Supervised discretization for decluttering classification models.

Presented here is the first demonstration of supervised discretization to 'declutter' multivariate classification data in chemical sensor applications. The performance of multivariate classification models is often limited by the non-informative chemical variance within each target class; decluttering methods seek to reduce within-class variance while retaining between-class variance. Supervised discretization is shown to declutter classes in a manner that is superior to the state-of-the-art External Parameter Orthogonalization (EPO) by constructing a more parsimonious model with fewer parameters to optimize and is, consequently, less susceptible to overfitting and information loss. The comparison of supervised discretization and EPO is performed on three classification applications: X-ray fluorescence spectra of pine ash where the pine was grown in three distinct soil types, laser induced breakdown spectroscopy of colored artisanal glasses, and laser induced breakdown spectroscopy of exotic hardwood species.

Raman Spectroscopy to Enhance Investigative Lead Information in Automotive Clearcoats.

A new method to determine the make and model of a vehicle from an automotive paint sample recovered at the crime scene of a vehicle-related fatality such as a hit-and-run using Raman microscopy has been developed. Raman spectra were collected from 118 automotive paint samples from six General Motors (GM) vehicle assembly plants to investigate the discrimination power of Raman spectroscopy for automotive clearcoats using a genetic algorithm for pattern recognition that incorporates model inference and sample error in the variable selection process. Each vehicle assembly plant pertained to a specific vehicle model. The spectral region between 1802 and 697 cm-1 was found to be supportive of the discrimination of these six GM assembly plants. By comparison, only one of the six automotive assembly plants could be differentiated from the other five assembly plants using Fourier transform infrared spectroscopy (FT-IR), which is the most widely used analytical method for the examination of automotive paint) and the genetic algorithm for pattern recognition. The results of this study indicate that Raman spectroscopy in combination with pattern recognition methods offers distinct advantages over FT-IR for the identification and discrimination of automotive clearcoats.

Analysis of Automotive Paint Smears Using Attenuated Total Reflection Infrared Microscopy.

Paint smears represent a type of automotive paint sample found at a crime scene that is problematic for forensic automotive paint examiners to analyze as there are no reference materials present in automotive paint databases to generate hit-lists of potential suspect vehicles. Realistic paint smears are difficult to create in a laboratory and have also proven challenging to analyze because of the mixing of the various automotive paint layers. A procedure based on an impact tester has been developed to create smears to simulate paint transfer between vehicles during a collision. Data collected from 24 original equipment manufacturer (OEM) paints in simulated collisions using an impact tester with a steel (inert) substrate to simulate vehicle to vehicle collisions shows that attenuated total reflection infrared microscopy can isolate individual paint layers. For each OEM paint sample, the corresponding smear obtained was dependent upon the conditions used. By varying these conditions, the number of distinct layers obtained could be tuned for each of the OEM paints investigated. Furthermore, the IR spectrum of each layer extracted from the paint smear using alternating least squares was found to compare favorably to an in-house OEM paint infrared spectral library for each layer as the correct match (make and model of the vehicle from which the smear originated) was always found as a top five hit in the hit-list. The results of this study indicate that paint smears developed using an impactor can serve as the basis of realistic proficiency tests for forensic laboratories.

Gluten Conformation at Different Temperatures and Additive Treatments.

The effect of temperature (25, 45, and 65 °C) on the gluten secondary structure was investigated by using Fourier transform infrared (FTIR) spectroscopy and modulation of disulfide and hydrogen bonds contributions (100 ppm ascorbic acid (AA), 0.6% diacetyl tartaric acid ester of monoglycerides (DATEM), and 0.25 mM dithiothreitol (DTT)). The results showed that additives heated at 65 °C altered most of the gluten matrix formation by changing structural secondary structures compared to the secondary structures of native gluten (control). The content of random coils, α-helices, and ß-sheet of gluten increased, while the extent of ß-turns and antiparallel ß-sheets decreased, which led to the transformation to a more stable secondary conformation. In addition, the rheological properties (%creep strain) revealed that gluten deformation increased during the heating process with all of the additives. The chemometric method could quantitate an overall alteration of gluten polymerization and gluten matrix formation during heating with additive treatments.

