Search

Small extracellular vesicles (sEVs) are cell-released vesicles ranging from 30-150nm in size. They have garnered increasing attention because of their potential for both the diagnosis and treatment of disease. The diversity of sEVs derives from their biological composition and cargo content. Currently, the isolation of sEV subpopulations is primarily based on bio-physical and affinity-based approaches. Since a standardized definition for sEV subpopulations is yet to be fully established, it is important to further investigate the correlation between the biomolecular composition of sEVs and their physical properties. In this study, we employed a platform combining single-vesicle surface-enhanced Raman spectroscopy (SERS) and machine learning to examine individual sEVs isolated by size-exclusion chromatography (SEC). The biomolecular composition of each vesicle examined was reflected by its corresponding SERS spectral features (biomolecular "fingerprints"), with their roots in the composition of their collective Raman-active bonds. Origins of the SERS spectral features were validated through a comparative analysis between SERS and mass spectrometry (MS). SERS fingerprinting of individual vesicles was effective in overcoming the challenges posed by EV population averaging, allowing for the possibility of analyzing the variations in biomolecular composition between the vesicles of similar and/or different sizes. Using this approach, we uncovered that each of the size-based fractions of sEVs contained particles with predominantly similar SERS spectral features. Indeed, more than 84% of the vesicles residing within a particular group were clearly distinguishable from that of the other EV sub-populations, despite some spectral variations within each sub-population. Our results suggest the possibility that size-based EV fractionation methods produce samples where similarly eluted sEVs are correlated with their respective biochemical contents, as reflected by their SERS spectra. Our findings therefore highlight the possibility that the biogenesis and respective biological functionalities of the various sEV fractions may be inherently different.

Next basket recommendation is a critical task in market basket data analysis. It is particularly important in grocery shopping, where grocery lists are an essential part of shopping habits of many customers. In this work, we first present a new grocery Recommender System available on the MyGroceryTour platform. Our online system uses different traditional machine learning (ML) and deep learning (DL) algorithms, and provides recommendations to users in a real-time manner. It aims to help Canadian customers create their personalized intelligent weekly grocery lists based on their individual purchase histories, weekly specials offered in local stores, and product cost and availability information. We perform clustering analysis to partition given customer profiles into four non-overlapping clusters according to their grocery shopping habits. Then, we conduct computational experiments to compare several traditional ML algorithms and our new DL algorithm based on the use of a gated recurrent unit (GRU)-based recurrent neural network (RNN) architecture. Our DL algorithm can be viewed as an extension of DREAM (Dynamic REcurrent bAsket Model) adapted to multi-class (i.e. multi-store) classification, since a given user can purchase recommended products in different grocery stores in which these products are available. Among traditional ML algorithms, the highest average F-score of 0.516 for the considered data set of 831 customers was obtained using Random Forest, whereas our proposed DL algorithm yielded the average F-score of 0.559 for this data set. The main advantage of the presented Recommender System is that our intelligent recommendation is personalized, since a separate traditional ML or DL model is built for each customer considered. Such a personalized approach allows us to outperform the prediction results provided by general state-of-the-art DL models.

Next basket recommendation is a critical task in market basket data analysis. It is particularly important in grocery shopping, where grocery lists are an essential part of shopping habits of many customers. In this work, we first present a new grocery Recommender System available on the MyGroceryTour platform. Our online system uses different traditional machine learning (ML) and deep learning (DL) algorithms, and provides recommendations to users in a real-time manner. It aims to help Canadian customers create their personalized intelligent weekly grocery lists based on their individual purchase histories, weekly specials offered in local stores, and product cost and availability information. We perform clustering analysis to partition given customer profiles into four non-overlapping clusters according to their grocery shopping habits. Then, we conduct computational experiments to compare several traditional ML algorithms and our new DL algorithm based on the use of a gated recurrent unit (GRU)-based recurrent neural network (RNN) architecture. Our DL algorithm can be viewed as an extension of DREAM (Dynamic REcurrent bAsket Model) adapted to multi-class (i.e. multi-store) classification, since a given user can purchase recommended products in different grocery stores in which these products are available. Among traditional ML algorithms, the highest average F-score of 0.516 for the considered data set of 831 customers was obtained using Random Forest, whereas our proposed DL algorithm yielded the average F-score of 0.559 for this data set. The main advantage of the presented Recommender System is that our intelligent recommendation is personalized, since a separate traditional ML or DL model is built for each customer considered. Such a personalized approach allows us to outperform the prediction results provided by general state-of-the-art DL models.

