Classification Tool Research Articles

Adversarial training for signal modulation classification based on Ulam stability theory

The modulation classification of radar, communication, and other radiation source signals under complex electromagnetic conditions is an important task in the field of military electronic countermeasures. Deep learning is an important tool for automatic modulation classification (AMC) tasks. However, AMC models based on deep neural networks are easily attacked by adversarial methods such as gradient attacks, and even under small perturbations, the model may exhibit significant differences in prediction. To address adversarial attacks, we proposed a Ulam stability adversarial training method, which could improve the robust accuracy and the natural accuracy of the adversarial training model. We introduced a class of Jensen type Ulam stability theorem, which transforms the Jensen condition into a special class of stability preserving regularization terms for adversarial training models, enhancing the model's adversarial robustness. More importantly, our method provided a framework for improving the stability of adversarial training, which can be combined with existing adversarial training methods to effectively enhance their adversarial defense capabilities. In the experimental section, we compared four types of adversarial training methods on three types of modulation recognition datasets, and the results showed that Ulam adversarial training improved robust accuracy by an average of 4.8% and natural accuracy by an average of 2.2% on three datasets. The difference between the best and final accuracies of the models decreased by an average of 4.5%.

Inequality relations for NMR-based polymer homoblock analysis and extended application: Reanalysis of historical data on alginates, chitosans, homogalacturonans, and galactomannans

There has been a long-standing bottleneck in the quantitative analysis of the frequencies of homoblock polyads beyond triads using 1H and 13C NMR for linear polysaccharides, primarily because monosaccharides within a long homoblock share similar chemical environments due to identical neighboring units, resulting in indistinct NMR peaks. In this study, through rigorous mathematical induction, inequality relations were established that enabled the calculation of frequency ranges of homoblock polyads from historically reported NMR-derived frequency values of diads and/or triads of alginates, chitosans, homogalacturonans, and galactomannans. The calculated homoblock frequency ranges were then applied to evaluate three chain growth statistical models, including the Bernoulli chain, first-order Markov chain, and second-order Markov chain, for predicting homoblock frequencies in these polysaccharides. Furthermore, based on the mathematically derived inequality relations, a novel 2D array was constructed, enabling the graphical visualization of homoblock features in polysaccharides. It was demonstrated, as a proof of concept, that the novel 2D array, along with a 1D code generated from it, could serve as an effective feature engineering tool for polymer classification using machine learning algorithms.

Validation of the self-report classification tool to determine language dominance in Kannada-English bilinguals

Background In recent times, the efforts to profile the language characteristics of bilinguals have been extended from mere documentation of proficiency in each language to the determination of language dominance that captures both proficiency and usage (i.e., frequency & contexts) of each language. In multilingual countries, individuals are immersed in various languages in different contexts. With the broader intention to improve the linguistic profiling of bilinguals in countries with similar characteristics, we aimed to adapt and validate the Self-report classification tool in Indian Kannada-English bilinguals. Methods A group of 88 adult Kannada-English bilingual participants self-rated their language proficiency. We measured their language dominance with the adapted tool. Finally, to objectively measure their language abilities, we used the short version of the Bilingual Aphasia Test. Results Discriminant analysis of the ratings showed that the self-report classification tool accurately classified our participants into three groups based on language dominance. Both the self-rating and the objective measure of language proficiency supported the (dominance) classifications by the adapted tool. Conclusion Findings show that the adapted self-report classification tool is valid for determining language dominance in Kannada-English bilinguals. Further, the current study shows that this tool is adaptable to novel bilingual language dyads.

