Chivalai Temiyasathit

An effective classification procedure for diagnosis of prostate cancer in near infrared spectra

The main purpose of this study is to develop an effective classification procedure that discriminates between normal spectra and cancerous spectra in near infrared (NIR) spectroscopic data in which the classes are highly imbalanced and overlapped. Our proposed procedure consists of several steps. First, to ensure the comparability between spectra, normalization was done by dividing each spectral point by the area of the total intensity of the spectrum. Second, clustering analysis was performed with these normalized spectra to separate the spectra that represent the normal pattern from a mixed group that contains both normal and tumor spectra. Third, we conducted two-stage classification, the first being an effort to construct a classification model with the labels obtained from the preceding clustering analysis and the second being a classification to focus on the mixed group classified from the first classification model. To increase the accuracy, the second classification model was constructed based on the selected features that capture important characteristics of the spectral data. Our proposed procedure was evaluated by its classification ability in testing samples using a leave-one-out cross validation technique, yielding acceptable classification accuracy.

Characterization of spatially homogeneous regions based on temporal patterns of fine particulate matter in the continental United States.

Abstract Statistical analyses of time-series or spatial data have been widely used to investigate the behavior of ambient air pollutants. Because air pollution data are generally collected in a wide area of interest over a relatively long period, such analyses should take into account both spatial and temporal characteristics. The objective of this study is 2-fold: (1) to identify an efficient way to characterize the spatial variations of fine particulate matter (PM2.5) concentrations based solely upon their temporal patterns, and (2) to analyze the temporal and seasonal patterns of PM2.5 concentrations in spatially homogenous regions. This study used 24-hr average PM2.5 concentrations measured every third day during a period between 2001 and 2005 at 522 monitoring sites in the continental United States. A k-means clustering algorithm using the correlation distance was used to investigate the similarity in patterns between temporal profiles observed at the monitoring sites. A k-means clustering analysis produced six clusters of sites with distinct temporal patterns that were able to identify and characterize spatially homogeneous regions of the United States. The study also presents a rotated principal component analysis (RPCA) that has been used for characterizing spatial patterns of air pollution and discusses the difference between the clustering algorithm and RPCA.

Using multivariate statistical methods to model the electrospray ionization response of GXG tripeptides based on multiple physicochemical parameters

Response factors were determined for twelve GXG peptides (where G stands for glycine and X is any of alanine [A], arginine [R], asparagine [N], aspartic acid [D], glycine [G], histidine [H], leucine [L], lysine [K], phenylalanine [F], serine [S], tyrosine [Y], valine [V]) by electrospray ionization mass spectrometry (ESI-MS). The response factors were measured using a novel flow injection method. This new method is based on the Gaussian distribution of analyte concentration resulting from band-broadening dispersion experienced by the analyte upon passage through an extended volume of PEEK tubing. This method removes the need for preparing a discrete series of standard solutions to assess concentration-dependent response. Relative response factors were calculated for each peptide with reference to GGG. The observed trends in the relative response factors were correlated with several analyte physicochemical parameters, chosen based on current understanding of ion release from charged droplets during the ESI process. These include analyte properties: nonpolar surface area; polar surface area; gas-phase basicity; proton affinity; and Log D. Multivariate statistical analysis using multiple linear regression, decision tree, and support vector regression models were investigated to assess their potential for predicting ESI response based on the analyte properties. The support vector regression model was more versatile and produced the least predictive error following 12-fold cross-validation. The effect of variation in solution pH on the relative response factors is highlighted, as evidenced by the different predictive models obtained for peptide response at two pH values (pH = 6.0 and 9.0). The relationship between physicochemical parameters and associated ionization efficiencies for GXG tripeptides is discussed based on the equilibrium partitioning model.

Laser desorption/ionization mass spectrometry fingerprinting of complex hydrocarbon mixtures: Application to crude oils using data mining techniques

Crude oil fingerprints were obtained from four crude oils by laser desorption/ionization mass spectrometry (LDI-MS) using a silver nitrate cationization reagent. Replicate analyses produced spectral data with a large number of features for each sample (>11,000 m/z values) which were statistically analyzed to extract useful information for their differentiation. Individual characteristic features from the data set were identified by a false discovery rate based feature selection procedure based on the analysis of variance models. The selected features were, in turn, evaluated using classification models. A substantially reduced set of 23 features was obtained through this procedure. One oil sample containing a high ratio of saturated/aromatic hydrocarbon content was easily distinguished from the others using this reduced set. The other three samples were more difficult to distinguish by LDI-MS using a silver cationization reagent; however, a minimal number of significant features were still identified for this purpose. Focus is placed on presenting this multivariate statistical method as a rapid and simple analytical procedure for classifying and distinguishing complex mixtures.

Spatial prediction of ozone concentration profiles

Ground level ozone is one of the major air pollutants in many urban areas. Ozone formation affects ecosystems and is known to be associated with many adverse health issues in humans. Effective modeling of ozone is a necessary step to develop a system to warn residents of high ozone levels. In the present study we propose a statistical procedure that uses multiscale and functional data analysis to improve the spatial prediction of ozone concentration profiles in the Dallas Fort Worth (DFW) area of Texas. This study uses daily eight-hour ozone concentrations and meteorological predictors during a period between 2003 and 2006 at 14 monitoring sites in the DFW area. Wavelet transformation was used as a means of multiscale data analysis, followed by functional modeling to reduce model complexity. Kriging was then used for spatial prediction. The experimental results with real data demonstrated that the proposed procedures achieved acceptable accuracy of spatial prediction.

