Comparison of spectral and spatial denoising techniques in the context of High Definition FT-IR imaging hyperspectral data

Application of Odd-Order Derivatives in Fourier Transform Nuclear Magnetic Resonance Spectroscopy toward Quantitative Deconvolution

When Fourier transform (FT) spectrum peaks are overlapped, primary maxima of odd-order derivatives can be used to evaluate their independent intensities. We studied the feasibility of higher odd-order derivatives on Lorentzian peak shape and magnitude peak shape. Simulation studies for FT nuclear magnetic resonance (NMR) spectroscopy demonstrated good results toward quantitative deconvolution of overlapping FT spectrum peaks. Although it is not so desirable to deconvolute special line shapes such as Gaussian, Voigt, and Tsallis profiles, the odd-order derivatives exhibit a bright future compared to even-order derivatives. An application example of practical NMR spectroscopy with ethylbenzene isomers is presented. White Gaussian noises were added to the simulated spectra at two different signal-to-noise ratios (20 and 40). Kauppinen's denoising and smoothing algorithms can effectively remove interference of the noise and help to have good deconvoluting results using the odd-order derivatives. We compared features of our approach with popular deconvolution sharpening algorithms and conducted a comparison study with Kauppinen's Fourier self-deconvolution. Our approach has a better dynamic range of peak intensities and is not sensitive to the sampling rates. Other common deconvolution methods are also discussed briefly.

Pancreatic intraepithelial neoplasia detection and duct pathology grading using FT-IR imaging and machine learning

Pancreatic intraepithelial neoplasia (PanIN) is manifested by noninvasive lesions in the epithelium of smaller pancreatic ducts. Generally, cancer development risk from low-grade PanIN is minor, whereas, invasive pancreatic ductal adenocarcinoma (PDAC) development is highly related to high-grade PanINs. Therefore, in the case of high-grade PanIN detection, additional surgical resection may be recommended. However, even the low-grade PanINs can indicate possible progression to PDAC. The definition of PanIN is constantly changing and there is a need for new tools to better characterize and understand its behavior. We have recently developed a comprehensive pancreatic cancer classification model with biopsies collected from over 600 biopsies from 250 patients. Here, we take the next step and employ Infrared (IR) spectroscopy to build the first classification model for PanINs detection. Furthermore, we created a Partial Least Squares Regression (PLSR) model to characterize ducts from benign to cancerous. This model was then used to predict and grade PanINs accordingly to their malignancy level.

Scattering correction for samples with cylindrical domains measured with polarized infrared spectroscopy

Scattering artifacts are one of the most common effects distorting transmission spectra in Fourier-Transform Infrared spectroscopy. Their increased impact, strongly diminishing the quantitative and qualitative power of IR spectroscopy, is especially observed for structures with a size comparable to the radiation wavelength. To tackle this problem, a wide range of preprocessing techniques based on the Extended Multiplicative Scattering Correction method was developed, using physical properties to remove scattering presence in the spectra. However, until recently those algorithms were mostly focused on spherically shaped samples, for example, cells. Here, an algorithm for samples with cylindrical domains is described, with additional implementation of a linearly polarized light case, which is crucial for the growing field of polarized IR imaging and spectroscopy. An open-source code with GPU based implementation is provided, with a calculation time of several seconds per spectrum. Optimizations done to improve the throughput of this algorithm allow the application of this method into the standard preprocessing pipeline of small datasets.

Secondary structure and stability of a gel-forming tardigrade desiccation-tolerance protein

Protein-based pharmaceuticals are increasingly important, but their inherent instability necessitates a "cold chain" requiring costly refrigeration during production, shipment, and storage. Drying can overcome this problem, but most proteins need the addition of stabilizers, and some cannot be successfully formulated. Thus, there is a need for new, more effective protective molecules. Cytosolically, abundant heat-soluble proteins from tardigrades are both fundamentally interesting and a promising source of inspiration; these disordered, monodisperse polymers form hydrogels whose structure may protect client proteins during drying. We used attenuated total reflectance Fourier transform infrared spectroscopy, differential scanning calorimetry, and small-amplitude oscillatory shear rheometry to characterize gelation. A 5% (wt/vol) gel has a strength comparable with human skin, and melts cooperatively and reversibly near body temperature with an enthalpy comparable with globular proteins. We suggest that the dilute protein forms α-helical coiled coils and increasing their concentration drives gelation via intermolecular β-sheet formation.

