially for polytocous species such as swine and poultry, for which the occurrence of full-sib families makes the inclusion of dam effects in the model necessary and tests are completed throughout the year. Formulas are developed for a model including sires, dams, individuals within full-sib families and records within individuals. This model has a fundamentally hierarchical structure but includes some cross-classification. The formulas for predictors combine information across contemporary groups within a herd and incorporate relationships between sires and(or) dams in that herd. Formulas to approximate prediction error variances also are developed.

Algorithms↗

PubMed1988

A numerical classification of the genus Bacillus.

Three hundred and sixty-eight strains of aerobic, endospore-forming bacteria which included type and reference cultures of Bacillus and environmental isolates were studied. Overall similarities of these strains for 118 unit characters were determined by the SSM, SJ and DP coefficients and clustering achieved using the UPGMA algorithm. Test error was within acceptable limits. Six cluster-groups were defined at 70% SSM, which corresponded to 69% SP and 48-57% SJ. Groupings obtained with the three coefficients were generally similar but there were some changes in the definition and membership of cluster-groups and clusters, particularly with the SJ coefficient. The Bacillus strains were distributed among 31 major (4 or more strains), 18 minor (2 or 3 strains) and 30 single-member clusters at the 83% SSM level. Most of these clusters can be regarded as taxospecies. The heterogeneity of several species, including Bacillus brevis, B. circulans, B. coagulans, B. megateriun, B. sphaericus and B. stearothermophilus, has been indicated and the species status of several taxa of hitherto uncertain validity confirmed. Thus on the basis of the numerical phenetic and appropriate (published) molecular genetic data, it is proposed that the following names be recognized; Bacillus flexus (Batchelor) nom. rev., Bacillus fusiformis (Smith et al.) comb. nov., Bacillus kaustophilus (Prickett) nom. rev., Bacillus psychrosaccharolyticus (Larkin & Stokes) nom. rev. and Bacillus simplex (Gottheil) nom. rev. Other phenetically well-defined taxospecies included 'B. aneurinolyticus', 'B. apiarius', 'B. cascainensis', 'B. thiaminolyticus' and three clusters of environmental isolates related to B. firmus and previously described as 'B. firmus-B. lentus intermediates'. Future developments in the light of the numerical phenetic data are discussed.

Bacillus↗

PubMed1991

Molecular modeling of antibody combining sites.

Each of the six CDRs of Gloop2 is shown with the modeled structure in. Overall, the results obtained using the combined algorithm are similar in accuracy to those achieved using the canonical method of Chothia et al. However, the canonical method is limited to those loops where the key residues identified by Chothia are present. With the number of antibody structures currently available, it is not possible to classify CDR-H3 into canonical ensembles. Additionally, a small percentage of examples in the remaining CDRs do not match the current canonical classifications and the protein engineer may well wish to mutate the key residues, precluding the use of Chothia's method for modeling the resulting conformation. Thus the best approach appears to be to use Chothia's method (at least to model the backbone conformation) when the loop to be modeled is represented in the database of canonical structures. Any other loops, either unrepresented among the known canonicals (including CDR-H3), or where mutations have been made to the key residues, may then be modeled by the combined algorithm presented here.

Algorithms↗

PubMed1990

An introduction to Anatomical Image Digitizing System (AIDS).

We designed, constructed, and evaluated a prototype of a digital data management system for anatomical studies and education, named AIDS (Anatomical Image Digitizing System). By implementing the prototype systems, we intended to determine the technical and anatomical parameters required for such a system, and to attempt digital imaging analysis of morphology. We demonstrated that such systems can be used as a new morphological tool on a macroscopic level, particularly in mathematical image analysis, in comparative anatomy by organizing a database and by modifying images with graphic techniques, and in statistical discussions by manipulating large numbers of illustrations quickly. A 512 X 512 X 8 bit matrix is judged to be satisfactory for analyzing images mathematically and producing illustrations of bones, but higher quality tools are required to store anatomical figures with halftones in a good condition, and to prepare fine anatomical illustrations. Classifications of figures and formats of figure legends are necessary to standardize the storage of figures in the form of a database.

Algorithms↗

PubMedDate not supplied

A conversational program for hierarchic and non-hierarchic cluster analysis.

