Comparative assessment of established and deep learning-based segmentation methods for hippocampal volume estimation in brain magnetic resonance imaging analysis.

In this study, our objective was to assess the performance of two deep learning-based hippocampal segmentation methods, SynthSeg and TigerBx, which are readily available to the public. We contrasted their performance with that of two established techniques, FreeSurfer-Aseg and FSL-FIRST, using three-dimensional T1-weighted MRI scans (n = 1447) procured from public databases. Our evaluation focused on the accuracy and reproducibility of these tools in estimating hippocampal volume. The findings suggest that both SynthSeg and TigerBx are on a par with Aseg and FIRST in terms of segmentation accuracy and reproducibility, but offer a significant advantage in processing speed, generating results in less than 1 min compared with several minutes to hours for the latter tools. In terms of Alzheimer's disease classification based on the hippocampal atrophy rate, SynthSeg and TigerBx exhibited superior performance. In conclusion, we evaluated the capabilities of two deep learning-based segmentation techniques. The results underscore their potential value in clinical and research environments, particularly when investigating neurological conditions associated with hippocampal structures.

Virtual MOLLI Target: Generative Adversarial Networks Toward Improved Motion Correction in MRI Myocardial T1 Mapping.

BACKGROUND: The modified Look-Locker inversion recovery (MOLLI) sequence is commonly used for myocardial T1 mapping. However, it acquires images with different inversion times, which causes difficulty in motion correction for respiratory-induced misregistration to a given target image. HYPOTHESIS: Using a generative adversarial network (GAN) to produce virtual MOLLI images with consistent heart positions can reduce respiratory-induced misregistration of MOLLI datasets. STUDY TYPE: Retrospective. POPULATION: 1071 MOLLI datasets from 392 human participants. FIELD STRENGTH/SEQUENCE: Modified Look-Locker inversion recovery sequence at 3 T. ASSESSMENT: A GAN model with a single inversion time image as input was trained to generate virtual MOLLI target (VMT) images at different inversion times which were subsequently used in an image registration algorithm. Four VMT models were investigated and the best performing model compared with the standard vendor-provided motion correction (MOCO) technique. STATISTICAL TESTS: The effectiveness of the motion correction technique was assessed using the fitting quality index (FQI), mutual information (MI), and Dice coefficients of motion-corrected images, plus subjective quality evaluation of T1 maps by three independent readers using Likert score. Wilcoxon signed-rank test with Bonferroni correction for multiple comparison. Significance levels were defined as P < 0.01 for highly significant differences and P < 0.05 for significant differences. RESULTS: The best performing VMT model with iterative registration demonstrated significantly better performance (FQI 0.88 ± 0.03, MI 1.78 ± 0.20, Dice 0.84 ± 0.23, quality score 2.26 ± 0.95) compared to other approaches, including the vendor-provided MOCO method (FQI 0.86 ± 0.04, MI 1.69 ± 0.25, Dice 0.80 ± 0.27, quality score 2.16 ± 1.01). DATA CONCLUSION: Our GAN model generating VMT images improved motion correction, which may assist reliable T1 mapping in the presence of respiratory motion. Its robust performance, even with considerable respiratory-induced heart displacements, may be beneficial for patients with difficulties in breath-holding. LEVEL OF EVIDENCE: 3 TECHNICAL EFFICACY: Stage 1.

Germicidal effect of intense pulsed light on Pseudomonas aeruginosa in food processing.

Background: Pseudomonas aeruginosa (P. aeruginosa) can cause serious infections in many parts of the body and is also an underestimated foodborne pathogen. Intense pulsed light sterilization is recognized for its high sterilization efficiency, flexible and safe operation and ease of installation on production lines, which makes up for the shortcomings of several other physical sterilization technologies. Methods: This experiment studied the killing efficiency of different capacitances (650 µF, 470 µF, and 220 µF) of intense pulsed light on foodborne pathogenic microorganisms P. aeruginosa in the models of liquid food models, 96-well cell plates, and polycarbonate membrane models at room temperature (25°C) and refrigerated (4°C) environments to provide data to support the application of IPL sterilization devices in food processing. Results: The IPL was very effective in killing P. aeruginosa in the planktonic state as well as in the early and mature biofilm states, meeting target kill rates of 100%, 99.99%, and 94.33% for a given number of exposures. The biofilms formed in the polycarbonate membrane model and the 96-well plate model were more resistant to killing compared to the planktonic state. To achieve the same bactericidal effect, the number of flashes increased with decreasing capacitance. Conclusion: The bactericidal effect of IPL on P. aeruginosa was significantly influenced by the state of the bacterium. The larger the capacitance the higher the number of pulses and the better the sterilization effect on P. aeruginosa.

