Comparison of deep learning methods for the radiographic detection of patients with different periodontitis stages.

OBJECTIVES: The objective of this study is to assess the accuracy of computer-assisted periodontal classification bone loss staging using deep learning (DL) methods on panoramic radiographs and to compare the performance of various models and layers. METHODS: Panoramic radiographs were diagnosed and classified into 3 groups, namely "healthy," "Stage1/2," and "Stage3/4," and stored in separate folders. The feature extraction stage involved transferring and retraining the feature extraction layers and weights from 3 models, namely ResNet50, DenseNet121, and InceptionV3, which were proposed for classifying the ImageNet dataset, to 3 DL models designed for classifying periodontal bone loss. The features obtained from global average pooling (GAP), global max pooling (GMP), or flatten layers (FL) of convolutional neural network (CNN) models were used as input to the 8 different machine learning (ML) models. In addition, the features obtained from the GAP, GMP, or FL of the DL models were reduced using the minimum redundancy maximum relevance (mRMR) method and then classified again with 8 ML models. RESULTS: A total of 2533 panoramic radiographs, including 721 in the healthy group, 842 in the Stage1/2 group, and 970 in the Stage3/4 group, were included in the dataset. The average performance values of DenseNet121 + GAP-based and DenseNet121 + GAP + mRMR-based ML techniques on 10 subdatasets and ML models developed using 2 feature selection techniques outperformed CNN models. CONCLUSIONS: The new DenseNet121 + GAP + mRMR-based support vector machine model developed in this study achieved higher performance in periodontal bone loss classification compared to other models in the literature by detecting effective features from raw images without the need for manual selection.

Quantitative determination of phenolic compounds in propolis samples from the Black Sea Region (Türkiye) based on HPTLC images using partial least squares and genetic inverse least squares methods.

The complex chemical composition of propolis is related to the plant source to be used by honeybees. Propolis type is defined based on the plant source with the highest proportion in its composition, which is determined by chromatographic techniques as high-performance thin-layer chromatography (HPTLC). In addition to marker component identification to specify the propolis type, quantification of its proportion is also significant for prediction and reproducible pharmacological activity. One drawback for propolis marker component quantitation is that during the chromatographical analysis, not the main but the other plant sources with less proportion may cause interferences during the chemical analysis. In this study, the amounts of marker components were compared with the reference analysis data obtained by high-performance liquid chromatography (HPLC) and from HPTLC images using Partial Least Squares (PLS) and Genetic Inverse Least Squares (GILS) regression methods. Firstly, HPTLC images of propolis samples were processed by an image algorithm (developed in MATLAB) where the bands of each standard and the samples were cut same dimensional pieces as 351 × 26 pixels in height and width, respectively. Simultaneously, reference analysis of the marker components in propolis samples was performed with a validated HPLC method. Consequently, the reference values obtained from HPLC versus PLS, and GILS predicted values of the eight compounds based on the digitized HPTLC images of the chromatograms were found to be matched successfully. The results of the multivariate calibration models demonstrated that HPTLC images could be used quantitatively for quality control of propolis used as a food supplement.

Rapid detection of green-pea adulteration in pistachio nuts using Raman spectroscopy and chemometrics.

BACKGROUND: Ground pistachio nut is prone to adulteration because of its high economic value and wide usage. Green pea is known as the main adulterant in frauds involving pistachio nuts. The present study developed a new, rapid, reliable and low-cost methodology by using a portable Raman spectrometer in combination with chemometrics for the detection of green pea in pistachio nuts. RESULTS: Three different methods of Raman spectroscopy-based chemometrics analysis were developed for the determination of green-pea adulteration in pistachio nuts. The first method involved the development of hierarchical cluster analysis (HCA) and principal component analysis (PCA), which differentiated authentic pistachio nuts from green pea and green pea-adulterated samples. The best classification pattern was observed in the adulteration range of 20-80% (w/w). In addition to classification methods, partial least squares regression (PLSR) and genetic algorithm-based inverse least squares (GILS) were also used to develop multivariate calibration models to determine quantitatively the degree of green-pea adulteration in grounded pistachio nuts. The spectral range of 1790-283 cm-1 was used in the case of multivariate data analysis. A green-pea adulteration level of 5-80% (w/w) was successfully identified by PLSR and GILS. The correlation coefficient of determination (R2 ) was determined as 0.91 and 0.94 for the PLSR and GILS analyses, respectively. CONCLUSION: A Raman spectrometer combined with chemometrics has a high capability with regard to the detection of adulteration in pistachio nuts, combined with low cost, strong reliability, a high level of accuracy, rapidity of analysis, and minimum sample preparation. © 2020 Society of Chemical Industry.

