Classification using partial least squares with penalized logistic regression

MOTIVATION: One important aspect of data-mining of microarray data is to discover the molecular variation among cancers. In microarray studies, the number n of samples is relatively small compared to the number p of genes per sample (usually in thousands). It is known that standard statistical methods in classification are efficient (i.e. in the present case, yield successful classifiers) particularly when n is (far) larger than p. This naturally calls for the use of a dimension reduction procedure together with the classification one. RESULTS: In this paper, the question of classification in such a high-dimensional setting is addressed. We view the classification problem as a regression one with few observations and many predictor variables. We propose a new method combining partial least squares (PLS) and Ridge penalized logistic regression. We review the existing methods based on PLS and/or penalized likelihood techniques, outline their interest in some cases and theoretically explain their sometimes poor behavior. Our procedure is compared with these other classifiers. The predictive performance of the resulting classification rule is illustrated on three data sets: Leukemia, Colon and Prostate.     

Investigation of the effect of clinically relevant interferents on glucose monitoring using near-infrared spectroscopy

Near infrared spectroscopy (NIR) is a promising technique for continuous blood glucose monitoring for diabetic patients. Four interferents, at physiological concentrations, were introduced to study how the glucose predictions varied with a standard multivariate calibration model. Lactate and ethanol were found to interfere strongly with the glucose predictions unless they were included in the calibration models. Lactate was mistaken for glucose and gave erroneously high glucose predictions, with a dose response of 0.46 mM/mM. The presence of ethanol resulted in too low glucose predictions, with a dose response of −0.43 mM/mM. Acetaminophen, a known interferent in the glucose monitoring devices used for diabetes management today, was not found to be an interferent in NIR spectroscopy, nor was caffeine. Thus, interferents that may appear in high concentrations, such as ethanol and lactate, must be included in the calibration or model building of future NIR-based glucose measurement devices for diabetes monitoring.     

Quantification of hydroxycinnamic acids and lignin in perennial forage and energy grasses by Fourier-transform infrared spectroscopy and partial least squares regression

Levels of lignin and hydroxycinnamic acid wall components in three genera of forage grasses (Lolium,Festuca and Dactylis) have been accurately predicted by Fourier-transform infrared spectroscopy using partial least squares models correlated to analytical measurements. Different models were derived that predicted the concentrations of acid detergent lignin, total hydroxycinnamic acids, total ferulate monomers plus dimers, p-coumarate and ferulate dimers in independent spectral test data from methanol extracted samples of perennial forage grass with accuracies of 92.8%, 86.5%, 86.1%, 59.7% and 84.7% respectively, and analysis of model projection scores showed that the models relied generally on spectral features that are known absorptions of these compounds. Acid detergent lignin was predicted in samples of two species of energy grass, (Phalaris arundinacea and Pancium virgatum) with an accuracy of 84.5%.     

In‐situ monitoring of Saccharomyces cerevisiae ITV01 bioethanol process using near‐infrared spectroscopy NIRS and chemometrics

The application feasibility of in‐situ or in‐line monitoring of S. cerevisiae ITV01 alcoholic fermentation process, employing Near‐Infrared Spectroscopy (NIRS) and Chemometrics, was investigated. During the process in a bioreactor, in the complex analytical matrix, biomass, glucose, ethanol and glycerol determinations were performed by a transflection fiber optic probe immersed in the culture broth and connected to a Near‐Infrared (NIR) process analyzer. The NIR spectra recorded between 800 and 2,200 nm were pretreated using Savitzky‐Golay smoothing and second derivative in order to perform a partial least squares regression (PLSR) and generate the calibration models. These calibration models were tested by external validation and then used to predict concentrations in batch alcoholic fermentations. The standard errors of calibration (SEC) for biomass, ethanol, glucose and glycerol were 0.212, 0.287, 0.532, and 0.296 g/L and standard errors of prediction (SEP) were 0.323, 0.369, 0.794, and 0.507 g/L, respectively. Calibration and validation criteria were defined and evaluated in order to generate robust and reliable models for an alcoholic fermentation process matrix. © 2016 American Institute of Chemical Engineers Biotechnol. Prog., 32:510–517, 2016     

Visible‐near‐infrared luminescent lanthanide ternary complexes based on beta‐diketonate using visible‐light excitation

