Classification Model Research Articles

Investigating the contribution of image time series observations to cauliflower harvest-readiness prediction.

Cauliflower cultivation is subject to high-quality control criteria during sales, which underlines the importance of accurate harvest timing. Using time series data for plant phenotyping can provide insights into the dynamic development of cauliflower and allow more accurate predictions of when the crop is ready for harvest than single-time observations. However, data acquisition on a daily or weekly basis is resource-intensive, making selection of acquisition days highly important. We investigate which data acquisition days and development stages positively affect the model accuracy to get insights into prediction-relevant observation days and aid future data acquisition planning. We analyze harvest-readiness using the cauliflower image time series of the GrowliFlower dataset. We use an adjusted ResNet18 classification model, including positional encoding of the data acquisition dates to add implicit information about development. The explainable machine learning approach GroupSHAP analyzes time points' contributions. Time points with the lowest mean absolute contribution are excluded from the time series to determine their effect on model accuracy. Using image time series rather than single time points, we achieve an increase in accuracy of 4%. GroupSHAP allows the selection of time points that positively affect the model accuracy. By using seven selected time points instead of all 11 ones, the accuracy improves by an additional 4%, resulting in an overall accuracy of 89.3%. The selection of time points may therefore lead to a reduction in data collection in the future.

Content-Based Image Retrieval and Feature Extraction: Analysing the Literature

A significant amount of multimedia data consists of digital images, and multimedia content analysis is used in many real-world computer vision applications. Multimedia information, especially photos, has become much more complicated in the last several years. Every day, millions of photos are posted to various websites, such as Instagram, Facebook, and Twitter. Finding a suitable image in an archive is a difficult research subject for the field of computer vision. Most search engines use standard text-based techniques that depend on metadata and captions in order to fetch photos. Over the past 20 years, a great deal of research has been conducted on content-based image retrieval (CBIR), picture categorization, and analysis. In image classification models and CBIR, high-level picture representations are represented as feature vectors made up of numerical values. Empirical evidence indicates a considerable disparity between picture feature representation and human visual understanding. Reducing the semantic gap between human visual understanding and picture feature representation is the aim of this study. This study aims to do a thorough analysis of the latest advancements in the domains of Content-Based picture Retrieval and picture representation. We performed a comprehensive analysis of many models for image retrieval and picture representation, encompassing the most recent advancements in semantic deep-learning methods and feature extraction. This paper provides an in-depth analysis of the key ideas and important studies related to image representation and content-based picture retrieval. In an effort to stimulate more research in this field, it also offers a preview of potential future study topics.

Hyperspectral Imaging for Phenotyping Plant Drought Stress and Nitrogen Interactions Using Multivariate Modeling and Machine Learning Techniques in Wheat

Accurate detection of drought stress in plants is essential for water use efficiency and agricultural output. Hyperspectral imaging (HSI) provides a non-invasive method in plant phenotyping, allowing the long-term monitoring of plant health due to sensitivity to subtle changes in leaf constituents. The broad spectral range of HSI enables the development of different vegetation indices (VIs) to analyze plant trait responses to multiple stresses, such as the combination of nutrient and drought stresses. However, known VIs may underperform when subjected to multiple stresses. This study presents new VIs in tandem with machine learning models to identify drought stress in wheat plants under varying nitrogen (N) levels. A pot wheat experiment was set up in the glasshouse with four treatments: well-watered high-N (WWHN), well-watered low-N (WWLN), drought-stress high-N (DSHN) and drought-stress low-N (DSLN). In addition to ensuring that plants were watered according to the experiment design, photosynthetic rate (Pn) and stomatal conductance (gs) (which are used to assess plant drought stress) were taken regularly, serving as the ground truth data for this study. The proposed VIs, together with known VIs, were used to train three classification models: support vector machines (SVM), random forest (RF), and deep neural networks (DNN) to classify plants based on their drought status. The proposed VIs achieved more than 0.94 accuracy across all models, and their performance further increased when combined with known VIs. The combined VIs were used to train three regression models to predict the stomatal conductance and photosynthetic rates of plants. The random forest regression model performed best, suggesting that it could be used as a stand-alone tool to forecast gs and Pn and track drought stress in wheat. This study shows that combining hyperspectral data with machine learning can effectively monitor and predict drought stress in crops, especially in varying nitrogen conditions.