Designing A Tool for Small Rural Community Classification Based on Traditional Belief and Its Implication for Future Tourism Development: A Case Study of North-Eastern Community

This research focused on the cultural and economic significance of traditional ghost beliefs in rural communities. The primary objectives of this study are threefold: 1) the development and contribution of tools and processes for classifying communities based on their ghost beliefs, 2) the testing and validation of these classification tools, and 3) the proposition of strategies aimed at leveraging ghost beliefs for local community space development to foster tourism. This study engaged 240 local residents in community classification testing, sought insights from three key informants, and collaborated with three academic experts for validation. A combination of quantitative and qualitative surveys was employed to rigorously test and refine the community classification tools. The findings demonstrated that the developed tools effectively classified rural communities, thereby increasing residents' awareness of traditional beliefs as valuable community assets. For more effectiveness, two innovative strategies-”Watch local” and “Be local”-are introduced to the cultural and economic development in rural communities. The tools expedited community assessment and the development of custom tourism formats for diverse tourist types. These contributed to coherent tourism plans that met community needs, thereby enhancing tourist destinations. Data analysis refined the tourism formats to efficaciously meet community and tourist expectations. This research furnished vital insights into capitalizing on ghost beliefs for the economic advancement of rural communities via sustainable tourism practices.

ScSNViz: a user-friendly toolset for visualization and analysis of Cell-Specific Expressed SNVs.

Understanding genetic variation at the single-cell level is crucial for insights into cellular heterogeneity, clonal evolution, and gene expression regulation, but there is a scarcity of tools for visualizing and analyzing cell-level genetic variants. We introduce scSNViz, a comprehensive R-based toolset for visualization and analysis of cell-specific expressed Single Nucleotide Variants (sceSNVs) within cell-barcoded single-cell RNA-sequencing (scRNA-seq) data. ScSNViz offers 3D sceSNV visualization capabilities for dimensionally reduced scRNA-seq gene expression data, compatibility with popular scRNA-seq processing tools like Seurat, cell-type classification tools such as SingleR and scType, and trajectory inference computation using Slingshot. Furthermore, scSNViz conducts estimation, summary, and graphical representation of statistical metrics pertaining to sceSNVs distribution and expression across individual cells. It also provides support for the analysis of individual sceSNVs as well as sets comprising multiple expressed sceSNVs of interest. ScSNViz is implemented as user-friendly R-scripts, freely available on https://horvathlab.github.io/NGS/scSNViz , supported by help utilities, and requiring no specialized bioinformatics skills for use.

Patients’ voices matter: Translating pretreatment patient-reported outcomes into high and low health-related quality of life provides a new clinical aspect on prognosis and survival in patients with previously untreated diffuse large B‐cell lymphoma.

e23165 Background: Patient-reported health-related quality of life (HRQoL) measures collected prior to treatment initiation have been shown to be prognostic of disease outcomes and survival in patients with previously untreated diffuse large B-cell lymphoma (DLBCL), including progression-free survival (PFS) and overall survival (OS), However, these HRQoL measures are usually expressed as numbers from a scale (e.g. from 0 to 100), which limit their ease of clinical utility compared to binary scales (i.e. high vs. low). Using HRQoL measures collected from the phase 3 study GOYA, we aim to convert these measures into binary scales by identifying optimal cutoffs and demonstrate their clinical utility in differentiating patient risks in DLBCL. Methods: HRQoL data from the physical functioning (PF2), global health status (QL2), and fatigue (FA) scales from the EORTC QLQ-C30 questionnaire and the lymphoma subscale (LYMS) from the FACT-LYM questionnaire were used in this analysis. The optimal cutoff points were identified using statistical metrics that evaluate the prognostic value and goodness-of-fit for each assessed cutoff point. The top performing binary HRQoL scale was further assessed in its clinical utility as an addition to the Internal Prognostic Index (IPI), a standard risk classification tool in DLBCL based on 5 binary measures of patient baseline characteristics, to evaluate whether adding the HRQoL scale would potentially improve the risk classification. Results: As shown in the table, while all binary scales can clearly differentiate patient overall survival (demonstrated by the hazard ratios), PF2 numerically exhibited the best performance, and was further assessed as an addition to IPI. After adding PF2 to the base OS model containing only IPI, we observed improvements in model performance metrics, including the concordance index and the goodness-of-fit. In addition, among patients classified as high risk by IPI (n = 185, 3-yr OS: 65% [95% CI 68-72%]), those with low HRQoL (n = 102, 3-yr OS: 56% [95% CI 47-67%]) had a substantially worse OS, suggesting that further refinement of patient-risk defined by IPI may be feasible with HRQoL measures. Conclusions: Our work highlights the potential of the binary HRQoL scales to provide additional prognostic value to IPI, and suggests that patient reported outcomes may be useful in further refining patient risk classification in addition to IPI in untreated DLBCL. [Table: see text]