Transmission Infrared Microscopy and Machine Learning Applied to the Forensic Examination of Original Automotive Paint.

Alternate least squares (ALS) reconstructions of the infrared (IR) spectra of the individual layers from original automotive paint were analyzed using machine learning methods to improve both the accuracy and speed of a forensic automotive paint examination. Twenty-six original equipment manufacturer (OEM) paints from vehicles sold in North America between 2000 and 2006 served as a test bed to validate the ALS procedure developed in a previous study for the spectral reconstruction of each layer from IR line maps of cross-sectioned OEM paint samples. An examination of the IR spectra from an in-house library (collected with a high-pressure transmission diamond cell) and the ALS reconstructed IR spectra of the same paint samples (obtained at ambient pressure using an IR transmission microscope equipped with a BaF2 cell) showed large peak shifts (approximately 10 cm-1) with some vibrational modes in many samples comprising the cohort. These peak shifts are attributed to differences in the residual polarization of the IR beam of the transmission IR microscope and the IR spectrometer used to collect the in-house IR spectral library. To solve the problem of frequency shifts encountered with some vibrational modes, IR spectra from the in-house spectral library and the IR microscope were transformed using a correction algorithm previously developed by our laboratory to simulate ATR spectra collected on an iS-50 FT-IR spectrometer. Applying this correction algorithm to both the ALS reconstructed spectra and in-house IR library spectra, the large peak shifts previously encountered with some vibrational modes were successfully mitigated. Using machine learning methods to identify the manufacturer and the assembly plant of the vehicle from which the OEM paint sample originated, each of the twenty-six cross-sectioned automotive paint samples was correctly classified as to the "make" and model of the vehicle and was also matched to the correct paint sample in the in-house IR spectral library.

Synthesis and Characterization of N-Isopropylacrylamide Microspheres as pH Sensors.

Swellable polymer microspheres that respond to pH were prepared by free radical dispersion polymerization using N-isopropylacrylamide (NIPA), N,N'-methylenebisacrylamide (MBA), 2,2-dimethoxy-2-phenylacetylphenone, N-tert-butylacrylamide (NTBA), and a pH-sensitive functional comonomer (acrylic acid, methacrylic acid, ethacrylic acid, or propacrylic acid). The diameter of the microspheres was between 0.5 and 1.0 µm. These microspheres were cast into hydrogel membranes prepared by mixing the pH-sensitive swellable polymer particles with aqueous polyvinyl alcohol (PVA) solutions followed by crosslinking with glutaric dialdehyde for use as pH sensors. Large changes in the turbidity of the PVA membrane were observed as the pH of the buffer solution in contact with the membrane was varied. These changes were monitored by UV-visible absorbance spectroscopy. Polymer swelling of many NIPA copolymers was reversible and independent of the ionic strength of the buffer solution in contact with the membrane. Both the degree of swelling and the apparent pKa of the polymer microspheres increased with temperature. Furthermore, the apparent pKa of the polymer particles could be tuned to respond sharply to pH in a broad range (pH 4.0-7.0) by varying the amount of crosslinker (MBA) and transition temperature modifier (NTBA), and the amount, pKa, and hydrophobicity of the pH-sensitive functional comonomer (alkyl acrylic acid) used in the formulation. Potential applications of these polymer particles include fiber optic pH sensing where the pH-sensitive material can be immobilized on the distol end of an optical fiber.

Comparison of Spectroscopic Techniques for Determining the Peroxide Value of 19 Classes of Naturally Aged, Plant-Based Edible Oils.