The theoretical basis and the features of some cluster analysis programs are described in this paper. These programs use as input data an N x M matrix. For the N' less than or equal to N set of individuals, the cross-correlation matrix, R = (rij), is then calculated. The dissimilarity matrix, D = (dhk), and the distribution histogram of the distances between each pair is worked out. Hierarchic and non-hierarchic clustering techniques are allowed. Using the non-hierarchic strategy, the non-overlapping clusters are then separated into different groups. The distances between these groups are calculated and statistical tests of significance are performed. In the hierarchic case, a dendrogram is drawn using an agglomerative algorithm. As an example the programs (written in FORTRAN, batch or TSS version) were used to classify human mitotic chromosomes.

Classification↗

PubMed1990

Rough sets approach to the analysis of the structure-activity relationship of quaternary imidazolium compounds.

The relationship between chemical structure and antimicrobial activity of 201 quaternary imidazolium compounds is analysed using the theory of rough sets. The compounds are described by 8 attributes concerning structure and are divided into 5 classes of activity. The description builds up on information system. Using the rough sets approach, first a minimum set of 4 attributes significant for a high quality of classification has been found. The analysis of distribution of values of significant attributes in the best and worst class led to the definition of typical representatives of best and worst compounds in terms of significant attributes. Finally, a decision algorithm has been driven from information system, showing important relations between structure and activity. It may be helpful in supporting decisions concerning the synthesis of new antimicrobial compounds.

Algorithms↗

PubMedDate not supplied

[Informatics in mental sciences: clinical data management and "expert systems"].

The role of information systems and particularly that of expert systems in medicine is analysed in brief. It is pointed out that, unlike other branches of medicine, in mental sciences the role of expert systems has so far been more limited because of certain intrinsic problems of programming on the one hand, and the scientific models of the psychiatrist on the other. An improvement in information technology and programming languages and better empirical classification of mental disturbances could provide useful bases for the realisation of computerised consultancy systems in mental sciences.

Algorithms↗

PubMed1992

Use of the International Classification of Diseases (ICD-9-CM) to identify hospitalizations for mechanical low back problems in administrative databases.

Large administrative databases are increasingly valuable tools for health care research. Although increased access to these databases provides valuable opportunities to study health care utilization, costs and outcomes and valid and comparable results require explicit and consistent analytic methods. Algorithms for identifying surgical and nonsurgical hospitalizations for "mechanical" low back problems in automated databases are described. Sixty-six ICD-9-CM diagnosis and 15 procedure codes that could be applied to patients with mechanical low back problems were identified. Twenty-seven diagnosis and two procedure codes identify hospitalizations for problems definitely in the lumbar or lumbosacral region. Exclusion criteria were developed to eliminate nonmechanical causes of low back pain, such as malignancies, infections, and major trauma. The use of the algorithms is illustrated using national hospital discharge data.

Algorithms↗

PubMed1982

Numerical and chemical classification of Nocardia amarae.

Twenty-one strains of Nocardia amarae and marker cultures of Mycobacterium, Nocardia, Rhodococcus and the 'aurantiaca' taxon were subjected to numerical phenetic analyses using 92 unit characters. The data were examined using the simple matching (SSM), Jaccard (SJ) and pattern (DP) coefficients and clustering was achieved using the unweighted average linkage algorithm. Neither cluster nor aggregate cluster composition was markedly affected by the coefficient used or by test error, estimated at 1.5%. The N. amarae strains formed a distinct and homogeneous cluster which showed its highest similarity to phena equated with Nocardia asteroides, Nocardia brasiliensis and Nocardia otitidis-caviarum. The non-hydroxylated fatty acid composition and overall size of the mycolic acids was similar to that found to be characteristic of Nocardia sensu stricto, though the long-chain in the 2-position of the mycolic acids was relatively much richer in monounsaturated components. Nocardia amarae, in containing dihydrogenated menaquinones with nine isoprene units, is clearly distinguished from established representatives of Nocardia.

Fatty Acids↗

PubMed1992

Clustering proteins into families using artificial neural networks.

An artificial neural network was used to cluster proteins into families. The network, composed of 7 x 7 neurons, was trained with the Kohonen unsupervised learning algorithm using, as inputs, matrix patterns derived from the bipeptide composition of 447 proteins, belonging to 13 different families. As a result of the training, and without any a priori indication of the number or composition of the expected families, the network self-organized the activation of its neurons into topologically ordered maps in which almost all the proteins (96.7%) were correctly clustered into the corresponding families. In a second computational experiment, a similar network was trained with one family of the previous learning set (76 cytochrome c sequences). The new neural map clustered these proteins into 25 different neurons (five in the first experiment), wherein phylogenetically related sequences were positioned close to each other. This result shows that the network can adapt the clustering resolution to the complexity of the learning set, a useful feature when working with an unknown number of clusters. Although the learning stage is time consuming, once the topological map is obtained, the classification of new proteins is very fast. Altogether, our results suggest that this novel approach may be a useful tool to organize the search for homologies in large macromolecular databases.