The role of input imaging combination and ADC threshold on segmentation of acute ischemic stroke lesion using U-Net.

BACKGROUND: To evaluate the effect of the weighting of input imaging combo and ADC threshold on the performance of the U-Net and to find an optimized input imaging combo and ADC threshold in segmenting acute ischemic stroke (AIS) lesion. METHODS: This study retrospectively enrolled a total of 212 patients having AIS. Four combos, including ADC-ADC-ADC (AAA), DWI-ADC-ADC (DAA), DWI-DWI-ADC (DDA), and DWI-DWI-DWI (DDD), were used as input images, respectively. Three ADC thresholds including 0.6, 0.8 and 1.8 × 10-3 mm2/s were applied. Dice similarity coefficient (DSC) was used to evaluate the segmentation performance of U-Nets. Nonparametric Kruskal-Wallis test with Tukey-Kramer post-hoc tests were used for comparison. A p < .05 was considered statistically significant. RESULTS: The DSC significantly varied among different combos of images and different ADC thresholds. Hybrid U-Nets outperformed uniform U-Nets at ADC thresholds of 0.6 × 10-3 mm2/s and 0.8 × 10-3 mm2/s (p < .001). The U-Net with imaging combo of DDD had segmentation performance similar to hybrid U-Nets at an ADC threshold of 1.8 × 10-3 mm2/s (p = .062 to 1). The U-Net using the imaging combo of DAA at the ADC threshold of 0.6 × 10-3 mm2/s achieved the highest DSC in the segmentation of AIS lesion. CONCLUSIONS: The segmentation performance of U-Net for AIS varies among the input imaging combos and ADC thresholds. The U-Net is optimized by choosing the imaging combo of DAA at an ADC threshold of 0.6 × 10-3 mm2/s in segmentating AIS lesion with highest DSC. KEY POINTS: • Segmentation performance of U-Net for AIS differs among input imaging combos. • Segmentation performance of U-Net for AIS differs among ADC thresholds. • U-Net is optimized using DAA with ADC = 0.6 × 10-3 mm2/s.

Biofilm formation of two genetically diverse Staphylococcus aureus isolates under beta-lactam antibiotics.

Purpose: Our aim was to evaluate the biofilm formation of 2 genetically diverse Staphylococcus aureus isolates, 10379 and 121940, under different concentrations of beta-lactam antibiotics on biomass content and biofilm viability. Methods: Biofilm formation and methicillin resistance genes were tested using PCR and multiplex PCR. PCR was combined with bioinformatics analysis to detect multilocal sequence typing (MLST) and SCCmec types, to study the genetical correlation between the tested strains. Then, the crystal violet (CV) test and XTT were used to detect biomass content and biofilm activity. Antibiotic susceptibility was tested using a broth dilution method. According to their specific MIC, different concentrations of beta-lactam antibiotics were used to study its effect on biomass content and biofilm viability. Results: Strain 10379 carried the icaD, icaBC, and MRSA genes, not the icaA, atl, app, and agr genes, and MLST and SCCmec typing was ST45 and IV, respectively. Strain 121940 carried the icaA, icaD, icaBC, atl, and agr genes, not the aap gene, and MLST and SCCmec typed as ST546 and IV, respectively. This suggested that strains 10379 and 121940 were genotypically very different. Two S. aureus isolates, 10379 and 121940, showed resistance to beta-lactam antibiotics, penicillin, ampicillin, meropenem, streptomycin and kanamycin, some of which promoted the formation of biofilm and biofilm viability at low concentrations. Conclusion: Despite the large differences in the genetic background of S. aureus 10379 and 121940, some sub-inhibitory concentrations of beta-lactam antibiotics are able to promote biomass and biofilm viability of both two isolates.

Single breath-hold MR T1 mapping in the heart: Hybrid MOLLI combining saturation and inversion recovery.