The peroxide value of edible oils is a measure of the degree of oxidation, which directly relates to the freshness of the oil sample. Several studies previously reported in the literature have paired various spectroscopic techniques with multivariate analyses to rapidly determine peroxide values using field portable and process instrumentation; those efforts presented "best-case scenarios" with oils from narrowly defined training and test sets. The purpose of this paper is to evaluate the use of near- and mid-infrared absorption and Raman scattering spectroscopies on oil samples from different oil classes, including seasonal and vendor variations, to determine which measurement technique or combination thereof is best for predicting peroxide values. Following peroxide value assays of each oil class using an established titration-based method, global and global-subset calibration models were constructed from spectroscopic data collected on the 19 oil classes used in this study. Spectra from each optical technique were used to create partial least squares regression calibration models to predict the peroxide value of unknown oil samples. A global peroxide value model based on near-infrared (8 mm optical path length) oil measurements produced the lowest RMSEP (4.9), followed by 24 mm optical path length near infrared (5.1), Raman (6.9) and 50 µm optical path length mid-infrared (7.3). However, it was determined that the Raman RMSEP resulted from chance correlations. Global peroxide value models based on low-level fusion of the NIR (8 and 24 mm optical path length) data and all infrared data produced the same RMSEP of 5.1. Global subset models, based on any of the spectroscopies and olive oil training sets from any class (pure, extra light, extra virgin), all failed to extrapolate to the non-olive oils. However, the near-infrared global subset model built on extra virgin olive oil could extrapolate to test samples from other olive oil classes. This work demonstrates the difficulty of developing a truly global method for determining peroxide value of oils.

Swellable Copolymers of N-isopropylacrylamide and Alkyl Acrylic Acids for Optical pH Sensing.

Swellable polymers that respond to pH (including a portion of the physiological pH range) have been prepared from N-isopropylacrylamide (NIPA) copolymerized with acrylic acid, methacrylic acid, ethacrylic acid or propacrylic acid by dispersion polymerization. When the swellable polymer particles are dispersed in a polyvinyl alcohol (PVA) hydrogel membrane, large changes occur in the turbidity of the membrane (which is measured using an absorbance spectrometer) as the pH of the buffer solution in contact with the hydrogel membrane is varied. The swelling of the NIPA copolymer is nonionic, as the ionic strength of the buffer solution in contact with the PVA membrane was increased from 0.1 to 1.0 M without a decrease in the swelling. For many of these NIPA copolymers, swelling was also reversible in both low- and high ionic strength pH-buffered media and at ambient and physiological temperatures. The composition of the formulation used to prepare these copolymers of NIPA can be correlated to the enthalpy and entropy of the pH-induced swelling.

Differentiation of Edible Oils by Type Using Raman Spectroscopy and Pattern Recognition Methods.

The application of Raman spectroscopy and pattern recognition methods to the problem of discriminating edible oils by type was investigated. Two-hundred and eighty-six Raman spectra obtained from 53 samples spanning 15 varieties of edible oils were collected for 90 s at 2 cm-1 resolution. Employing a Whittaker filter, all Raman spectra were baseline corrected after removing the high-intensity fluorescent background in each spectrum. The Raman spectral data were then examined using the three major types of pattern recognition methodology: mapping and display, discriminant development and clustering. The 15 varieties of edible oils could be partitioned into five distinct groups based on their degree of saturation and the ratio of polyunsaturated fatty acids to monounsaturated fatty acids. Edible oils assigned to one group could be readily differentiated from those assigned to other groups, whereas Raman spectra within the same group more closely resembled each other and therefore would be more difficult to classify by type.

pH sensing using whispering gallery modes of a silica hollow bottle resonator.

The limitations of electrochemical pH sensors have stimulated the development of optical pH sensing methods. In the method reported here, swellable pH-sensitive polymer particles are deposited on the interior surface of a silica hollow bottle resonator. As the pH of the buffer solution in contact with the particles increases, the refractive index of the particles decreases. As a result, whispering gallery modes with internal evanescent components shift in frequency as a function of pH. This shift is monitored by the throughput of tunable diode laser light coupled into the whispering gallery modes using a tapered fiber. Plots of selected mode frequencies vs. pH yielded sigmoid shaped titration curves similar to those obtained using turbidity to monitor refractive index changes of the particles as a function of pH. The response time of 10-15 s and best resolution of 0.06 pH unit represent improvements over previous optical pH sensing methods.

