A novel method for human gender classification using Raman spectroscopy of fingernail clippings

Estimation of age and sex from fingernail clippings by using ATR-FTIR spectroscopy coupled with chemometric interpretation

Fingernails can act as important forensic evidence as they can be a source of DNA that may link the victim or accused to the crime scene and may also contain traces of drugs such as cocaine and heroin, in regular users. Moreover, previous studies have shown that analyzing fingernails with various techniques can reveal important information, such as age and sex. In this work, ATR-FTIR spectroscopy with chemometric tools has been used to estimate the age and sex from fingernails by analyzing 140 fingernail samples (70 males, and 70 females) collected from volunteers aged between 10 and 70 years old. The amide bands obtained from spectra confirmed the presence of keratin proteins in the samples. PCA and PLS-R were used for the classification of samples. For sex estimation, samples were divided into four categories based on age groups, followed by the differentiation of sex in each group. Similarly, for age estimation, all samples were divided into two sets based on male and female followed by differentiation of age groups in each set. The result showed that PLS-R was able to differentiate fingernail samples based on sex in groups G1, G2, G3, and G4 with R-square values of 0.972, 0.993, 0.991, and 0.996, respectively, and based on age in females, and males with R-square values of 0.93 and 0.97, respectively. External validation and blind tests were also performed which showed results with 100% accuracy. This approach has proved to be effective for the estimation of sex and age from fingernail samples.

Sex Determination of Human Nails Based on Attenuated Total Reflection Fourier Transform Infrared Spectroscopy in Forensic Context

This study reports on the successful use of a machine learning approach using attenuated total reflectance Fourier transform infrared (ATR FT-IR) spectroscopy for the classification and prediction of a donor's sex from the fingernails of 63 individuals. A significant advantage of ATR FT-IR is its ability to provide a specific spectral signature for different samples based on their biochemical composition. The infrared spectrum reveals unique vibrational features of a sample based on the different absorption frequencies of the individual functional groups. This technique is fast, simple, non-destructive, and requires only small quantities of measured material with minimal-to-no sample preparation. However, advanced multivariate techniques are needed to elucidate multiplex spectral information and the small differences caused by donor characteristics. We developed an analytical method using ATR FT-IR spectroscopy advanced with machine learning (ML) based on 63 donors' fingernails (37 males, 26 females). The PLS-DA and ANN models were established, and their generalization abilities were compared. Here, the PLS scores from the PLS-DA model were used for an artificial neural network (ANN) to create a classification model. The proposed ANN model showed a greater potential for predictions, and it was validated against an independent dataset, which resulted in 92% correctly classified spectra. The results of the study are quite impressive, with 100% accuracy achieved in correctly classifying donors as either male or female at the donor level. Here, we underscore the potential of ML algorithms to leverage the selectivity of ATR FT-IR spectroscopy and produce predictions along with information about the level of certainty in a scientifically defensible manner. This proof-of-concept study demonstrates the value of ATR FT-IR spectroscopy as a forensic tool to discriminate between male and female donors, which is significant for forensic applications.

Unsupervised convolutional variational autoencoder deep embedding clustering for Raman spectra

Unsupervised deep learning methods place increased emphasis on the process of cluster analysis of unknown samples without requiring sample labels. Clustering algorithms based on deep embedding networks have been recently developed and are widely used in data mining, speech processing and image recognition, but barely any of them have been used on spectra data. This study presents an unsupervised clustering algorithm for Raman spectra, called the convolutional variational autoencoder deep embedding clustering method (CVDE). It improves the network structure of the multi-layer perception (MLP) that is commonly used in other methods based on the VAE-GMM model, like VaDE, by replacing the hidden fully connected layer in the MLP with three convolution layers and two pooling layers for better clustering on the Raman spectra. The three convolution layers extend vertical channels to learn features, while pooling layers directly reduce the horizontal coding dimensions to prevent gradient explosion and overfitting. Furthermore, such network structures can easily incorporate the gradient-weighted class activation mapping (Grad-Cam) method to visualise the importance of spectral features for clustering, facilitating network tuning and spectral difference analysis. Moreover, through comparative experiments, CVDE has proven that it affords better clustering performance than current advanced clustering methods on not only the MNIST dataset but also two sets of Raman spectra: soybean oil Raman spectra with very small Raman feature differences and drug Raman spectra with a small data size. The clustering accuracies of these three datasets reach 94.48%, 90.43% and 98.70% respectively. Thus, CVDE is suitable for applications in static spectra, such as Raman spectra and LIBS spectra, and is more versatile than supervised methods in the spectral and chemical analysis fields.