The impact of socioeconomic factors on the healthcare costs of people living with HIV in Turkey.

BACKGROUND: This study addresses an important field within HIV research, the impact of socioeconomic factors on the healthcare costs of people living with HIV/AIDS (PLHIV). We aimed to understand how different socioeconomic factors could create diverse healthcare costs for PLHIV in Turkey. METHODS: Data were collected between January 2017 and December 2017. HIV-positive people attending the clinic who had been referred to the national ART programme from January 1992 until December 2017 were surveyed. The questionnaire collected socioeconomic data. The cost data for the same patients was taken from the electronic database Probel Hospital Information Management System (PHIMS) for the same period. The PHIMS data include costs for medication (highly active antiretroviral therapy or HAART), laboratory, pathology, radiology, polyclinic, examination and consultation, hospitalisation, surgery and intervention, blood and blood products, supplies and other costs. Data were analysed using STATA 14.2 to estimate the generalised linear model (GLM). RESULTS: The findings of our GLM indicate that age, gender, marital and parental status, time since diagnosis, employment, wealth status, illicit drug use and CD4 cell count are the factors significantly related to the healthcare cost of patients. We found that compared with people who have AIDS (CD4 cells < 200 cells/mm3), people who have a normal range of CD4 cells (≥ 500 cells/mm3) have $1046 less in expenditures on average. Compared to younger people (19-39 years), older people (≥ 55) have $1934 higher expenditures on average. Costs are $644 higher on average for married people and $401 higher on average for people who have children. Healthcare costs are $518 and $651 higher on average for patients who are addicted to drugs and who use psychiatric drug(s), respectively. Compared to people who were recently diagnosed with HIV, people who were diagnosed ≥10 years ago have $743 lower expenditures on average. CONCLUSION: Our results suggest that in addition to immunological status, socioeconomic factors play a substantial role in the healthcare costs of PLHIV. The key factors influencing the healthcare costs of PLHIV are also critical for public policy makers, healthcare workers, health ministries and employment community programs.

The determinants of employability of people living with HIV/AIDS in Turkey.

BACKGROUND: This study addresses an important field within HIV research, the factors affecting the determinants of the employability of people living with HIV/AIDS (PLHIV) in Turkey. The employability of PLHIV is now even more vital because the use of antiretroviral therapy improves the quality of life of patients. In spite of this, the related literature suggests that there are serious impediments to the employment of PLHIV who face considerable levels of discrimination based on their HIV status. METHODS: This is a cohort study of 170 PLHIV of working age, treated at the Izmir Bozyaka Education and Training Hospital. We use a univariate logistic model to determine the effects of all determinants of interest with probit/logit modeling and penalized maximum likelihood estimation to avoid bias and to test the robustness of results. RESULTS: Age, time since diagnosis, work status at diagnosis, wealth status, illicit drug use, and CD4 cell count were significantly related to the employability of PLHIV. Younger individuals had a higher probability of workforce participation. HIV-infected patients aged 19 to 39 and 40 to 54 years were 32% and 20% more likely, respectively, to be employed. Economically better-off PLHIV were more likely to participate in the labor force and HIV patients who were working at the time of diagnosis were more likely to be re-employed. Time since diagnosis was negatively associated with the employment status. Compared to recently diagnosed patients, PLHIV for more than a decade were less likely to be employed. Those with high CD4 cell counts were more likely to be employed. Illicit drug use was negatively associated with employment and drug-addicted HIV patients were less likely to be employed. Higher education did not significantly predict the employability of PLHIV. CONCLUSIONS: Our results suggest that besides immunological status, socioeconomic factors play a substantial role in the employability of PLHIV. We suggest that even if a patient is skilled, educated, and qualified for the job, other factors such as stigma and employment discrimination in the workplace may hinder employment even among highly educated PLHIV.

