Prompt diagnosis of epilepsy relies on accurate classification of automated electroencephalogram (EEG) signals. Several approaches have been developed to characterize epileptic EEG data; however, none of them have exploited time-frequency data to evaluate the effect of tweaking parameters in pretrained frameworks for EEG data classification. This study compares the performance of several pretrained convolutional neural networks (CNNs) namely, AlexNet, GoogLeNet, MobileNetV2, ResNet-18 and SqueezeNet for the localization of epilepsy EEG data using various time-frequency data representation algorithms. Continuous wavelet transform (CWT), empirical Fourier decomposition (EFD), empirical mode decomposition (EMD), empirical wavelet transform (EWT), and variational mode decomposition (VMD) were exploited for the acquisition of 2D scalograms from 1D data. The research evaluates the effect of multiple factors, including noisy versus denoised scalograms, different optimizers, learning rates, single versus dual channels, model size, and computational time consumption. The benchmark Bern-Barcelona EEG dataset is used for testing purpose. Results obtained show that the combination of MobileNetV2, Continuous Wavelet Transform (CWT) and Adam optimizer at a learning rate of 10-4, coupled with dual-data channels, provides the best performance metrics. Specifically, these parameters result in optimal sensitivity, specificity, f1-score, and classification accuracy, with respective values of 96.06%, 96.15%, 96.08%, and 96.10%. To further corroborate the efficacy of opted pretrained models on exploited Signal Decomposition (SD) algorithms, the classifiers are also being simulated on Temple University database at pinnacle modeling composition. A similar pattern in the outcome readily validate the findings of our study and robustness of deep learning models on epilepsy EEG scalograms.The conclusions drawn emphasize the potential of pretrained CNN-based models to create a robust, automated system for diagnosing epileptiform. Furthermore, the study offers insights into the effectiveness of varying time-frequency techniques and classifier parameters for classifying epileptic EEG data.

To compare changes in the spherical component, regular astigmatism, and irregular astigmatism of the anterior surface of the cornea after small-incision lenticule extraction (SMILE) and transepithelial photorefractive keratectomy (TransPRK). Fifty-six patients underwent SMILE in 56 eyes, and 68 patients underwet TransPRK in 68 eyes. The right eye was chosen to enter the group. Six months after the procedure, Scheimpflug images were acquired, and Fourier analysis of the anterior surface of patients\' corneas was performed using the Pentacam built-in software. Fourier parameters encompass various measurements such as the steepest radius of the curvature and average eccentricity of the spherical components (SphRmin and SphEcc), maximum decentration (MaxDec), central and peripheral regular astigmatism (regular astigmatism at the center [AstC] and regular astigmatism at the periphery [AstP]), and irregularity (Irr). At 6 months postoperatively, SphEcc decreased significantly (P < .001), MaxDec increased significantly (P < .001), and Irr increased insignificantly (P = .254) in the SMILE group. SphEcc decreased significantly (P < .001) and MaxDec and Irr increased significantly (P < .001) in the TransPRK group. TransPRK caused greater changes in SphEcc, MaxDec, and Irr on the anterior corneal surface than SMILE (P < .05). The amount of MaxDec-induced changes in SMILE and TransPRK was significantly correlated with the amount of higher-order aberrations and spherical aberration changes (P < .05). SMILE and TransPRK increase overall irregular astigmatism on the anterior surface of the cornea, more so with TransPRK, where changes in decentration are associated with with increased higher-order aberrations.

As2O3 has shown significant anti-gastric cancer effects, but the mechanism is still unclear. Thus, biomacromolecular changes induced by As2O3 were investigated by using human gastric cancer AGS cells as the model. Flow cytometry results confirmed that As2O3 induced AGS cells apoptosis. Fourier transform infrared (FTIR) microspectroscopy detected biomacromolecular changes during As2O3-induced AGS cells apoptosis sensitively: IR spectra showed significant changes in the lipids content and the proteins and DNA structure. Peak-area ratios indicated obvious changes in the lipids and DNA content and the proteins structure, while also showing a relatively good linear relationship between A1733/A969 and the apoptosis rate. PCA exhibited significant alteration in nucleic acids while curve fitting further revealed the changes in nucleic acids and proteins. On the whole, our study explored As2O3-induced gastric cancer cells apoptosis in depth on the basis of analyzing biomacromolecular changes, in addition, it also suggested FTIR microspectroscopy to be possibly useful in the research of apoptosis.