Unsupervised Classification Identifies Warm, Fresh, and Dense Regimes of the Antarctic Margins

Abstract The ocean surrounding Antarctica, also known as the Antarctic margins, is characterized by complex and heterogeneous process interactions, which have major impacts on the global climate. A common way to understand changes in the Antarctic margins is to categorize regions into similar “regimes,” thereby guiding process-based studies and observational analyses. However, this categorization is traditionally largely subjective and based on temperature, density, and bathymetric criteria that are bespoke to the dataset being analyzed. In this work, we introduce a method to classify Antarctic shelf regimes using unsupervised learning. We apply Gaussian mixture modeling to the across-shelf temperature and salinity properties along the Antarctic margins from a high-resolution ocean model, ACCESS-OM2-01. Three clusters are found to be optimum based on the Bayesian information criterion and an assessment of regime properties. The three clusters correspond to the fresh, dense, and warm regimes identified canonically via subjective approaches. Our analysis allows us to track changes to these regimes in a future projection of the ACCESS-OM2-01 model. We identify the future collapse of dense water formation, and the merging of dense and fresh shelf regions into a single fresh regime that covers the entirety of the Antarctic shelf except for the West Antarctic. Our assessment of these clusters indicates that the Antarctic margins may shift into a two-regime system in the future, consisting only of a strengthening warm shelf in the West Antarctic and a fresh shelf regime everywhere else. Significance Statement The Antarctic margins are characterized by complex interactions of surface and ocean processes, producing distinct regions or “regimes.” Understanding where these regimes are and their future state is critical to understanding climate change. Based on a subjective assessment of ocean conditions, past research has identified fresh, dense, and warm regimes in the Antarctic margins. In this work, we use an unsupervised classification tool, Gaussian mixture modeling, to objectively identify the location of regimes around the Antarctic margins. Our method detects three regimes in an ocean model, which match the location of subjectively identified fresh, dense, and warm regimes, and indicates a future shrinking of the dense regime. Our method is adaptable to multiple datasets, enabling us to identify trends and processes in the Antarctic margins.

SIMS: A deep-learning label transfer tool for single-cell RNA sequencing analysis

Cell atlases serve as vital references for automating cell labeling in new samples, yet existing classification algorithms struggle with accuracy. Here we introduce SIMS (scalable, interpretable machine learning for single cell), a low-code data-efficient pipeline for single-cell RNA classification. We benchmark SIMS against datasets from different tissues and species. We demonstrate SIMS's efficacy in classifying cells in the brain, achieving high accuracy even with small training sets (<3,500 cells) and across different samples. SIMS accurately predicts neuronal subtypes in the developing brain, shedding light on genetic changes during neuronal differentiation and postmitotic fate refinement. Finally, we apply SIMS to single-cell RNA datasets of cortical organoids to predict cell identities and uncover genetic variations between cell lines. SIMS identifies cell-line differences and misannotated cell lineages in human cortical organoids derived from different pluripotent stem cell lines. Altogether, we show that SIMS is a versatile and robust tool for cell-type classification from single-cell datasets.

Classification of gram-positive and gram-negative bacteria using Few-shot learning algorithm

Our research introduces a novel approach to classify bacteria as Gram-positive or Gram-negative using few-shot learning. We employ deep neural networks, specifically Prototypical Networks, to learn distinctive features from bacterial images, enabling accurate classification even with limited data. Experimental results on diverse datasets demonstrate the model's effectiveness and potential for real-world applications in microbiology and healthcare. We also address interpretability, ethics, and data privacy, making it a valuable tool for bacterial classification and diagnostics.