Increase performance of four-class classification for motor-imagery based brain-computer interface

Brain computer interface (BCI) is a system that provide a direct communication between human brain and external devices. BCIs which based on mental tasks of users are widely used for disabled or paralyzed patients, in order to help their mobility. Preprocessing techniques have been extensively developed to increase the signal-to-noise ratio and spatial distribution of the signals. Common Spatial Pattern (CSP) has shown to be a robust and effective method for processing Electroencephalogram (EEG) data. However, the results of CSP filter are still far from being completely explored. CSP was originally designed for two-class problem despite the fact that a practical application of Motor-imagery (MI) based BCI contains numbers of activities. It is necessary to design the classification algorithm which applicable to more than two-class problem. In this paper we investigate the performance of CSP by selecting optimal time slice and components for training CSP filters in four-class BCI by separating the four-class problem into multiple binary classifications. Our method is verified in the testing phase with four different types of classification approaches which are linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), linear support vector machines (LSVM), and support vector machines with radial basis function kernel (RBF-SVM). The result showed that, under the optimal time slice and components, the classification accuracy reach 78.82% for the best untrained subject in this dataset.

Classification of high-resolution NMR spectra based on complex wavelet domain feature selection and kernel-induced random forest

High-resolution nuclear magnetic resonance (NMR) spectra contain important biomarkers that have potentials for early diagnosis of disease and subsequent monitoring of its progression. Traditional features extraction and analysis methods have been carried out in the original frequency spectrum domain. In this study, we conduct feature selection based on a complex wavelet transform by making use of its energy shift-insensitive property in a multiresolution signal decomposition. A false discovery rate based multiple testing procedure is employed to identify important metabolite features. Furthermore, a novel kernel-induced random forest algorithm is used for the classification of NMR spectra based on the selected features. Our experiments with real NMR spectra showed that the proposed method leads to significant reduction in misclassification rate.

Regularizing multi-bands Common Spatial Patterns (RMCSP): A data processing method for brain-computer interface

In this paper, we propose a novel approach which is called the Regularizing Multi-bands Common Spatial Patterns (RMCSP) that particularly used for processing motor-imagery based Electroencephalography (EEG) data in Brain-computer Interface (BCI). The usage of BCI is severely limited due to the inconvenience of large number of channels used in recording devices. Moreover, Common Spatial Patterns (CSP) is a very well-known algorithm for its efficiency, but it just can extract the spatial information of the brain signals. To address these issues, we introduce the RMCSP method that exploits data in spectral, temporal and spatial domains in order to increase the classification accuracy in BCI. In addition, RMCSP is designed to handle EEG with small number of channels. To verify the efficacy of our approach, we rigorously tested the performances of the method in 17 subjects, from BCI competition datasets, in both two-class and four-class problems. Results show that RMCSP approach can outperform normal CSP method by nearly 10% in terms of median classification accuracy. It also enables us to significantly reduce the number of channels used in the datasets without decreasing the performances of the subjects.

Incomplete Nucleic Acid Sequences Visualization: A Case Study in Virus Sequences

Dengue virus (DENV) is one of the most widespread infectious diseases in the world, especially in the South East Asian regions. Transmitting the virus through mosquitoes, Dengue is an infectious viral borne disease. The virus sequences are assembled as series of nucleic acid, making the task of diagnosing virus sequences burdensome. Graphical representations are then proposed to represent Dengue virus to sustain the studies in virus sequences diagnosis. However, graphically representing sequences remained a crucified task especially for the incomplete genome sequences due to the missing nucleic acids. Although a number of studies provide methodologies on virus sequence visualization, in Dengue virus researches, those methodologies provide the visualization solely for complete genome sequences while neglecting the incomplete genome sequences. With the unaccommodating availabilities of research inputs, our study proposes a methodology for graphically representing the incomplete Dengue virus sequences, as well as complete virus sequences, by imputing in the incomplete part of a sequence with created reference sequences. The proposed methodology employs the use of database technology and majority voting technique to create reference sequences for each serotype of Dengue. Experimental results show that incomplete sequences are visualized realistically according to its respective serotype, thus providing flexibilities in Dengue virus researches to compensate incomplete sequences as inputs.

An Investigation of Lean Implementation Status in Thai Manufacturing

Lean Manufacturing has been widely adopted in various business and industries worldwide. To determine the current stage of Lean implementation in Thailand and the barrier Lean implementation, the qualitative research involving interview with 10 industrial companies practicing Lean manufacturing was conducted. The result from this study shows that most Lean practitioners understand the principal of Lean manufacturing. Different organization functions adopt different types of Lean tools and techniques as well as the performance measures depending on their business characteristics. However, there are three performance measures, which are manufacturing cost per unit, total sales and part per million (PPM) in defective products shipped to customers, that were adopted by all 10 companies. The most important barrier in Lean implementation is the cultural change since it requires the entire company participation. Finally, the companies participated in this study confirmed that they are satisfy with the result of Lean manufacturing though some companies have not completely adopted the Lean approach into their manufacturing process. To excellent the Lean manufacturing, companies are required to satisfy customer needs, improve the manufacturing process, and increase their flexibility. Furthermore, it is necessary for the top management to provide a clear policies as well as plan and direction. If Lean implementation is to be successful, the communication and human resource department are also the main keys.