We used the synthesized dinaphthylmethane (Hdnm) ligand whose absorption extends to the visible‐light wavelength, to prepare a family of ternary lanthanide complexes, named as [Ln(dnm)₃phen] (Ln = Sm, Nd, Yb, Er, Tm, Pr). The properties of these complexes were investigated by Fourier transform infrared (FT‐IR) spectroscopy, diffuse reflectance (DR) spectroscopy, thermogravimetric analyses, and excitation and emission spectroscopy. Generally, excitation with visible light is much more advantageous than UV excitation. Importantly, upon excitation with visible light (401–460 nm), the complexes show characteristic visible (Sm³⁺) as well as near‐infrared (Sm³⁺, Nd³⁺, Yb³⁺, Er³⁺, Tm³⁺, Pr³⁺) luminescence of the corresponding lanthanide ions, attributed to the energy transfer from the ligands to the lanthanide ions, an antenna effect. Now, using these near‐infrared luminescent lanthanide complexes, the luminescent spectral region from 800 to 1650 nm, can be covered completely, which is of particular interest for biomedical imaging applications, laser systems, and optical amplification applications. Copyright © 2015 John Wiley & Sons, Ltd.     

Measurement of hemoglobin in whole blood using a partial least squares regression model with selected second derivative near infrared transmission spectral signals

In this work, we propose a signal selection procedure for determination of hemoglobin (Hb) concentration in whole blood using near infrared (NIR) transmission spectral signals. A dataset of 190 whole blood NIR transmission spectra with reference Hb concentrations was used to evaluate the method. Spectral signals were selected based on the squared correlation coefficient (R(2)) between the signal and the Hb concentration. An improved uninformative variable elimination (UVE) procedure was performed to remove redundant signals from the primary selected signal set. A partial least squares (PLS) regression model was built with the final selected signals and the corresponding Hb concentrations. The results indicate that the proposed method is effective at increasing the predictive power of the NIR-PLS spectral model for determining Hb concentration in whole blood samples.     

Real‐time monitoring prefrontal activities during online video game playing by functional near‐infrared spectroscopy

A growing body of literature has suggested that video game playing can induce functional and structural plasticity of the brain. The underlying mechanisms, however, remain poorly understood. In this study, functional near‐infrared spectroscopy (fNIRS) was used to record prefrontal activities in 24 experienced game players when they played a massively multiplayer online battle arena video game, League of Legends (LOL), under naturalistic conditions. It was observed that game onset was associated with significant activations in the ventrolateral prefrontal cortex (VLPFC) and concomitant deactivations in the dorsolateral prefrontal cortex (DLPFC) and frontal pole area (FPA). Game events, such as slaying an enemy and being slain by an enemy evoked region‐specific time‐locked hemodynamic/oxygenation responses in the prefrontal cortex (PFC). It was proposed that the VLPFC activities during LOL playing are likely responses to visuo‐motor task load of the game, while the DLPFC/FPA activities may be involved in the constant shifts of attentional states and allocation of cognitive resources required by game playing. The present study demonstrated that it is feasible to use fNIRS to monitor real‐time prefrontal activity during online video game playing.      

The utility of far‐infrared illumination in oxygenation dynamics as measured with near‐infrared spectroscopy

Near‐infrared spectroscopy (NIRS) is a noninvasive method for measuring the oxygenation in muscle and other tissues in vivo. For quantitative NIRS measurement of oxygenation dynamics, the vessel‐occlusion test was usually applied as physiological intervention. There are several drawbacks of the vessel‐occlusion method that include skin contact, uncomfortable and microcirculation block of patients. Thus, we propose the far‐infrared (FIR) illumination as a new physiological intervention method in this paper. Our preliminary result shows a linear correlation of oxygenation dynamic signals between FIR illumination and arterial‐occlusion test (AOT) that implies the FIR illumination could be applied for hemodynamic response measurement in clinical diagnosis. (© 2012 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)     

Altered near‐infrared spectroscopy response to breath‐holding in patients with fibromyalgia

Fibromyalgia (FM) is a complex syndrome characterized by chronic widespread pain and a heightened response to pressure. Most medical researches pointed out that FM patients with endothelial dysfunction and arterial stiffness. A continuous‐wave near‐infrared spectroscopy (NIRS) system is used in present study to measure the hemodynamic changes elicited by breath‐holding task in patients with FM. Each patient completed a questionnaire survey including demographics, characteristics of body pain, associated symptoms, headache profiles and Hospital Anxiety and Depression Scale. A total of 27 FM patients and 26 health controls were enrolled. In comparison with healthy controls, patients with FM showed lower maximal and averaged change of oxyhemoglobin concentration in both the left (1.634 ±0.890 and 0.810 ±0.525 μM) and the right (1.576 ±0.897 and 0.811 ±0.601 μM) prefrontal cortex than healthy controls (P < .05 for both sides) during the breath‐holding task. In conclusion, FM is associated with altered cerebrovascular reactivity measured by NIRS and breath‐holding task, which may reflect endothelial dysfunction or arterial stiffness. Oxygenated hemoglobin concentration changes of healthy controls and FM patients.      

