Three most commonly used primary models (Gompertz, Baranyi and three-phase linear models) to describe the microbial growth curves were applied to three different isothermal growth data of Listeria monocytogenes. Further Monte Carlo analysis was performed with 100, 1000 and 10000 simulations. The results indicated that there was no reason to use higher number of simulations since the simulations produced almost identical means of the model parameter values for all models. Moreover, the models had similar coefficient of variation values for the initial (log10N0) and maximum (log10Nmax) number of bacteria. On the other hand, the Gompertz model had the highest coefficient of variation for the growth rate (µmax) and the Baranyi model had the highest coefficient of variation for the lag time (λ). Correlations between the parameters log10N0 and λ, and µmax and λ could be easily observed after the Monte Carlo analysis for all models. Deviation from normal distribution for the parameter λ for the three-phase linear model was evident, other than that all parameters for all models had normal distribution. It was concluded that Monte Carlo analysis can be used as a simple yet an effective method to describe the uncertainty in model parameters and correlation between the parameters as well as the spread of the possible parameter values.
9 February 2023 in International Journal of Biological Macromolecules
Bu çalışmada mikroorganizmaların (bakteriler, mayalar, küfler) büyüme/gelişme eğrilerini tanımlamak için Excel® tabanlı ücretsiz bir araç olan ÖK-BUZ (Öksüz-Buzrul) GRoFiT (büyüme eğrileri tanımlama) uygulamasının kullanımı gösterilmiştir. Günümüzde birçok büyüme modeli kullanılmakta olup, bunların içerisinden belki de en fazla kullanılan büyüme modellerinden olan Baranyi, Gompertz ve Üç Fazlı Doğrusal (ÜFD) modelleri Excel içerisine yerleştirilerek, modelleme konusunda çok yetkin olmayan kişilerin de rahatlıkla kullanabileceği ve sonuçları yorumlayabileceği kullanıcı dostu bir araç oluşturulmuştur. Bu araç kullanılarak her üç modelin parametreleri yani mikroorganizmaların başlangıç sayısı (log10N0), mikroorganizmaların ulaşabileceği azami sayı (log10Nmaks), azami özgül büyüme hızı (µmaks veya µ) ve uyum (lag) zamanı (λ) standart hatalarıyla birlikte bulunabilir. Dahası model parametrelerinin üst ve alt sınırları (%95 güven aralıkları) da elde edilebilmektedir. Bu değerlerin hemen yanında model uyum göstergeleri listelenmektedir. Bunlar belirleme katsayısı (R2), ayarlı belirleme katsayısı (ayarlı R2) ve modelin standart hatası veya karesel ortalama hatadır (RMSE). Araç içerisindeki modeller birçok farklı büyüme veri setine uygulanmış ve sonuçlar ücretli bir yazılım olan SigmaPlot® programı ile karşılaştırılmıştır. Karşılaştırma sonucunda rahatlıkla söylenebilir ki bu uygulamanın mikroorganizmaların büyüme eğrilerini tanımlaması bakımından paralı yazılımlardan farkı yoktur: ÖK-BUZ GRoFiT kullanılarak aynı veri seti için aynı parametre değerlerini ve aynı model uyum göstergelerini elde etmek mümkündür. Ayrıca, ÖK-BUZ GRoFiT Monte Carlo benzetimi yaparak ücretli programların birçoğunda bulunmayan simetrik olmayan %95 güven aralıklarını da hesaplamaktadır. Parametrelerin güven aralıklarını hesaplayabilen ücretli yazılımlar ise genel olarak simetrik güven aralıklarını vermektedir. ÖK-BUZ GRoFiT uygulaması tamamen Türkçe olup, “https://mmf.gidatarim.edu.tr/ok-buz-grofit” adresinden ücretsiz olarak indirilerek kullanılabilir. Kullanıcılardan gelecek olan talepler doğrultusunda uygulama içerisine yeni büyüme modelleri de eklenebilecek ve farklı disiplinlerde çalışan araştırmacıların (gıda, kimya, biyoloji, ziraat, biyoteknoloji vb.) bu uygulamayı kullanması mümkün olacaktır.
Fine-grained wheat variety classification involving a large number of cultivars remains challenging because different varieties often exhibit highly similar morphological and textural characteristics. In this study, a dataset comprising 124 wheat varieties was used to systematically evaluate convolutional and Transformer-based architectures and investigate the contribution of foreground-aware preprocessing to wheat variety classification. DenseNet201, MobileNetV2, InceptionV3, and Swin Transformer Tiny (SwinT) were first evaluated as baseline architectures. SwinT achieved the highest baseline performance, with a weighted precision of 92.81% and a weighted F1-score of 91.91%, and was therefore selected for subsequent experiments. A foreground-aware preprocessing pipeline, termed CropWheatSwinT, was developed using foreground localization, safety-margin cropping, aspect-ratio-preserving resizing, and background-color padding to increase the proportion of informative kernel pixels while reducing irrelevant background information. CropWheatSwinT achieved a weighted precision of 93.92%, a weighted recall of 93.07%, and a weighted F1-score of 93.15%. At the class level, the Macro F1-score increased from 0.9196 ± 0.0451 for SwinT to 0.9317 ± 0.0437 for CropWheatSwinT, corresponding to a mean improvement of 0.0121. This improvement was statistically significant according to the Wilcoxon signed-rank test (W = 2075.0, p < 0.001), with a moderate effect size (Cohen’s dz = 0.40). In contrast, the CA-, CBAM-, and RGB–Sobel-based variants achieved lower Macro F1-scores than SwinT, with significant negative differences relative to the baseline (p < 0.001). These findings suggest that foreground-focused preprocessing can provide a more effective improvement in fine-grained wheat variety classification than the investigated attention- and edge-enhancement modifications, highlighting the potential of data-centric strategies for improving Transformer-based agricultural image classification.
A user-friendly MS Excel® spreadsheet as a freeware (R-BioXL) was developed to fit mathematical models to experimental data. (R-BioXL is available to everyone at https://drive.google.com/drive/folders/1GyjT3Z_CJQZu6ASb4LQBlS-ajLa_nF6X?usp=sharing) Initially, users are expected to enter their X-Y data and define their parameters of the model. Then, a model equation should also be entered again by users. Users can visualize data (scatter plot) and model fit (line plot) with the defined initial estimates of parameters on the same graph by default. Squared differences between experimental data and model estimates are calculated automatically. Users can change the initial estimates of the parameters to make the model closer to the data instantly, and Solver Add-In of Excel® should be used to minimize the sum of squared error by changing the parameter values. After the parameters are obtained, standard errors (by using “SolverAid” macro), 95 and 99% confidence intervals of the parameters, p values to determine the statistical significance of the parameters, and goodness-of-fit indices are calculated as the last step. All results can be saved on a different Excel® working page. Whole procedure takes a couple of minutes (~3 to 10 min) depending on the Excel® experience of the user. The utility, accuracy and reliability of the spreadsheet was shown by applying two-parameter (non-linear) Michealis-Menten equation for enzyme kinetics, three-parameter (linear) van Deemter equation for chromatography, and four-parameter (non-linear) modified Gompertz equation for microbial growth. In conclusion, R-BioXL can be safely and freely used to describe the experimental data with Excel® knowledge, without any skills in programming and without additional cost for other software package.