Deep Learning for Raman Spectroscopy: A Review

A comparative Raman study of some transition metal fullerides

PurposePrimary squamous cell carcinoma is a rare malignancy in the thyroid gland (SCCTh). The overall prognosis of this carcinoma is poor. This study aimed to explore the application of Raman spectroscopy in investigating the expression of CK5/6 and P63 in SCCTh.Patients and MethodsTissues of the SCCTh and adjacent normal thyroid, as well as blood serum, were collected from a patient with pathology-confirmed SCCTh. Whole genome sequencing analysis was performed with the tissue of the SCCTh. The expressions of keratins and TP53 family gene were investigated by the Raman spectroscopy in tissues of the SCCTh and adjacent normal thyroid. The serum was also investigated by the Raman spectroscopy for the expression of keratins and TP53 family gene.ResultsThe whole genome sequencing analysis identified the mutation of the TP53 gene (42%) in the tissues of SCCTh. Accordingly, the Raman spectra analyses showed higher expression of keratins and TP53 family gene in the tissues of SCCTh compared with that in the adjacent normal thyroid. Raman spectra analyses of the serum of the patient also showed the expressions of the keratins and TP53 family gene.ConclusionThe expressions of the keratins and TP53 are different in the tissues of SCCTh and adjacent normal thyroid, and the difference could be identified with high sensitivity by the Raman spectra analyses.

Improving the performance of the echinococcosis diagnosis model based on serum Raman spectroscopy via the integration of convolutional neural network and support vector machine.

We determine the fraction f of B atoms belonging to boroxol rings in vitreous boron oxide through a first-principles analysis. After generating a model structure of vitreous B2O3 by first-principles molecular dynamics, we address a large set of properties, including the neutron structure factor, the neutron density of vibrational states, the infrared spectra, the Raman spectra, and the 11B NMR spectra, and find overall good agreement with corresponding experimental data. From the analysis of Raman and 11B NMR spectra, we yield consistently for both probes a fraction f of approximately 0.75. This result indicates that the structure of vitreous boron oxide is largely dominated by boroxol rings.

Multivariate analysis of X-ray absorption and Raman spectra from samples prepared by a high-throughput melting system was conducted to investigate the effects of ZrO 2 concentration ( x ) on the Zr coordination environment and the glass network structure in ZrO 2 -doped R 2 O–CaO–SiO 2 glasses ( R = Na, K). The hundreds of samples were synthesized using a slurry-based combinatorial method involving Pt-Au microwells, followed by measurements of their Zr K-edge X-ray absorption near-edge structure and Raman spectra. Principal component analysis and non-negative matrix factorization revealed distinct structural evolution depending on the alkali species. For R = K , at x < 5 mol%, Zr predominantly adopted a highly symmetric six-coordinated environment. The spectral analysis suggests that this environment is stabilized by preferential charge compensation by Ca 2+ ions. This regime showed an increase in Q 3 (Zr) units without Q 2 formation. As x increased, the results indicate that K + ions participated in charge compensation, reducing Zr site symmetry and promoting Q 2 formation. Consequently, only tetragonal ZrO 2 precipitated at x > 13 mol%. In contrast, for R = Na, a less symmetric Zr environment and the simultaneous formation of Q 2 and Q 3 (Zr) units were observed across all concentrations, implying mixed charge compensation by Na + and Ca 2+ ions. These structural differences led to the precipitation of both monoclinic and tetragonal ZrO 2 phases at x > 16 mol%. The results demonstrate that the competition between network modifiers for charge compensation governs the glass structure and crystallization behavior.

.SignificanceRaman spectroscopy (RS) provides an automated approach for assisting Mohs micrographic surgery for skin cancer diagnosis; however, the specificity of RS is limited by the high spectral similarity between tumors and normal tissues structures. Reflectance confocal microscopy (RCM) provides morphological and cytological details by which many features of epidermis and hair follicles can be readily identified. Combining RS with deep-learning-aided RCM has the potential to improve the diagnostic accuracy of RS in an automated fashion, without requiring additional input from the clinician.AimThe aim of this study is to improve the specificity of RS for detecting basal cell carcinoma (BCC) using an artificial neural network trained on RCM images to identify false positive normal skin structures (hair follicles and epidermis).ApproachOur approach was to build a two-step classification model. In the first step, a Raman biophysical model that was used in prior work classified BCC tumors from normal tissue structures with high sensitivity. In the second step, 191 RCM images were collected from the same site as the Raman data and served as inputs for two ResNet50 networks. The networks selected the hair structure and epidermis images, respectively, within all images corresponding to the positive predictions of the Raman biophysical model with high specificity. The specificity of the BCC biophysical model was improved by moving the Raman spectra corresponding to these selected images from false positive to true negative.ResultsDeep-learning trained on RCM images removed 52% of false positive predictions from the Raman biophysical model result while maintaining a sensitivity of 100%. The specificity was improved from 84.2% using Raman spectra alone to 92.4% by integrating Raman spectra with RCM images.ConclusionsCombining RS with deep-learning-aided RCM imaging is a promising tool for guiding tumor resection surgery.