The native T1 values of the myocardium provide valuable information for tissue characterization and assessment of cardiomyopathies. In this study, we proposed a novel hybrid MOLLI sequence for myocardial T1 mapping. Unlike the two groups of inversion-recovery sampling of the conventional MOLLI5(3 s)3 sequence, the hybrid MOLLI sequence consisted of an inversion-recovery block followed by a saturation-recovery block. Since the second block employed a saturation pulse to spoil the longitudinal magnetization, it did not require a waiting period as MOLLI5(3 s)3 did. As a result, the hybrid MOLLI required less acquisition time leading to a practical application for patients with breath-hold difficulties. Phantom and healthy subject experiments were performed to evaluate the proposed sequence against the MOLLI5(3 s)3 sequence. The phantom study showed that the heart-rate dependency of one variant of the hybrid MOLLI sequences, hbMOLLI4, was comparable to that of MOLLI5(3 s)3. In addition, both hbMOLLI4 and MOLLI53 derived T1 values under 2% variations with simulated heart rates from 50 to 90 beats-per-minute within the range of T1 values for myocardium and blood before contrast administration. Simulation results suggested slightly reduced T1 fitting precision in hbMOLLI4 compared with MOLLI5(3 s)3, but prominently better than saturation recovery. Bland-Altman analysis on accuracy assessment revealed that hbMOLLI4 partially reduced the T1 underestimation of MOLLI5(3 s)3. In the human study, The T1 values of both methods were consistent (hbMOLLI4 vs. MOLLI5(3 s)3, slope = 1.14, R2 > 0.97), with equal reproducibility. The results supported that hybrid MOLLI produced comparable T1 mapping results in terms of accuracy, reproducibility, and heart-rate dependency, at the expense of slightly reduced precision. We concluded that the hybrid MOLLI sequence presents a competitive alternative to the MOLLI5(3 s)3 sequence when a speedy acquisition is required.

Deriving a robust deep-learning model for subcortical brain segmentation by using a large-scale database: Preprocessing, reproducibility, and accuracy of volume estimation.

Increasing the accuracy and reproducibility of subcortical brain segmentation is advantageous in various related clinical applications. In this study, we derived a segmentation method based on a convolutional neural network (i.e., U-Net) and a large-scale database consisting of 7039 brain T1-weighted MRI data samples. We evaluated the method by using experiments focused on three distinct topics, namely, the necessity of preprocessing steps, cross-institutional and longitudinal reproducibility, and volumetric accuracy. The optimized model, MX_RW-where "MX" is a mix of RW and nonuniform intensity normalization data and "RW" is raw data with basic preprocessing-did not require time-consuming preprocessing steps, such as nonuniform intensity normalization or image registration, for brain MRI before segmentation. Cross-institutional testing revealed that MX_RW (Dice similarity coefficient: 0.809, coefficient of variation: 4.6%, and Pearson's correlation coefficient: 0.979) exhibited a performance comparable with that of FreeSurfer (Dice similarity coefficient: 0.798, coefficient of variation: 5.6%, and Pearson's correlation coefficient: 0.973). The computation time per dataset of MX_RW was generally less than 5 s (even when tested without graphics processing units), which was notably faster than FreeSurfer. Thus, for time-restricted applications, MX_RW represents a competitive alternative to FreeSurfer.

Deep learning-based classification for lung opacities in chest x-ray radiographs through batch control and sensitivity regulation.

In this study, we implemented a system to classify lung opacities from frontal chest x-ray radiographs. We also proposed a training method to address the class imbalance problem presented in the dataset. We participated in the Radiological Society of America (RSNA) 2018 Pneumonia Detection Challenge and used the datasets provided by the RSNA for further research. Using convolutional neural networks, we implemented a training procedure termed batch control to manipulate the data distribution of positive and negative cases in each training batch. The batch control method regulated and stabilized the performance of the deep-learning models, allowing the adaptive sensitivity of the network models to meet the specific application. The convolutional neural network is practical for classifying lung opacities on chest x-ray radiographs. The batch control method is advantageous for sensitivity regulation and optimization for class-unbalanced datasets.

Microbial Interaction between Lactiplantibacillus plantarum and Saccharomyces cerevisiae: Transcriptome Level Mechanism of Cell-Cell Antagonism.