Analysis of gentisic acid and related renal cell carcinoma biomarkers using reversed-phase liquid chromatography with water-rich mobile phases.

The problem of longer retention times using water-rich mobile phases in reversed phase liquid chromatography (RPLC) has been addressed using hydrophobic alcohols such as butanol in very low quantities (approximately 0.1%) as the organic modifier. Advantages of water-rich mobile phases in RPLC for the separation of water-soluble and weakly retained compounds are improved separation of congeners and better tuning of RPLC separations. This is demonstrated in the separation of gentisic acid and related renal cell carcinoma (RCC) biomarkers in urine with a Zorbax C18 column and a mobile phase of 0.1% (volume/volume) butanol in water with 0.6% (volume/volume) acetic acid. Calibration curves for the RCC biomarkers were linear over the concentration range investigated (5 ppm to 1000 ppm). Detection limits for the RCC biomarkers were 0.85ppm (quinolinic acid), 1.75ppm (gentisic acid), and 1.25ppm (4-hydroxybenzoic acid). Recovery tests using synthetic urine samples containing 20 ppm, 100 ppm, and 700 pm of each RCC biomarker were successful for all compounds.

Transmission infrared imaging microscopy and multivariate curve resolution applied to the forensic examination of automotive paints.

In the forensic examination of automotive paint, each layer of paint is analyzed individually by infrared (IR) spectroscopy. Laboratories in North America typically hand section each layer and present each separated layer to the spectrometer for analysis, which is time consuming. In addition, sampling too close to the boundary between adjacent layers can pose a problem as it produces an IR spectrum that is a mixture of the two layers. Not having a "pure" spectrum of each layer will prevent a meaningful comparison between each paint layer or in the situation of searching an automotive database will prevent the forensic paint examiner from developing an accurate hit list of potential suspects. These two problems can be addressed by collecting concatenated IR data from all paint layers in a single analysis by scanning across the cross sectioned layers of the paint sample using a FTIR imaging microscope. Decatenation of the IR data is achieved by multivariate curve resolution using a Varimax extended rotation to select the starting point (i.e., initial estimates of the reconstructed IR spectra of each layer) for the alternating least squares algorithm to obtain a pure IR spectrum of each automotive paint layer. Comparing the reconstructed IR spectrum of each layer against the IR spectral library of the PDQ database demonstrated that it is possible to identify the correct model of the vehicle from these reconstructed spectra. This imaging approach to IR analysis of automotive paint, not only saves time and eliminates the need to analyze each layer separately, but also ensures that the final spectrum of each layer is "pure" and not a mixture.

Development of Infrared Library Search Prefilters for Automotive Clear Coats from Simulated Attenuated Total Reflection (ATR) Spectra.

A previously published study featuring an attenuated total reflection (ATR) simulation algorithm that mitigated distortions in ATR spectra was further investigated to evaluate its efficacy to enhance searching of infrared (IR) transmission libraries. In the present study, search prefilters were developed from transformed ATR spectra to identify the assembly plant of a vehicle from ATR spectra of the clear coat layer. A total of 456 IR transmission spectra from the Paint Data Query (PDQ) database that spanned 22 General Motors assembly plants and served as a training set cohort were transformed into ATR spectra by the simulation algorithm. These search prefilters were formulated using the fingerprint region (1500 cm-1 to 500 cm-1). Both the transformed ATR spectra (training set) and the experimental ATR spectra (validation set) were preprocessed for pattern recognition analysis using the discrete wavelet transform, which increased the signal-to-noise of the ATR spectra by concentrating the signal in specific wavelet coefficients. Attenuated total reflection spectra of 14 clear coat samples (validation set) measured with a Nicolet iS50 Fourier transform IR spectrometer were correctly classified as to assembly plant(s) of the automotive vehicle from which the paint sample originated using search prefilters developed from 456 simulated ATR spectra. The ATR simulation (transformation) algorithm successfully facilitated spectral library matching of ATR spectra against IR transmission spectra of automotive clear coats in the PDQ database.