The objective of this study was to non-destructively characterize samples of fresh beef loin by low-intensity ultrasound inspection at various frequencies and to correlate the acoustic parameters of these inspections with quality parameters. In this regard, ultrasonic parameters such as ultrasound pulse velocity (UPV) and variables related to attenuation and frequency components obtained from fast Fourier transform (FFT) were considered. For this, pulsed ultrasonic signal transducers with a frequency of 0.5 and 1.0 MHz were used. Acoustic parameters and those obtained through traditional instrumental analyses (physicochemical and texture) underwent a Pearson correlation analysis. The acoustic determinations revealed numerous significant correlations with the rest of the studied parameters. The results demonstrate that ultrasonic inspection has the ability to characterize samples with a non-destructive nature, and likewise, this methodology can be postulated as a promising predictive tool for determining quality parameters in beef loin samples.

The nonlinear rheological behaviors of three different classes of foods (emulsion, suspension, and elastic network) were studied and analyzed using the Rogers Sequence of Physical Processes (SPP) method and the Ewoldt-McKinley method of coupling Fourier Transform with Chebyshev Decomposition (FTC). SPP analysis led to instantaneous rheological parameters G\'t and G″t at any point in time, providing a more accurate picture of the linear viscoelastic region and crossover points by the 3D amplitude sweep. When G\'t is plotted against G″t, the resulting graph is a deltoid which offers a detailed and distinctive intracycle behavior of each class of food. Analyzing the revolution of deltoids with increasing strain allows for the determination of a critical strain, beyond which irreversible network breakdown occurs. The strain range between the linear viscoelastic limit and the critical strain found in SPP is comparable to the MAOS region as determined with FTC. Under increasing amplitude, predominantly elastic networks showed a gradual structural rearrangement, while more erratic and abrupt changes were observed in the suspension and emulsion we studied. Under increasing frequency, elastic responses dominate viscous responses in all samples due to the shorter experimental time, allowing less relaxation.

The article reports the results of experimental study of vibratory surface grinding in the range of low excitation frequencies and variable directions of excited vibrations in the plane of the table, and investigates the effect of these directions on the roughness and waviness of the ground surface. The tests were conducted on a production surface grinder with a vibrating table on which the samples were mounted. The table made it possible to change the direction for the introduction of vibrations to the workpiece (longitudinally, transversely, and obliquely to the longitudinal feed of the table) and the parameters of the introduced vibrations, frequency and amplitude. In the course of the study, selected parameters of surface roughness and waviness of samples ground conventionally and with vibrations introduced on the workpiece were compared. The results show an improvement in the roughness and waviness parameters of the vibration-ground surfaces compared to surfaces ground without vibration (conventionally). The profile of the ground surface was subjected to Fourier analysis and the harmonic components of the surface shape of the ground samples were determined to characterize the effect of the introduced vibrations on the surface roughness. It was determined that the direction of vibration introduction, which is most favorable in terms of the parameters of the geometric structure of the ground surface, is the direction perpendicular to the longitudinal feed of the grinding table. In other directions of vibration introduction, the simultaneous effect of improving both parameters of the geometric structure of the ground surface profile was not obtained.

The management of cancer patients has markedly improved with the advent of personalised medicine where treatments are given based on tumour antigen expression amongst other. Within this remit, liquid biopsies will no doubt improve this personalised cancer management. Identifying circulating tumour cells in blood allows a better assessment for tumour screening, staging, response to treatment and follow up. However, methods to identify/capture these circulating tumour cells using cancer cells\' antigen expression or their physical properties are not robust enough. Thus, a methodology that can identify these circulating tumour cells in blood regardless of the type of tumour is highly needed. Fourier Transform Infrared (FTIR) microspectroscopy, which can separate cells based on their biochemical composition, could be such technique. In this feasibility study, we studied lung cancer cells (squamous cell carcinoma and adenocarcinoma) mixed with peripheral blood mononuclear cells (PBMC). The data obtained shows, for the first time, that FTIR microspectroscopy together with Random Forest classifier is able to identify a single lung cancer cell in blood. This separation was easier when the region of the IR spectra containing lipids and the amide A (2700 to 3500 cm-1) was used. Furthermore, this work was carried out using glass coverslips as substrates that are widely used in pathology departments. This allows further histopathological cell analysis (staining, immunohistochemistry, …) after FTIR spectra are obtained. Hence, although further work is needed using blood samples from patients with cancer, FTIR microspectroscopy could become another tool to be used in liquid biopsies for the identification of circulating tumour cells, and in the personalised management of cancer.