Machine-Learning Predictions of Critical Temperatures from Chemical Compositions of Superconductors.

In the quest for advanced superconducting materials, the accurate prediction of critical temperatures (Tc) poses a formidable challenge, largely due to the complex interdependencies between superconducting properties and the chemical and structural characteristics of a given material. To address this challenges, we have developed a machine-learning framework that aims to elucidate these complicated and hitherto poorly understood structure-property and property-property relationships. This study introduces a novel machine-learning-based workflow, termed the Gradient Boosted Feature Selection (GBFS), which has been tailored to predict Tc for superconductors by employing a distributed gradient-boosting framework. This approach integrates exploratory data analyses, statistical evaluations, and multicollinearity reduction techniques to select highly relevant features from a high-dimensional feature space, derived solely from the chemical composition of materials. Our methodology was rigorously tested on a data set comprising approximately 16,400 chemical compounds with around 12,000 unique chemical compositions. The GBFS workflow enabled the development of a classification model that distinguishes compositions likely to exhibit Tc values greater than 10 K. This model achieved a weighted average F1-score of 0.912, an AUC-ROC of 0.986, and an average precision score of 0.919. Additionally, the GBFS workflow underpinned a regression model that predicted Tc values with an R2 of 0.945, an MAE of 3.54 K, and an RMSE of 6.57 K on a test set obtained via random splitting. Further exploration was conducted through out-of-sample Tc predictions, particularly those exceeding the liquid nitrogen temperature, and out-of-distribution predictions for (Ca1-xLax)FeAs2 based on varying lanthanum content. The outcome of our study underscores the significance of systematic feature analysis and selection in enhancing predictive model performance, offering various advantages over models that rely primarily on algorithmic complexity. This research not only advances the field of superconductivity but also sets a precedent for the application of machine learning in materials science.

Development of message passing-based graph convolutional networks for classifying cancer pathology reports

BackgroundApplying graph convolutional networks (GCN) to the classification of free-form natural language texts leveraged by graph-of-words features (TextGCN) was studied and confirmed to be an effective means of describing complex natural language texts. However, the text classification models based on the TextGCN possess weaknesses in terms of memory consumption and model dissemination and distribution. In this paper, we present a fast message passing network (FastMPN), implementing a GCN with message passing architecture that provides versatility and flexibility by allowing trainable node embedding and edge weights, helping the GCN model find the better solution. We applied the FastMPN model to the task of clinical information extraction from cancer pathology reports, extracting the following six properties: main site, subsite, laterality, histology, behavior, and grade.ResultsWe evaluated the clinical task performance of the FastMPN models in terms of micro- and macro-averaged F1 scores. A comparison was performed with the multi-task convolutional neural network (MT-CNN) model. Results show that the FastMPN model is equivalent to or better than the MT-CNN.ConclusionsOur implementation revealed that our FastMPN model, which is based on the PyTorch platform, can train a large corpus (667,290 training samples) with 202,373 unique words in less than 3 minutes per epoch using one NVIDIA V100 hardware accelerator. Our experiments demonstrated that using this implementation, the clinical task performance scores of information extraction related to tumors from cancer pathology reports were highly competitive.

Machine Learning for the Prediction of Surgical Morbidity in Placenta Accreta Spectrum.