Automatic prediction of isocitrate dehydrogenase mutation status of low-grade gliomas using radiomics and domain knowledge inspired features in magnetic resonance imaging

Aim: Most common and most deadly primary central nervous tumors, glial tumors harbor many heterogeneous clones of cells. Noninvasive determination of the genomic profiles of these tumors would have important implications regarding the classification, management, and prognostication of these tumors. Isocitrate dehydrogenase mutation is a key genomic signature that can downgrade the expected dismal course of these tumors. In this study we aimed to build a performant prediction model which can determine the Isocitrate Dehydrogenase (IDH) mutation status of glial tumors, using radiomics and leveraging automatic computation of domain knowledge-inspired features. Methods: Radiomics methods based on high throughput feature extraction and application of data science principles to these extracted features are promising tools for the noninvasive classification of lesions. Domain knowledge-inspired features besides radiomics features can contribute positively to the performance of the models. Some efforts particularly a joint approach to standardize the magnetic resonance imaging (MRI), reporting of glial tumors are mainstay for domain knowledge-inspired features. However, this requires active involvement and reporting of the radiologist which hampers automatization efforts. Additionally, this feature set evaluates a small subset of all possible signal and spatial-based computations. In this study, we combined domain knowledge-inspired features with radiomics features along with a multiparametric multihabitat comprehensive lesion description strategy. Results: Our best model which consisted of a combination of radiomics, and radiologist knowledge-inspired features reached a 0.93 f1 score (standard deviation (SD): 0.03), 0.93 accuracy (SD:0.03), and 0.98 area under curve (AUC), (SD:0.02). Conclusion: The multiparametric and multiregional approach employed in this study coupled with the integration of both radiomics and domain knowledge-inspired features resulted in a high-performance model emphasizing the contribution of each strategy to the outcome.

The potential role of RNA sequencing in diagnosing unexplained insensitivity to conventional chemotherapy in pediatric patients with B-cell acute lymphoblastic leukemia

Pediatric B-cell acute lymphoblastic leukemia (B-ALL) is a highly heterogeneous disease. According to large-scale RNA sequencing (RNA-seq) data, B-ALL patients can be divided into more than 10 subgroups. However, many genomic defects associated with resistance mechanisms have not yet been identified. As an individual clinical tool for molecular diagnostic risk classification, RNA-seq and gene expression pattern-based therapy could be potential upcoming strategies. In this study, we retrospectively analyzed the RNA-seq gene expression profiles of 45 children whose molecular diagnostic classifications were inconsistent with the response to chemotherapy. The relationship between the transcriptome and chemotherapy response was analyzed. Fusion gene identification was conducted for the included patients who did not have known high-risk associated fusion genes or gene mutations. The most frequently detected fusion gene pair in the high-risk group was the DHRSX duplication, which is a novel finding. Fusions involving ABL1, LMNB2, NFATC1, PAX5, and TTYH3 at onset were more frequently detected in the high-risk group, while fusions involving LFNG, TTYH3, and NFATC1 were frequently detected in the relapse group. According to the pathways involved, the underlying drug resistance mechanism is related to DNA methylation, autophagy, and protein metabolism. Overall, the implementation of an RNA-seq diagnostic system will identify activated markers associated with chemotherapy response, and guide future treatment adjustments.

Classification of rare earths mineral resources using the Jequié/BA complex rock weathering chemical index

The most common economic mineral deposits of Rare Earths are associated to magmatic calc-alkaline to alkaline rocks and their products. This work presents data from drilling carried out in rocks of the Jequié/BA Complex, mineralized in rare earths, which were subjected to an intense weathering process, similar to what occurred in the Ion Adsorption Clays deposits in southern China. The interpretation of the survey data allowed the classification of rare earths mineral resources with similar, sometimes higher, levels than those of Chinese deposits, contained in the lateritic regolithic profile studied. The chemical index of rock alteration (CIA) was adopted as a classification tool, horizontally separating the mineralized weather profile into six zones. The data indicate that there was supergenic enrichment REO in the lower portions of the profile and sometimes in the upper zones. The average levels ranged between 0.09% and 0.22% REO, with maximum values of 1.6%. A model section representing the deposit of rare earths minerals of supergenic origin in rocks of the Jequié/BA Complex was developed.