A rapid ATR-FTIR spectroscopic method for classification of gelatin gummy candies in relation to the gelatin source.

Gelatin is widely used in gummy candies because of its unique functional properties. Generally, porcine and bovine gelatins are used in the food industry. FTIR-ATR combined with chemometrics analysis such as hierarchical cluster analysis (HCA) (OPUS Version 7.2 software), principal component analysis (PCA) (OPUS Version 7.2 software) and partial least squares-discriminant analysis (PLS-DA) (Matlab R2017b) were used for classification and discrimination of gelatin gummy candies related to their gelatin source. The spectral region between 1734 and 1528 cm-1 was selected for chemometric analysis. The potential of FTIR spectroscopy for determination of bovine and porcine source in gummy candies was examined and validated by a real-time polymerase chain reaction (PCR) method. Twenty commercial samples were tested by developed ATR-FTIR methodology and RT-PCR technique, mutually confirming and supporting results were obtained. Gummy candies were classified and discriminated in relation to the bovine or porcine source of gelatin with 100% success without any sample preparation using FTIR-ATR technique.

The impact of economic and social factors on the prevalence of hepatitis B in Turkey.

BACKGROUND: Viral Hepatitis is one of the major global health problems, affecting millions of people every year. Limited information is available on the impact of social and economic factors on the prevalence of Hepatitis B virus (HBV) in Turkey. This study, contrary to other studies in the literature, was undertaken with the aim of examining the Majority of the excluded data come from the volunteers. METHODS: There are medical and the social-economic factors affecting the prevalence of HBV. This research, while taking medical factors as control variables, clarify the social and economic factors affecting the prevalence of HBV by utilising clinical data with the use of the Binary Probit Model (BPM). The BPM estimation is a powerful tool to determine not only the factors but explain also the exact impacts of each factor. RESULTS: The estimations of the BPM shows that economic and social variables such as age, gender, migration, education, awareness, social welfare, occupation are very important factors for determining HBV prevalence. Compared to the youngest population, the 46 to 66+ age group has a higher prevalence of HBV. The male respondents were 5% more likely to develop HBV compared to females. When region-specific differences are taken into account, migrating from the poorest parts of the country such as the eastern and south-eastern regions of Turkey are approximately 16% more likely to be infected. The welfare indicators such as a higher number of rooms in the respondent's house or flat decreases the probability of having HBV and, relatively higher income groups are less likely to develop HBV compared to labourers. The Self-employed/Business owner/Public sector worker category are approximately 10% less likely to develop HBV. When people are aware of the methods of prevention of HBV, they are 6% less likely to be infected. Previous HBV infection history increases the probability of having HBV again B by 17%. CONCLUSIONS: These findings strongly suggest that the impact of social and economic factors on the prevalence of HBV is vital. Any improvements in these factors are likely to reduce prevalence of HBV.

Determination of honey adulteration with beet sugar and corn syrup using infrared spectroscopy and genetic-algorithm-based multivariate calibration.