We sought to create a machine learning (ML) model to identify variables that would aid in the prediction of surgical morbidity in cases of placenta accreta spectrum (PAS). A multicenter analysis including all cases of PAS identified by pathology specimen confirmation, across five tertiary care perinatal centers in New York City from 2013 to 2022. We developed models to predict operative morbidity using 213 variables including demographics, obstetrical information, and limited prenatal imaging findings detailing placental location. Our primary outcome was prediction of a surgical morbidity composite defined as including any of the following: blood loss (>1,500 mL), transfusion, intensive care unit admission, vasopressor use, mechanical ventilation/intubation, and organ injury. A nested, stratified, cross-validation approach was used to tune model hyperparameters and estimate generalizability. Gradient boosted tree classifier models incorporated preprocessing steps of standard scaling for numerical variables and one-hot encoding for categorical variables. Model performance was evaluated using area under the receiver operating characteristic curve (AUC), positive and negative predictive values (PPV, NPV), and F1 score. Variable importance ranking was also determined. Among 401 PAS cases, 326 (81%) underwent hysterectomy. Of the 401 cases of PAS, 309 (77%) had at least one event defined as surgical morbidity. Our predictive model had an AUC of 0.79 (95% confidence interval: 0.69, 0.89), PPV 0.79, NPV 0.76, and F1 score of 0.88. The variables most predictive of surgical morbidity were completion of a hysterectomy, prepregnancy body mass index (BMI), absence of a second trimester ultrasound, socioeconomic status zip code, BMI at delivery, number of prenatal visits, and delivery time of day. By identifying social and obstetrical characteristics that increase patients' risk, ML models are useful in predicting PAS-related surgical morbidity. Utilizing ML could serve as a foundation for risk and complexity stratification in cases of PAS to optimize surgical planning. · ML models are useful models are useful in predicting PAS-related surgical morbidity.. · Optimal management for PAS remains unclear.. · Utilizing ML can serve as a foundation for risk and complexity stratification in cases of PAS..

MobilenetV2-RC: A lightweight network model for retinopathy classification in retinal OCT images

Abstract Retinopathy is an important ophthalmic disease that causes blindness in the elderly population. With the global old peoples increasing, the ophthalmic healthcare system is more and significant to pre-diagnosis. Optical Coherence Tomography (OCT) is considered the gold standard for ophthalmic treatment and diagnosis. OCT technologies and equipment continue to develop towards the intelligence and convenience for requirements of rapid diagnosis in the remote and poverty-stricken areas. We proposed an improved MobilenetV2 lightweight model for retinopathy classification (MobilenetV2-RC), which incorporates Spatial and Channel Reconstruction Convolution (SCConv) and the improved Convolutional Block Attention Module (CBAM) attention mechanism into the framework. Not only can it effectively limit feature redundancy to reduce model parameters, but also enhance the ability of feature representation to improve classification accuracy. The parameters of the proposed model are only 1.96 M with an overall accuracy of 98.96%, which is higher 3.32% than the original MobilenetV2. Compared with ResNet18, InceptionV3, and VGG16_BN, the overall accuracy is increased by 4.6%, 6.3%, and 3.9%, respectively. The test results of UCSD and Duke open-source datasets are more remarkable. Experimental results show that our proposed algorithm has strong reliability and generalization for the accurate classification of retinopathy, and a greater application prospect in the intelligent diagnosis of ophthalmology and mobile detection terminals.

Medical image analysis using improved SAM-Med2D: segmentation and classification perspectives

Recently emerged SAM-Med2D represents a state-of-the-art advancement in medical image segmentation. Through fine-tuning the Large Visual Model, Segment Anything Model (SAM), on extensive medical datasets, it has achieved impressive results in cross-modal medical image segmentation. However, its reliance on interactive prompts may restrict its applicability under specific conditions. To address this limitation, we introduce SAM-AutoMed, which achieves automatic segmentation of medical images by replacing the original prompt encoder with an improved MobileNet v3 backbone. The performance on multiple datasets surpasses both SAM and SAM-Med2D. Current enhancements on the Large Visual Model SAM lack applications in the field of medical image classification. Therefore, we introduce SAM-MedCls, which combines the encoder of SAM-Med2D with our designed attention modules to construct an end-to-end medical image classification model. It performs well on datasets of various modalities, even achieving state-of-the-art results, indicating its potential to become a universal model for medical image classification.

Alzheimer’s disease (AD) classification using MRI: A deep ensemble model with modified local pattern feature set

The Alzheimer disease (AD) is a neurologic brain condition, which affects the cells in the brain and eventually renders a patient incapable of performing routine daily tasks. Due to the outstanding spatial clarity, high access, and strong contrast, MRI has been utilized in analyses pertaining to AD. This work develops an AD classification model using MRI images. Here, preprocessing is done by the Gabor filter. Subsequently, the Improved U-net segmentation model is employed for image segmentation. The features extracted comprises of modified LGXP features, LTP features, and LBP features as well. Finally, the Deep ensemble classifier (DEC) model is proposed for AD classification which combines classifiers such as RNN, DBN, and Deep Maxout Network (DMN). For enhancing the efficiency for classification of AD, the optimal weight of DMN is adjusted using the Self Customized BWO (SC-BWO) model. The outputs from DEC are averaged and the final result is obtained. Finally, the analysis of dice, Jaccard scores is performed to show the betterment of the SC-BWO scheme.