Lactiplantibacillus plantarum and Saccharomyces cerevisiae are frequently co-isolated in food, although playing different roles. This study aimed at investigating the microbial interaction between L. plantarum and S. cerevisiae, especially cell-cell direct interaction and their mechanism. Cell-cell and supernatant-cell coculture models were set up, with CFU counting, OD600 measurement, optical and atomic force microscopy performed to examine the growth and morphology of L. plantarum and S. cerevisiae cells. In cell-cell coculture model, L. plantarum cells inhibited S. cerevisiae growth (inhibition rate ~80%) with its own growth pattern unaffected. Cell-cell aggregation happened during coculture with surface roughness changed and partial S. cerevisiae cell lysis. Mature (24 h) L. plantarum cell-free culture supernatant showed inhibition (35%-75%) on S. cerevisiae growth independent of pH level, while supernatant from L. plantarum-S. cerevisiae coculture showed relatively stronger inhibition. Upon transcriptomics analysis, hypothesis on the mechanism of microbial interaction between L. plantarum and S. cerevisiae was demonstrated. When L. plantarum cell density reached threshold at 24 h, all genes in lamBDCA quorum sensing (QS) system was upregulated to potentially increase adhesion capability, leading to the aggregation to S. cerevisiae cell. The downregulation of whole basic physiological activity from DNA to RNA to protein, cell cycle, meiosis, and mitogen-activated protein kinase (MAPK) signaling pathways, as well as growth maintenance essential genes ari1, skg6, and kex2/gas1 might induce the decreased growth and proliferation rate and partial death of S. cerevisiae cells in coculture. IMPORTANCE L. plantarum and S. cerevisiae are frequently co-isolated in food, although playing different roles. The co-existence of L. plantarum and S. cerevisiae could result in variable effects, raising economic benefits and safety concerns in food industry. Previous research has reported the microbial interaction between L. plantarum and S. cerevisiae mainly rely on the signaling through extracellular metabolites. However, cell-cell aggregation has been observed with mechanism remain unknown. In the current study, the microbial interaction between L. plantarum and S. cerevisiae was investigated with emphasis on cell-cell direct interaction and further in-depth transcriptome level study showed the key role of lamBDCA quorum sensing system in L. plantarum. The results yield from this study demonstrated the antagonistic effect between L. plantarum and S. cerevisiae.

Byproduct formation during the biosynthesis of spinosyn A and evidence for an enzymatic interplay to prevent its formation.

Biosynthesis of spinosyn A in Saccharopolyspora spinosa involves a 1,4-dehydration followed by an intramolecular [4 + 2]-cycloaddition catalyzed by SpnM and SpnF, respectively. The cycloaddition also takes place in the absence of SpnF leading to questions regarding its mechanism of catalysis and biosynthetic role. Substrate analogs were prepared with an unactivated dienophile or an acyclic structure and found to be unreactive consistent with the importance of these features for cyclization. The SpnM-catalyzed dehydration reaction was also found to yield a byproduct corresponding to the C11 = C12 cis isomer of the SpnF substrate. This byproduct is stable both in the presence and absence of SpnF; however, relative production of the SpnM product and byproduct could be shifted in favor of the former by including SpnF or the dehydrogenase SpnJ in the reaction. This result suggests a potential interplay between the enzymes of spinosyn A biosynthesis that may help to improve the efficiency of the pathway.

Improving interobserver agreement and performance of deep learning models for segmenting acute ischemic stroke by combining DWI with optimized ADC thresholds.