BACKGROUND: Fourier transform infrared spectroscopy (FTIR) equipped with attenuated total reflectance accessory was used to determine honey adulteration. Adulterated honey samples were prepared by adding corn syrup, beet sugar and water as adulterants to the pure honey samples in various amounts. The spectra of adulterated and pure honey samples (n = 209) were recorded between 4000 and 600 cm-1 wavenumber range. RESULTS: Genetic-algorithm-based inverse least squares (GILS) and partial least squares (PLS) methods were used to determine honey content and amount of adulterants. Results indicated that the multivariate calibration generated with GILS could produce successful models with standard error of cross-validation in the range 0.97-2.52%, and standard error of prediction between 0.90 and 2.19% (% w/w) for all the components contained in the adulterated samples. Similar results were obtained with PLS, generating slightly larger standard error of cross-validation and standard error of prediction values. CONCLUSION: The fact that the models were generated with several honey samples coming from various different botanical and geographical origins, quite successful results were obtained for the detection of adulterated honey samples with a simple Fourier transform infrared spectroscopy technique. Having a genetic algorithm for variable selection helped to build somewhat better models with GILS compared with PLS. © 2018 Society of Chemical Industry.

[Evaluation of Malassezia species by Fourier transform infrared (FT-IR) spectroscopy]. / Fourier Dönüsümlü Kizilötesi (FT-IR) Spektroskopisi ile Malassezia Türlerinin Degerlendirilmesi.

Malassezia species which are lipophilic exobasidiomycetes fungi, have been accepted as members of normal cutaneous flora as well as causative agent of certain skin diseases. In routine microbiology laboratory, species identification based on phenotypic characters may not yield identical results with taxonomic studies. Lipophilic and lipid-dependent Malassezia yeasts require lipid-enriched complex media. For this reason, Fourier transform infrared (FT-IR) spectroscopy analysis focused on lipid window may be useful for identification of Malassezia species. In this study, 10 different standard Malassezia species (M.dermatis CBS 9145, M.furfur CBS 7019, M.japonica CBS 9432, M.globosa CBS 7966, M.nana CBS 9561, M.obtusa CBS 7876, M.pachydermatis CBS 1879, M.slooffiae CBS 7956, M.sympodialis CBS 7222 and M.yamatoensis CBS 9725) which are human pathogens, have been analyzed by FT-IR spectroscopy following standard cultivation onto modified Dixon agar medium. Results showed that two main groups (M1; M.globosa, M.obtusa, M.sympodialis, M.dermatis, M.pachydermatis vs, M2; M.furfur, M.japonica, M.nana, M.slooffiae, M.yamatoensis) were discriminated by whole spectra analysis. M.obtusa in M1 by 1686-1606 cm-1 wavenumber ranges and M.japonicum in M2 by 2993-2812 cm-1 wavenumber ranges were identified with low level discrimination power. Discriminatory areas for species differentiation of M1 members as M.sympodialis, M.globosa and M.pachydermatis and M2 members as M.furfur and M.yamatoensis could not be identified. Several spectral windows analysis results revealed that FT-IR spectroscopy was not sufficient for species identification of culture grown Malassezia species.

Determination of thiamine HCl and pyridoxine HCl in pharmaceutical preparations using UV-visible spectrophotometry and genetic algorithm based multivariate calibration methods.

Simultaneous determination of binary mixtures pyridoxine hydrochloride and thiamine hydrochloride in a vitamin combination using UV-visible spectrophotometry and classical least squares (CLS) and three newly developed genetic algorithm (GA) based multivariate calibration methods was demonstrated. The three genetic multivariate calibration methods are Genetic Classical Least Squares (GCLS), Genetic Inverse Least Squares (GILS) and Genetic Regression (GR). The sample data set contains the UV-visible spectra of 30 synthetic mixtures (8 to 40 microg/ml) of these vitamins and 10 tablets containing 250 mg from each vitamin. The spectra cover the range from 200 to 330 nm in 0.1 nm intervals. Several calibration models were built with the four methods for the two components. Overall, the standard error of calibration (SEC) and the standard error of prediction (SEP) for the synthetic data were in the range of <0.01 and 0.43 microg/ml for all the four methods. The SEP values for the tablets were in the range of 2.91 and 11.51 mg/tablets. A comparison of genetic algorithm selected wavelengths for each component using GR method was also included.