The rapid development and progress in deep machine-learning techniques have become a key factor in solving the future challenges of humanity. Vision-based target detection and object classification have been improved due to the development of deep learning algorithms. Data fusion in autonomous driving is a fact and a prerequisite task of data preprocessing from multi-sensors that provide a precise, well-engineered, and complete detection of objects, scene or events. The target of the current study is to develop an in-vehicle information system to prevent or at least mitigate traffic issues related to parking detection and traffic congestion detection. In this study we examined to solve these problems described by (1) extracting region-of-interest in the images (2) vehicle detection based on instance segmentation, and (3) building deep learning model based on the key features obtained from input parking images. We build a deep machine learning algorithm that enables collecting real video-camera feeds from vision sensors and predicting free parking spaces. Image augmentation techniques were performed using edge detection, cropping, refined by rotating, thresholding, resizing, or color augment to predict the region of bounding boxes. A deep convolutional neural network F-MTCNN model is proposed that simultaneously capable for compiling, training, validating and testing on parking video frames through video-camera. The results of proposed model employing on publicly available PK-Lot parking dataset and the optimized model achieved a relatively higher accuracy 97.6% than previous reported methodologies. Moreover, this article presents mathematical and simulation results using state-of-the-art deep learning technologies for smart parking space detection. The results are verified using Python, TensorFlow, OpenCV computer simulation frameworks.

After curing, phenol‐formaldehyde resins were post‐cured at 160°C, and then carbonized and graphitized from 300°C to 2400°C. The structure of the resulting carbonized and graphitized resins were studied using X‐ray diffraction and Raman spectroscopy. Thermal fragmentation and condensation of the polymer structure occurred above 300°C. The crystal size of the cured phenolic resins increased with an increase in temperature. The crystal size increased from 0.997 nm to 1.085 nm when the heat‐treatment temperature rose from 160°C to 500°C. Above 600°C, the original resin structures disappeared completely. Below 1000°C, the stack size (Lc) increased very slowly. The values increased from 0.992 to 1.192 nm when the heattreatment temperature rose from 600°C to 1000°C. Above 1000°C, the stack size showed an increase with the increase in heat‐treatment temperature. The values increased from 1.192 to 2.366 nm when the temperature rose from 1000°C to 2400°C. The carbonized and graphitized resins were characterized using Raman spectroscopy. The Raman spectrra were recorded between 700 and 2000 cm−1. Below 400°C, there were no carbon structures in the Raman spectra analysis. Above 500°C, G and D bands appeared. Raman spectra confirmed progressive structure ordering as heat‐treatment temperature increased. The frequency of the G band of all carbonized and graphitized samples shifted to 1600 cm−1 from the 1582 cm−1 of graphite. At the same temperature, the D band shifted to 1330 cm−1 from the 1357 cm−1 of the imperfect carbon. In the curve fitting analysis of the Raman spectra, a Gaussian shaped band centered at 1165 cm−1 was included. This band has not been described before in the literature and is attributed to disordered structures, which are formed from the original polymeric structures. These polymeric structures formed unknown disordered structures and remained in the carbonized phenolic resins. Above 1800°C, this band disappeared completely. But, a weak peak is present near 1620 cm−1. This indicated that those disoriented molecules and some disordered carbons were removed as volatiles or repacked into the glassy carbon structures during graphitization. The carbonized and graphitized phenolic resins were found to correspond to low order sp2 bonded carbon, but cannot be considered as truly glassy or amorphous carbon materials since they have some degree of order in the basal plane.

