Hans Grahn

Techniques and applications of hyperspectral image analysis

Preface. List of Contributors. List of Abbreviations. 1 Multivariate Images, Hyperspectral Imaging: Background and Equipment (Paul L. M. Geladi, Hans F. Grahn and James E. Burger). 2 Principles of Multivariate Image Analysis (MIA) in Remote Sensing, Technology and Industry (Kim H. Esbensen and Thorbjorn T. Lied). 3 Clustering and Classification in Multispectral Imaging for Quality Inspection of Postharvest Products (Jacco C. Noordam and Willie H. A. M. van den Broek). 4 Self-modeling Image Analysis with SIMPLISMA (Willem Windig, Sharon Markel and Patrick M. Thompson). 5 Multivariate Analysis of Spectral Images Composed of Count Data (Michael R. Keenan). 6 Hyperspectral Image Data Conditioning and Regression Analysis (James E. Burger and Paul L. M. Geladi). 7 Principles of Image Cross-validation (ICV): Representative Segmentation of Image Data Structures (Kim H. Esbensen and Thorbjorn T. Lied). 8 Detection, Classification, and Quantification in Hyperspectral Images Using Classical Least Squares Models (Neal B. Gallagher). 9 Calibration Standards and Image Calibration (Paul L. M. Geladi). 10 Multivariate Movies and their Applications in Pharmaceutical and Polymer Dissolution Studies (Jaap van der Weerd and Sergei G. Kazarian). 11 Multivariate Image Analysis of Magnetic Resonance Images: Component Resolution with the Direct Exponential Curve Resolution Algorithm (DECRA) (Brian Antalek, Willem Windig and Joseph P. Hornak). 12 Hyperspectral Imaging Techniques: an Attractive Solution for the Analysis of Biological and Agricultural Materials (Vincent Baeten, Juan Antonio Fernandez Pierna and Pierre Dardenne). 13 Application of Multivariate Image Analysis in Nuclear Medicine: Principal Component Analysis (PCA) on Dynamic Human Brain Studies with Positron Emission Tomography (PET) for Discrimination of Areas of Disease at High Noise Levels (Pasha Razifar and Mats Bergstrom). 14 Near Infrared Chemical Imaging: Beyond the Pictures (E. Neil Lewis, Janie Dubois, Linda H. Kidder and Kenneth S. Haber). Index.

Strategies for multivariate image regression

Abstract Esbensen, K.H., Geladi, P.L. and Grahn, H.F., 1992. Strategies for multivariate image regression. Chemometrics and Inteligent Laboratory Systems, 14: 357–374. We present multivariate image regression (MIR) as a set of typically problem-dependent strategies for image decomposition guided by the nature of the Y variable and/or training data set delineation in the ( X , Y ) image domains. Regression techniques common in chemometrics may be applied also to the image regimen (in this paper we treat mainly two-dimensional images). We present applications of both IMPCR and IMPLS-DISCRIM in an effort to delineate the various possibilities for image regression. IMPCR builds directly on our earlier bilinear multivariate image analysis projection approach, while IMPLS-DISCRIM is trained on scene space binary classification masking with subsequent off-screen partial least squares analysis; the results are back-projected as images in the original scene space. Regression may either be carried out for modelling purposes and/or for subsequent prediction purposes. In the image domain this duality is accompanied by several optional training data set delineations in the scene space and/or in the spectral domain. We try to cover as complete a survey as possible of typical, representative regression problem types. We illustrate some of these MIR strategies with an MR-imaging example as well as a simple didactic MIR calibration from analytical chemistry.

Image analysis in chemistry I. Properties of images, greylevel operations, the multivariate image

Abstract The use of imaging and images in chemical and technological applications is treated in two parts. Part I focuses on the generation and properties of images. The result of image registration is an analog or digital image, a representation of intensities as a function of position coordinates x , y (and sometimes also z ) in a non-homogeneous material. Certain univariate operations for reducing noise and errors, improving contrast, measuring and counting particles etc. can be carried out on digitized images. A simple example from bioenergy research is given. In this first part a multivariate image is defined and an explanation is given of how it can be obtained. Operations on multivariate images and interpretations of multivariate results are presented in part II.

Clustering of aryl carbon-13 nuclear magnetic resonance substituent chemical shifts. A multivariate data analysis using principal components

A principal component analysis of the 13C substituent-induced chemical shifts of 82 monosubstituted benzenes shows that ca. 90% of the substituents belong to one of four clusters, acceptors, alkyls, donors, or halogens. This grouping is confirmed statistically. The extensions of the subclasses are not parallel. It is also shown that the predictive capability of the single-parameter models for each subclass is better than any multiparameter model applied on the whole data set. The observed grouping of substituents provides an explanation to the apparent correlation frequently found between 13C n.m.r. chemical shifts and dual substituent parameters. The ability of the statistical method to discover incorrect shift data is also illustrated.

