Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Discriminant Analysis”

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 1,063 records · Page 59Linked to original sources
PubMed1999

A pattern classification approach to characterizing solitary pulmonary nodules imaged on high resolution CT: preliminary results.

The purpose of this research is to characterize solitary pulmonary nodules as benign or malignant based on quantitative measures extracted from high resolution CT (HRCT) images. High resolution CT images of 31 patients with solitary pulmonary nodules and definitive diagnoses were obtained. The diagnoses of these 31 cases (14 benign and 17 malignant) were determined from either radiologic follow-up or pathological specimens. Software tools were developed to perform the classification task. On the HRCT images, solitary nodules were identified using semiautomated contouring techniques. From the resulting contours, several quantitative measures were extracted related to each nodule's size, shape, attenuation, distribution of attenuation, and texture. A stepwise discriminant analysis was performed to determine which combination of measures were best able to discriminate between the benign and malignant nodules. A linear discriminant analysis was then performed using selected features to evaluate the ability of these features to predict the classification for each nodule. A jackknifed procedure was performed to provide a less biased estimate of the linear discriminator's performance. The preliminary discriminant analysis identified two different texture measures--correlation and difference entropy--as the top features in discriminating between benign and malignant nodules. The linear discriminant analysis using these features correctly classified 28/31 cases (90.3%) of the training set. A less biased estimate, using jackknifed training and testing, yielded the same results (90.3% correct). The preliminary results of this approach are very promising in characterizing solitary nodules using quantitative measures extracted from HRCT images. Future work involves including contrast enhancement and three-dimensional measures extracted from volumetric CT scans, as well as the use of several pattern classifiers.

Biophysical Phenomena↗
PubMed2003

Geographic classification of extra virgin olive oils from the eastern Mediterranean by chemometric analysis of visible and near-infrared spectroscopic data.

Visible and near-infrared reflectance spectra have been examined for their ability to classify extra virgin olive oils from the eastern Mediterranean on the basis of their geographic origin. Classification strategies investigated were partial least-squares regression, factorial discriminant analysis, and k-nearest neighbors analysis. Discriminant models were developed and evaluated using spectral data in the visible (400-750 nm), near-infrared (1100-2498 nm), and combined (400-2498 nm) wavelength ranges. A variety of data pretreatments was applied. Best results were obtained using factorial discriminant analysis on raw spectral data over the combined wavelength range; a correct classification rate of 93.9% was obtained on a prediction sample set. Though the overall sample set was limited in numbers, these results demonstrate the potential of near-infrared spectroscopy to classify extra virgin olive oils on the basis of their geographic origin.

Algorithms↗
PubMed1997

Computer-Assisted Microscope Analysis of Feulgen-Stained Nuclei in Gonadotroph Adenomas and Null-Cell Adenomas of the Pituitary Gland.

The current classification of clinically nonfunctioning pituitary adenomas is based on immunocytochemical and ultrastructural studies. However, a number of cases have less distinctive features that cannot be easily conformed with the prevailing morphologic classifications. The diagnostic information contributed by the determination of the nuclear DNA content (DNA ploidy level) and quantitative chromatic pattern description as opposed to the morphofunctional diagnosis in clinically nonfunctioning adenomas was consequently investigated in a series of 71 pituitary adenomas, including 31 null-cell adenomas, 35 gonadotropin adenomas, and 5 nonfunctioning adenomas that were not examined by electron microscopy. DNA ploidy level (8 variables) and quantitative chromatin pattern description (30 variables) were carried out by means of the computer-assisted microscope analysis of 80-1600 Feulgen-stained nuclei analyzed/case. The diagnostic information contributed by the 38 quantitative variables was determined by multifactorial statistical analysis (i.e., Discriminant Analysis). This computer-assisted classification significantly differentiated nulLcell adenomas from gonadotropin adenomas (p = 0.0025). In addition, it was able to differentiate three major subtypes of nonfunctioning adenomas on the basis of their immunohistochemical profiles. These were the immunonegative adenomas, the follicle-stimulating hormone (FSH)-positive adenomas, and the a-subunit (a) and/or luteiwizing hormone (LH)-positive adenomas (p < 0.0001 to p < 0.001). We thus suggest that the cytometric image analysis of Feulgen-stained nuclei can contribute on the discrimination of subtypes of clinically nonfunctioning pituitary adenomas.