Library Search Prefilters for Vehicle Manufacturers to Assist in the Forensic Examination of Automotive Paints.

Pattern recognition techniques have been applied to the infrared (IR) spectral libraries of the Paint Data Query (PDQ) database to differentiate between nonidentical but similar IR spectra of automotive paints. To tackle the problem of library searching, search prefilters were developed to identify the vehicle make from IR spectra of the clear coat, surfacer-primer, and e-coat layers. To develop these search prefilters with the appropriate degree of accuracy, IR spectra from the PDQ database were preprocessed using the discrete wavelet transform to enhance subtle but significant features in the IR spectral data. Wavelet coefficients characteristic of vehicle make were identified using a genetic algorithm for pattern recognition and feature selection. Search prefilters to identify automotive manufacturer through IR spectra obtained from a paint chip recovered at a crime scene were developed using 1596 original manufacturer's paint systems spanning six makes (General Motors, Chrysler, Ford, Honda, Nissan, and Toyota) within a limited production year range (2000-2006). Search prefilters for vehicle manufacturer that were developed as part of this study were successfully validated using IR spectra obtained directly from the PDQ database. Information obtained from these search prefilters can serve to quantify the discrimination power of original automotive paint encountered in casework and further efforts to succinctly communicate trace evidential significance to the courts.

Chemometric Methods for Estimating the Strain Hardening Modulus in Polyethylene Resins.

The feasibility of using multiway or N-way partial least square (NPLS) methods to estimate physical properties of 1-butene and 1-hexene polyethylene (PE) copolymers directly from multidimensional data obtained from size exclusion chromatography coupled to a Fourier transform infrared detector (SEC FT-IR) was explored. Digital sample sets of horizontal slices (slabs) of two-dimensional data simulating the molecular weight distribution and the corresponding orthogonal FT-IR spectra were correlated to a particular Y-block response using NPLS. The NPLS results were compared to those obtained through separate estimations using various algorithms and exploratory response surface methods. The estimated strain hardening modulus () for bimodal PE-like digital structures could adequately be modeled using both the linear response surface method (RSM) and NPLS. Although different input values were used, the predicted values for by NPLS was found to mirror both the analytical results and the expected structural effects obtained using linear RSM models.

Boosting the Performance of Genetic Algorithms for Variable Selection in Partial Least Squares Spectral Calibrations.

A genetic algorithm (GA) for variable selection in partial least squares (PLS) regression that incorporates adaptive boosting to identify informative wavelengths in near-infrared (NIR) spectra has been developed. Three studies demonstrating the advantages of incorporating an adaptive boosting routine into a GA that employs the root mean square error of calibration as its fitness function are highlighted: (1) prediction of hydroxyl number of terpolymers from NIR diffuse reflectance spectra; (2) calibration of acetone from NIR transmission spectra of mixtures of water, acetone, t-butyl alcohol and isopropyl alcohol; and (3) determination of the active pharmaceutical ingredients in drug tablets from NIR diffuse reflectance spectra. The performance of the GA with adaptive boosting to select wavelengths was compared with one without adaptive boosting. For all three NIR data sets, variable selected PLS models developed by a GA with adaptive boosting performed better. Analysis of the wavelengths selected by the GA with adaptive boosting also demonstrate that chemical information indicative of the analyte was captured by the selected wavelengths.

Pattern Recognition-Assisted Infrared Library Searching of the Paint Data Query Database to Enhance Lead Information from Automotive Paint Trace Evidence.