Cu‐doped quantum dots: a new class of near‐infrared emitting fluorophores for bioanalysis and bioimaging

Transition metal ion‐doped quantum dots (QDs) exhibit unique optical and photophysical properties that offer significant advantages over undoped QDs, such as larger Stokes shift to avoid self‐absorption/energy transfer, longer excited‐state lifetimes, wider spectral window, and improved chemical and thermal stability. Among the doped QDs emitters, Cu is widely introduced into the doped QDs as novel, efficient, stable, and tunable optical materials that span a wide spectrum from blue to near‐infrared (NIR) light. Their unique physical and chemical characteristics enable the use of Cu‐doped QDs as NIR labels for bioanalysis and bioimaging. In this review, we discuss doping mechanisms and optical properties of Cu‐doped QDs that are capable of NIR emission. Applications of Cu‐doped QDs in in vitro biosensing and in in vivo bioimaging are highlighted. Moreover, a prospect of the future of Cu‐doped QDs for bioanalysis and bioimaging are also summarized.     

On-line monitoring of the transesterification reaction between triglycerides and ethanol using near infrared spectroscopy combined with gas chromatography

Many analytical procedures have been developed to determine the composition of reaction mixtures during transesterification of vegetable oils with alcohols. However, despite their accuracy, these methods are time consuming and cannot be easily used for on-line monitoring. In this work, a fast analytical method was developed to on-line monitor the transesterification reaction of high oleic sunflower oil with ethanol using Near InfraRed spectroscopy and a multivariate approach. The reactions were monitored through sequential scans of the reaction medium with a probe in a one-liter batch reactor without collecting and preparing samples. To calibrate the NIR analytical method, gas chromatography-flame ionization detection was used as a reference method. The method was validated by studying the kinetics of the EtONa-catalyzed transesterification reaction. Activation energy (51.0 kJ/mol) was also determined by considering a pseudo second order kinetics model.     

Three‐dimensional cellular imaging in thick biological tissue with confocal detection of one‐photon fluorescence in the near‐infrared II window

Fluorescence imaging in the second near‐infrared optical window (NIR‐II, 900‐1700 nm) has become a technique of choice for noninvasive in vivo imaging in recent years. Greater penetration depths with high spatial resolution and low background can be achieved with this NIR‐II window, owing to low autofluorescence within this optical range and reduced scattering of long wavelength photons. Here, we present a novel design of confocal laser scanning microscope tailored for imaging in the NIR‐II window. We showcase the outstanding penetration depth of our confocal setup with a series of imaging experiments. HeLa cells labeled with PbS quantum dots with a peak emission wavelength of 1276 nm can be visualized through a 3.5‐mm‐thick layer of scattering medium, which is a 0.8% Lipofundin solution. A commercially available organic dye IR‐1061 (emission peak at 1132 nm), in its native form, is used for the first time, as a NIR‐II fluorescence label in cellular imaging. Our confocal setup is capable of capturing optically sectioned images of IR‐1061 labeled chondrocytes in fixed animal cartilage at a depth up to 800 μm, with a superb spatial resolution of around 2 μm.      

Oxygenation dynamics of sepsis patients undergoing far‐infrared intervention based on near‐infrared spectroscopy

Near‐infrared spectroscopy (NIRS; continuous wave type) is a noninvasive tool for detecting the relative change of oxyhemoglobin and deoxyhemoglobin. To make this change, intervention methods must be applied. This study determined the hemodynamics of 44 healthy participants and 35 patients with sepsis during exposure to FIR as a novel physical intervention approach. Local microcirculation of their brachioradialis was monitored during exposure and recovery through NIRS. The variations in blood flow and microvascular reaction were determined by conducting paired and unpaired t tests. The oxyhemoglobin levels of the healthy participants increased continuously, even during recovery. In contrast to expextations, the oxyhemoglobin levels of the patients plateaued after only 5 min of FIR illumination. The proposed method has potential applications for ensuring efficient treatment and facilitating doctors in diagnosing the functions of vessels in intensive care units.

Mapping diagrams of HbO₂ in healthy males and males with sepsis illustrated unique scenarios during the process.     