Super-Resolved 3D Mapping of Molecular Orientation Using Vibrational Techniques

When a sample has an anisotropic structure, it is possible to obtain additional information controlling the polarization of incident light. With their straightforward instrumentation approaches, infrared (IR) and Raman spectroscopies are widely popular in this area. Single-band-based determination of molecular in-plane orientation, typically used in materials science, is here extended by the concurrent use of two vibration bands, revealing the orientational ordering in three dimension. The concurrent analysis was applied to IR spectromicroscopic data to obtain orientation angles of a model polycaprolactone spherulite sample. The applicability of this method spans from high-resolution, diffraction-limited Fourier transform infrared (FT-IR) and Raman imaging to super-resolved optical photothermal infrared (O-PTIR) imaging. Due to the nontomographic experimental approach, no image distortion is visible and nanometer scale orientation domains can be observed. Three-dimensional (3D) bond orientation maps enable in-depth characterization and consequently precise control of the sample’s physicochemical properties and functions.

Low-Cost Probabilistic 3D Denoising with Applications for Ultra-Low-Radiation Computed Tomography

We propose a pipeline for synthetic generation of personalized Computer Tomography (CT) images, with a radiation exposure evaluation and a lifetime attributable risk (LAR) assessment. We perform a patient-specific performance evaluation for a broad range of denoising algorithms (including the most popular deep learning denoising approaches, wavelets-based methods, methods based on Mumford−Shah denoising, etc.), focusing both on accessing the capability to reduce the patient-specific CT-induced LAR and on computational cost scalability. We introduce a parallel Probabilistic Mumford−Shah denoising model (PMS) and show that it markedly-outperforms the compared common denoising methods in denoising quality and cost scaling. In particular, we show that it allows an approximately 22-fold robust patient-specific LAR reduction for infants and a 10-fold LAR reduction for adults. Using a normal laptop, the proposed algorithm for PMS allows cheap and robust (with a multiscale structural similarity index >90%) denoising of very large 2D videos and 3D images (with over 107 voxels) that are subject to ultra-strong noise (Gaussian and non-Gaussian) for signal-to-noise ratios far below 1.0. The code is provided for open access.

Deep Learning-Based Phenological Event Modeling for Classification of Crops

Classification of crops using time-series vegetation index (VI) curves requires appropriate modeling of phenological events and their characteristics. The current study explores the use of capsules, a group of neurons having an activation vector, to learn the characteristic features of the phenological curves. In addition, joint optimization of denoising and classification is adopted to improve the generalizability of the approach and to make it resilient to noise. The proposed approach employs reconstruction loss as a regularizer for classification, whereas the crop-type label is used as prior information for denoising. The activity vector of the class capsule is applied to sample the latent space conditioned on the cell state of a Long Short-Term Memory (LSTM) that integrates the sequences of the phenological events. Learning of significant phenological characteristics is facilitated by adversarial variational encoding in conjunction with constraints to regulate latent representations and embed label information. The proposed architecture, called the variational capsule network (VCapsNet), significantly improves the classification and denoising results. The performance of VCapsNet can be attributed to the suitable modeling of phenological events and the resilience to outliers and noise. The maxpooling-based capsule implementation yields better results, particularly with limited training samples, compared to the conventional implementations. In addition to the confusion matrix-based accuracy measures, this study illustrates the use of interpretability-based evaluation measures. Moreover, the proposed approach is less sensitive to noise and yields good results, even at shallower depths, compared to the main existing approaches. The performance of VCapsNet in accurately classifying wheat and barley crops indicates that the approach addresses the issues in crop-type classification. The approach is generic and effectively models the crop-specific phenological features and events. The interpretability-based evaluation measures further indicate that the approach successfully identifies the crop transitions, in addition to the planting, heading, and harvesting dates. Due to its effectiveness in crop-type classification, the proposed approach is applicable to acreage estimation and other applications in different scales.