Optical coherence tomography-guided confocal Raman microspectroscopy for rapid measurements in tissues

We report a joint system with both confocal Raman spectroscopy (CRS) and optical coherence tomography (OCT) modules capable of quickly addressing the region of interest in a tissue for targeted Raman measurements from OCT. By using an electrically tunable lens in the Raman module, the focus of the module can be adjusted to address any specific depth indicated in an OCT image in a few milliseconds. We demonstrate the performance of the joint system in the depth dependent measurements of an ex vivo swine tissue and in vivo human skin. This system can be useful in measuring samples embedded with small targets, for example, to identify tumors in skin in vivo and assessment of tumor margins, in which OCT can be used to perform initial real-time screening with high throughput based on morphological features to identify suspicious targets then CRS is guided to address the targets in real time and fully characterize their biochemical fingerprints for confirmation.

Development of Crime Scene Intelligence Using a Hand-Held Raman Spectrometer and Transfer Learning

The classification of ignitable liquids, such as gasoline, is critical crime scene intelligence to assist arson investigations. Rapid field gasoline classification is challenging because the current forensic testing standard requires gas chromatography-mass spectrometry analysis of evidence in an accredited laboratory. In this work, we reported a new intelligent analytical platform for field identification and classification of gasoline evidence. A hand-held Raman spectrometer was utilized to collect Raman spectra of reference gasoline samples with various octane numbers. The Raman spectrum pattern was converted into image presentations by continuous wavelet transformation (CWT) to facilitate artificial intelligence development using the transfer learning technique. GoogLeNet, a pretrained convolutional neural network (CNN), was adapted to train the classification model. Six different classification models were also developed from the same data set using conventional machine learning algorithms to evaluate the performance of our new approach. The experimental results indicated that the pretrained CNN model developed by our new data workflow outperformed other models in several performance benchmarks, such as accuracy, precision, recall, F1, Cohen's Kappa, and Matthews correlation coefficient. When the transfer learning model was challenged with the data collected from weathered gasoline samples, the classifier could still offer 73 and 53% accuracy for 50 and 25% weathered gasoline samples, respectively. In conclusion, wavelet transforms combined with transfer learning successfully processed and classified complex Raman spectral data without feature engineering. We envision that this nondestructive, automated, and accurate platform will accelerate crime scene intelligence development based on evidence's chemical signatures detected by hand-held Raman spectrometers.

Raman Spectroscopy, X-ray Diffraction, and Scanning Electron Microscopy as Noninvasive Methods for Microstructural Alterations in Psoriatic Nails

Psoriasis is a chronic inflammatory disease associated with immune system dysfunction that can affect nails, with a negative impact on patient life quality. Usually, nail psoriasis is associated with skin psoriasis and is therefore relatively simple to diagnose. However, up to 10% of nail psoriasis occurs isolated and may be difficult to diagnose by means of current methods (nail biopsy, dermoscopy, video dermoscopy, capillaroscopy, ultrasound of the nails, etc.). Since the nail is a complex biological tissue, mainly composes of hard α-keratins, the structural and morphological techniques can be used to analyze the human fingernails. The aim of this study was to corroborate the information obtained using Raman spectroscopy with those obtained by scanning electron microscopy (SEM) and X-ray diffractometry and to assess the potential of these techniques as non-invasive dermatologic diagnostic tools and an alternative to current methods.