Journal Article↗
PubMed1978

Distinguishing between malignant and cirrhotic ascites by computerized step-wise discriminant functional analysis of its biochemistry.

This study was designed to evaluate the ability of a specifically programmed computer to select those biochemical substances most capable of distinguishing "cirrhotic ascites" from "malignant ascites". After simultaneously performing selected biochemical and electrophoretic studies on fresh unstored serum and ascites of 23 patients with documented cirrhosis and 18 patients with proven malignancies, computerized step-wise discriminant analysis of the multiple input revealed that the serum-to-ascites LDH ratio was able to distinguish "cirrhotic ascites" from "malignant ascites" with greater than 86% accuracy. Assignment to proper groups was increased to 89% with the addition of the ratio to serum total protein-to-ascites total protein. The predictive value of a positive result was 100%; specificity was 100%; the predictive value of a negative result was 85%. This type of computer analysis also permits incorporation of both additional cases and new substances thus increasing predictability and reducing type II statistical errors.

Albumins↗
PubMed1988

[Multivariate analysis of hematological and serum biochemical values in female cynomolgus monkeys (Macaca fascicularis) aged less than a year].

Hematological and serum biochemical data obtained from 91 laboratory-bred cynomolgus monkeys (Macaca fascicularis) aged 11 to 362 days were analyzed by the discriminant analysis. All animals used had been bred and reared under uniform environmental conditions at Tsukuba Primate Center for Medical Science, N. I. H., Japan. The examination items were as follows: red blood cell count (RBC), hematocrit value (Ht), hemoglobin concentration (Hb), mean corpuscular volume (MCV), white blood cell count (WBC), glutamic oxaloacetic transaminase activity (GOT), glutamic pyruvic transaminase activity (GPT), total protein concentration (TP), albumin concentration (ALB), albumin-globulin ratio (A/G), blood urea nitrogen (BUN), glucose concentration (GLU), total cholesterol concentration (TCHO), free cholesterol concentration (FCHO), triglyceride concentration (TG), alkaline phosphatase activity (ALP) and calcium concentration (Ca). The animals were divided into four groups (A: Suckling infants, less than 176 days old. B, C, D: Weanlings and juveniles, 121 to 220 days old, 221 to 280 days old, and 281 to 362 days old, respectively). Discrimination was possible among these four groups on the basis of the Mahalanobis' generalized distance. Regarding the canonical discriminant analysis (discriminant analysis with reduction of dimensionality), discrimination was possible. The suckling infant group could be discriminated from the juvenile groups by the first canonical variate. Concerning the juvenile groups, age was highly correlated to the value of the second canonical variate.2+ Judging from an approximate relative eigenvector value for the second canonical variate, the effective discriminant variables were WBC, TP, ALB, A/G, FCHO, TG, and ALP. It can be concluded that these eight parameters are important and useful for monitoring the physiologicals conditions of growing juvenile monkeys.

Aging↗
PubMed1994

EEG-based, neural-net predictive classification of Alzheimer's disease versus control subjects is augmented by non-linear EEG measures.

Attempts to classify Alzheimer's disease (AD) subjects versus controls using spectral-band measures of electroencephalographic (EEG) data typically achieve around 80% success. This study assessed the ability of adding non-linear EEG measures and using a neural-net classification procedure to improve this performance level. The non-linear EEG measures were estimated correlation dimension ("dimensional complexity," or DCx) and saturation (degree of leveling-off of DCx with increasing embedding dimension). In a sample of 39 subjects (14 ADs, 25 controls), it was found that (a) the addition of non-linear EEG measures improved the classification accuracy of the AD/control status of subjects, and (b) a back-percolation neural net predictively classified the subjects much better than the standard linear techniques of multivariate discriminant analysis or nearest-neighbor discriminant analysis.