Algorithms↗

PubMed1983

[Early cancer diagnosis through automated cytoanalysis (Fazytan system)].

A prescreening system for the early detection of uterine cancer was developed. For the diagnosis of each input specimen (PAP-stained), a figure of malignancy has to be calculated automatically. Methods, problems and results of the system are summarized in this paper. Some thousand microscopic subfields of a specimen are successively scanned by an optimized TV-camera with high spatial resolution. The automated specimen analysis can be described by a two step procedure: single cell classification and evaluation of cell population. For the single cell classifier only features derived from the nucleus are used. The pattern recognition procedures are based on a processor-oriented strategy, and can be adapted to other cytological specimen. The algorithms have been tested with 3 . 10(5) cell images of about 300 specimens.

Autoanalysis↗

PubMedDate not supplied

Specialized PC software package for creation of computer systems supporting partial diagnostics based on numerical results of medical examinations.

Computer scientists creating computer systems supporting partial diagnostics based on numerical results of medical examinations always follow a similar method. Why not algorithmize this procedure? This was the main reason for preparation of the specialized PC software presented in the paper. The software is addressed to clinicians-scientists and enables such users to create by themselves (with no need of programming) practically useful computer systems for differential diagnosis of chosen diseases. These generated systems are assigned to an end-user, i.e. any physician interested in a partial diagnosis of the considered type. They perform classification of patients on the basis of numerical results of the examinations selected for the training databases. The software is based on a statistical approach; multi-dimensional variance analysis and discriminant analysis methods have been used. The pilot version of the software (experimentally implemented in a Warsaw clinic) and an outline of planned work are presented.

Algorithms↗

PubMed1987

Prediction of exposure degree diagram and sites of limited proteolysis in globular proteins as an approach to computer-aided design of protein bioregulators with prolonged action.

In order to prolong the lifetime of protein bioregulators in blood it is possible to engineer analogs with protected sites of limited proteolysis. To determine the sites, primarily accessible to trypsin-like proteases, a computer procedure has been developed, including a prediction algorithm, to produce the residue diagram of a globular protein and a discriminant algorithm to determine the sites most liable to proteolysis. The accuracy of prediction of amino acid residue exposure is characterised by correlation coefficients between experimental and theoretical exposure values, the coefficients being about 0.7 as calculated for 10 globular proteins. The classification of Arg and Lys residues into two groups, susceptible or insusceptible to protease, has an error percentage of about 25.

Arginine↗

PubMed1977

Automatisation of a localising technique in aphasia based upon averaged evoked potentials.

Averaged evoked potentials were utilised as a localising technique in aphasia. Based on a set of measurements extracted from various evoked potentials, subjects from four diagnostic categories were correctly classified in more than 70% of the cases. The measurements were derived from amplitudes and latencies of significant peaks which were selected by human judgement from plots of the recorded evoked potentials. An algorithm has been developed which simulates the manual procedure and reduces the processing time per patient by several magnitudes. The automated method yields results which are more consistent with expected results than those from the manual method. The percentage of correct classification is in both cases essentially the same.

Aphasia↗

PubMed1983

Qualitative and quantitative evaluation of liver diseases: functional classifications.

The problem of the best use of experimental data for qualitative and quantitative evaluation of liver disfunction is analysed. Only statistical procedures for classification are considered. Emphasis is placed on a complete description of the most important steps that must be performed (degree of clustering of the data, validity of the hypotheses underlying the statistical methods used, meaning and adequate use of the reference classification, reliability of the different statistical methods used, information content of the laboratory tests studied) since the omission of any of them may contribute to a misleading interpretation of the overall results or may prevent significant practical utilization. An application to BSP tests is carried out: results obtained from 350 subjects are presented and discussed. A classification related to the main functional aspects of liver impairment was chosen. Different combinations of the parameters obtained from BSP blood disappearance curve--initial disappearance rate (K1), 45 min-retention of total (RT45) and conjugated (RC45) BSP- and different mathematical algorithms are compared. Five classes (N, normal subjects; PN, paranormal subjects; H, prevalent hepatocellular damage; C, prevalent cholestatic damage; HC, heavy combined damages) can be satisfactorily discriminated with all three parameters (K1 + RT45 + RC45); however K1 + RC45 give almost the same results and represent an interesting compromise between data information content and laboratory complexity.

Diagnosis, Computer-Assisted↗