The potential of Raman and FT-IR spectroscopic methods for the detection of chlorine in human nail samples

Although chlorine (Cl2) has been used as a chemical warfare agent since World War I there is no known specific and reliable biomarker to indicate the presence of chlorine. We distinguished chlorinated human nails from unchlorinated ones using Raman spectroscopy and Fourier Transform Infrared (FT-IR) Spectroscopy. This research was carried out between October 2018 and July 2019 on two nail samples taken from 55 male and 104 female volunteers. One sample from each participant was chlorinated, while the second sample was used as a control. Spectral data were collected from chlorinated and unchlorinated (control) human nails using Raman and FT-IR spectroscopy. Raman measurements were made between 100 and 3200 cm^-1, while FT-IR measurements were recorded over the range of 650 to 4000 cm^-1. Partial least squares regression-discriminant analysis (PLS-DA) was used to develop classification models for each spectral instrument. Results showed that the control and chlorinated nail samples were successfully discriminated with similar results achieved with both instruments. Minor differences were observed in the performance of classification models. The FT-IR spectroscopy model (sensitivity = 95%, specificity = 99%, accuracy = 97%) was found to be more successful with a smaller margin of error (sensitivity = 95%, specificity = 99%, accuracy = 96%) compared to the Raman spectroscopy model. This method can be used successfully for both ante-mortem and post-mortem diagnosis of chlorine exposure.

Mechanism of human nail poration by high-repetition-rate, femtosecond laser ablation

Optical poration, or drilling, of the human nail has the potential to drastically improve transungual drug delivery. However, this approach is accompanied by thermal damage to the nail tissue surrounding the laser radiation-created pore. In this paper, fluorescence microscopy has been employed to quantitatively evaluate thermal damage to the nail induced by laser ablation with 80 MHz, nanojoule, femtosecond pulses delivered via a hollow-core fibre. An empirical relation has been established between the intensity of the resulting fluorescence signal and temperature to which the nail was exposed. Using this relationship, detailed temperature maps have been created of the areas surrounding the pores, enabling the mechanism of poration to be better understood. It was deduced that plasma-mediated ablation is primarily responsible for nail tissue ablation at the centre of the pore, while cumulative photothermal processes dominate at the pore edges. It is concluded, furthermore, that temperature mapping represents a useful new tool with which to optimise the process of nail poration. The method is potentially generic and may be applicable to other biological materials.

Improved Vancouver Raman Algorithm Based on Empirical Mode Decomposition for Denoising Biological Samples

A novel method based on the Vancouver Raman algorithm (VRA) and empirical mode decomposition (EMD) for denoising Raman spectra of biological samples is presented. The VRA is one of the most used methods for denoising Raman spectroscopy and is composed of two main steps: signal filtering and polynomial fitting. However, the signal filtering step consists in a simple mean filter that could eliminate spectrum peaks with small intensities or merge relatively close spectrum peaks into one single peak. Thus, the result is often sensitive to the order of the mean filter, so the user must choose it carefully to obtain the expected result; this introduces subjectivity in the process. To overcome these disadvantages, we propose a new algorithm, namely the modified-VRA (mVRA) with the following improvements: (1) to replace the mean filter step by EMD as an adaptive parameter-free signal processing method; and (2) to automate the selection of polynomial degree. The denoising capabilities of VRA, EMD, and mVRA were compared in Raman spectra of artificial data based on Teflon material, synthetic material obtained from vitamin E and paracetamol, and biological material of human nails and mouse brain. The correlation coefficient (ρ) was used to compare the performance of the methods. For the artificial Raman spectra, the denoised signal obtained by mVRA (ρ>0.91) outperforms VRA (ρ>0.86) for moderate to high noise levels whereas mVRA outperformed EMD (ρ>0.90) for high noise levels. On the other hand, when it comes to modeling the underlying fluorescence signal of the samples (i.e., the baseline trend), the proposed method mVRA showed consistent results (ρ>0.94). For Raman spectra of synthetic material, good performance of the three methods (ρ=0.99 for VRA, ρ=0.93 for EMD, and ρ=0.99 for mVRA) was obtained. Finally, in the biological material, mVRA and VRA showed similar results (ρ=0.96 for VRA, ρ=0.85 for EMD, and ρ=0.91 for mVRA); however, mVRA retains valuable information corresponding to relevant Raman peaks with small amplitude. Thus, the application of EMD as a filter in the VRA method provides a good alternative for denoising biological Raman spectra, since the information of the Raman peaks is conserved and parameter tuning is not required. Simultaneously, EMD allows the baseline correction to be automated.