OBJECTIVES: To examine the role of ADC threshold on agreement across observers and deep learning models (DLMs) plus segmentation performance of DLMs for acute ischemic stroke (AIS). METHODS: Twelve DLMs, which were trained on DWI-ADC-ADC combination from 76 patients with AIS using 6 different ADC thresholds with ground truth manually contoured by 2 observers, were tested by additional 67 patients in the same hospital and another 78 patients in another hospital. Agreement between observers and DLMs were evaluated by Bland-Altman plot and intraclass correlation coefficient (ICC). The similarity between ground truth (GT) defined by observers and between automatic segmentation performed by DLMs was evaluated by Dice similarity coefficient (DSC). Group comparison was performed using the Mann-Whitney U test. The relationship between the DSC and ADC threshold as well as AIS lesion size was evaluated by linear regression analysis. A p < .05 was considered statistically significant. RESULTS: Excellent interobserver agreement and intraobserver repeatability in the manual segmentation (all ICC > 0.98, p < .001) were achieved. The 95% limit of agreement was reduced from 11.23 cm2 for GT on DWI to 0.59 cm2 for prediction at an ADC threshold of 0.6 × 10-3 mm2/s combined with DWI. The segmentation performance of DLMs was improved with an overall DSC from 0.738 ± 0.214 on DWI to 0.971 ± 0.021 on an ADC threshold of 0.6 × 10-3 mm2/s combined with DWI. CONCLUSIONS: Combining an ADC threshold of 0.6 × 10-3 mm2/s with DWI reduces interobserver and inter-DLM difference and achieves best segmentation performance of AIS lesions using DLMs. KEY POINTS: • Higher Dice similarity coefficient (DSC) in predicting acute ischemic stroke lesions was achieved by ADC thresholds combined with DWI than by DWI alone (all p < .05). • DSC had a negative association with the ADC threshold in most sizes, both hospitals, and both observers (most p < .05) and a positive association with the stroke size in all ADC thresholds, both hospitals, and both observers (all p < .001). • An ADC threshold of 0.6 × 10-3 mm2/s eliminated the difference of DSC at any stroke size between observers or between hospitals (p = .07 to > .99).

Improving diagnosing performance for malignant parotid gland tumors using machine learning with multifeatures based on diffusion-weighted magnetic resonance imaging.

In this study, the performance of machine learning in classifying parotid gland tumors based on diffusion-related features obtained from the parotid gland tumor, the peritumor parotid gland, and the contralateral parotid gland was evaluated. Seventy-eight patients participated in this study and underwent magnetic resonance diffusion-weighted imaging. Three regions of interest, including the parotid gland tumor, the peritumor parotid gland, and the contralateral parotid gland, were manually contoured for 92 tumors, including 20 malignant tumors (MTs), 42 Warthin tumors (WTs), and 30 pleomorphic adenomas (PMAs). We recorded multiple apparent diffusion coefficient (ADC) features and applied a machine-learning method with the features to classify the three types of tumors. With only mean ADC of tumors, the area under the curve of the classification model was 0.63, 0.85, and 0.87 for MTs, WTs, and PMAs, respectively. The performance metrics were improved to 0.81, 0.89, and 0.92, respectively, with multiple features. Apart from the ADC features of parotid gland tumor, the features of the peritumor and contralateral parotid glands proved advantageous for tumor classification. Combining machine learning and multiple features provides excellent discrimination of tumor types and can be a practical tool in the clinical diagnosis of parotid gland tumors.

Pseudomonas aeruginosa: A typical biofilm forming pathogen and an emerging but underestimated pathogen in food processing.

Pseudomonas aeruginosa (P. aeruginosa) is a notorious gram-negative pathogenic microorganism, because of several virulence factors, biofilm forming capability, as well as antimicrobial resistance. In addition, the appearance of antibiotic-resistant strains resulting from the misuse and overuse of antibiotics increases morbidity and mortality in immunocompromised patients. However, it has been underestimated as a foodborne pathogen in various food groups for instance water, milk, meat, fruits, and vegetables. Chemical preservatives that are commonly used to suppress the growth of food source microorganisms can cause problems with food safety. For these reasons, finding effective, healthy safer, and natural alternative antimicrobial agents used in food processing is extremely important. In this review, our ultimate goal is to cover recent advances in food safety related to P. aeruginosa including antimicrobial resistance, major virulence factors, and prevention measures. It is worth noting that food spoilage caused by P. aeruginosa should arouse wide concerns of consumers and food supervision department.

Preclinical Studies of OBI-999: A Novel Globo H-Targeting Antibody-Drug Conjugate.