Cross-Domain Image Classification via Multimodal Augmentations and Bacterial Foraging Optimization

Identification of diseases from body scans requires design of pre-processing, filtering, segmentation, feature representation, classification & post-processing operations. Existing deep learning-based classification models use context-specific segmentation followed by convolutions or transformer-based classification techniques. However, the majority of these models do not perform correlative analysis between different disease types. This limitation restricts their scalability and applicability in clinical scenarios. To address these limitations, this research work proposed IMAC-MOC, a model that integrates multimodal augmentations to enhance performance in cross-domain and multi-organ image classification. The proposed model initially collects large-scale scans at organ-level like MRI, CT scan, Kidney Scans, etc. and represents them via multimodal feature sets. These feature maps are combined and processed via a Bacterial Foraging Optimizer (BFO) which assists in identification of high variance inter-class feature sets.

Automated Surgical Wound Classification for Intelligent Emergency Care Applications

Developing AI-powered solutions for emergency care is critical to improving patient outcomes. This study presents an advanced AI model designed to accurately classify surgical wounds to facilitate prompt and appropriate emergency response. We used two state-of-the-art image classification models, ResNeXt and Vision Transformer (ViT), to evaluate their effectiveness in classifying wound images. These models were selected based on preliminary evaluations of several popular models, and their superior performance metrics justified their final selection. The models were trained on a combined dataset of approximately 1,000 images from the Medetec and AZH(Advancing the Zenith of Healthcare) datasets. To improve classification accuracy, the dataset was preprocessed, including image resizing, flipping, normalization, and augmentation. The ViT model demonstrated superior performance, achieving an accuracy of 92.78%, precision of 94.89%, recall of 91.87%, and an F1 score of 92.44%. These results surpass those of existing studies such as Shenoy et al. (85.1% accuracy), Rostami et al. (68.7% accuracy), and Gao et al. (76.6% accuracy). Our proposed AI system not only accelerates life-saving first aid by providing timely and accurate wound classification but also enhances the skills of emergency responders through continuous learning and real-time feedback. By integrating AI into emergency service protocols, we aim to improve response times and collaboration among medical personnel, ultimately contributing to better patient survival rates.

Septic Arthritis Modeling Using Sonographic Fusion with Attention and Selective Transformation: a Preliminary Study.

Conventionally diagnosing septic arthritis relies on detecting the causal pathogens in samples of synovial fluid, synovium, or blood. However, isolating these pathogens through cultures takes several days, thus delaying both diagnosis and treatment. Establishing a quantitative classification model from ultrasound images for rapid septic arthritis diagnosis is mandatory. For the study, a database composed of 342 images of non-septic arthritis and 168 images of septic arthritis produced by grayscale (GS) and power Doppler (PD) ultrasound was constructed. In the proposed architecture of fusion with attention and selective transformation (FAST), both groups of images were combined in a vision transformer (ViT) with the convolutional block attention module, which incorporates spatial, modality, and channel features. Fivefold cross-validation was applied to evaluate the generalized ability. The FAST architecture achieved the accuracy, sensitivity, specificity, and area under the curve (AUC) of 86.33%, 80.66%, 90.25%, and 0.92, respectively. These performances were higher than using conventional ViT (82.14%) and significantly better than using one modality alone (GS 73.88%, PD 72.02%), with the p-value being less than 0.01. Through the integration of multi-modality and the extraction of multiple channel features, the established model provided promising accuracy and AUC in septic arthritis classification. The end-to-end learning of ultrasound features can provide both rapid and objective assessment suggestions for future clinic use.