Raman spectroscopy as a predictive tool for monitoring osteoporosis therapy in a rat model of postmenopausal osteoporosis

Pharmacological therapy of osteoporosis reduces bone loss and risk of fracture in patients. Modulation of bone mineral density cannot explain all effects. Other aspects of bone quality affecting fragility and ways to monitor them need to be better understood. Keratinous tissue acts as surrogate marker for bone protein deterioration caused by oestrogen deficiency in rats. Ovariectomised rats were treated with alendronate (ALN), parathyroid hormone (PTH) or estrogen (E2). MicroCT assessed macro structural changes. Raman spectroscopy assessed biochemical changes. Micro CT confirmed that all treatments prevented ovariectomy-induced macro structural bone loss in rats. PTH induced macro structural changes unrelated to ovariectomy. Raman analysis revealed ALN and PTH partially protect against molecular level changes to bone collagen (80% protection) and mineral (50% protection) phases. E2 failed to prevent biochemical change. The treatments induced alterations unassociated with the ovariectomy; increased beta sheet with E2, globular alpha helices with PTH and fibrous alpha helices with both ALN and PTH. ALN is closest to maintaining physiological status of the animals, while PTH (comparable protective effect) induces side effects. E2 is unable to prevent molecular level changes associated with ovariectomy. Raman spectroscopy can act as predictive tool for monitoring pharmacological therapy of osteoporosis in rodents. Keratinous tissue is a useful surrogate marker for the protein related impact of these therapies.The results demonstrate utility of surrogates where a clear systemic causation connects the surrogate to the target tissue. It demonstrates the need to assess broader biomolecular impact of interventions to examine side effects.

Label-free spectrochemical probe for determination of hemoglobin glycation in clinical blood samples

Glycated hemoglobin, HbA1c, is an important biomarker that reveals the average value of blood glucose over the preceding 3 months. While significant recent attention has been focused on the use of optical and direct molecular spectroscopic methods for determination of HbA1c, a facile test that minimizes sample preparation needs and turnaround time still remains elusive. Here, we report a label-free approach for identifying low, mid and high-HbA1c groups in hemolysate and in whole blood samples featuring resonance Raman (RR) spectroscopy and support vector machine (SVM)-based classification of spectral patterns. The diagnostic power of RR measurements stems from its selective enhancement of hemoglobin-specific features, which simultaneously minimizes the blood matrix spectral interference and permits detection in the native solution. In this pilot study, our spectroscopic observations reveal that glycation of hemoglobin results in subtle but reproducible changes even when detected in the whole blood matrix. Leveraging SVM analysis of the principal component scores determined from the RR spectra, we show high degree of accuracy in classifying clinical specimen. We envisage that the promising findings will pave the way for more extensive clinical specimen investigations with the ultimate goal of translating molecular spectroscopy for routine point-of-care testing.