PURPOSE:The five point Deauville (D) scale is widely used to assess interim PET metabolic response to chemotherapy in Hodgkin lymphoma (HL) patients. An International Validation Study reported good concordance among reviewers in ABVD treated advanced stage HL patients for the binary discrimination between score D1,2,3 and score D4,5. Inter-reader reliability of the whole scale is not well characterised. METHODS:Five international expert readers scored 100 interim PET/CT scans from paediatric HL patients. Scans were acquired in 51 European hospitals after two courses of OEPA chemotherapy (according to the EuroNet-PHL-C1 study). Images were interpreted in direct comparison with staging PET/CTs. RESULTS:The probability that two random readers concord on the five point D score of a random case is only 42% (global kappa = 0.24). Aggregating to a three point scale D1,2 vs. D3 vs. D4,5 improves concordance to 60% (kappa = 0.34). Concordance if one of two readers assigns a given score is 70% for score D1,2 only 36% for score D3 and 64% for D4,5. Concordance for the binary decisions D1,2 vs. D3,4,5 is 67% and 86% for D1,2,3 vs D4,5 (kappa = 0.36 resp. 0.56). If one reader assigns D1,2,3 concordance probability is 92%, but only 64% if D4,5 is called. Discrepancies occur mainly in mediastinum, neck and skeleton. CONCLUSION:Inter-reader reliability of the five point D-scale is poor in this interobserver analysis of paediatric patients who underwent OEPA. Inter-reader variability is maximal in cases assigned to D2 or D3. The binary distinction D1,2,3 versus D4,5 is the most reliable criterion for clinical decision making.

Analytic approximate solutions using Optimal Homotopy Perturbation Method (OHPM) are given for steady boundary layer flow over a nonlinearly stretching wall in presence of partial slip at the boundary. The governing equations are reduced to nonlinear ordinary differential equation by means of similarity transformations. Some examples are considered and the effects of different parameters are shown. OHPM is a very efficient procedure, ensuring a very rapid convergence of the solutions after only two iterations.

Gliomas are common brain tumours, but obtaining tissue for definitive diagnosis can be difficult. There is, therefore, interest in the use of non-invasive methods to diagnose and grade the disease. Although positron emission tomography (PET) with 18F-fluorethyltyrosine (18F-FET) can be used to differentiate between low-grade (LGG) and high-grade (HGG) gliomas, the optimal parameters to measure and their cut-points have yet to be established. We therefore assessed the value of single and dual time-point acquisition of 18F-FET PET parameters to differentiate between primary LGGs (n = 22) and HGGs (n = 24). PET examination was considered positive for glioma if the metabolic activity was 1.6-times higher than that of background (contralateral) brain, and maximum tissue-brain ratios (TBRmax) were calculated 10 and 60 min after isotope administration with their sums and differences calculated from individual time-point values. Using a threshold-based method, the overall sensitivity of PET was 97%. Several analysed parameters were significantly different between LGGs and HGGs. However, in a receiver operating characteristics analysis, TBR sum had the best diagnostic accuracy of 87% and sensitivity, specificity, and positive and negative predictive values of 100%, 72.7%, 80%, and 100%, respectively. 18F-FET PET is valuable for the non-invasive determination of glioma grade, especially when dual time-point metrics are used. TBR sum shows the greatest accuracy, sensitivity, and negative predictive value for tumour grade differentiation and is a simple method to implement. However, the cut-off may differ between institutions and calibration strategies would be useful.

Wolfram syndrome (WFS) is inherited as an autosomal recessive disease with main clinical features of diabetes mellitus, optic atrophy, diabetes insipidus and deafness. However, various neurological defects may also be detected. The aim of this study was to evaluate aspects of brain structure and function using PET-CT (positron emission tomography and computed tomography) and MRI (magnetic resonance imaging) in pediatric patients with WFS. Regional changes in brain glucose metabolism were measured using standardized uptake values (SUVs) based on images of (18F) fluorodeoxyglucose (FDG) uptake in 7 WFS patients aged 10.1-16.0 years (mean 12.9±2.4) and in 20 healthy children aged 3-17.9 years (mean 12.8±4.1). In all patients the diagnosis of WFS was confirmed by DNA sequencing of the WFS1 gene. Hierarchical clustering showed remarkable similarities of glucose uptake patterns among WFS patients and their differences from the control group. SUV data were subsequently standardized for age groups 13 years old to account for developmental differences. Reduced SUVs in WFS patients as compared to the control group for the bilateral brain regions such as occipital lobe (-1.24±1.20 vs. -0.13±1.05; p = 0.028) and cerebellum (-1.11±0.69 vs. -0.204±1.00; p = 0.036) were observed and the same tendency for cingulate (-1.13±1.05 vs. -0.15±1.12; p = 0.056), temporal lobe (-1.10±0.98 vs. -0.15±1.10; p = 0.057), parietal lobe (-1.06±1.20 vs. -0.08±1.08; p = 0.058), central region (-1.01±1.04 vs. -0.09±1.06; p = 0.060), basal ganglia (-1.05±0.74 vs. -0.20±1.07; p = 0.066) and mesial temporal lobe (-1.06±0.82 vs. -0.26±1.08; p = 0.087) was also noticed. After adjusting for multiple hypothesis testing, the differences in glucose uptake were non-significant. For the first time, regional differences in brain glucose metabolism among patients with WFS were shown using PET-CT imaging.