Efficient Gene Expression Data Analysis using ES-DBN For Microarray Cancer Data Classification

INTRODUCTION: DNA microarray has become a promising means for classification of various cancer types via the creation of various Gene Expression (GE) profiles, with the advancement of technologies. But, it is challenging to classify the GE profile since not all genes contribute to the presence of cancer and might lead to incorrect diagnoses. Thus an efficient GE data analysis for microarray cancer data classification using Exponential Sigmoid-Deep Belief Network (ES-DBN) is proposed in this work.OBJECTIVES: The study aims to develop an efficient GE data analysis using Exponential Sigmoid-Deep Belief Network (ES-DBN) for microarray cancer data classification.METHODS: The proposed methodology starts with pre-processing to compact data. Afterward, by utilizing Min-Max feature scaling technique, the pre-processed data is normalized. The normalized data is further encoded and feature ranking is performed. The subset values are selected using Cauchy Mutation-Coral Reefs Optimization (CM-CRO) in feature ranking. The feature vector is calculated by Pearson Correlation Coefficient based GloVe (PCC-GloVe) algorithm since different subsets return the same fitness value. Statistical and Biological validations take place after feature vector calculation. Lastly, for effective classification of the type of cancer, the vector features obtained are fed to ES-DBN.RESULTS: The outcomes of the proposed technique are evaluated with various datasets, which exhibited that the proposed technique performed well with the Ovarian cancer dataset and outperforms other conventional approaches.CONCLUSION: This study presents a comprehensive methodology for efficiently classifying cancer types using GE profile. The proposed GE data analysis using ES-DBN shows promising results, highlighting its potential as a valuable tool for cancer diagnosis and classification.

Bayesian causal forests for multivariate outcomes: application to Irish data from an international large scale education assessment

Abstract Bayesian Causal Forests (BCF) is a causal inference machine learning model based on the flexible non-parametric regression and classification tool, Bayesian Additive Regression Trees (BART). Motivated by data from the Trends in International Mathematics and Science Study (TIMSS), which includes data on student achievement in both mathematics and science, we present a multivariate extension of the BCF algorithm. With the help of simulation studies, we show that our approach can accurately estimate causal effects for multiple outcomes subject to the same treatment. We apply our model to Irish data from TIMSS 2019. Our findings reveal the positive effects of having access to a study desk at home (Mathematics ATE 95% CI: [−0.50, 10.14]) while also highlighting the negative consequences of students often feeling hungry at school (Mathematics ATE 95% CI: [−8.86, −1.56] , Science ATE 95% CI: [−10.35, −0.94]) or often being absent (Mathematics ATE 95% CI: [−11.88, −2.27]). Code for replicating the results can be found at https://github.com/Nathan-McJames/MVBCF-Paper.

Classification of bioactive peptides: A systematic benchmark of models and encodings

Bioactive peptides are short amino acid chains possessing biological activity and exerting physiological effects relevant to human health. Despite their therapeutic value, their identification remains a major problem, as it mainly relies on time-consuming in vitro tests. While bioinformatic tools for the identification of bioactive peptides are available, they are focused on specific functional classes and have not been systematically tested on realistic settings. To tackle this problem, bioactive peptide sequences and functions were here gathered from a variety of databases to generate a unified collection of bioactive peptides from microbial fermentation. This collection was organized into nine functional classes including some previously studied and some unexplored such as immunomodulatory, opioid and cardiovascular peptides. Upon assessing their sequence properties, four alternative encoding methods were tested in combination with a multitude of machine learning algorithms, from basic classifiers like logistic regression to advanced algorithms like BERT. Tests on a total of 171 models showed that, while some functions are intrinsically easier to detect, no single combination of classifiers and encoders worked universally well for all classes. For this reason, we unified all the best individual models for each class and generated CICERON (Classification of bIoaCtive pEptides fRom micrObial fermeNtation), a classification tool for the functional classification of peptides. State-of-the-art classifiers were found to underperform on our realistic benchmark dataset compared to the models included in CICERON. Altogether, our work provides a tool for real-world peptide classification and can serve as a benchmark for future model development.