Microstructural alterations in the onychomycotic and psoriatic nail: Relevance in drug delivery

Despite the important nail alterations caused by onychomycosis and psoriasis few studies have characterized the microstructure of the diseased nail plate and the diffusion and penetration of drugs through this altered structure. This work aimed to characterize the microstructure of the healthy, onychomycotic and psoriatic human nail using Raman spectroscopy, scanning electron microscopy, optical microscope profilometry and mercury intrusion porosimetry followed by analysis of the structure with PoreCor® software. The results showed that onychomycotic nails have higher porosity and lower amounts of disulphide bonds compared to healthy nails. This suggests that the presence and action of fungi on the nail plate makes this structure more permeable to water and drugs. Psoriatic nails had increased porosity compared to healthy nails but lower than fungal infected specimens. In vitro permeation studies showed that diseased nails were more permeable to ciclopirox (onychomycosis) and clobetasol (psoriasis) although drug permeation was highly variable and likely to be influenced by the degree of alteration of the nail structure. On the whole, this work provides new and valuable information about the microstructure and porosity of diseased nails and a plausible explanation of the increased drug permeability observed in this work and elsewhere.

Towards Fingermark Dating: A Raman Spectroscopy Proof-of-Concept Study

Fingermarks have, for a long time, been vital in the forensic community for the identification of individuals, and a possibility to non‐destructively date the fingermarks would of course be beneficial. Raman spectroscopy is, herein, evaluated for the purpose of estimating the age of fingermarks deposits. Well‐resolved spectra were non‐destructively acquired to reveal spectral uniqueness, resembling those of epidermis, and several molecular markers were identified that showed different decay kinetics: carotenoids > squalene > unsaturated fatty acids > proteins. The degradation rates were accelerated, less pronounced for proteins, when samples were stored under ambient light conditions, likely owing to photo‐oxidation. It is hypothesized that fibrous proteins are present and that oxidation of amino acid side chains can be observed both through Raman and fluorescence spectroscopy. Clearly, Raman spectroscopy is a useful technique to non‐destructively study the aging processes of fingermarks.

Sex-related chemical differences in keratin from fingernail plates: a solid-state carbon-13 NMR study

The carbon-13 solid-state NMR reveals chemical differences in fingernail keratin between young, healthy males and females.

Changes in nail keratin observed by Raman spectroscopy after Nd:YAG laser treatment

Lasers and photodynamic therapy have been considered a convergence treatment for onychomycosis, which is a fungal infection on the nail bed and nail plate. Laser therapies have shown satisfactory results without significant complications for onychomycosis; however, the mechanism of clearing remains unknown. In this work, we investigated changes in the chemical structure of nail keratin induced by Nd:YAG laser using Raman spectroscopy. Toe nails with onychomycosis were treated with 1064 nm Nd:YAG laser. After laser treatment, the disulfide band (490-590 cm^-1 ) of nail keratin was rarely observed or was reduced in intensity. The amide I band (1500-1700 cm^-1 ) also showed changes induced by the laser. The α-helical (1652 cm^-1 ) structures dominated the β-sheet (1673 cm^-1 ) in nontreated nail, but the opposite phenomenon was observed after laser treatment.

Discerning the differential molecular pathology of proliferative middle ear lesions using Raman spectroscopy

Despite its widespread prevalence, middle ear pathology, especially the development of proliferative lesions, remains largely unexplored and poorly understood. Diagnostic evaluation is still predicated upon a high index of clinical suspicion on otoscopic examination of gross morphologic features. We report the first technique that has the potential to non-invasively identify two key lesions, namely cholesteatoma and myringosclerosis, by providing real-time information of differentially expressed molecules. In addition to revealing signatures consistent with the known pathobiology of these lesions, our observations provide the first evidence of the presence of carbonate- and silicate-substitutions in the calcium phosphate plaques found in myringosclerosis. Collectively, these results demonstrate the potential of Raman spectroscopy to not only provide new understanding of the etiology of these conditions by defining objective molecular markers but also aid in margin assessment to improve surgical outcome.