The Impact of Preprocessing Methods for a Successful Prostate Cell Lines Discrimination Using Partial Least Squares Regression and Discriminant Analysis Based on Fourier Transform Infrared Imaging

Fourier transform infrared spectroscopy (FT-IR) is widely used in the analysis of the chemical composition of biological materials and has the potential to reveal new aspects of the molecular basis of diseases, including different types of cancer. The potential of FT-IR in cancer research lies in its capability of monitoring the biochemical status of cells, which undergo malignant transformation and further examination of spectral features that differentiate normal and cancerous ones using proper mathematical approaches. Such examination can be performed with the use of chemometric tools, such as partial least squares discriminant analysis (PLS-DA) classification and partial least squares regression (PLSR), and proper application of preprocessing methods and their correct sequence is crucial for success. Here, we performed a comparison of several state-of-the-art methods commonly used in infrared biospectroscopy (denoising, baseline correction, and normalization) with the addition of methods not previously used in infrared biospectroscopy classification problems: Mie extinction extended multiplicative signal correction, Eiler’s smoothing, and probabilistic quotient normalization. We compared all of these approaches and their effect on the data structure, classification, and regression capability on experimental FT-IR spectra collected from five different prostate normal and cancerous cell lines. Additionally, we tested the influence of added spectral noise. Overall, we concluded that in the case of the data analyzed here, the biggest impact on data structure and performance of PLS-DA and PLSR was caused by the baseline correction; therefore, much attention should be given, especially to this step of data preprocessing.

Influence of interference effects on the spectral quality and histological classification by FT-IR imaging in transflection geometry

Infrared (IR) imaging can be used for fast, accurate and non-destructive pathology recognition of biopsies when supported by machine learning algorithms. Transflection mode of measurements has the potential to be translated into the clinic due to economic reasons of large-scale imaging with the need for inexpensive substrates. Unfortunately, in this mode spectral distortions originating from light interference appear. Due to this fact transmission measurement mode is more frequently used in pathology recognition. Nevertheless, this measurement mode also is not devoid of spectral distortion effects like scattering. However, this effect is better understood and there are preprocessing algorithms to minimize it. In this work, we investigated the influence of interference effects on spectral quality of pancreatic tissues measured in transmission and transflection mode with Fourier tranform IR (FT-IR) microscopy using samples embedded with and without paraffin. The removal of paraffin leads to an altered magnitude of interference in transflection and provides a platform for a detailed analysis of its effect on the spectra of biological material, since the same sample is measured with different interference conditions. Moreover, the potential of transflection mode measurements in histological classification of analyzed samples was investigated and compared with classification results for transmission mode.

Raman mapping coupled to self-modelling MCR-ALS analysis to estimate active cosmetic ingredient penetration profile in skin

Confocal Raman mapping (CRM) is a powerful, label free, non-destructive tool, enabling molecular characterization of human skin with applications in the dermo-cosmetic field. Coupling CRM to multivariate analysis can be used to monitor the penetration and permeation of active cosmetic ingredients (ACI) after topical application. It is presently illustrated how multivariate curve resolution alternating least squares (MCR-ALS) can be applied to detect and semi-quantitatively describe the diffusion profile of Delipidol, a commercially available slimming ACI, from Raman spectral maps. Although the analysis outcome can be critically dependent on the a priori selection of the number of regression components, it is demonstrated that profiling of the kinetics of diffusion into the skin can be established with or without additionnal spectral equality constraints in the multivariate analysis, with similar results. Ultimately, MCR-ALS, applied without spectral equality contraints, specifically identifies the ACI as one of main spectral components enabling to investigate its distribution and penetration into the stratum corneum and underlying epidermis layers.