Predicting Air Quality Based on Multiclass Machine Learning Techniques

India is among the most polluted nations with severe environmental implications of pollution increase in several bas cities. For the past few years’ Indian cities have witnessed an alarming drop in the Air Quality (AQ) which is a result of the rapid economic growth. The lives of billions of people are deeply affected by Air Pollution (AP) every year. The economic sectors that source poor urban AQ include transportation, agriculture, construction, forestry, and logging, emitting dangerous gases and particulates such as NO2, NH3, SO3, and Benzene. This research attempts to implement four different classifiers independently on the AQ dataset. The four classifiers that are being implemented are the Multiclass Decision Jungle (MDJ), Multiclass Logistic Regression (MLR), Multiclass Decision Forest (MDF), and Multiclass Neural Network (MNN). This project objective is to identify the most suitable among the four classification models in building the best model for AQ. The performance of a model is judged by accuracy, precision, and recall. Likewise, about other three modes MDF performs best, where it obtained 99.96% accuracies, 98.91% precision, and 99.76% recall.

A Spatio-Temporal Capsule Neural Network with Self-Correlation Routing for EEG Decoding of Semantic Concepts of Imagination and Perception Tasks.

Decoding semantic concepts for imagination and perception tasks (SCIP) is important for rehabilitation medicine as well as cognitive neuroscience. Electroencephalogram (EEG) is commonly used in the relevant fields, because it is a low-cost noninvasive technique with high temporal resolution. However, as EEG signals contain a high noise level resulting in a low signal-to-noise ratio, it makes decoding EEG-based semantic concepts for imagination and perception tasks (SCIP-EEG) challenging. Currently, neural network algorithms such as CNN, RNN, and LSTM have almost reached their limits in EEG signal decoding due to their own short-comings. The emergence of transformer methods has improved the classification performance of neural networks for EEG signals. However, the transformer model has a large parameter set and high complexity, which is not conducive to the application of BCI. EEG signals have high spatial correlation. The relationship between signals from different electrodes is more complex. Capsule neural networks can effectively model the spatial relationship between electrodes through vector representation and a dynamic routing mechanism. Therefore, it achieves more accurate feature extraction and classification. This paper proposes a spatio-temporal capsule network with a self-correlation routing mechaninsm for the classification of semantic conceptual EEG signals. By improving the feature extraction and routing mechanism, the model is able to more effectively capture the highly variable spatio-temporal features from EEG signals and establish connections between capsules, thereby enhancing classification accuracy and model efficiency. The performance of the proposed model was validated using the publicly accessible semantic concept dataset for imagined and perceived tasks from Bath University. Our model achieved average accuracies of 94.9%, 93.3%, and 78.4% in the three sensory modalities (pictorial, orthographic, and audio), respectively. The overall average accuracy across the three sensory modalities is 88.9%. Compared to existing advanced algorithms, the proposed model achieved state-of-the-art performance, significantly improving classification accuracy. Additionally, the proposed model is more stable and efficient, making it a better decoding solution for SCIP-EEG decoding.

Advanced Multi-Label Fire Scene Image Classification via BiFormer, Domain-Adversarial Network and GCN

Detecting wildfires presents significant challenges due to the presence of various potential targets in fire imagery, such as smoke, vehicles, and people. To address these challenges, we propose a novel multi-label classification model based on BiFormer’s feature extraction method, which constructs sparse region-indexing relations and performs feature extraction only in key regions, thereby facilitating more effective capture of flame characteristics. Additionally, we introduce a feature screening method based on a domain-adversarial neural network (DANN) to minimize misclassification by accurately determining feature domains. Furthermore, a feature discrimination method utilizing a Graph Convolutional Network (GCN) is proposed, enabling the model to capture label correlations more effectively and improve performance by constructing a label correlation matrix. This model enhances cross-domain generalization capability and improves recognition performance in fire scenarios. In the experimental phase, we developed a comprehensive dataset by integrating multiple fire-related public datasets, and conducted detailed comparison and ablation experiments. Results from the tenfold cross-validation demonstrate that the proposed model significantly improves recognition of multi-labeled images in fire scenarios. Compared with the baseline model, the mAP increased by 4.426%, CP by 4.14% and CF1 by 7.04%.