Multilayered automotive paint fragments, which are one of the most complex materials encountered in the forensic science laboratory, provide crucial links in criminal investigations and prosecutions. To determine the origin of these paint fragments, forensic automotive paint examiners have turned to the paint data query (PDQ) database, which allows the forensic examiner to compare the layer sequence and color, texture, and composition of the sample to paint systems of the original equipment manufacturer (OEM). However, modern automotive paints have a thin color coat and this layer on a microscopic fragment is often too thin to obtain accurate chemical and topcoat color information. A search engine has been developed for the infrared (IR) spectral libraries of the PDQ database in an effort to improve discrimination capability and permit quantification of discrimination power for OEM automotive paint comparisons. The similarity of IR spectra of the corresponding layers of various records for original finishes in the PDQ database often results in poor discrimination using commercial library search algorithms. A pattern recognition approach employing pre-filters and a cross-correlation library search algorithm that performs both a forward and backward search has been used to significantly improve the discrimination of IR spectra in the PDQ database and thus improve the accuracy of the search. This improvement permits inter-comparison of OEM automotive paint layer systems using the IR spectra alone. Such information can serve to quantify the discrimination power of the original automotive paint encountered in casework and further efforts to succinctly communicate trace evidence to the courts.

Evidential significance of automotive paint trace evidence using a pattern recognition based infrared library search engine for the Paint Data Query Forensic Database.

A prototype library search engine has been further developed to search the infrared spectral libraries of the paint data query database to identify the line and model of a vehicle from the clear coat, surfacer-primer, and e-coat layers of an intact paint chip. For this study, search prefilters were developed from 1181 automotive paint systems spanning 3 manufacturers: General Motors, Chrysler, and Ford. The best match between each unknown and the spectra in the hit list generated by the search prefilters was identified using a cross-correlation library search algorithm that performed both a forward and backward search. In the forward search, spectra were divided into intervals and further subdivided into windows (which corresponds to the time lag for the comparison) within those intervals. The top five hits identified in each search window were compiled; a histogram was computed that summarized the frequency of occurrence for each library sample, with the IR spectra most similar to the unknown flagged. The backward search computed the frequency and occurrence of each line and model without regard to the identity of the individual spectra. Only those lines and models with a frequency of occurrence greater than or equal to 20% were included in the final hit list. If there was agreement between the forward and backward search results, the specific line and model common to both hit lists was always the correct assignment. Samples assigned to the same line and model by both searches are always well represented in the library and correlate well on an individual basis to specific library samples. For these samples, one can have confidence in the accuracy of the match. This was not the case for the results obtained using commercial library search algorithms, as the hit quality index scores for the top twenty hits were always greater than 99%.

Pattern recognition-assisted infrared library searching of automotive clear coats.

Pattern recognition techniques have been developed to search the infrared (IR) spectral libraries of the paint data query (PDQ) database to differentiate between similar but nonidentical IR clear coat paint spectra. The library search system consists of two separate but interrelated components: search prefilters to reduce the size of the IR library to a specific assembly plant or plants corresponding to the unknown paint sample and a cross-correlation searching algorithm to identify IR spectra most similar to the unknown in the subset of spectra identified by the prefilters. To develop search prefilters with the necessary degree of accuracy, IR spectra from the PDQ database were preprocessed using wavelets to enhance subtle but significant features in the data. Wavelet coefficients characteristic of the assembly plant of the vehicle were identified using a genetic algorithm for pattern recognition and feature selection. A search algorithm was then used to cross-correlate the unknown with each IR spectrum in the subset of library spectra identified by the search prefilters. Each cross-correlated IR spectrum was simultaneously compared to an autocorrelated IR spectrum of the unknown using several spectral windows that span different regions of the cross-correlated and autocorrelated data from the midpoint. The top five hits identified in each search window are compiled, and a histogram is computed that summarizes the frequency of occurrence for each selected library sample. The five library samples with the highest frequency of occurrence are selected as potential hits. Even in challenging trials where the clear coat paint samples evaluated were all the same make (e.g., General Motors) within a limited production year range, the model of the automobile from which the unknown paint sample was obtained could be identified from its IR spectrum.