Image analysis in chemistry II. Multivariate image analysis

Abstract Part I [Trends Anal. Chem., 11 (1992) 41] focused on the generation of univariate digital images in the chemical laboratory and in industrial situations and on the possible use of operations on univariate images. The concept of a multivariate image was introduced and an example given. This part focuses on the use of multivariate methods to extract problem-dependent, useful information from multivariate images. The example of the powder mixtures given in Part I is further analyzed by principal component analysis. Concepts of exploratory analysis, classification and regression are explained. The use of visual interpretation and statistical diagnostics is emphasized. Ideas about future developments of multivariate image analysis are introduced.

Multivariate image regression and analysis: Useful techniques for the evaluation of clinical magnetic resonance images

Abstract Multivariate image analysis (MIA) and multivariate image regression (MIR) techniques are useful tools in the extraction of information from magnetic resonance images. They aid the characterization of different tissues and can be used to describe their size and distribution. The obtained new information can be used to monitor growth, progression and effects of a treatment. The methodology is illustrated by a clinical example. The ongoing development of MIA and MIR, combining parameters from different imaging modalities, is commented on.

Multivariate data analysis of NMR data

Multivariate methods based on principal components (PCA and PLS) have been used to reduce NMR spectral information, to predict NMR parameters of complicated structures, and to relate shift data sets to dependent descriptors of biological significance. Noise reduction and elimination of instrumental artifacts are easily performed on 2D NMR data. Configurational classification of triterpenes and shift predictions in disubstituted benzenes can be obtained using PCA and PLS analysis. Finally, the shift predictions of tripeptides from descriptors of amino acids open the possibility of automatic analysis of multidimensional data of complex structures.

Hyperspectral Imaging and Data Analysis for Detecting and Determining Plastic Contamination in Seawater Filtrates

One possible way of monitoring plastic particles in sea water is by imaging spectroscopic measurements on filtrates. The idea is that filters from seawater sampling can be imaged in many wavelengths and that a multivariate data analysis can give information on (1) spatial location of plastic material on the filter and (2) composition of the plastic materials. This paper reports on simulated samples, with spiked reference plastic particles and real seawater filtrates containing microplastic pollutants. These real samples were previously identified through visual examination in a microscope. The samples were imaged using three different imaging systems. The different wavelength ranges were 375–970 nm, 960–1662 nm and 1000–2500 nm. Data files from all three imaging systems were analysed by hyperspectral image analysis. The method using the wavelength span 1000–2500 nm was shown to be the most applicable to this specific type of samples and gave a 100% particle recognition on reference plastic, above 300 μm, and an 84% pixel recognition on household polyethylene plastic. When applied to environmental samples the technique showed an increase in identified particles compared with visual investigations. These initial tests indicate a potential underestimation of microplastics in environmental samples. This is the first study to demonstrate that hyperspectral imaging techniques can be used to study microplastics down to 300 μm, which is a common size limit used in microplastic surveys.

Diagnostic subgrouping of depressed patients by principal component analysis and visualized pattern recognition

A data-analytical method is described for identifying behavioral and biological variables in psychiatric patients with predictive value in defining clinical subgroups. The procedure, based on principal component analysis (PCA) and graphical analysis, was applied in a group of 28 depressed patients. The 28 depressed patients of unipolar type were observed for up to 15 years for re-evaluation of the diagnoses at the start of the study. Platelet monoamine oxidase activity, post-dexamethasone serum cortisol and serum melatonin predicted two main clinical subgroups as well as a smaller subgroup of bipolar patients. The selection procedure revealed which of several variables were predictive of subgroups that were not possible to identify by univariate methods. The three biological variables may thus be useful in further assessment of clinical subgroups of unipolar depressed patients studied by other research groups.

Chemical Multivariate Image Analysis: Some Case Studies

The basic principles of multivariate image analysis were presented in (1), together with a simple example that explained the concepts of principal component analysis and an introduction to regression on multivariate images. Figure 1 gives an overview of the methods of multivariate image analysis. This chapter contains two more advanced examples, that are closer to chemical practice. One example is from a visual and near-infrared camera digitization of powder mixtures in 13 wavelengths. It serves as an illustration of a chemical classification using spectroscopic information and feedback to the imaging process. It is shown that mixtures of different chemical composition and physical status can be classified using the spectral information in the multivariate image. The second example is from Magnetic Resonance Imaging (MRI) where different experimental pulse parameter combinations were used to create 12 differently weighted images for the same slice of an egg. Principal Component Analysis and Partial Least Squares (PLS) regression were applied to this multivariate image. The technique of MRI is very useful in medical applications, but industrial applications in product quality control can also be envisioned with increasing instrument capabilities and decreasing price/performance ratios.