Macromolecular Orientation in Biological Tissues Using a Four-Polarization Method in FT-IR Imaging

Fourier transform infrared spectroscopy has emerged as a powerful tool for tissue specimen investigation. Its nondestructive and label-free character enables direct determination of biochemical composition of samples. Furthermore, the introduction of polarization enriches this technique by the possibility of molecular orientation study apart from purely quantitative analysis. Most of the molecular orientation studies focused on polymer samples with a well-defined molecular axis. Here, a four-polarization approach for Herman’s in-plane orientation function and azimuthal angle determination was applied to a human tissue sample investigation for the first time. Attention was focused on fibrous tissues rich in collagen because of their cylindrical shape and established amide bond vibrations. Despite the fact that the tissue specimen contains a variety of molecules, the presented results of molecular ordering and orientation agree with the theoretical prediction based on sample composition and vibration directions.

Translation of an esophagus histopathological FT-IR imaging model to a fast quantum cascade laser modality

The technical progress in fast quantum cascade laser (QCL) microscopy offers a platform where chemical imaging becomes feasible for clinical diagnostics. QCL systems allow the integration of previously developed FT-IR-based pathology recognition models in a faster workflow. The translation of such models requires a systematic approach, focusing only on the spectral frequencies that carry crucial information for discrimination of pathologic features. In this study, we optimize an FT-IR-based histopathological method for esophageal cancer detection to work with a QCL system. We explore whether the classifier's performance is affected by paraffin presence from tissue blocks compared to removing it chemically. Working with paraffin-embedded samples reduces preprocessing time in the lab and allows samples to be archived after analysis. Moreover, we test, whether the creation of a QCL model requires a preestablished FTIR model or can be optimized using solely QCL measurements.

Comparison between high definition FT-IR, Raman and AFM-IR for subcellular chemical imaging of cholesteryl esters in prostate cancer cells

The family of vibrational spectroscopic imaging techniques grows every few years and there is a need to compare and contrast new modalities with the better understood ones, especially in the case of demanding biological samples. Three vibrational spectroscopy techniques (high definition Fourier-transform infrared [FT-IR], Raman and atomic force microscopy infrared [AFM-IR]) were applied for subcellular chemical imaging of cholesteryl esters in PC-3 prostate cancer cells. The techniques were compared and contrasted in terms of image quality, spectral pattern and chemical information. All tested techniques were found to be useful in chemical imaging of cholesterol derivatives in cancer cells. The results obtained from FT-IR and Raman imaging showed to be comparable, whereas those achieved from AFM-IR study exhibited higher spectral heterogeneity. It confirms AFM-IR method as a powerful tool in local chemical imaging of cells at the nanoscale level. Furthermore, due to polarization effect, p-polarized AFM-IR spectra showed strong enhancement of lipid bands when compared to FT-IR.

Noise-free simulation of an FT-IR imaging hyperspectral dataset of pancreatic biopsy core bound by experiment

A noise-free hyperspectral FT-IR imaging dataset of a pancreatic tissue core was simulated based on experimental data that allows to test the performance of various data analysis and processing algorithms. A set of experimental noise levels was also added and used for denoising approaches comparison, which due to the noise-free reference signal enables to truly observe signal distortion caused by different approaches.

Denoising influence on discrete frequency classification results for quantum cascade laser based infrared microscopy

Currently, there is great interest in bringing the application of IR spectroscopy into the clinic. This however will require a significant reduction in measurement time as Fourier Transform Infrared (FT-IR) imaging takes hours to days to scan a clinically relevant specimen. A potential remedy for this issue is the use of Quantum Cascade Laser Infrared (QCL IR) microscopy performed in Discrete Frequency (DF) mode for maximum speed gain. This gain could be furthermore improved by applying a proper denoising algorithm that takes into account the specific data structure. We have recently compared spectral and spatial denoising techniques in the context of Fourier Transform IR (FT-IR) imaging and showed that the optimal methods depend heavily on the exact data structure. In general multivariate denoising methods such as Principal Component Analysis (PCA) and Minimum Noise Fraction (MNF) are the most effective for a dataset containing multiple bands. Histologic classification of QCL IR images of pancreatic tissue using Random Forest was therefore performed to investigate which denoising schemes are the most optimal for such experimental data structure. This work is the first to show the effects of denoising on classification accuracy of QCL data and is likely to be transferable to other QCL microscopes and other modalities using DF imaging, e.g. AFM-IR or CARS/SRS imaging.