Monitoring galenical process development by near infrared chemical imaging: One case study

Evaluation of the Miscibility of Novel Cocoa Butter Equivalents by Raman Mapping and Multivariate Curve Resolution-Alternating Least Squares

Cocoa butter (CB) is an ingredient traditionally used in the manufacturing of chocolates, but its availability is decreasing due to its scarcity and high cost. For this reason, other vegetable oils, known as cocoa butter equivalents (CBE), are used to replace CB partially or wholly. In the present work, two Peruvian vegetable oils, coconut oil (CNO) and sacha inchi oil (SIO), are proposed as novel CBEs. Confocal Raman microscopy (CRM) was used for the chemical differentiation and polymorphism of these oils with CB based on their Raman spectra. To analyze their miscibility, two types of blends were prepared: CB with CNO, and CB with SIO. Both were prepared at 5 different concentrations (5%, 15%, 25%, 35%, and 45%). Raman mapping was used to obtain the chemical maps of the blends and analyze their miscibility through distribution maps, histograms and relative standard deviation (RSD). These values were obtained with multivariate curve resolution-alternating least squares. The results show that both vegetable oils are miscible with CB at high concentrations: 45% for CNO and 35% for SIO. At low concentrations, their miscibility decreases. This shows that it is possible to consider these vegetable oils as novel CBEs in the manufacturing of chocolates.

Comparison of Raman and Near-Infrared Chemical Mapping for the Analysis of Pharmaceutical Tablets

Raman and near-infrared (NIR) chemical mapping are widely used methods in the pharmaceutical industry to understand the distribution of components within a drug product. Recent advancements in instrumentation have enabled the rapid acquisition of high-resolution images. The comparison of these techniques for the analysis of pharmaceutical tablets has not recently been explored and thus the relative performance of each technique is not currently well defined. Here, the differences in the chemical images obtained by each method are assessed and compared with scanning electron microscopy with energy dispersive X-ray microanalysis (SEM-EDX), as an alternative surface imaging technique to understand the ability of each technique to acquire a chemical image representative of the sample surface. It was found that the Raman data showed the best agreement with the spatial distribution of components observed in the SEM-EDX images. Quantitative and qualitative comparison of the Raman and NIR images revealed a very different spatial distribution of components with regards to domain size and shape. The Raman image exhibited sharper and better discriminated domains of each component, whereas the NIR image was heavily dominated by large pixelated domains. This study demonstrated the superiority of using Raman chemical mapping compared with NIR chemical mapping to produce a chemical image representative of the sample surface using routinely available instrumentation to obtain a better approximation of domain size and shape. This is fundamental for understanding knowledge gaps in current manufacturing processes; particularly relating the relationship between components in the formulation, processing condition, and final characteristics. By providing a means to more accurately visualize the components within a tablet matrix, these areas can all be further understood.

Evaluation of an autoencoder as a feature extraction tool for near-infrared spectroscopic discriminant analysis

The utility of an autoencoder (AE) as a feature extraction tool for near-infrared (NIR) spectroscopy-based discrimination analysis has been explored and the discrimination of the geographic origins of 8 different agricultural products has been performed as the case study. The sample spectral features were broad and insufficient for component distinction due to considerable overlap of individual bands, so AE enabling of extracting the sample-descriptive features in the spectra would help to improve discrimination accuracy. For comparison, four different inputs of AE-extracted features, raw NIR spectra, principal component (PC) scores, and features extracted using locally linear embedding were employed for sample discrimination using support vector machine. The use of AE-extracted feature improved the accuracy in the discrimination of samples in all 8 products. The improvement was more substantial when the sample spectral features were indistinct. It demonstrates that AE is expandable for vibrational spectroscopic discriminant analysis of other samples with complex composition.