Support vector machine regression (SVR/LS-SVM)--an alternative to neural networks (ANN) for analytical chemistry? Comparison of nonlinear methods on near infrared (NIR) spectroscopy data

In this study, we make a general comparison of the accuracy and robustness of five multivariate calibration models: partial least squares (PLS) regression or projection to latent structures, polynomial partial least squares (Poly-PLS) regression, artificial neural networks (ANNs), and two novel techniques based on support vector machines (SVMs) for multivariate data analysis: support vector regression (SVR) and least-squares support vector machines (LS-SVMs). The comparison is based on fourteen (14) different datasets: seven sets of gasoline data (density, benzene content, and fractional composition/boiling points), two sets of ethanol gasoline fuel data (density and ethanol content), one set of diesel fuel data (total sulfur content), three sets of petroleum (crude oil) macromolecules data (weight percentages of asphaltenes, resins, and paraffins), and one set of petroleum resins data (resins content). Vibrational (near-infrared, NIR) spectroscopic data are used to predict the properties and quality coefficients of gasoline, biofuel/biodiesel, diesel fuel, and other samples of interest. The four systems presented here range greatly in composition, properties, strength of intermolecular interactions (e.g., van der Waals forces, H-bonds), colloid structure, and phase behavior. Due to the high diversity of chemical systems studied, general conclusions about SVM regression methods can be made. We try to answer the following question: to what extent can SVM-based techniques replace ANN-based approaches in real-world (industrial/scientific) applications? The results show that both SVR and LS-SVM methods are comparable to ANNs in accuracy. Due to the much higher robustness of the former, the SVM-based approaches are recommended for practical (industrial) application. This has been shown to be especially true for complicated, highly nonlinear objects.

Development of robust calibration models using support vector machines for spectroscopic monitoring of blood glucose

Sample-to-sample variability has proven to be a major challenge in achieving calibration transfer in quantitative biological Raman spectroscopy. Multiple morphological and optical parameters, such as tissue absorption and scattering, physiological glucose dynamics and skin heterogeneity, vary significantly in a human population introducing nonanalyte specific features into the calibration model. In this paper, we show that fluctuations of such parameters in human subjects introduce curved (nonlinear) effects in the relationship between the concentrations of the analyte of interest and the mixture Raman spectra. To account for these curved effects, we propose the use of support vector machines (SVM) as a nonlinear regression method over conventional linear regression techniques such as partial least-squares (PLS). Using transcutaneous blood glucose detection as an example, we demonstrate that application of SVM enables a significant improvement (at least 30%) in cross-validation accuracy over PLS when measurements from multiple human volunteers are employed in the calibration set. Furthermore, using physical tissue models with randomized analyte concentrations and varying turbidities, we show that the fluctuations in turbidity alone causes curved effects which can only be adequately modeled using nonlinear regression techniques. The enhanced levels of accuracy obtained with the SVM based calibration models opens up avenues for prospective prediction in humans and thus for clinical translation of the technology.

ProRaman: a program to classify Raman spectra

The program ProRaman, developed for the Matlab platform, provides an interactive and flexible graphic interface to develop efficient algorithms to classify Raman spectra into two or three different classes. A set of preprocessing algorithms to decrease the variable dimensionality and to extract the main features which improve the correct classification ratio was implemented. The implemented classification algorithms were based on the Mahalanobis distance and neural network. To verify the functionality of the developed program, 72 spectra from human artery samples, 36 of which had been histopathologically diagnosed as non-diseased and 36 as having an atherosclerotic lesion, were processed using a combination of different preprocessing and classification techniques. The best result was accomplished when the variables were selected from the Raman spectrum shift range from 1200 to 1700 cm(-1), then preprocessed using wavelets for compression and principal component analysis for feature extraction and, finally, classified by a multilayer perceptron with one hidden layer with eight neurons.