Aged↗
PubMed1989

Immunity to Salmonella typhi: considerations relevant to measurement of cellular immunity in typhoid-endemic regions.

Experiments were performed in Baltimore, Maryland and in Santiago, Chile, to determine the level of Salmonella typhi antigen-driven in vitro lymphocyte replication response which signifies specific acquired immunity to this bacterium and to determine the best method of data analysis and form of data presentation. Lymphocyte replication was measured as incorporation of 3H-thymidine into desoxyribonucleic acid. Data (ct/min/culture) were analyzed in raw form and following log transformation, by non-parametric and parametric statistical procedures. A preference was developed for log-transformed data and discriminant analysis. Discriminant analysis of log-transformed data revealed 3H-thymidine incorporation rates greater than 3,433 for particulate S. typhi, Ty2 antigen stimulated cultures signified acquired immunity at a sensitivity and specificity of 82.7; for soluble S. typhi O polysaccharide antigen-stimulated cultures, ct/min/culture values of greater than 1,237 signified immunity (sensitivity and specificity 70.5%).

Adolescent↗
PubMed1993

Linear body measurements of cattle before and after 20 years of selection for postweaning gain when fed two different diets.

At weaning and again after postweaning gain tests, height and width at hips, height at withers, body length, girth, head length and width, muzzle width, and cannon bone circumference measurements were obtained from Angus and Hereford bulls and heifers born in 1964, 1984, and 1985 (n = 989). The cattle were from the initial and final two calf crops selected for postweaning gain when fed either a high-concentrate diet or an all-hay diet. Analysis of variance and canonical discriminant analysis were used to examine the relationships among body measurements and major sources of variation (breed, year of birth, sire within breed and year, diet, sex of calf, age of dam, and the regression on age of calf). Canonical discriminant analysis indicated that one underlying variate explained nearly 90% of the total variation among the weaning measurements, whereas three variates were required to account for that proportion in the end-of-test measurements. At both measurement times, the first canonical variate was associated with year of birth, the second with sex of calf, and the third with dietary energy. Correlations between each canonical variate and the original body measurements indicated that year of birth (variate 1) was most closely associated with body length and cannon bone circumference at weaning and with body length and height at hips (but not withers) at end of test. Sex of calf (variate 2) was associated most closely with width of muzzle and head. Diet (variate 3) was associated with heart girth. Faster-growing cattle were longer in body, but not necessarily taller.

Analysis of Variance↗
PubMed1997

Statistical evaluation of ecosystem properties influencing the uptake of As, Cd, Co, Cu, Hg, Mn, Ni, Pb and Zn in seaweed (Eucus vesiculosus) and common mussel (Mytilus edulis).

In order to evaluate the influence of geographically varying marine ecosystem properties on the uptake of trace elements in bioindicators, samples were taken of seaweed (Fucus vesiculosus) and common mussel (Mytilus edulis) along the North Sea and Baltic Sea coast. Seasonal variations of the bioindicator status were minimized by sampling within 1 month. Ecosystem properties considered were the geographical position, the salinity and the concentrations of the macroelements Ca, Fe, K, Mg, Na, P and S in the bioindicators. Trace elements studied were As, Cd, Co, Cu, Hg, Mn, Ni, Pb and Zn. Factor analysis of the concentration patterns in the bioindicators and of salinity as a function of location confirmed the influence of the geographically varying salinity on the biological uptake of macroelements and trace elements. This influence of salinity was higher in the case of seaweed than in the case of mussel. Comparison of the geographical courses of the macroelement and trace-element concentrations by cluster analysis revealed corresponding courses for As and Hg in both bioindicators. All other elements showed different courses in seaweed and mussel. Subsequent cluster analysis comparing locations with respect to the macroelement or trace-element concentration patterns in the bioindicators, indicated a clear separation of North and Baltic Sea locations. However, the trace-element concentration patterns provided a regionally less distinctive ecosystem arrangement than those of the macroelement ones. The results of the cluster analysis were verified by discriminant analysis forming groups of locations with respect to geographical position and salinity. Results of discriminant analysis demonstrated, both for seaweed and for mussel as bioindicators, that the location groups formed according to the macroelement concentration patterns corresponded well with the geographical regions in the order of salinity. On the other hand, location groups based on the trace-element concentration patterns again showed a modified less distinctive ecosystem arrangement than the location groups based on macroelement concentration patterns. This confirms modified conditions for the uptake of trace elements in seaweed or mussel in comparison to the uptake of macroelements.