Exploring the Complexity of Processing-Induced Dehydration during Hot Melt Extrusion Using In-Line Raman Spectroscopy

The specific aim in this study was to understand the effect of critical process parameters on the solid form composition of model drug compounds during hot melt extrusion using in-line Raman spectroscopy combined with Multivariate Curve Resolution-Alternating Least Squares (MCR-ALS) modeling for semi-quantitative kinetic profiling. It was observed that the hydrate and anhydrate solid forms of two model drugs in the melts of nitrofurantoin (NF):polyethylene oxide (PEO) and piroxicam (PRX):PEO could be resolved from a MCR-ALS model without an external calibration dataset. Based on this model, the influence of two critical process parameters (shear and temperature) on the solid form composition could be evaluated in a real-time mode and the kinetics of complex transformation pathways could be explored. Additionally, the dehydration pathways of NF monohydrate and PRX monohydrate in molten PEO could be derived. It can be concluded that dehydration of both hydrates in PEO occurs via competing mechanisms—a solution-mediated transformation pathway and a solid–solid transformation, and that the balance between these mechanisms is determined by the combined effect of both temperature and shear. Another important observation was that the water released from these hydrate compounds has a detectable effect on the rheological characteristics of this mixture.

InSituAnalyze: A Python Framework for Multicomponent Synchronous Analysis of Spectral Imaging

Spectral imaging is visualization of high precision and high sensitivity and suitable for analyzing the spatial distribution of complex materials. While providing rich and detailed information, it makes higher demands on feature extraction and information mining of high-dimensional data. For the convenience of further utilization, our research team has developed a Python framework for the multicomponent synchronous analysis of spectral imaging based on a characteristic band method and fast-NNLS algorithm, helping to handle spectrum data from complex samples and gaining semiquantitative information on the sample on the scale of pixel based on target components. With the help of the easy-to-use framework, users are leading to choose suitable pretreatment methods for images and spectra, extract spatial information on tissues/structures account of multispace, and conduct analysis on target components in an intuitive and timesaving way. The sophisticated functional architecture also makes the framework expedited to add algorithms and supported data formats.

Visualization and Semiquantitative Study of the Distribution of Major Components in Wheat Straw in Mesoscopic Scale using Fourier Transform Infrared Microspectroscopic Imaging

Understanding the biochemical heterogeneity of plant tissue linked to crop straw anatomy is attractive to plant researchers and researchers in the field of biomass refinery. This study provides an in situ analysis and semiquantitative visualization of major components distribution in internodal transverse sections of wheat straw based on Fourier transform infrared (FTIR) microspectroscopic imaging, with a fast non-negativity-constrained least squares (fast NNLS) fitting. This paper investigates changes in biochemical components of tissue during stages of elongation, booting, heading, flowering, grain-filling, milk-ripening, dough, and full-ripening. Visualization analysis was carried out with reference spectra for five components (microcrystalline cellulose, xylan, lignin, pectin, and starch) of wheat straw. Our result showed that (a) the cellulose and lignin distribution is consistent with that from tissue-dyeing with safranin O-fast green and (b) the distribution of cellulose, lignin, and starch is consistent with chemical images for characteristic wavelength at 1432, 1507, and 987 cm^-1, respectively, showing no interference from the other components analyzed. With the validation from biochemical images using characteristic wavelength and tissue-dyeing techniques, further semiquantitative analysis in local tissues based on fast NNLS was carried out, and the result showed that (a) the contents of cellulose in various tissues are very different, with most in parenchyma tissue and least in the epidermis and (b) during plant development, the fluctuation of each component in tissues follows nearly the same trend, especially within vascular bundles and parenchyma tissue. Thus, FTIR microspectroscopic imaging combined with suitable chemometric methods can be successfully applied to study chemical distributions within the internodes transverse sections of wheat straw, providing semiquantitative chemical information.

Discrimination of Radix Isatidis and Rhizoma et Radix Baphicacanthis Cusia samples by near infrared spectroscopy with the aid of chemometrics

A novel method for the discrimination of the three kinds of Indigowoad Root sample, Radix Isatidis (RI), Rhizoma et Radix Baphicacanthis Cusia (RRBC) and simulated adulterated samples (AD) was researched and developed with the use of near infrared spectroscopy (NIR) and chemometrics. Principal component analysis (PCA) was applied to process the NIR data of 75 collected Indigowoad Root samples, and the three kinds of such sample were discriminated along the first principal component (PC1) axis. In addition, the data pretreatment methods - genetic algorithm-partial least squares (GA-PLS), successive projections algorithm (SPA), and wavelet transform (WT), were employed to select the key analytical wavelengths, and then, these were used as input variables for the three kinds of the pattern recognition method, such as K-nearest neighbor (KNN), radial basis function-artificial neural network (RBF-ANN), least squares-support vector machine (LS-SVM) and back propagation-artificial neural network (BP-ANN). The WT was the method of choice for data pretreatment, and three pretreatment-prediction method combinations performed well (basis: %recognition rate) - WT-KNN (98.2%) and BP-ANN (97.3%) as well as GA-PLS - LS-SVM (97.2). A BP-ANN calibration model was built for the quantitative discrimination of the three types of the complex Indigowoad Root samples, and it was successfully validated.