A consensus molecular subtypes classification strategy for clinical colorectal cancer tissues.

Consensus Molecular Subtype (CMS) classification of colorectal cancer (CRC) tissues is complicated by RNA degradation upon formalin-fixed paraffin-embedded (FFPE) preservation. Here, we present an FFPE-curated CMS classifier. The CMSFFPE classifier was developed using genes with a high transcript integrity in FFPE-derived RNA. We evaluated the classification accuracy in two FFPE-RNA datasets with matched fresh-frozen (FF) RNA data, and an FF-derived RNA set. An FFPE-RNA application cohort of metastatic CRC patients was established, partly treated with anti-EGFR therapy. Key characteristics per CMS were assessed. Cross-referenced with matched benchmark FF CMS calls, the CMSFFPE classifier strongly improved classification accuracy in two FFPE datasets compared with the original CMSClassifier (63.6% versus 40.9% and 83.3% versus 66.7%, respectively). We recovered CMS-specific recurrence-free survival patterns (CMS4 versus CMS2: hazard ratio 1.75, 95% CI 1.24-2.46). Key molecular and clinical associations of the CMSs were confirmed. In particular, we demonstrated the predictive value of CMS2 and CMS3 for anti-EGFR therapy response (CMS2&3: odds ratio 5.48, 95% CI 1.10-27.27). The CMSFFPE classifier is an optimized FFPE-curated research tool for CMS classification of clinical CRC samples.

Simple Behavioral Analysis (SimBA) as a platform for explainable machine learning in behavioral neuroscience.

The study of complex behaviors is often challenging when using manual annotation due to the absence of quantifiable behavioral definitions and the subjective nature of behavioral annotation. Integration of supervised machine learning approaches mitigates some of these issues through the inclusion of accessible and explainable model interpretation. To decrease barriers to access, and with an emphasis on accessible model explainability, we developed the open-source Simple Behavioral Analysis (SimBA) platform for behavioral neuroscientists. SimBA introduces several machine learning interpretability tools, including SHapley Additive exPlanation (SHAP) scores, that aid in creating explainable and transparent behavioral classifiers. Here we show how the addition of explainability metrics allows for quantifiable comparisons of aggressive social behavior across research groups and species, reconceptualizing behavior as a sharable reagent and providing an open-source framework. We provide an open-source, graphical user interface (GUI)-driven, well-documented package to facilitate the movement toward improved automation and sharing of behavioral classification tools across laboratories.

Distinguishing Turkish pine honey from multi-floral honey through MALDI-MS-based N-glycomics and machine learning

Honey, a multifaceted blend of sugars, amino acids, vitamins, proteins, and minerals, exhibits compositional variability dependent upon the floral source. While previous studies have attempted to categorize honey, the use of glycomic profiles for honey classification remains an unexplored avenue. This investigation seeks to establish a methodology for distinguishing honey types, specifically multi-floral and pine honey, employing mass spectrometry-based glycomic analysis in tandem with machine learning. In this search, seven samples of pine honey and eight samples of multi-floral honey were obtained from diverse regions of Turkey. Subsequently, the proteins within these honey samples were extracted, and glycans were enzymatically released. The released glycans were labeled with 2-aminobenzoic acid (2-AA) and subjected to analysis via matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS). The glycan profiles of pine and multi-floral honey were determined through these analytical procedures, revealing a total of 76 distinct N-glycan structures. Among these, 13 N-glycan profiles consistently established at high levels across experimental replicates and were incorporated in subsequent analyses. Following the quantification of individual glycan abundances, statistically significant differences in glycan profiles were determined. Notably, N-glycans Hex5HexNAc2, Hex4HexNAc3, and Hex5HexNAc3 displayed considerable differences. Using the 13 N-glycan profiles, an accuracy rate of 93.5% was obtained from machine learning analysis, which increased to 100% when incorporating the identified significantly changed glycans. The most productive models were identified as “subspace and fine k-nearest neighbors (KNN).” The findings underscore the potential of mass spectrometry-based glycomics in conjunction with machine learning as a robust tool for precise honey type classification and its prospective utility in quality control and honey product authentication.