Assessing Mobile Network Fraud Threats and Prevention Strategies in Kenya

Background: Mobile network technology has exponentially advanced in the last decade and with this development, fraud activities have risen in equal measure resulting in companies and customers losing huge amounts of money as a result of, especially in developing economies that lag in the regulatory frameworks when it comes to Mobile network fraud. The purpose of the study was to explore Mobile network fraud in Kenya identifying the most common types of fraud, ways which service providers and regulators are employing to prevent or reduce fraud, methods currently used to detect fraud, and gaps thereof. Finally, the effect of concept drift in the automated fraud detection process. Method: A qualitative research method was adopted for the study and using a semi-structured question guide, four focus group discussions composed of 23 participants were conducted. The criteria used for selecting and placing participants into focus groups considered the following: The expert area of the participant, years of experience in the fraud ecosystem of the participant, and the organization to which the participant is attached. The availability and willingness of the participants were also considered in the selection process. The focus group approach was selected as it facilitated balanced discussion amongst all the players in the Kenyan fraud ecosystem, harnessing the power of group dynamics as it involved the regulators and the service providers who were drawn from different organizations. Results: The mobile network fraud ecosystem was stratified into three dimensions namely Fraud prevention which looked at the policies and methods used by both regulators and service providers to reduce fraud, Fraud categorization which aimed at categorizing different types of mobile frauds, and finally the Fraud detection which looked at the current tools being used to detect fraud. From the study, it emerged that although the regulators have provided strict guidelines on the customer onboarding process, not all service providers are currently using biometric approaches while onboarding new customers as this was highlighted as the entry point of most fraud cases. The study established five major types of Mobile network fraud in Kenya: SIM swap, SIM boxing, Wangiri, Commission arbitrage, and Hoax SMS and scams. Most of this fraud is committed using either SMS or voice channels; in some cases, both channels are used. Different matrixes derived from multiple factors are used by service providers while evaluating the criticality of fraud cases though not enforced by the regulators. The study also revealed that most of the fraud detection processes amongst the service providers still use manual tools that constantly require human input. While some of the detection processes are automated, concept drift is a major challenge for automated classification models due to the constant evolution of fraud patterns. Conclusion: The study revealed gaps in Mobile Network fraud prevention processes in Kenya as service providers still use non-biometric customer validation processes that are open to forgery and exploitation. A strict customer onboarding process that is fully automated and integrated should be used to address this gap. In the fraud categorization, there is no clear universal categorization matrix to guide the service providers while assessing the criticality of fraud and in this regard, a qualitative scientific model should be developed and used by all the stakeholders as a reference point. When it comes to fraud detection, concept drift is a major challenge, and service providers in Kenya still rely on manual processes due to the dynamic nature of mobile fraud. This exposes a huge gap in the detection process and there is a need to address this by developing systems and processes that will automatically detect and react to concept drift while automating the detection processes.

PSIII-29 Evaluating the efficacy of bioelectrical impedance analysis using machine learning models for the classification of parasitized goats

Abstract Rapid identification and evaluation of animal health condition is one of the most important concerns when it comes to livestock productivity, particularly for small ruminants susceptible to infection from blood-feeding gastrointestinal nematodes, like Haemonchus contortus. The present study was designed to optimize the process of differentiating parasite infected goats from healthy goats. This study used male intact Spanish goats (n = 65), aiming to establish a proof of concept for the efficacy of bioelectrical impedance, specifically through the metrics of Resistance, and Reactance, in identifying the health status of goats. The research involved the development and comparative analysis of seven machine learning (ML) models, divided into four classification-based models and three regression models, to assess their capabilities in accurately classifying goats based on bioelectrical properties. The data for this study were gathered from live animals using CQR 3.0 (Sea food analytics), with measurements taken from both ear and tail ends to ensure comprehensive and reliable bioelectrical readings. Among the classification models evaluated, the Random Forest Classification model emerged as the most effective, demonstrating a validation accuracy of 71.8% and a testing accuracy of 77.1%. This was closely followed by the K-NN Classification model, which showed a promising testing accuracy of 86.7%. Conversely, the Support Vector Machines and Boosting Classification models displayed decreased accuracies, indicating variability in the performance of classification models in this context. The regression models, evaluated through metrics such as Mean Squared Error (MSE) and R-square, revealed the K-NN Regression model as the most proficient, with the least Test MSE of 0.13 and an R-square of 0.482. This suggests a moderate correlation and predictive capability in determining the health status of goats based on bioelectrical impedance measurements. The Decision Tree Regression and Support Vector Regressor models showed varying degrees of effectiveness, with the latter indicating a minimal R-square value, highlighting the challenges in applying regression models to this domain. The conducted proof of concept research not only underscores the potential of bioelectrical impedance analysis in veterinary diagnostics, but also opens avenues for further refinement of ML algorithms in classifying and predicting the health status of farm animals. The outcomes of this study could significantly influence management strategies for small ruminants, offering a non-invasive, rapid, and potentially cost-effective tool for early detection of parasitism, thereby enhancing animal welfare and productivity in the agricultural sector.