A criterion for assessing homogeneity distribution in hyperspectral images. Part 2: application of homogeneity indices to solid pharmaceutical dosage forms

This article is the second of a series of two articles detailing the application of mixing index to assess homogeneity distribution in oral pharmaceutical solid dosage forms by image analysis. Chemical imaging (CI) is an emerging technique integrating conventional imaging and spectroscopic techniques with a view to obtaining spatial and spectral information from a sample. Near infrared chemical imaging (NIR-CI) has proved an excellent analytical tool for extracting high-quality information from sample surfaces. The primary objective of this second part was to demonstrate that the approach developed in the first part could be successfully applied to near infrared hyperspectral images of oral pharmaceutical solid dosage forms such as coated, uncoated and effervescent tablets, as well as to powder blends. To this end, we assessed a new criterion for establishing mixing homogeneity by using four different methods based on a three-dimensional (M×N×λ) data array of hyperspectral images (spectral standard deviations and correlation coefficients) or a two-dimensional (M×N) data array (concentration maps and binary images). The four methods were used applying macropixel analysis to the Poole (M(P)) and homogeneity (H%(Poole)) indices. Both indices proved useful for assessing the degree of homogeneity of pharmaceutical samples. The results testify that the proposed approach can be effectively used in the pharmaceutical industry, in the finished products (e.g., tablets) and in mixing unit operations for example, as a process analytical technology tool for the blending monitoring (see part 1).

Visualizing excipient composition and homogeneity of Compound Liquorice Tablets by near-infrared chemical imaging

This study demonstrated that near-infrared chemical imaging (NIR-CI) was a promising technology for visualizing the spatial distribution and homogeneity of Compound Liquorice Tablets. The starch distribution (indirectly, plant extraction) could be spatially determined using basic analysis of correlation between analytes (BACRA) method. The correlation coefficients between starch spectrum and spectrum of each sample were greater than 0.95. Depending on the accurate determination of starch distribution, a method to determine homogeneous distribution was proposed by histogram graph. The result demonstrated that starch distribution in sample 3 was relatively heterogeneous according to four statistical parameters. Furthermore, the agglomerates domain in each tablet was detected using score image layers of principal component analysis (PCA) method. Finally, a novel method named Standard Deviation of Macropixel Texture (SDMT) was introduced to detect agglomerates and heterogeneity based on binary image. Every binary image was divided into different sizes length of macropixel and the number of zero values in each macropixel was counted to calculate standard deviation. Additionally, a curve fitting graph was plotted on the relationship between standard deviation and the size length of macropixel. The result demonstrated the inter-tablet heterogeneity of both starch and total compounds distribution, simultaneously, the similarity of starch distribution and the inconsistency of total compounds distribution among intra-tablet were signified according to the value of slope and intercept parameters in the curve.

Confocal micro-X-ray fluorescence analysis as a new tool for the non-destructive study of the elemental distributions in pharmaceutical tablets

Chemical imaging studies of pharmaceutical tablets are currently an important emerging field in the pharmaceutical industry. Finding the distribution of the different compounds inside the tablet is an important issue for production quality control but also for counterfeit detection. Most of the currently used techniques are limited to the study of the surface of the compacts, whereas the study of the bulk requires a time-consuming sample preparation. In this paper, we present the use of 3D micro-X-ray fluorescence analysis (3D μXRF) for the non-destructive study of pharmaceutical tablets. Based on two different examples, it was shown that it was possible to measure the distribution of several inorganic elements (Zn, Fe, Ti, Mn, Cu) from the surface to a depth of several hundred microns under the surface. The X-ray absorption, depending on both matrix composition and energy, is one of the most critical factors of this analytical method while performing depth profiling or mapping. Therefore, an original method to correct the absorption, in order to accurately measure the true elemental distribution, was proposed. Moreover, by using the presence of titanium dioxide in a pharmaceutical coating, we proved that this technique is also suited to the non-destructive measurement of coating thickness.