Correlation between Raman spectroscopy and mechanical properties of As-Sb-S-I chalcogenide glasses

SERS spectrum of gallic acid obtained from a modified silver colloid

Resonance Raman spectroscopy data of 2-aminoadenine and its deuterated species (C8-deuterated, N-deuterated and C8-, N-deuterated derivatives) in aqueous solution have been collected in the spectral region between 400 and 1800 cm−1, by using ultraviolet excitation wavelengths (λexc = 222, 257 and 281 nm) located in the three main UV absorption bands corresponding to the strongly allowed electronic transitions of the molecule of interest. Moreover, a Raman spectrum has been recorded under off-resonance conditions with a visible excitation (λexc= 488 nm). In order to assign the 2-aminoadenine in-plane vibrational bands displayed in the RRS spectra, a normal coordinate analysis has been performed by means of an empirical internal valence force field. These calculations are based on our recent normal mode analysis of adenine and guanine nucleic bases and their deuterated species, which was based on the joint use of resonance Raman spectroscopy and neutron inelastic scattering data. In the 2-aminoadenine force field proposed here, the diagonal force constants have been directly transferred from those recently obtained for adenine (and from guanine as concerns the 2-amino group), the interaction force constants (off-diagonal) then being adjusted on the basis of the actual experimental data from 2-aminoadenine and its deuterated species. The current force field is also able to assign infrared and Raman data obtained by other authors from polycrystalline samples of the pure species.

Spin-orbit entangled pseudospins hold promise for a wide array of exotic magnetism ranging from a Heisenberg antiferromagnet to a Kitaev spin liquid depending on the lattice and bonding geometry, but many of the host materials suffer from lattice distortions and deviate from idealized models in part due to inherent strong pseudospin-lattice coupling. Here, we report on the synthesis of a magnetic superlattice comprising the single ($n$=1) and the double ($n$=2) layer members of the Ruddlesden-Popper series iridates Sr$_{n+1}$Ir$_{n}$O$_{3n+1}$ alternating along the $c$-axis, and provide a comprehensive study of its lattice and magnetic structures using scanning transmission electron microscopy, resonant elastic and inelastic x-ray scattering, third harmonic generation measurements and Raman spectroscopy. The superlattice is free of the structural distortions reported for the parent phases and has a higher point group symmetry, while preserving the magnetic orders and pseudospin dynamics inherited from the parent phases, featuring two magnetic transitions with two symmetry-distinct orders. We infer weaker pseudospin-lattice coupling from the analysis of Raman spectra and attribute it to frustrated magnetic-elastic couplings. Thus, the superlattice expresses a near ideal network of effective spin-one-half moments on a square lattice.

A tutorial review on the use of Raman spectroscopy for studying conducting polymers is presented. The principles of Raman spectroscopy, instrumentation, and analysis of Raman spectra are described. Special emphasis is laid on the characterization of polarons and bipolarons formed upon doping. Two important points are described: (1) Raman spectroscopy with not only visible but also near-infrared laser excitation; (2) an “oligomer approach” for the analysis of Raman spectra. An application of this method to Na-doped poly(1,4-phenylene vinylene) is demonstrated.KeywordsRaman SpectrumRaman SpectroscopyRadical AnionNormal VibrationRaman IntensityThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Results of a method is presented for synthesizing ZnS:Mn nanoparticles capped in situ by ZnO. Analysis of Raman spectra and x-ray photoelectron spectra results have reinforced claim of the formation of ZnO capping layer on the surface of ZnS:Mn nanoparticles. Raman spectra results also showed presence of stress at an optimum ZnO capping thickness. In brief, the only variation within samples is in their ZnO capping thickness. Phase formation was analyzed and confirmed from powder x-ray diffraction. ZnS:Mn particle size is about 4nm. The change in photoluminescent properties with ZnO capping thickness variation is presented. It is shown that the variation in ZnO thickness and the resultant stress leads to an enhanced photoluminescence intensity/efficiency of nano-ZnS:Mn.

We present an optical spectroscopic technique, making use of both Raman signals and fluorescence spectroscopy, for the identification of five brands of commercially available extra-virgin olive-oil (EVOO). We demonstrate our technique on both a ‘bulk-optics’ free-space system and a compact device. Using the compact device, which is capable of recording both Raman and fluorescence signals, we achieved an average sensitivity and specificity of 98.4% and 99.6% for discrimination, respectively. Our approach demonstrates that both Raman and fluorescence spectroscopy can be used for portable discrimination of EVOOs which obviates the need to use centralised laboratories and opens up the prospect of in-field testing. This technique may enable detection of EVOO that has undergone counterfeiting or adulteration. One of the main challenges facing Raman spectroscopy for use in quality control of EVOOs is that the oxidation of EVOO, which naturally occurs due to aging, causes shifts in Raman spectra with time, which implies regular retraining would be necessary. We present a potential method of analysis to minimize the effect that aging has on discrimination efficiency; we show that by discarding the first principal component, which contains information on the variations due to oxidation, we can improve discrimination efficiency thus improving the robustness of our technique.