Animals↗
PubMed1985

Sandhoff disease heterozygote detection: a component of population screening for Tay-Sachs disease carriers. I. Statistical methods.

Serum and leukocyte hexosaminidase profiles (total activity and percent heat-labile activity levels) in obligate Sandhoff disease (SHD) heterozygotes differ from those of obligate Tay-Sachs disease (TSD) heterozygotes and noncarrier individuals. We have developed a procedure to identify, with 95% sensitivity, carriers of the allele(s) for SHD among individuals screened in a TSD heterozygote identification program. Using multivariate statistical methods of cluster analysis and discriminant analysis on serum and leukocyte hexosaminidase profiles from 102 potential SHD carriers, a linear discriminant function to classify individuals as SHD carriers or SHD noncarriers was constructed. This function classifies the serum and leukocyte profiles from all 15 obligate SHD heterozygotes studied, as those of SHD carriers. A 95% isodensity ellipse derived from only the serum hexosaminidase profiles of the 15 SHD obligate carriers has been applied to a TSD screened sample of 37,843 Jewish and non-Jewish individuals. A potential recall rate of screened individuals for serum retests and leukocyte assays of 2.01% has been estimated. These statistical methods enhance the TSD heterozygote screening program by permitting one to detect SHD heterozygotes within the screened population.

Genetic Carrier Screening↗
PubMed2003

Probabilistic disease classification of expression-dependent proteomic data from mass spectrometry of human serum.

We have developed an algorithm called Q5 for probabilistic classification of healthy versus disease whole serum samples using mass spectrometry. The algorithm employs principal components analysis (PCA) followed by linear discriminant analysis (LDA) on whole spectrum surface-enhanced laser desorption/ionization time of flight (SELDI-TOF) mass spectrometry (MS) data and is demonstrated on four real datasets from complete, complex SELDI spectra of human blood serum. Q5 is a closed-form, exact solution to the problem of classification of complete mass spectra of a complex protein mixture. Q5 employs a probabilistic classification algorithm built upon a dimension-reduced linear discriminant analysis. Our solution is computationally efficient; it is noniterative and computes the optimal linear discriminant using closed-form equations. The optimal discriminant is computed and verified for datasets of complete, complex SELDI spectra of human blood serum. Replicate experiments of different training/testing splits of each dataset are employed to verify robustness of the algorithm. The probabilistic classification method achieves excellent performance. We achieve sensitivity, specificity, and positive predictive values above 97% on three ovarian cancer datasets and one prostate cancer dataset. The Q5 method outperforms previous full-spectrum complex sample spectral classification techniques and can provide clues as to the molecular identities of differentially expressed proteins and peptides.

Algorithms↗
PubMed1992

Vibrations--discriminant models and possibilities for prognosticating specific and non-specific effects on the organism.

The present study involved 290 workers from branches of industry in which they were exposed to intense vibration and 102 patients suffering from disease due to vibration. Some specific (physiological) indices for exposure to vibration, such as the sensation of vibration, cold provocation test, Doppler-ultrasound, etc., and some non-specific (biochemical) indices, such as blood lipids, electrolytes and trace elements were studied. A discriminant analysis was made and 17 discriminant models (including different numbers of physiological and biochemical indices, age, and length of service) were elaborated. These models were used to predict the presence or absence of concomitant diseases, such as hypertension, coronary heart disease and arthritic lesions in persons exposed to vibrations; the stage of the vibration disease; the occurrence of exposure to vibration or of vibration disease. It was shown that physiological (specific) and biochemical (non-specific) indices affected by the exposure to vibration were interrelated, and that their estimation by means of the discriminant analysis revealed the distinction between the states mentioned above.