Variable selection in near-infrared spectroscopy: benchmarking of feature selection methods on biodiesel data

During the past several years, near-infrared (near-IR/NIR) spectroscopy has increasingly been adopted as an analytical tool in various fields from petroleum to biomedical sectors. The NIR spectrum (above 4000 cm(-1)) of a sample is typically measured by modern instruments at a few hundred of wavelengths. Recently, considerable effort has been directed towards developing procedures to identify variables (wavelengths) that contribute useful information. Variable selection (VS) or feature selection, also called frequency selection or wavelength selection, is a critical step in data analysis for vibrational spectroscopy (infrared, Raman, or NIRS). In this paper, we compare the performance of 16 different feature selection methods for the prediction of properties of biodiesel fuel, including density, viscosity, methanol content, and water concentration. The feature selection algorithms tested include stepwise multiple linear regression (MLR-step), interval partial least squares regression (iPLS), backward iPLS (BiPLS), forward iPLS (FiPLS), moving window partial least squares regression (MWPLS), (modified) changeable size moving window partial least squares (CSMWPLS/MCSMWPLSR), searching combination moving window partial least squares (SCMWPLS), successive projections algorithm (SPA), uninformative variable elimination (UVE, including UVE-SPA), simulated annealing (SA), back-propagation artificial neural networks (BP-ANN), Kohonen artificial neural network (K-ANN), and genetic algorithms (GAs, including GA-iPLS). Two linear techniques for calibration model building, namely multiple linear regression (MLR) and partial least squares regression/projection to latent structures (PLS/PLSR), are used for the evaluation of biofuel properties. A comparison with a non-linear calibration model, artificial neural networks (ANN-MLP), is also provided. Discussion of gasoline, ethanol-gasoline (bioethanol), and diesel fuel data is presented. The results of other spectroscopic techniques application, such as Raman, ultraviolet-visible (UV-vis), or nuclear magnetic resonance (NMR) spectroscopies, can be greatly improved by an appropriate feature selection choice.

Monitoring of API particle size during solid dosage form manufacturing process by chemical imaging and particle sizing

Agglomeration of API during a solid dosage form manufacturing process is followed from the bulk API, through the initial blend with the excipients to the ribbons by a combination of chemical imaging and particle sizing experiments. Particle size of the ingoing API was characterized using a Sympatec HELOS laser diffractometer. Chemical images of the API were obtained from the blends, granules, and ribbons using near-infrared (NIR) and Raman mapping instruments. All the chemical images are obtained in the univariate fashion through the API-characteristic wavenumbers. Light microscopy and laser diffraction were used to assess presence of large agglomerates in the bulk API. NIR chemical images of the sparsely distributed blend particles confirmed that the large agglomerates were not dispersed during the blending. Also, it was found that normal microscopy may be efficient at detecting those API agglomerates due to their distinct appearance (whiteness and size). The agglomerates were not detected in the NIR chemical images of the granules and the ribbons. This was more reliably confirmed by Raman chemical images in which small API domains were clearly identified which was not attainable by NIR mapping.

Risperidone solid dispersion for orally disintegrating tablet: its formulation design and non-destructive methods of evaluation

The focus of present investigation was to assess the utility of non-destructive techniques in the evaluation of risperidone solid dispersions (SD) with methyl-β-cyclodextrin (MBCD) and subsequent incorporation of the SD into orally disintegrating tablets (ODT) for a faster release of risperidone. The SD was prepared by a solvent evaporation method and evaluated by scanning electron microscopy (SEM), Fourier transform infrared (FTIR), near infrared spectroscopy (NIR), NIR-chemical imaging (NIR-CI), powder X-ray diffraction (PXRD) and differential scanning calorimetry (DSC). DSC and XRD analysis indicated that crystallinity of SD has reduced significantly. FTIR showed no interaction between risperidone and MBCD. Partial least square (PLS) was applied to the NIR data for the construction of chemometric models to determine both components of the SD. Good correlations were obtained for calibration and prediction as indicated by correlation coefficients >0.9965. The model was more accurate and less biased in predicting the MBCD than risperidone as indicated by its lower mean accuracy and mean bias values. SD-3 (risperidone:MBCD, 1:3) was incorporated into ODT tablets containing diluent (D-mannitol, FlowLac(®) 100 or galenIQ™-721) and superdisintegrant (Kollidon(®) CL-SF, Ac-Di-Sol or sodium starch glycolate). Disintegration time, T(50) and T(90) were decreased in the formulations containing mannitol and Kollidon(®) CL-SF, but increased with galenIQ™-721 and sodium starch glycolate, respectively. NIR-CI images confirmed the homogeneity of SD and ODT formulations.