Globo H (GH), a hexasaccharide, is expressed at low levels in normal tissues but is highly expressed in multiple cancer types, rendering it a promising target for cancer immunotherapy. OBI-999, a novel antibody-drug conjugate, is derived from a conjugation of a GH-specific mAb with a monomethyl auristatin E (MMAE) payload through a site-specific ThioBridge and a cleavable linker. OBI-999 high homogeneity with a drug-to-antibody ratio of 4 (>95%) was achieved using ThioBridge. OBI-999 displayed GH-dependent cellular internalization and trafficked to endosome and lysosome within 1 and 5 hours, respectively. Furthermore, OBI-999 showed low nanomolar cytotoxicity in the assay with high GH expression on tumor cells and exhibited a bystander killing effect on tumor cells with minimal GH expression. Tissue distribution indicated that OBI-999 and free MMAE gradually accumulated in the tumor, reaching maximum level at 168 hours after treatment, whereas OBI-999 and free MMAE decreased quickly at 4 hours after treatment in normal organs. Maximum MMAE level in the tumor was 16-fold higher than in serum, suggesting that OBI-999 is stable during circulation and MMAE is selectively released in the tumor. Excellent tumor growth inhibition of OBI-999 was demonstrated in breast, gastric, and pancreatic cancer xenograft or lung patient-derived xenograft models in a dose-dependent manner. The highest nonseverely toxic dose in cynomolgus monkeys is 10 mg/kg determined by a 3-week repeated-dose toxicology study demonstrating an acceptable safety margin. Taken together, these results support further clinical development of OBI-999, which is currently in a phase I/II clinical study in multiple solid tumors (NCT04084366). OBI-999, the first GH-targeting ADC, displayed excellent tumor inhibition in animal models across multiple cancer types, including breast, gastric, pancreatic, and lung cancers, warranting further investigation in the treatment of solid tumors.

Categorizing SHR and WKY rats by chi2 algorithm and decision tree.

Classifying mental disorder is a big issue in psychology in recent years. This article focuses on offering a relation between decision tree and encoding of fMRI that can simplify the analysis of different mental disorders and has a high ROC over 0.9. Here we encode fMRI information to the power-law distribution with integer elements by the graph theory in which the network is characterized by degrees that measure the number of effective links exceeding the threshold of Pearson correlation among voxels. When the degrees are ranked from low to high, the network equation can be fit by the power-law distribution. Here we use the mentally disordered SHR and WKY rats as samples and employ decision tree from chi2 algorithm to classify different states of mental disorder. This method not only provides the decision tree and encoding, but also enables the construction of a transformation matrix that is capable of connecting different metal disorders. Although the latter attempt is still in its fancy, it may have a contribution to unraveling the mystery of psychological processes.

Classification of parotid gland tumors by using multimodal MRI and deep learning.

Various MRI sequences have shown their potential to discriminate parotid gland tumors, including but not limited to T2 -weighted, postcontrast T1 -weighted, and diffusion-weighted images. In this study, we present a fully automatic system for the diagnosis of parotid gland tumors by using deep learning methods trained on multimodal MRI images. We used a two-dimensional convolution neural network, U-Net, to segment and classify parotid gland tumors. The U-Net model was trained with transfer learning, and a specific design of the batch distribution optimized the model accuracy. We also selected five combinations of MRI contrasts as the input data of the neural network and compared the classification accuracy of parotid gland tumors. The results indicated that the deep learning model with diffusion-related parameters performed better than those with structural MR images. The performance results (n = 85) of the diffusion-based model were as follows: accuracy of 0.81, 0.76, and 0.71, sensitivity of 0.83, 0.63, and 0.33, and specificity of 0.80, 0.84, and 0.87 for Warthin tumors, pleomorphic adenomas, and malignant tumors, respectively. Combining diffusion-weighted and contrast-enhanced T1 -weighted images did not improve the prediction accuracy. In summary, the proposed deep learning model could classify Warthin tumor and pleomorphic adenoma tumor but not malignant tumor.

"One-step" characterization platform for pathogenic genetics of Staphylococcus aureus.

Staphylococcus aureus (S. aureus) is an important human pathogen causing a variety of life-threatening diseases. In recent years, the health problem caused by S. aureus contaminated food has become a global health problem. S. aureus can express various pathogenic factors, mainly used for adhesion, colonization, invasion and infection of the host. Therefore, rapid and accurate detection of virulence genes in S. aureus is necessary to prevent outbreaks caused by this pathogen. PCR is a useful tool for rapid detection of foodborne pathogens. The objective of this study was to detect the presence of major toxin genes in S. aureus, including sea, seb, sec, see, pvl and tsst, by using a PCR plate. Of the 13 strains tested, 12 (92.3%) were found to be positive for one or more toxin genes. This study realized the one-step detection of main toxin factors in S. aureus.

Study on the Viable but Non-culturable (VBNC) State Formation of Staphylococcus aureus and Its Control in Food System.