Adult↗
PubMed1988

Architectural and nuclear morphometrical features together are more important prognosticators in endometrial hyperplasias than nuclear morphometrical features alone.

Previous studies have shown that 80-90 per cent of cases of atypical hyperplasia of the endometrium do not progress to cancer. Criteria to predict the outcome in an individual patient with hyperplasia are lacking, and hysterectomy is the usual (over)treatment in order to avoid a 10-20 per cent chance of confrontation with cancer later on in the course of the disease. A recent study has shown that using a nuclear morphometric classification rule, 15 per cent of patients without progression can be accurately separated from patients with progression. However, as it is unlikely that nuclear morphometrical features are the only morphological factors reflecting the outcome of the disease, other quantitative parameters describing the architecture of the glands have also been studied for their potential value in selecting patients who will progress to cancer. In total, 10 nuclear features and 12 glandular architectural features were studied in 39 cases of atypical endometrial hyperplasia. Among these cases, seven (18 per cent) progressed to cancer. Using linear stepwise regression analysis and discriminant analysis, the volume percentage stroma and the standard deviation of the shortest nuclear axis are the best discriminators, although the outer surface density of the glands also adds to the discriminating power. The volume percentage stroma is the best single prognosticator; this feature is highly reproducible. In total, using these combined architectural and nuclear morphometrical features, 20 of the 32 cases without progression were separated from those who subsequently progressed (62.5 per cent). This is a considerable improvement over nuclear morphometrical features alone (15 per cent separated).

Cell Nucleus↗
PubMed1975

A statistical prognostic study of pump failure in acute myocardial infarction.

Repeated measurements of maemodynamic status were made and were subsequently subjected to analysis to assess prognosis in a group of patients with acute myocardial infarction and pump failure. Two multivariate methods were used: correspondence analysis and discriminant analysis. Both these methods agreed with the highest prognostic reliability of left ventricular function indices. Using various discriminant functions a prospective study was carried out on a test sample and a rate for well-classified subjects ranging from 80% to 90% was obtained. For each discriminant function the different causes of misclassification were analysed; among them a right ventricular dysfunction was allowed to account for discrepancies between expected and observed data on inferior infarction.

Acute Disease↗
PubMed1993

Improved exercise test accuracy using discriminant function analysis and "recovery ST slope".

The objective of the study was to optimize the accuracy of the exercise test for predicting the presence of significant angiographic coronary artery disease. A retrospective analysis of stored digital exercise electrocardiographic data on 147 men who had undergone exercise testing and cardiac catheterization was performed. With significant coronary artery disease defined as > or = 70% stenosis, 95 patients had one or more vessel(s) diseased. None were receiving digoxin, had a myocardial infarction or previous coronary artery bypass graft, or exhibited left bundle branch block, left ventricular hypertrophy, Q waves, or ST depression on their resting electrocardiogram. Analysis was performed using the authors' averaging and measurement software at rest and at each 30 seconds throughout the exercise and recovery in leads II, V2, and V5. Discriminant function analysis was used to analyze pretest variables, as well as hemodynamic and electrocardiographic changes and symptoms during exercise. A discriminant function score was developed and compared to other treadmill scores. The setting was a 1,000 bed Veterans Affairs Medical Center (Long Beach, CA). Discriminant function analysis chose age, smoking status, presenting chest pain characteristics, and lead V5 ST slope in recovery to have independent power for separating those with and without coronary artery disease. A discriminant function score using these four variables was used to form a receiver operating characteristics curve (and derive receiver operating characteristics curve areas) for comparison to other exercise test methods and scores: (discriminant function score = .81; slope 3.5 minutes into recovery in lead V5 = .73; traditional ST amplitude method = .72; ST60/HR index (amplitude of ST depression 60 ms after the J point/delta heart rate) = .66; traditional ST amplitude/HR index (traditional method/delta heart rate) = .75; Hollenberg score = .68; Hollenberg areas only = .66; and ST integral = .66. Receiver operating characteristics curve analysis revealed a trend for the discriminant function score to be superior to all other measurements and scores. Recovery ST slope in lead V5 performed as well as or better than all other electrocardiographic criteria or treadmill scores except for the authors' discriminant function score.