Animal models of cirrhosis are of great interest to investigate the pathological process leading to the final stage of cirrhosis. The aim of this study was to analyze the different steps involved in the progressive development of cirrhosis using Fourier transform infrared spectral histology in 2 mouse models of cirrhosis, the STAM model of metabolic cirrhosis, and the carbon tetrachloride-induced cirrhosis model. Formalin-fixed, paraffin-embedded liver samples were obtained from 3 mice at 5 time points in each model to analyze the course of hepatic lesions up to the formation of cirrhosis. For each time point, adjacent 3-μm-thick liver sections were obtained for histologic stains and spectral histology. Fourier transform infrared acquisitions of liver sections were performed at projected pixel sizes of 25 μm × 25 μm and 6.25 μm × 6.25 μm. Spectral images were then preprocessed with an extended multiplicative signal correction and analyzed with common k-means clustering, including all stages in each model. In both models, the 2- and 4-class common k-means clustering in the 1000 to 1350 cm-1 range showed that spectral classes characterized by higher absorbance peaks of glycogen were predominant at baseline, then decreased markedly in early stages of hepatic damage, and almost disappeared in cirrhotic tissues. Concomitantly, spectral classes characterized by higher absorbance peaks of nucleic acids became progressively predominant during the course of hepatic lesions. These results were confirmed using k-means clustering on the peaks of interest identified for glycogen and nucleic acid content. Our study showed that the glycogen depletion previously described at the stage of cirrhosis is an early event in the pathological process, independently of the cause of cirrhosis. In addition, there was a progressive increase in the nucleic acid content, which may be linked to increased proliferation and polyploidy in response to cellular lesions.

This paper aims to understand the binding strategies of a nanobody-protein pair by studying known complexes. Rigid body protein-ligand docking programs produce several complexes, called decoys, which are good candidates with high scores of shape complementarity, electrostatic interactions, desolvation, buried surface area, and Lennard-Jones potentials. However, the decoy that corresponds to the native structure is not known. We studied 36 nanobody-protein complexes from the single domain antibody database, sd-Ab DB, http://www.sdab-db.ca/. For each structure, a large number of decoys are generated using the Fast Fourier Transform algorithm of the software ZDOCK. The decoys were ranked according to their target protein-nanobody interaction energies, calculated by using the Dreiding Force Field, with rank 1 having the lowest interaction energy. Out of 36 protein data bank (PDB) structures, 25 true structures were predicted as rank 1. Eleven of the remaining structures required Ångstrom size rigid body translations of the nanobody relative to the protein to match the given PDB structure. After the translation, the Dreiding interaction (DI) energies of all complexes decreased and became rank 1. In one case, rigid body rotations as well as translations of the nanobody were required for matching the crystal structure. We used a Monte Carlo algorithm that randomly translates and rotates the nanobody of a decoy and calculates the DI energy. Results show that rigid body translations and the DI energy are sufficient for determining the correct binding location and pose of ZDOCK created decoys. A survey of the sd-Ab DB showed that each nanobody makes at least one salt bridge with its partner protein, indicating that salt bridge formation is an essential strategy in nanobody-protein recognition. Based on the analysis of the 36 crystal structures and evidence from existing literature, we propose a set of principles that could be used in the design of nanobodies.

Biofilms represent the main mode of existence of bacteria and play very significant roles in many industrial, medical and agricultural fields. Analysis of biofilms is a challenging task owing to their sophisticated composition, heterogeneity and variability. In this study, biofilms formed by the rhizobacterium Azospirillum baldaniorum (strain Sp245), isolated biofilm matrix and its macrocomponents have for the first time been studied in detail, using Fourier transform infrared (FTIR) spectroscopy, with a special emphasis on the methodology. The accompanying novel data of comparative chemical analyses of the biofilm matrix, its fractions and lipopolysaccharide isolated from the outer membrane of the cells of this strain, as well as their electrophoretic analyses (SDS-PAGE) have been found to be in good agreement with the FTIR spectroscopic results.