A chemotaxonomic assessment of four indigenous South African Lippia species using GC–MS and vibrational spectroscopy of the essential oils

Lippia species are widely used as traditional remedies by many South African communities. The identification of Lippia spp. indigenous to South Africa is currently based on the morphology and geographical origin of the specimen. However, morphological similarities are evident and due to sympatric distributions, different species may co-occur in the same natural environment, thereby presenting difficulties in the taxonomic delimitation of these taxa. In this paper, gas chromatography–mass spectrometry (GC–MS) and vibrational spectroscopic analyses were used to investigate the chemotaxonomic differentiation of four Lippia species (L. javanica, L. scaberrima, L. rehmannii and L. wilmsii), based on the essential oil composition. The results demonstrate that multivariate classification of chromatographic data enabled separation of Lippia oils into three distinct clusters representing L. scaberrima, L. rehmannii and L. javanica. With the aid of chemometric algorithms, it was also possible to use both mid infrared (MIR) and near infrared (NIR) spectroscopy for a clear distinction of the Lippia spp., with results comparable to those obtained by GC–MS. Orthogonal projections to latent structures (OPLS) presented a better classification algorithm than principal component analysis (PCA), in both GC–MS and spectroscopy analyses. Chemical variations within the Lippia spp. complex were evident, particularly for L. javanica and L. scaberrima, which exhibited significant intra-species variation.     

Multispectral photoacoustic imaging of cancer with broadband CuS nanoparticles covering both near‐infrared I and II biological windows

In this study, CuS nanoparticles with optical absorption covering both near‐infrared I (NIR‐I) and NIR‐II biological windows were prepared and served as the contrast agents for multispectral photoacoustic imaging. The physiological parameters including concentrations of deoxyhemoglobin and oxyhemoglobin as well as the water content in the tumor location were quantified based on the multispectral photoacoustic reconstruction method. More importantly, the concentration of CuS nanoparticles/drugs accumulated in the tumor was also recovered after intravenously injection, which are essential for image‐guided cancer theranostics. In addition, phantom and in vivo experimental tests were performed to inspect and compare the imaging depth and signal‐to‐noise ratio (SNR) between the two NIR biological windows. Interestingly, we discovered that a higher SNR was obtained in the NIR‐II window than that in the NIR‐I window. Meanwhile, the multispectral imaging results also demonstrated that the imaging contrast and penetration depth in the NIR‐II window were also significantly improved as compared to those from the NIR‐I window.      

Application of near‐infrared (NIR) spectroscopy for screening of raw materials used in the cell culture medium for the production of a recombinant therapeutic protein

Control of raw materials based on an understanding of their impact on product attributes has been identified as a key aspect of developing a control strategy in the Quality by Design (QbD) paradigm. This article presents a case study involving use of a combined approach of Near‐infrared (NIR) spectroscopy and Multivariate Data Analysis (MVDA) for screening of lots of basal medium powders based on their impact on process performance and product attributes. These lots had identical composition as per the supplier and were manufactured at different scales using an identical process. The NIR/MVDA analysis, combined with further investigation at the supplier site, concluded that grouping of medium components during the milling and blending process varied with the scale of production and media type. As a result, uniformity of blending, impurity levels, chemical compatibility, and/or heat sensitivity during the milling process for batches of large‐scale media powder were deemed to be the source of variation as detected by NIR spectra. This variability in the raw materials was enough to cause unacceptably large variability in the performance of the cell culture step and impact the attributes of the resulting product. A combined NIR/MVDA approach made it possible to finger print the raw materials and distinguish between good and poor performing media lots. © 2009 American Institute of Chemical Engineers Biotechnol. Prog., 2010     

Tendinopathy diagnosis and treatment monitoring using attenuated total reflectance‐Fourier transform infrared spectroscopy

Tendinopathy, an important sports injury afflicting athletes and general public, is associated with huge economic losses. The currently used diagnostic tests are subjective, show moderate sensitivity and specificity; while treatment failures persist despite advances in therapy. This highlights the need for tendinopathy diagnostic and treatment monitoring tools. This study investigates tendon injury, natural healing and effect of treatment using ATR‐FTIR complemented with histopathology. Control (C), injured (I) and treated (T) rat tendons were extracted 3, 7, 14 and 28 days post‐injury/treatment, representing phases of healing; and subjected to hematoxylin & eosin staining as well as spectroscopy. While C showed no change, I‐ and T‐related histological changes could be clearly observed in stained sections. ATR‐FTIR spectra highlighted the biochemical changes within groups. Multivariate analysis could classify C, I and T with 75%; different days between groups with 84%; and different days within group with 65% efficiency. Results suggest that such analysis can not only identify C, I or T but also different phases of healing. Difference between I and T at different time points also suggest change in rate of healing. Further studies may help develop this technique for clinical diagnosis and treatment monitoring in future.