Cardiac Catheterization↗
PubMed2001

An EEG severity index of traumatic brain injury.

EEG spectral analyses were conducted from 19 scalp locations for patients with mild (n=40), moderate (n=25), and severe (n=43) traumatic brain injury (TBI), 15 days to 4 years after injury. Severity of TBI was judged by emergency hospital admission records (Glasgow Coma Score and duration of coma and amnesia). Highest-loading EEG variables on each factor that differed significantly between severe and mild TBI by univariate t-test were entered into a multivariate discriminant analysis, yielding 16 variables. Discriminant analysis between mild and severe TBI groups showed classification accuracy of 96.39%, sensitivity 95.45%, and specificity 97.44%. The EEG discriminant score also measured intermediate severity in moderate TBI patients. Results were cross-validated in 503 VA patients. Significant correlations between EEG discriminant scores, emergency admission measures, and post-trauma neuropsychological test scores validated the discriminant function as an index of severity of injury and a classifier of the extremes of severity.

Adolescent↗
PubMed1997

Characterization of the nuclear deoxyribonucleic acid content and nuclear morphometry in 71 primary cutaneous melanomas.

BACKGROUND: While the determination of nuclear deoxyribonucleic acid (DNA) content (DNA ploidy level) and nuclear morphometry characterization has proved to be of prognostic value in melanocytic lesions, there are several ways of performing these determinations. OBJECTIVE: To identify which of 9 DNA ploidy- and 2 nuclear morphometry-related variables are of prognostic and/or diagnostic value in 71 primary melanomas. METHODS: Histological typing, Breslow depth determination, the evaluation of Clark's level of invasion and the 11 quantitative variables (calculated in Feulgen-stained nuclei using computer-assisted microscope analysis) determined for each melanoma were submitted to discriminant analysis. RESULTS: The discriminant analysis of image cytometric variables enabled specific cell subpopulations to be identified in histological and the Breslow-related groups, but not in the Clark-related ones. CONCLUSION: The characterization of melanoma heterogeneity by means of the identification of specific DNA ploidy level-related cell subpopulations in specific Breslow-related groups enables the problem of intra- and interobserver variability in Breslow depth determination to be reduced and therefore can help dermatologists in their daily routine.

Adult↗
PubMed2005

Spatio-temporal variation in serum chemistry of the lobster, Homarus americanus Milne-Edwards.

Monthly variations in serum chemistry of the American lobster, Homarus americanus Milne-Edwards, were investigated at one location in Long Island Sound (LIS). Comparisons between three locations within and outside LIS were also made for a single time point. Most serum analytes displayed significant fluctuation over the study period and between locations. Temporal patterns could be classified as: low in cool months/high in warm months, i.e. Na, Cl, Na:K ratio, Ca, albumin:globulin ratio, percentage Fe saturation; high in cool months/low in warm months, i.e. pH, K, urea, total protein, albumin, globulin, alanine aminotransferase (ALT), aspartate aminotransferase (AST), lipaemia; June spike, i.e. glucose, cholesterol, creatine kinase, iron, transferrin iron-binding capacity; other less obvious fluctuations, i.e. Mg, PO4; and no apparent fluctuation, i.e. HCO3, alkaline phosphatase. The proportion of samples correctly classified into month of collection by a subset of 13 analytes using discriminant analysis improved as the months progressed from May (0.75) to October (>0.95). Discriminant analysis also resolved 96.5% of samples by location. The significant depression of serum calcium at the eastern LIS site correlates with excretory calcinosis, a calcium storage disease described from lobsters at this site, but contrasts with a seasonal elevation in serum calcium recorded in the temporal component of the study. Serum proteins, the electrolytes Ca and K and the enzymes ALT and AST proved to have the strongest spatio-temporal patterns of variation. Serum chemistry is a useful research tool for lobster populations, but the dearth of information on the homology of analyte functions in this species with those in vertebrate species makes interpretation of the results challenging. Late summer/autumn water conditions appear to cause stress for lobsters in LIS.

Animals↗