Deep-learning model for background parenchymal enhancement classification in contrast-enhanced mammography

Background. Breast background parenchymal enhancement (BPE) is correlated with the risk of breast cancer. BPE level is currently assessed by radiologists in contrast-enhanced mammography (CEM) using 4 classes: minimal, mild, moderate and marked, as described in breast imaging reporting and data system (BI-RADS). However, BPE classification remains subject to intra- and inter-reader variability. Fully automated methods to assess BPE level have already been developed in breast contrast-enhanced MRI (CE-MRI) and have been shown to provide accurate and repeatable BPE level classification. However, to our knowledge, no BPE level classification tool is available in the literature for CEM. Materials and methods. A BPE level classification tool based on deep learning has been trained and optimized on 7012 CEM image pairs (low-energy and recombined images) and evaluated on a dataset of 1013 image pairs. The impact of image resolution, backbone architecture and loss function were analyzed, as well as the influence of lesion presence and type on BPE assessment. The evaluation of the model performance was conducted using different metrics including 4-class balanced accuracy and mean absolute error. The results of the optimized model for a binary classification: minimal/mild versus moderate/marked, were also investigated. Results. The optimized model achieved a 4-class balanced accuracy of 71.5% (95% CI: 71.2–71.9) with 98.8% of classification errors between adjacent classes. For binary classification, the accuracy reached 93.0%. A slight decrease in model accuracy is observed in the presence of lesions, but it is not statistically significant, suggesting that our model is robust to the presence of lesions in the image for a classification task. Visual assessment also confirms that the model is more affected by non-mass enhancements than by mass-like enhancements. Conclusion. The proposed BPE classification tool for CEM achieves similar results than what is published in the literature for CE-MRI.

The Nature and Origin of Plant Viruses with an Incursion into Virus Nomenclature

Eversince, the chemical nature of not only the anisometric tobacco mosaic virus (TMV) but also the crystalline viruses like bushy stunt of tomato, tobacco necrosis virus, turnip yellow mosaic virus and others have been characterized, sap transmissible viruses have been shown to be good antigens and that their antisera had group specificity aiding identification. Whilst symptomatology alone was found unreliable and of insufficient means for diagnosis and classification of viruses, serology, chemical and physical structure of viruses and the modes of transmission have played an important role in virus nomenclature. However, it is premature to think of orthodox binomial and trinomial nomenclature as practised in fungal classification playing a role in virus terminology, Anisometric and crystalline viruses are comparatively simple chemical entities (as compared with animal viruses) and are essentially nucleoproteins where the RNA (ribonucleic acid) core has a shell of protein in a helical pattern serving as a protective coating. The infectivity of the nucleic acid alone (which is hardly 6% of the nucleoproteiri) has been proved beyond doubt whereas, the protein (which is the bulk of the virus) does not count for infectivity but plays an important role in serological reactions and consequently is of immense value in grouping viruses. Nevertheless, size, shapes and structure of viruses are of considerable significance as a tool in virus classification of rodshaped particles. Spherical virus particles, however, have not lent themselves to this approach. A somewhat similar approach to classification has been made by using specific vector-virus relationships.In the speculative field of origin of viruses, the free gene or plasmagene theory of Darlington seems to find little support at present. Nonetheless, a biochemical approach is still valid and as all the ribonucleic acid (RNA) in the TMV appears to be present in a single polynucleotide chain, a break in this chain has to be transmitted intact to the recipient cell to enable reduplication. That the RNA may be replicated in the cells is, therefore, not to be regarded as a shortlived messenger carrying genetic information from nuclear deoxyribonucleic acid (DNA) to protein. Stretching this argument one could conceive of a further step to the self-perpetuating RNA molecule capable of directing protein synthetic mechanisms in recipient cells to specific products.