Near-infrared chemical imaging (NIR-CI) for counterfeit drug identification--a four-stage concept with a novel approach of data processing (Linear Image Signature)

A new stage concept was developed to reliably identify counterfeit tablets which are very similar to the genuine drug product. This concept combines single-point near-infrared spectroscopy (NIRS) and near-infrared chemical imaging (NIR-CI) with statistical variance analysis. The advantage of NIR-CI over NIRS is the potential to determine not only the amount, but also the spatial distribution of ingredients within a single tablet. Previously published NIR-CI studies used homogeneity as a key indicator for the identification of counterfeits. The state of the art approach for estimating homogeneity is to record the average and % standard deviation of predicted classification scores (i.e. concentrations) for a given component within a specimen. A disadvantage of this approach is the partial loss of spatial information. In view of this, we developed a new method using much more of the spatial information for the estimation of homogeneity. The method is based on (1) summation and unfolding of multidimensional predicted classification scores, which results in a Linear Image Signature (LIS) and (2) multivariate LIS data analysis (LIS-MVA). It could be demonstrated that this kind of NIR-CI data analysis represents an innovative approach for the identification of counterfeit tablets. Moreover, this procedure is applicable to determine the product variability, i.e. process signature of a given product thus being a valuable tool within the Quality by Design (QbD) approach of the ICH Q8 guideline.

Pharmaceutical applications of vibrational chemical imaging and chemometrics: a review

The emergence of chemical imaging (CI) has gifted spectroscopy an additional dimension. Chemical imaging systems complement chemical identification by acquiring spatially located spectra that enable visualization of chemical compound distributions. Such techniques are highly relevant to pharmaceutics in that the distribution of excipients and active pharmaceutical ingredient informs not only a product's behavior during manufacture but also its physical attributes (dissolution properties, stability, etc.). The rapid image acquisition made possible by the emergence of focal plane array detectors, combined with publication of the Food and Drug Administration guidelines for process analytical technology in 2001, has heightened interest in the pharmaceutical applications of CI, notably as a tool for enhancing drug quality and understanding process. Papers on the pharmaceutical applications of CI have been appearing in steadily increasing numbers since 2000. The aim of the present paper is to give an overview of infrared, near-infrared and Raman imaging in pharmaceutics. Sections 2 and 3 deal with the theory, device set-ups, mode of acquisition and processing techniques used to extract information of interest. Section 4 addresses the pharmaceutical applications.

Near-infrared chemical imaging (NIR-CI) on pharmaceutical solid dosage forms-comparing common calibration approaches

Near-infrared chemical imaging (NIR-CI) is the fusion of near-infrared spectroscopy and image analysis. It can be used to visualize the spatial distribution of the chemical compounds in a sample (providing a chemical image). Each sample measurement generates a hyperspectral data cube containing thousands of spectra. An important part of a NIR-CI analysis is the data processing of the hyperspectral data cube. The aim of this study was to compare the ability of different commonly used calibration methods to generate accurate chemical images. Three common calibration approaches were compared: (1) using single wavenumber, (2) using classical least squares regression (CLS) and (3) using partial least squares regression (PLS1). Each method was evaluated using two different preprocessing methods. A calibration data set of tablets with five constituents was used for analysis. Chemical images of the active pharmaceutical ingredient (API) and the two major excipients cellulose and lactose in the formulation were made. The accuracy of the generated chemical images was evaluated by the concentration prediction ability. The most accurate predictions for all three compounds were generated by PLS1. The drawback of PLS1 is that it requires a calibration data set and CLS, which does not require a calibration data set, therefore proved to be an excellent alternative. CLS also generated accurate predictions and only requires the pure compound spectrum of each constituent in the sample. All three calibration approaches were found applicable for hyperspectral image analysis but their relevance of use depends on the purpose of analysis and type of data set. As expected, the single wavenumber method was primarily found useful for compounds with a distinct spectral band that was not overlapped by bands of other constituents. This paper also provides guidance for hyperspectral image (or NIR-CI) analysis describing each of the typical steps involved.