A Viable but non-culturable (VBNC) state is a bacterial survival strategy under reverse conditions. It poses a significant challenge for public health and food safety. In this study, the effect of external environmental conditions including acid, nutrition, and salt concentrations on the formation of S. aureus VBNC states at low temperatures were investigated. Different acidity and nutritional conditions were then applied to food products to control the VBNC state formation. Four different concentration levels of each factor (acid, nutrition, and salt) were selected in a total of 16 experimental groups. Nutrition showed the highest influence on the VBNC state formation S. aureus, followed by acid and salt. The addition of 1% acetic acid could directly kill S. aureus cells and inhibit the formation of the VBNC state with a nutrition concentration of 25, 50, and 100%. A propidium monoazide-polymerase chain reaction (PMA-PCR) assay was applied and considered as a rapid and sensitive method to detect S. aureus in VBNC state with the detection limit of 104 CFU/mL.

Effect of Environmental Conditions on the Formation of the Viable but Nonculturable State of Pediococcus acidilactici BM-PA17927 and Its Control and Detection in Food System.

Objective: This study aimed to investigate the effect of environmental conditions including nutrient content, acetic acid concentration, salt concentration, and temperature on the formation of viable but nonculturable (VBNC) state of Pediococcus acidilactici, as well as its control and detection in food system. Methods: Representing various environmental conditions in different food systems, 16 induction groups were designed for the formation of VBNC state of P. acidilactici. Traditional plate counting was applied to measure the culturable cell numbers, and Live/Dead Bacterial Viability Kit combined with fluorescent microscopy was used to identify viable cells numbers. The inhibition of bacterial growth and VBNC state formation by adjusting the environmental conditions were investigated, and the clearance effect of VBNC cells in crystal cake system was studied. In addition, a propidium monoazide-polymerase chain reaction (PMA-PCR) assay was applied to detect the VBNC P. acidilactici cells in crystal cake food system. Results: Among the environmental conditions included in this study, acetic acid concentration had the greatest effect on the formation of VBNC state of P. acidilactici, followed by nutritional conditions and salt concentration. Reducing nutrients in the environment and treating with 1.0% acetic acid can inhibit P. acidilactici from entering the VBNC state. In the crystal cake system, the growth of P. acidilactici and the formation of VBNC state can be inhibited by adding 1.0% acetic acid and storing at -20°C. In crystal cake system, the PMA-PCR assay can be used to detect VBNC P. acidilactici cells at a concentration higher than 104 cells/ml. Conclusion: The VBNC state of P. acidilactici can be influenced by the changing of environmental conditions, and PMA-PCR assay can be applied in food system for the detection of VBNC P. acidilactici cells.

Leveraging Multimodal Deep Learning Architecture with Retina Lesion Information to Detect Diabetic Retinopathy.

Purpose: To improve disease severity classification from fundus images using a hybrid architecture with symptom awareness for diabetic retinopathy (DR). Methods: We used 26,699 fundus images of 17,834 diabetic patients from three Taiwanese hospitals collected in 2007 to 2018 for DR severity classification. Thirty-seven ophthalmologists verified the images using lesion annotation and severity classification as the ground truth. Two deep learning fusion architectures were proposed: late fusion, which combines lesion and severity classification models in parallel using a postprocessing procedure, and two-stage early fusion, which combines lesion detection and classification models sequentially and mimics the decision-making process of ophthalmologists. Messidor-2 was used with 1748 images to evaluate and benchmark the performance of the architecture. The primary evaluation metrics were classification accuracy, weighted κ statistic, and area under the receiver operating characteristic curve (AUC). Results: For hospital data, a hybrid architecture achieved a good detection rate, with accuracy and weighted κ of 84.29% and 84.01%, respectively, for five-class DR grading. It also classified the images of early stage DR more accurately than conventional algorithms. The Messidor-2 model achieved an AUC of 97.09% in referral DR detection compared to AUC of 85% to 99% for state-of-the-art algorithms that learned from a larger database. Conclusions: Our hybrid architectures strengthened and extracted characteristics from DR images, while improving the performance of DR grading, thereby increasing the robustness and confidence of the architectures for general use. Translational Relevance: The proposed fusion architectures can enable faster and more accurate diagnosis of various DR pathologies than that obtained in current manual clinical practice.