Explainable rotation-invariant self-supervised representation learning

This paper describes a method that can perform robust detection and classification in out-of-distribution rotated images in the medical domain. In real-world medical imaging tools, noise due to the rotation of the body part is frequently observed. This noise reduces the accuracy of AI-based classification and prediction models. Hence, it is important to develop models which are rotation invariant. To that end, the proposed method - RISC (rotation invariant self-supervised vision framework) addresses this issue of rotational corruption. We present state-of-the-art rotation-invariant classification results and provide explainability for the performance in the domain. The evaluation of the proposed method is carried out on real-world adversarial examples in Medical Imagery-OrganAMNIST, RetinaMNIST and PneumoniaMNIST. It is observed that RISC outperforms the rotation-affected benchmark methods by obtaining 22\\%, 17\\% and 2\\% accuracy boost on OrganAMNIST, PneumoniaMNIST and RetinaMNIST rotated baselines respectively. Further, explainability results are demonstrated.This methods paper describes:•a representation learning approach that can perform robust detection and classification in out-of-distribution rotated images in the medical domain.•It presents a method that incorporates self-supervised rotation invariance for correcting rotational corruptions.•GradCAM-based explainability for the rotational SSL pretext task and the downstream classification outcomes for the three benchmark datasets are presented

The method of the etiological factor′s evaluation of the gingival recession, with count of their′s separately and common complicity in the pathogenesis

Introduction. Gingival recession is a highly relevant and common dental pathology. The prevalence of recession ranges from 45,5 to 99,3 % in people of different ages, with the risk of gingival recession increasing with age. The true etiology of gingival recession, its components and the degree of their influence, jointly or separately; the influence of the environment and lifestyle on the development of pathology are poorly understood. In the literature, for example, there is no classification of etiological factors according to the degree of their influence as triggers of gingival recession. It is of scientific interest to study the causes of gingival recession, their systematisation and assessment of the degree of influence of each factor separately and together.The aim of the study is to develop a classification system of etiological factors taking into account their significance and degree of influence on the formation of gingival recessions.Materials and methods. Analysis of scientific literature, methods of classification and structuring; methods of decomposition, integration and determination of cause-and-effect relationships.Results. A model of classification of factors of the etiology of gingival recessions was proposed, elements of the structure of etiological factors of gingival recessions were determined by the degree of their influence on pathogenesis.Conclusion. Conclusions. The study and understanding of the etiology of gingival recessions allows the clinician to select a personalised treatment taking into account individual characteristics that influence the recession occurrence and its condition: congenital and acquired factors; living conditions, habits, level of hygiene, as well as other components that aggravate or compensate for the degree and level of complexity of any recession - single, multiple or generalised. It is advisable to continue the study of literature data to expand the classification and detail the structure of etiological factors.

Construction of a stable YOLOv8 classification model for apple bruising detection based on physicochemical property analysis and structured-illumination reflectance imaging

Effective and accurate detection of bruises at all stages has always been a challenge in non-destructive grading of apples. In this study, the visible structured-illumination reflectance imaging (SIRI) combing with deep learning method was proposed to identify bruised ‘Fuji’ apples at four different time stages (0, 6, 12 and 24 h). The macroscopic/microscopic structures and physicochemical properties of bruised tissue were measured and analyzed to determine the relationship between bruising time and these properties, as well as how they affect the accuracy of bruising detection. Results indicated that classification accuracy increased with the decrease of water and total phenolic content of the bruised tissue, as well as with the increase of color browning and bruised area. The YOLOv8 model achieved the highest detection accuracy (99.5 %) and stability. This research enhances understanding of apple bruise optics and aids in developing advanced nondestructive testing techniques.